GRECO CLASSIC 0153I
Popular in Course
Popular in Classical Studies
verified elite notetaker
This 442 page Class Notes was uploaded by Hayden Braun on Friday September 4, 2015. The Class Notes belongs to CLASSIC 0153I at University of California - Los Angeles taught by Staff in Fall. Since its upload, it has received 84 views. For similar materials see /class/177874/classic-0153i-university-of-california-los-angeles in Classical Studies at University of California - Los Angeles.
Reviews for GRECO
Report this Material
What is Karma?
Karma is the currency of StudySoup.
You can buy or earn more Karma at anytime and redeem it for class notes, study guides, flashcards, and more!
Date Created: 09/04/15
Experiments and Legislative Bargaining Rebecca B Morton New York University EITM UCLA El 5 3 RCquot Experiments and Legislative Bargaining Plan forthe Talk o The research questions this literature address Plan for the Talk Experiments and Legislative Bargaining o The research questions this literature address 0 How experiments have been used ta address these questions Experiments and Legislative Bargaining Plan for the Talk o The research questions this literature address a How experiments have been used to address these questions 0 An idea that comes from this research that should be of interest to ALL political scientists regardless of whether they are experimentalists or study legislative bargaining Experiments and Legislative Bargaining Plan for the Talk The research questions this literature address How experiments have been used to address these questions An idea that comes from this research that should be of interest to ALL political scientists regardless of whether they are experimentalists or study legislative bargaining Howl got involved in this literature amp research amp the choices I made in doing research Experiments and Legislative Bargaining Background for this Talk o Morton quotWhy the Centipede Game Experiment Is Important for Political Sciencequot Experiments and Legislative Bargaining Background for this Talk O Morton Why the Centipede Game Experiment Is Important for Political Science a Frechette Kagel amp Morelli Behavioral Identification in Coalitional Bargaining An Experimental Analysis of Demand Bargaining and Alternating Offers Experiments and Legislative Bargaining Background for this Talk O Morton Why the Centipede Game Experiment Is Important for Political Science a Frechette Kagel amp Morelli Behavioral Identification in Coalitional Bargaining An Experimental Analysis of Demand Bargaining and Alternating Offers 0 Morton amp Diermeier Proportionality versus Perfectness W wyithta is Hmm am wamn W W Game 2560 640 040 020 L60 080 0 lo 089 040 320 640 060 320 1280 FIGURE 2 The six move centipede game WNW m Why the C nt p ede G a m e Ex39p me nt is Important for Political Science F V 9 hcoming in A39 o luential 6 Change in Political Science The Leg acyoiRicl1av39l D itings Edited by J amesAlt and Arthur Lupia Univevsjty 0 Available on my web page 6 Change in Pol calScience ThelL eg acy Ings gEdIted l 39 mes Alt a nd 0 Available on my web page 0 Paper puts the centipede game experiments in context of a research agenda in political science studying how legislatures make decisions 6 Change in Pol calScience ThelL eg acy ings gEdIted l mesAt and 39 0 Available on my web page 0 Paper puts the centipede game experiments in context of a research agenda in political science studying how legislatures make decisions 0 Argument Why the Game is Drum Pd iilmell l 160 at Fm W 3 WE M NEW Wing Elm Wm l m WNW W7 PM 0 Available on my web page 0 Paper puts the centipede game experiments in context of a research agenda in political science studying how legislatures make decisions 0 Argument 0 An experiment on an unreal game can be an important 84 consequential piece in a general literature that studies a highly applied political science mainstream substantive question mum was Why the Es lmmmantc Pdltlesxl Fwtnm mamgefm wlg w Mmewg M GMW Wisng Edam my WM ml Mm um GI PM 0 Available on my web page 0 Paper puts the centipede game experiments in context of a research agenda in political science studying how legislatures make decisions 0 Argument 0 An experiment on an unreal game can be an important 84 consequential piece in a general literature that studies a highly applied political science mainstream substantive question o It is a mistake to evaluate the experiments 84 the game in isolation not recognizing these ties I WMW 51 as th Pd u Putting the Cemipede Game Experiments in Context Some Experimental Results that Fail to Suppo the Co mpetitive Solution McKelvey Ordeshook I WWW rs a a W m WWW AW Putting the Centipede Game Experiments in Context Same Experimental Resuits i that Fail to Support the Competitive Solution McKelvey Ordeshuuk I WWW M P J ii ll mm m Tki Putting the Centipede Game Experiments in Context Some Experimental Results 39 r that Fail to Support the Competitive Solnlion McKelvey Ordeshook An Experimental Test of a Stochastic Game Model of Committee Bargaining MCKelvey I Emmi WM 5 if m is P 1J g1 mm m me i 6m Mm Timmy iligamira Cmf h i w 5 Putting the Centipede Game Experiments in Context Some Experimental Results inai Fail to Support the Competitive Solution McKelvey Ordeshoo An Exparimantal Test of a 1 Stochastic Game 0 o ommitlee Bargaining MCKelvey McKelvey Palfrey EWMI WM 3391 w quotProportionality versus Peifectness My Personal Contribution Comes in Here in Social Choice and Strategic Behavior Essays in Honor of Jeffrey Banks edited by Timothy Feddersen and David AustenSmith 2005 SpringerVerlag co authored with Daniel Diermeier My Personal Contribution Comes in Here quotProportionality versus Peifectness o in Social Choice and Strategic Behavior Essays in Honor of Jeffrey Banks edited by Timothy Feddersen and David AustenSmith 2005 SpringerVerlag co authored with Daniel Diermeier 0 available on my web page My Personal Contribution Comes in here quotProportionality versus Perfectness Flaur 1 came 1w by swam5 Furled PIMIMAMMIECOIWHII 00a9s74H v 1 9 x 4 5 s 7 x 91nn171a141515171n alumFurled mum an Puma 95x whim Munum magmas My Personal Contribution Comes in here quotProportionality versus Perfectness ngmz mamem 1 Proposera a 5 m 15 20 25 Mmaunn m hotsns umuun quot 952 mm My Personal Contribution Comes in here quotProportionality versus Perfectness Flgwei p mum r Prupauer b 25 E zu lt x lt5 I m 5 m 35 4a 4 m 5 mumm m n mummy My Personal Contribution Comes in here quotProportionality versus Perfectness Figure 4 Treatment 1 Proposer 35 30 3 25 3 2a 2 u 5 u 15 m 25 AHucaULm m c mm A Why the is Hmpmem Pd ii eexl Sdem quot Mimi am Miami Team Putting the Centipede Game Experiments in Context Some Experimental Results lliat Fail to Support the Competitive SaluIion McKelvey Ordesllookl An Experimental Test of a stochastic Game Model oi Committee Bargaining MCKerey An Experimental Behavioral identi cation in Coalitional Stud of the Bargaining An x rimental Analysis Centipede Game of Demand Barg g ampAlternaing McKelvey Palfrey Offers Freenette Kagel Mnrelli 35 as Behavioral Identification in Coa JitiOrial Bargai39i ng Ah Experi39 ental A allysis pfDemand Bargaining and Alternating Offers Erechette Guillaum e doh an d Massimo Mo relJi E39cono metrica voquot 73 no 6 November 2005 pages 16 39 0 Many empirical studies try to infer rules of game in government coalition formation 0 Many empirical studies try to infer rules of game in government coalition formation 0 Estimate regressions on Final allocations for which different models of multilateral bargaining make speci c predictions about coefficient estimates 0 Many empirical studies try to infer rules of game in government coalition formation 9 Estimate regressions on final allocations for which different models of multilateral bargaining make speci c predictions about coefficient estimates 0 ln lab control rules of game amp thus study behavior in different games rel m am m Em mmtal Wand m n ng and Almat fears Fmadlm e Eel ame Km and mm an will 7 aim 6 Wm images mama 0 Many empirical studies try to infer rules of game in government coalition formation 0 Estimate regressions on final allocations for which different models of multilateral bargaining make speci c predictions about coefficient estimates 0 ln lab control rules of game amp thus study behavior in different games I Use experiments to test estimation strategy used on field data I WWALA 39 w Behavioral I39denxi fica tign in CoaquotJitiOr1al Bargaini g An Experimental Analysis QfD39em a39nclj Bargaining and Alternati n Offers Pinlpp se of tlie Tli bi ies o A lot of what legislatures do is divide xed resources pork barrel politics portfolios using majority rule B eha 39Ioriail Idenxi fi 39atinn in CoaJiitiVOrial Bargaihi ng Ah Exper39 ntal A I siVSJQfaD39em a d Bargaining and Alterhatl ng39dOffers PL 6 of the Theories o A lot of what legislatures do is divide xed resources pork barrel politics portfolios using majority rule 0 However core is empty o A lot of what legislatures do is divide xed resources pork barrel politics portfolios using majority rule 9 However core is empty 0 Noncooperative legislative bargaining models assume process follows speci c structure o A lot of what legislatures do is divide xed resources pork barrel politics portfolios using majority rule 9 However core is empty 0 Noncooperative legislative bargaining models assume process follows speci c structure 0 Can derive equilibrium predictions Problems with the Theories Behavioral Identification in Coalitional Bargaining 0 Predictions very sensitive to variations in rules of game Problems with the Theories Behavioral Identification in Coalitional Bargaining parts of agents 0 Predictions very sensitive to variations in rules of game 0 Equilibrium solutions may require unrealistic degree of rationality on Behavioral Identification in Coalitional Bargaining Problems with the Theories 0 Predictions very sensitive to variations in rules of game 0 Equilibrium solutions may require unrealistic degree of rationality on parts of agents 0 ls actual behavior as sensitive to rules of game as theory predicts Behavioral Identification in Coalitional Bargaining Problems with the Theories 0 Predictions very sensitive to variations in rules of game a Equilibrium solutions may require unrealistic degree of rationality on parts of agents o Is actual behavior as sensitive to rules of game as theory predicts 0 Especially given evidence from experiments such as centipede game Behavioral Identification in Coalitional Bargaining Problems with the Theories 0 Predictions very sensitive to variations in rules of game a Equilibrium solutions may require unrealistic degree of rationality on parts of agents o Is actual behavior as sensitive to rules of game as theory predicts 0 Especially given evidence from experiments such as centipede game 0 Paper presents experiment analyzing 2 very different games amp use lab data to test methods used on field data Offers vs Demands Behavioral Identification in Coalitional Bargaining as making 0 In classic Rubinstein bargaining model rst mover can be thought of Offers vs Demands Behavioral Identification in Coalitional Bargaining as making 0 In classic Rubinstein bargaining model first mover can be thought of a An offer to agent or Offers vs Demands Behavioral Identification in Coalitional Bargaining 0 In classic Rubinstein bargaining model first mover can be thought of as making a An offer to agent or 0 As making a demand of a share leaving choice between accepting the residual or disagreeing Behavioral Identification in Coalitional Bargaining Offers vs Demands o In classic Rubinstein bargaining model first mover can be thought of as making a An offer to agent or 0 As making a demand of a share leaving choice between accepting the residual or disagreeing a As soon as there is a group with at least 3 members offers amp demands no longer equivalent Behavioral Identification in Coalitional Bargaining Offers vs Demands o In classic Rubinstein bargaining model first mover can be thought of as making a An offer to agent or 0 As making a demand of a share leaving choice between accepting the residual or disagreeing a As soon as there is a group with at least 3 members offers amp demands no longer equivalent 0 If proposer makes a distributive offer others basically face a voting decision Behavioral Identification in Coalitional Bargaining Offers vs Demands o In classic Rubinstein bargaining model first mover can be thought of as making a An offer to agent or 0 As making a demand of a share leaving choice between accepting the residual or disagreeing a As soon as there is a group with at least 3 members offers amp demands no longer equivalent a If proposer makes a distributive offer others basically face a voting decision 0 If first mover only makes her own her demand subsequent movers also have to decide what demand to make reducing asymmetry between movers II Behavioral Identification in Coalitional Bargaining Offers vs Demand II 0 Number of empirical studies employing field data to infer rules of game in legislative bargaining settings Offers vs Demand II Behavioral Identification in Coalitional Bargaining 0 Number of empirical studies employing field data to infer rules of game in legislative bargaining settings 0 The most recent empirical studies Behavioral Identification in Coalitional Bargaining Offers vs Demand II 0 Number of empirical studies employing field data to infer rules of game in legislative bargaining settings 0 The most recent empirical studies 0 Warwick 84 Druckman 2001 Behavioral Identification in Coalitional Bargaining Offers vs Demand II a Number of empirical studies employing field data to infer rules of game in legislative bargaining settings 0 The most recent empirical studies 0 Warwick 84 Druckman 2001 o Ansolabehere Snyder Strauss and Ting 2003 Behavioral Identification in Coalitional Bargaining Offers vs Demand II a Number of empirical studies employing field data to infer rules of game in legislative bargaining settings 0 The most recent empirical studies 0 Warwick 84 Druckman 2001 o Ansolabehere Snyder Strauss and Ting 2003 I Compare offers 84 demand models Behavioral Identification in Coalitional Bargaining Offers vs Demand II a Number of empirical studies employing field data to infer rules of game in legislative bargaining settings 0 The most recent empirical studies 0 Warwick 84 Druckman 2001 o Ansolabehere Snyder Strauss and Ting 2003 a Compare offers 84 demand models 0 Paper first experiment to compare legislative bargaining models Behavioral Identification in Coalitional Bargaining Baroanerejohn Alternating Offer Example I 3 players AB 8 C Behavioral Identification in Coalitional Bargaining Baroanerejohn Alternating Offer Example 0 3 players AB amp C 0 A is selected first randomly she proposes 0801 01 El 5 3 RCquot Behavioral Identification in Coalitional Bargaining Baroanerejohn Alternating Offer Example 0 3 players AB amp C 0 A is selected first randomly she proposes 0801 01 O B amp C reject El 5 3 RCquot Behavioral Identification in Coalitional Bargaining Baroanerejohn Alternating Offer Example o 3 players AB amp C 0 A is selected first randomly she proposes 0801 01 9 B 84 C reject O B is selected second he proposes 0066 034 El 5 3 RCquot Behavioral Identification in Coalitional Bargaining Baroanerejohn Alternating Offer Example o 3 players AB amp C 0 A is selected first randomly she proposes 0801 01 9 B 84 C reject 9 B is selected second he proposes 0066 034 o A rejects B amp C accept final payoffs 0 066034 Behavioral Identification in Coalitional Bargaining Baroanerejohn Alternating Offer Example o 3 players AB amp C 0 A is selected first randomly she proposes 0801 01 9 B amp C reject 9 B is selected second he proposes 0066 034 9 A rejects B amp C accept final payoffs 0 066 034 0 Focus on Stationary Subgame Perfect Equilibria Demand Bargaining Example Behavioral Identification in Coalitional Bargaining 0 A is selected first she requests 08 Demand Bargaining Example Behavioral Identification in Coalitional Bargaining 0 A is selected first she requests 08 O B is selected he requests 07 Demand Bargaining Example Behavioral Identification in Coalitional Bargaining o A is selected first she requests 08 9 B is selected he requests 07 C C is selected she requests 06 Behavioral Identification in Coalitional Bargaining Demand Bargaining Example 0 A is selected first she requests 08 9 B is selected he requests 07 e C is selected she requests 06 o No possible coalition formed all requests erased Behavioral Identification in Coalitional Bargaining Demand Bargaining Example 0 A is selected first she requests 08 9 B is selected he requests 07 e C is selected she requests 06 9 No possible coalition formed all requests erased B B is selected first in 2nd stage he requests 05 Behavioral Identification in Coalitional Bargaining Demand Bargaining Example 0 A is selected first she requests 08 9 B is selected he requests 07 9 C is selected she requests 06 9 No possible coalition formed all requests erased 9 B is selected first in 2nd stage he requests 05 G C is selected she requests 05 Behavioral Identification in Coalitional Bargaining Demand Bargaining Example 0 A is selected first she requests 08 9 B is selected he requests 07 9 C is selected she requests 06 9 No possible coalition formed all requests erased 9 B is selected first in 2nd stage he requests 05 a C is selected she requests 05 0 The coalition is closed final payoffs are 0 05 05 Behavioral Identification in Coalitional Bargaining Demand Bargaining Example 0 A is selected first she requests 08 9 B is selected he requests 07 9 C is selected she requests 06 9 No possible coalition formed all requests erased 9 B is selected first in 2nd stage he requests 05 a C is selected she requests 05 o The coalition is closed final payoffs are 0 05 05 0 Only requires Subgame Perfection Experiments Behavioral Identification in Coalitional Bargaining 0 All games involve Experiments Behavioral Identification in Coalitional Bargaining o All games involve 0 Groups of five subjects Experiments Behavioral Identification in Coalitional Bargaining o All games involve 3 Groups of five subjects Majority rule Experiments Behavioral Identification in Coalitional Bargaining o All games involve 3 Groups of five subjects 0 Majority rule 0 No shrinking of pie Experiments Behavioral Identification in Coalitional Bargaining o All games involve 3 Groups of five subjects 0 Majority rule a No shrinking of pie 0 Test comparative static predictions wihtin model by varying bargaining power Experiments Behavioral Identification in Coalitional Bargaining o All games involve 3 Groups of five subjects a Majority rule a No shrinking of pie 0 Test comparative static predictions wihtin model by varying bargaining power 0 Also compare Behavioral Identification in Coalitional Bargaining Experiments o All games involve 9 Groups of five subjects a Majority rule a No shrinking of pie 0 Test comparative static predictions wihtin model by varying bargaining power 0 Also compare 0 All players have equal voting power equal weight game Behavioral Identification in Coalitional Bargaining Experiments o All games involve 9 Groups of five subjects a Majority rule a No shrinking of pie 0 Test comparative static predictions wihtin model by varying bargaining power 0 Also compare a All players have equal voting power equal weight game n One player controls 3 votes while others one vote each Apex game Results from Experiments Behavioral Identification in Coalitional Bargaining O mportant behavioral regularities make treatments more similar in outcomes than predicted by theory Results from Experiments Behavioral Identification in Coalitional Bargaining 0 Important behavioral regularities make treatments more similar in outcomes than predicted by theory formateur power 0 Onevote base players receive a small extra benefit from moving first Behavioral Identification in Coalitional Bargaining Results from Experiments 0 Important behavioral regularities make treatments more similar in outcomes than predicted by theory a Onevote base players receive a small extra benefit from moving first formateur power 0 Whereas Apex formateur is typically at or below DB prediction Behavioral Identification in Coalitional Bargaining Results from Experiments 0 Important behavioral regularities make treatments more similar in outcomes than predicted by theory a Onevote base players receive a small extra benefit from moving first formateur power 0 Whereas Apex formateur is typically at or below DB prediction 0 To verify lack of formateur power for Apex players derives from an equity consideration effect add third treatment Behavioral Identification in Coalitional Bargaining Results from Experiments 0 Important behavioral regularities make treatments more similar in outcomes than predicted by theory a Onevote base players receive a small extra benefit from moving first formateur power 0 Whereas Apex formateur is typically at or below DB prediction 0 To verify lack of formateur power for Apex players derives from an equity consideration effect add third treatment 0 Apex13 like Apex treatment except Apex player only paid of his earnings Results from Experiments Behavioral Identification in Coalitional Bargaining 0 Important behavioral regularities make treatments more similar in outcomes than predicted by theory a Onevote base players receive a small extra benefit from moving first formateur power 0 Whereas Apex formateur is typically at or below DB prediction 0 To verify lack of formateur power for Apex players derives from an equity consideration effect add third treatment 9 Apex13 like Apex treatment except Apex player only paid of his earnings 0 Apex player in this treatment exhibits some power Behavioral Identification in Coalitional Bargaining Comparison with Field Data Analysis 0 Estimate regressions like ones performed with field data Behavioral Identification in Coalitional Bargaining Comparison with Field Data Analysis 0 Find 0 Estimate regressions like ones performed with field data Behavioral Identification in Coalitional Bargaining Comparison with Field Data Analysis 0 Estimate regressions like ones performed with field data 0 Find 0 Number of remarkable similarities between experimental amp field data Behavioral Identification in Coalitional Bargaining Comparison with Field Data Analysis 0 Estimate regressions like ones performed with field data 0 Find a Number of remarkable similarities between experimental 84 field data 0 Impossible looking at experimental data to distinguish 2 games using criteria employed in field Behavioral Identification in Coalitional Bargaining Comparison with Field Data Analysis 0 Estimate regressions like ones performed with field data 0 Find a Number of remarkable similarities between experimental 84 field data 0 Impossible looking at experimental data to distinguish 2 games using criteria employed in field 0 Impossible to distinguish between 2 bargaining models solely on the basis of payoff data means Behavioral Identification in Coalitional Bargaining Comparison with Field Data Analysis 0 Estimate regressions like ones performed with field data 0 Find a Number of remarkable similarities between experimental 84 field data 0 Impossible looking at experimental data to distinguish 2 games using criteria employed in field 0 Impossible to distinguish between 2 bargaining models solely on the basis of payoff data means o Behavioral Ideniti cation Problem in field analysis Related Literature Behavioral Identification in Coalitional Bargaining Field Data Analysis of coalition governments minitries vs seats 0 Warwick and Druckman 2001 Related Literature Behavioral Identification in Coalitional Bargaining Field Data Analysis of coalition governments minitries vs seats 0 Warwick and Druckman 2001 saHence 0 find proportional relation except with weighted portfolios portfolio Behavioral Identification in Coalitional Bargaining Related Literature Field Data Analysis of coalition governments minitries vs seats 0 Warwick and Druckman 2001 0 find proportional relation except with weighted portfolios portfolio saHence o Ansolabehere Snyder Strauss amp Ting 2003 Behavioral Identification in Coalitional Bargaining Related Literature Field Data Analysis of coalition governments minitries vs seats 0 Warwick and Druckman 2001 a find proportional relation except with weighted portfolios portfolio saHence O Ansolabehere Snyder Strauss amp Ting 2003 u proposer power even wo controlling for portfolio salience Behavioral Identification in Coalitional Bargaining Related Literature Field Data Analysis of coalition governments minitries vs seats 0 Warwick and Druckman 2001 a find proportional relation except with weighted portfolios portfolio saHence O Ansolabehere Snyder Strauss amp Ting 2003 a proposer power even wo controlling for portfolio salience a real weights vs seat shares Related Literature Expe ii In ents Behavioral Identification in Coalitional Bargaining o Baron Ferejohn Related Literature Expe ii In ents Behavioral Identification in Coalitional Bargaining 0 Baron Ferejohn McKelvey 1991 Related Literature Expe ii In ents Behavioral Identification in Coalitional Bargaining 0 Baron Ferejohn o McKelvey 1991 0 Diermeier amp Morton 2000 Related Literature Expe ii In ents Behavioral Identification in Coalitional Bargaining 0 Baron Ferejohn o McKelvey 1991 a Diermeier 84 Morton 2000 I Frechette Kagei Lehrer 2003 Behavioral Identification in Coalitional Bargaining Related Literature Expe ii In ents 0 Baron Ferejohn o McKelvey 1991 a Diermeier 84 Morton 2000 a Frechette Kagel Lehrer 2003 a Frechette Kagal Morelli 2003 Behavioral Identification in Coalitional Bargaining Related Literature Expe ii In ents 0 Baron Ferejohn o McKelvey 1991 a Diermeier 84 Morton 2000 a Frechette Kagel Lehrer 2003 o Frechette Kagel Morelli 2003 o No experiment of Demand Bargaining Model Behavioral Identification in Coalitional Bargaining Related Literature Expe ii In ents O Baron Ferejohn a McKelvey 1991 a Diermeier 84 Morton 2000 a Frechette Kagel Lehrer 2003 o Frechette Kagel Morelli 2003 a No experiment of Demand Bargaining Model 0 Apex game tested within framework of cooperative game theory Le loose rules on subject play Behavioral Identification in Coalitional Bargaining Related Literature Expe ii In ents O Baron Ferejohn a McKelvey 1991 a Diermeier 84 Morton 2000 a Frechette Kagel Lehrer 2003 o Frechette Kagel Morelli 2003 a No experiment of Demand Bargaining Model 0 Apex game tested within framework of cooperative game theory Le loose rules on subject play a Selten 84 Schuster 1968 Behavioral Identification in Coalitional Bargaining Related Literature Expe ii In ents O Baron Ferejohn a McKelvey 1991 a Diermeier 84 Morton 2000 a Frechette Kagel Lehrer 2003 o Frechette Kagel Morelli 2003 a No experiment of Demand Bargaining Model 0 Apex game tested within framework of cooperative game theory Le loose rules on subject play a Selten 84 Schuster 1968 u Horowitz 84 Rapoport 1974 Behavioral Identification in Coalitional Bargaining Experimental Design for Both Games 0 5 subjects per group 10 to 15 in lab Behavioral Identification in Coalitional Bargaining Experimental Design for Both Games 0 5 subjects per group 10 to 15 in lab 0 Divide 60 between 5 voting blocks with one subject representing each voting block Behavioral Identification in Coalitional Bargaining Experimental Design for Both Games 0 5 subjects per group 10 to 15 in lab Divide 60 between 5 voting blocks with One subject representing each voting block 11 bargaining rounds including 1 dryrun Behavioral Identification in Coalitional Bargaining Experimental Design for Both Games O 5 subjects per group 10 to 15 in lab 0 Divide 60 between 5 voting blocks with one subject representing each voting block 0 ll bargaining rounds including 1 dryrun I l bargaining round selected at random for payment Behavioral Identification in Coalitional Bargaining Experimental Design for Both Games 0 5 subjects per group 10 to 15 in lab Divide 60 between 5 voting blocks with one subject representing each voting block 11 bargaining rounds including 1 dryrun l bargaining round selected at random for payment 8 showup fee Behavioral Identification in Coalitional Bargaining Experimental Design for Both Games O 5 subjects per group 10 to 15 in lab 0 Divide 60 between 5 voting blocks with one subject representing each voting block 11 bargaining rounds including 1 dryrun l bargaining round selected at random for payment 8 showup fee Every treatment has 2 inexperienced amp l experienced sessions In Baron Ferejohn Treatments Behavioral Identification in Coalitional Bargaining 0 All subjects enter a proposal allocating 60 In Baron Ferejohn Treatments Behavioral Identification in Coalitional Bargaining 0 All subjects enter a proposal allocating 60 0 One proposal randomly selected as standing proposal Behavioral Identification in Coalitional Bargaining In Baron Ferejohn Treatments 0 All subjects enter a proposal allocating 60 0 One proposal randomly selected as standing proposal 0 Posted on subjects screens giving amounts allocated to each voting block by subject number along with number of votes controlled by subject Behavioral Identification in Coalitional Bargaining In Baron Ferejohn Treatments 0 All subjects enter a proposal allocating 60 0 One proposal randomly selected as standing proposal 0 Posted on subjects screens giving amounts allocated to each voting block by subject number along with number of votes controlled by subject 0 Proposals are voted up or down Behavioral Identification in Coalitional Bargaining In Baron Ferejohn Treatments 0 All subjects enter a proposal allocating 60 0 One proposal randomly selected as standing proposal 0 Posted on subjects screens giving amounts allocated to each voting block by subject number along with number of votes controlled by subject 0 Proposals are voted up or down 0 If simple majority accepts proposal payoffs implemented amp bargaining round ends Behavioral Identification in Coalitional Bargaining In Baron Ferejohn Treatments 0 0 All subjects enter a proposal allocating 60 One proposal randomly selected as standing proposal Posted on subjects screens giving amounts allocated to each voting block by subject number along with number of votes controlled by subject Proposals are voted up or down If simple majority accepts proposal payoffs implemented amp bargaining round ends If proposal rejected process repeats itsehc Behavioral Identification in Coalitional Bargaining In Demand Bargaining Treatments 0 All subjects enter a demand between 0 and 60 Behavioral Identification in Coalitional Bargaining In Demand Bargaining Treatments 0 All subjects enter a demand between 0 and 60 0 One request is randomly selected 3984 that request is posted on the screens of the other subjects in that group Behavioral Identification in Coalitional Bargaining In Demand Bargaining Treatments o All subjects enter a demand between 0 and 60 0 One request is randomly selected amp that request is posted on the screens of the other subjects in that group 0 The remaining subjects place a new request between 0 and 60 Behavioral Identification in Coalitional Bargaining In Demand Bargaining Treatments All subjects enter a demand between 0 and 60 One request is randomly selected amp that request is posted on the screens of the other subjects in that group The remaining subjects place a new request between 0 and 60 If sum of a set of requests made so far is less than or equal to 60 amp sum of votes controlled by subjects who have placed those requests is enough for a majority then subject who made last request decides to close or not amp with whom if he decides to close payoffs are implemented amp bargaining round ends Behavioral Identification in Coalitional Bargaining In Demand Bargaining Treatments All subjects enter a demand between 0 and 60 One request is randomly selected amp that request is posted on the screens of the other subjects in that group The remaining subjects place a new request between 0 and 60 If sum of a set of requests made so far is less than or equal to 60 amp sum of votes controlled by subjects who have placed those requests is enough for a majority then subject who made last request decides to close or not amp with whom if he decides to close payoffs are implemented amp bargaining round ends Otherwise process continues Behavioral Identification in Coalitional Bargaining In Demand Bargaining Treatments All subjects enter a demand between 0 and 60 One request is randomly selected amp that request is posted on the screens of the other subjects in that group The remaining subjects place a new request between 0 and 60 If sum of a set of requests made so far is less than or equal to 60 amp sum of votes controlled by subjects who have placed those requests is enough for a majority then subject who made last request decides to close or not amp with whom if he decides to close payoffs are implemented amp bargaining round ends 0 Otherwise process continues If no coalition is formed by end of stage new round starts amp process repeats itself II Behavioral Identification in Coalitional Bargaining Let39s Get the Terms Right o EaCh Session has Behavioral Identification in Coalitional Bargaining Let39s Get the Terms Right 0 Each Session has 0 multiple ll bargaining Rounds Behavioral Identification in Coalitional Bargaining Let39s Get the Terms Right 0 Each Session has a multiple ll bargaining Rounds 0 that can have any number of Stages Behavioral Identification in Coalitional Bargaining Let39s Get the Terms Right 0 Each Session has a multiple ll bargaining Rounds 0 that can have any number of Stages 0 which in demand bargaining each have up to 5 Steps Treatments Behavioral Identification in Coalitional Bargaining 0 Original design 2 Treatments Treatments Behavioral Identification in Coalitional Bargaining 0 Original design 2 Treatments 0 Equal Weight A total of 5 votes 1 per subject Treatments Behavioral Identification in Coalitional Bargaining 0 Original design 2 Treatments 31111 0 Equal Weight A total of 5 votes 1 per subject 0 Apex A total of 7 votes Treatments Behavioral Identification in Coalitional Bargaining 0 Original design 2 Treatments 0 Equal Weight A total of 5 votes 1 per subject a Apex A total of 7 votes 31ll 0 Additional treatment ApeX13 same as Apex but Apex player is only paid of his payost Behavioral Identification in Coalitional Bargaining Treatments 0 Original design 2 Treatments 0 Equal Weight A total of 5 votes 1 per subject a Apex A total of 7 votes 31ll l 0 Additional treatment Apex13 same as Apex but Apex player is only paid of his payoffs 0 Equity fairness considerations Behavioral Identification in Coalitional Bargaining Treatments 0 Original design 2 Treatments 0 Equal Weight A total of 5 votes 1 per subject a Apex A total of 7 votes 3 l l 11 0 Additional treatment Apex13 same as Apex but Apex player is only paid of his payoffs 9 Equity fairness considerations I Represent case where power equals number of representatives Behavioral Identification in Coalitional Bargaining Treatments 0 Original design 2 Treatments 0 Equal Weight A total of 5 votes 1 per subject a Apex A total of 7 votes 3 l l 11 0 Additional treatment Apex13 same as Apex but Apex player is only paid of his payoffs 9 Equity fairness considerations 9 Represent case where power equals number of representatives 0 All use proportional selection protocol Predicted Shares Behavioral Identification in Coalitional Bargaining Behavioral Identification in Coalitional Bargaining Predicted Shares by 3 for base players Behavioral Identification in Coalitional Bargaining Predicted Shares Behavioral Identification in Coalitional Bargaining Results Does Bargaining End in Stage 17 Conclusion 1 Behavioral Identification in Coalitional Bargaining Sig O Majority of allocations completed in stage 1 for both BF 84 DB stat Behavioral Identification in Coalitional Bargaining Conclusion 1 o Majority of allocations Completed in stage 1 for both BF amp DB stat sig 0 quot 39 quot more Sig r39 J in stage 1 under DB stat Behavioral Identification in Coalitional Bargaining Conclusion 1 o Majority of allocations completed in stage 1 for both BF amp DB stat sig 0 qllh tantiall more allocation r39 t J in stage 1 under DB stat sig 0 However far from all DB bargaining rounds ended in minimal number of steps Behavioral Identification in Coalitional Bargaining Conclusion 1 Majority of allocations completed in stage 1 for both BF amp DB stat sig qlll fanfiall sig However far from all DB bargaining rounds ended in minimal number of steps more allocation r39 t J in stage 1 under DB stat ln equal wt treatment 45 require more than 3 steps Behavioral Identification in Coalitional Bargaining Conclusion 1 0 Majority of allocations completed in stage 1 for both BF amp DB stat sis Hh tantiall more allocation r39 t J in stage 1 under DB stat sis However far from all DB bargaining rounds ended in minimal number of steps In equal wt treatment 45 require more than 3 steps Cause Early requests are too high 054 for excluded vs 029 for included when there are 4 steps Behavioral Identification in Coalitional Bargaining Frequency of Minimal Winning Coalitions MWC Conclusion 2 Behavioral Identification in Coalitional Bargaining o Majority of proposals were For MWCs stat Sig Conclusion 2 Behavioral Identification in Coalitional Bargaining o Majority of propOSals were for MWCs stat sig 0 FrequenCy of MWCs is slightly higher under DB not stat Sig El 5 3 RCquot Behavioral Identification in Coalitional Bargaining Conclusion 2 o Majority of prop0sas were for MWCs stat sig 0 Frequency of MWCs is slightly higher under DB not stat sig 0 NonMWCS in DB treatment consists of cases where a subject closed but left money for later movers Behavioral Identification in Coalitional Bargaining Conclusion 2 o Majority of proposals were for MWCs stat sig 9 Frequency of MWCs is slightly higher under DB not stat sig 0 NonMWCs in DB treatment consists of cases where a subject closed but left money for later movers O The sum leftover averaged 815 for DB as compared to 1385 given to redundant coalition partners in BF game Behavioral Identification in Coalitional Bargaining BF Allocations Passed for MWCs Behavioral Identification in Coalitional Bargaining BF Allocations Passed for MWCs Behavioral Identification in Coalitional Bargaining BF Allocations Passed for MWCs Behavioral Identification in Coalitional Bargaining DB Allocations Passed for MWCs Behavioral Identification in Coalitional Bargaining DB Allocations Passed for MWCs Behavioral Identification in Coalitional Bargaining DB Allocations Passed for MWCs Conclusion 3 Behavioral Identification in Coalitional Bargaining stat sig but only vs worst off subject in DB El 5 3 RCquot I Base formateurs have a rst mover advantage in both BF amp DB Behavioral Identification in Coalitional Bargaining Conclusion 3 0 Base formateurs have a first mover advantage in both BF amp DB stat sig but only vs worst OFF subject in DB 0 Although base formateurs do not take nearly as much as predicted under BF as predicted they take more than under DB stat sig Behavioral Identification in Coalitional Bargaining Conclusion 3 0 Base formateurs have a first mover advantage in both BF amp DB stat sig but only vs worst OFF subject in DB 0 Although base formateurs do not take nearly as much as predicted under BF as predicted they take more than under DB stat sig I Apex formateurs have no formateur power in both games thus behavior is much more similar than predicted between BF amp DB not stat sig dif Behavioral Identification in Coalitional Bargaining Aboth Formateur Power in Baron Ferejohn 0 Limited formateur power results from voting behavior Behavioral Identification in Coalitional Bargaining Aboth Formateur Power in Baron Ferejohn a Limited formateur power results from voting behavior 0 High rejection rates of shares close to Stationary Subgame Perfect Equilibrium Prediction SSPE imply that SSPE share does not maximize expected income Behavioral Identification in Coalitional Bargaining Aboth Formateur Power in Demand Bargaining 0 Why did base players accept smaller shares Behavioral Identification in Coalitional Bargaining Aboth Formateur Power in Demand Bargaining 0 Why did base players accept smaller shares 0 Risk aversion Behavioral Identification in Coalitional Bargaining Aboth Formateur Power in Demand Bargaining 0 Why did base players accept smaller shares 0 Risk aversion Inability to perform backward induction Behavioral Identification in Coalitional Bargaining Aboth Formateur Power in Demand Bargaining 0 Why did base players accept smaller shares 0 Risk aversion 9 Inability to perform backward induction 0 Belief that others cannot perform backward induction Behavioral Identification in Coalitional Bargaining About Formateur Power in Demand Bargaining 0 Why did base players accept smaller shares 0 Risk aversion 9 Inability to perform backward induction 9 Belief that others cannot perform backward induction 0 However this is not very costly lowest share in equal wt treatmetn is 290 lower than predicted Behavioral Identification in Coalitional Bargaining About Formateur Power in Demand Bargaining o 900 o Why did base players accept smaller shares Risk aversion Inability to perform backward induction Belief that others cannot perform backward induction However this is not very costly lowest share in equal wt treatmetn is 290 lower than predicted Cannot explain Apex taking smaller shares since they are always in However they accept 648 less than predicted Aboth Coalition Choice Behavioral Identification in Coalitional Bargaining o Baron Ferejohn Aboth Coalition Choice Behavioral Identification in Coalitional Bargaining o Baron Ferejohn I Base player earns more as format eurs when partnering with Apex Aboth Coalition Choice Behavioral Identification in Coalitional Bargaining o Baron Ferejohn 0 Base player earns more as formateurs when partnering with Apex They invite Apex player 70 of time as compared to predicted rate of 25 Aboth Coalition Choice Behavioral Identification in Coalitional Bargaining o Baron Ferejohn 9 Base player earns more as formateurs when partnering with Apex 0 They invite Apex player 70 of time as compared to predicted rate of 25 0 Demand Bargaining Behavioral Identification in Coalitional Bargaining Aboth Coalition Choice 0 Baron Ferejohn a Base player earns more as formateurs when partnering with Apex a They invite Apex player 70 of time as compared to predicted rate of 25 0 Demand Bargaining 0 Base forms coalition with Apex 100 of the time Behavioral Identification in Coalitional Bargaining Aboth Coalition Choice 0 Baron Ferejohn a Base player earns more as formateurs when partnering with Apex a They invite Apex player 70 of time as compared to predicted rate of 25 0 Demand Bargaining a Base forms coalition with Apex 100 of the time a Only 4 bargaining rounds where Apex was not selected by fourth step Behavioral Identification in Coalitional Bargaining could explain 0 Literature on bilateral bargaining games suggest equity considerations Behavioral Identification in Coalitional Bargaining could explain 0 Literature On bilateral bargaining games suggest equity COnsiderations Why Apex proposers do not exhibit formataur power in BF Behavioral Identification in Coalitional Bargaining could explain 0 Literature On bilateral bargaining games suggest equity COnsiderations 9 Why Apex proposers do notexhibit formateur power in BF I Why Base proposers rarely propose 4way coalitions Behavioral Identification in Coalitional Bargaining 0 Literature on bilateral bargaining games suggest equity considerations could explain 0 Why Apex proposers do not exhibit formateur power in BF 8 Why Base proposers rarely propose 4way coalitions 0 Why Base formateurs exhibit formateur power in DB Behavioral Identification in Coalitional Bargaining 0 Literature on bilateral bargaining games suggest equity considerations could explain 0 Why Apex proposers do not exhibit formateur power in BF 3 Why Base proposers rarely propose 4way coalitions a Why Base formateurs exhibit formateur power in DB 0 To verify this hypothesis ranApex13 treatment identical to Apex treatment except Apex player only gets of his earnings at end Behavioral Identification in Coalitional Bargaining 0 Literature on bilateral bargaining games suggest equity considerations could explain 0 Why Apex proposers do not exhibit formateur power in BF 3 Why Base proposers rarely propose 4way coalitions a Why Base formateurs exhibit formateur power in DB 0 To verify this hypothesis ranApex13 treatment identical to Apex treatment except Apex player only gets of his earnings at end I Equalizes exante payoffs of base amp Apex players Behavioral Identification in Coalitional Bargaining Allocations Passed for MWCs in Third Treatment Behavioral Identification in Coalitional Bargaining Allocations Passed for MWCs in Third Treatment Apex versus Third Treatment Behavioral Identification in Coalitional Bargaining 0 Under both BF amp DB Apex players obtain a small formateur advantage in the Apexlg stat Sig Behavioral Identification in Coalitional Bargaining Apex versus Third Treatment 0 Under both BF amp DB Apex players obtain a small formateur advantage in the ApeX13 stat sig 0 Apex players require larger nominal shares under both BF amp DB in ApeX13 when invited by base players more details later Behavioral Identification in Coalitional Bargaining Apex versus Third Treatment 0 Under both BF amp DB Apex players obtain a small formateur advantage in the ApeX13 stat sig 0 Apex players require larger nominal shares under both BF amp DB in ApeX13 when invited by base players more details later a Base players invite Apex players into MWCs less often in BF ApeX13 only 39 of the time as opposed to 70 of the time Behavioral Identification in Coalitional Bargaining Apex versus Third Treatment 0 Under both BF amp DB Apex players obtain a small formateur advantage in the ApeX13 stat sig 0 Apex players require larger nominal shares under both BF amp DB in ApeX13 when invited by base players more details later 0 Base players invite Apex players into MWCs less often in BF ApeX13 only 39 of the time as opposed to 70 of the time o In BF there is no impact on share base formateurs obtain when forming coalitions with all base players not stat sig Behavioral Identification in Coalitional Bargaining Apex versus Third Treatment 0 Under both BF amp DB Apex players obtain a small formateur advantage in the ApeX13 stat sig 0 Apex players require larger nominal shares under both BF amp DB in ApeX13 when invited by base players more details later 0 Base players invite Apex players into MWCs less often in BF ApeX13 only 39 of the time as opposed to 70 of the time o In BF there is no impact on share base formateurs obtain when forming coalitions with all base players not stat sig 0 Base formateurs in both BF amp DB display a small formateur advantage stat sig Conclusion 4 Behavioral Identification in Coalitional Bargaining 0 Voting power accounts for of difference in average share of Apex treatment formateurs in Apex13 treatment 4 a base player in equal weight Conclusion 4 Behavioral Identification in Coalitional Bargaining 0 Voting power accounts for of difference in average share of Apex treatment formateurs in ApeX13 treatment amp a base player in equal weight 0 The remaining represents an equity adjustment Behavioral Identification in Coalitional Bargaining allocation Conclusions 5 8 6 7 Analysis of Voting Patterns see paper for statistical analysis 0 Own share of benefits is key factor aFFecting voting for or against an Behavioral Identification in Coalitional Bargaining Conclusions 5 8 6 7 Analysis of Voting Patterns see paper for statistical analysis 0 Own share of bene ts is key factor affecting voting for or against an allocation 0 As predicted average shares required to vote in favor of an allocation are larger under DB than BF Behavioral Identification in Coalitional Bargaining Conclusions 5 8 6 7 Analysis of Voting Patterns see paper for statistical analysis 0 Own share of bene ts is key factor affecting voting for or against an allocation 0 As predicted average shares required to vote in favor of an allocation are larger under DB than BF 0 However these differences not as large as predicted Behavioral Identification in Coalitional Bargaining Conclusions 5 8 6 7 Analysis of Voting Patterns see paper for statistical analysis Own share of bene ts is key factor affecting voting for or against an allocation As predicted average shares required to vote in favor of an allocation are larger under DB than BF However these differences not as large as predicted Apex players required a higher share to support an allocation in ApeX13 consistent with notion that subjects have some lower bound on payoffs they are willing to accept Behavioral Identification in Coalitional Bargaining Conclusions 5 8 6 7 Analysis of Voting Patterns see paper for statistical analysis Own share of bene ts is key factor affecting voting for or against an allocation As predicted average shares required to vote in favor of an allocation are larger under DB than BF However these differences not as large as predicted Apex players required a higher share to support an allocation in ApeX13 consistent with notion that subjects have some lower bound on payoffs they are willing to accept However acceptance thresholds are sensitive to strategic considerations Behavioral Identification in Coalitional Bargaining Conclusions 5 8 6 7 Analysis of Voting Patterns see paper for statistical analysis 0 Own share of bene ts is key factor affecting voting for or against an allocation 0 As predicted average shares required to vote in favor of an allocation are larger under DB than BF However these differences not as large as predicted Apex players required a higher share to support an allocation in ApeX13 consistent with notion that subjects have some lower bound on payoffs they are willing to accept 0 However acceptance thresholds are sensitive to strategic considerations 0 See paper also for analysis of profits made by subjects as compared to predictions II Comparison to Field Data Behavioral Identification in Coalitional Bargaining 0 Empirical analysis use postworld war II European coalition governments to distinguish between BF amp DB Comparison to Field Data Behavioral Identification in Coalitional Bargaining 0 Empirical analysis use postworld war II European coalition 0 Two recent approaches governments to distinguish between BF amp Comparison to Field Data Behavioral Identification in Coalitional Bargaining 0 Empirical analysis use postworld war II European coalition 0 Two recent approaches governments to distinguish between BF amp 0 Warwickamp Druckman Comparison to Field Data Behavioral Identification in Coalitional Bargaining 0 Empirical analysis use postworld war II European coalition 0 Two recent approaches governments to distinguish between BF amp 0 Warwickamp Druckman I Ansolabeh ere Snyder Strauss amp Ting Warwick 8 Druckman Behavioral Identification in Coalitional Bargaining 0 Measure a party39s strength by share of seats they contribute to winning coalition Behavioral Identification in Coalitional Bargaining Warwick 8 Druckman Behavioral Identification in Coalitional Bargaining Warwick 8 Druckman Behavioral Identification in Coalitional Bargaining Warwick 8 Druckman Behavioral Identification in Coalitional Bargaining Warwick 8 Druckman Behavioral Identification in Coalitional Bargaining Warwick 8 Druckman Behavioral Identification in Coalitional Bargaining Ansolabehere Snyder Strauss 8 Ting a Seat shares do not generally equal voting weight Behavioral Identification in Coalitional Bargaining Ansolalehere Snyder Strauss 8 Ting 0 Seat shares do not generally equal voting weight 0 BF amp DB make predictions based on voting weight Behavioral Identification in Coalitional Bargaining Ansolalehere Snyder Strauss 8 Ting 0 Seat shares do not generally equal voting weight 0 BF amp DB make predictions based on voting weight 0 They develop a framework that nests BF amp DB Behavioral Identification in Coalitional Bargaining Ansolabehere Snyder Strauss 8 Ting Behavioral Identification in Coalitional Bargaining Ansolabehere Snyder Strauss 8 Ting Behavioral Identification in Coalitional Bargaining Ansolabehere Snyder Strauss 8 Ting Conclusions Behavioral Identification in Coalitional Bargaining o Expost distribution of bene ts under BF amp DB are predicted to be very diFFerent Conclusions Behavioral Identification in Coalitional Bargaining o EXpost distribution of benefits under BF amp DB are predicted to be very different BF sharply skewed in favor of proposer Conclusions Behavioral Identification in Coalitional Bargaining a EXpost distribution of benefits under BF amp DB are predicted to be very different 3 BF sharply skewed in favor of proposer I BD predicts shares proportional to bargaining power Behavioral Identification in Coalitional Bargaining Conclusions o EXpost distribution of benefits under BF amp DB are predicted to be very different 3 BF sharply skewed in favor of proposer 9 BD predicts shares proportional to bargaining power 9 As predicted experimental data shows Behavioral Identification in Coalitional Bargaining Conclusions o EXpost distribution of benefits under BF amp DB are predicted to be very different 3 BF sharply skewed in favor of proposer 9 BD predicts shares proportional to bargaining power I As predicted experimental data shows 0 Proposer power in BF games Behavioral Identification in Coalitional Bargaining Conclusions o EXpost distribution of benefits under BF amp DB are predicted to be very different 3 BF sharply skewed in favor of proposer 9 BD predicts shares proportional to bargaining power I As predicted experimental data shows a Proposer power in BF games 0 Benefits shift substantially in favor of player with greater real voting power Conclusions Continued Behavioral Identification in Coalitional Bargaining materialize 0 However sharp difFerences in expost shares between BF 8 DB fail to Conclusions Continued Behavioral Identification in Coalitional Bargaining materialize 0 However sharp differences in eXpos t shares between BF amp DB fail to I Less formateur advantage than predicted in BF Conclusions Continued Behavioral Identification in Coalitional Bargaining materialize 0 However sharp differences in eXpos t shares between BF amp DB fail to 0 Less formateur advantage than predicted in BF Unpredicted formateur advantage in D Behavioral Identification in Coalitional Bargaining Conclusions Continued 0 However sharp differences in eXpost shares between BF amp DB fail to materialize 9 Less formateur advantage than predicted in BF 9 Unpredicted formateur advantage in 0 Using epxeirmental data to conduct regressions similar to those reported with field data for distinguishing between BF amp DB unable to distinguish which game subjects are playing 113007 Week 9 Notes Section 1C Classes Structs and classes textbook reference Ch 6 Ch 7 up to half of page 291 most of Ch 103 1 Classes Introduction So far we have seen different types of data such as those oftype int doubl e char etc What if we want a type that contains other types because we want to associate multiple types of data with a certain type Perhaps we want a Date type that has a month day and year We can do this with an array of ints However keeping track of Date 0 Datel and Date2 can be cumbersome This is where classesstructs come to the rescue We39ll use the term 39class39 in this discussion We can create types called classes that keep track of these data elds in a more structured and exible way and classes let us de ne our custom types These classes can also have functions as well and we can add functions that give us the date in various forms Without further ado let39s start with a simple Candy class and add functionality as we go along 11 De nitions Member Variables Member Functions A class is a type that group together related variables and functions known as member variables and member functions Let s create a 39Candy39 class containing the variables 39category string type and 39age int type and the function growMoreStale of void type QSTN For this problem use either the struct or the class keyword struct Candy string category int age void growMoreStale remember that semicolonl class Candy public string category int age void growMoreStale l One difference by default the member variables and functions of a struct are public while those for a class are private More on public and private later Having the above code let39s say we have a Candy variable called 39trident Next we want to set the member variable 39category to be the string quotgumquot and call the growMoreStale function QSTN Complete int main i Candy trident fill in below tridentcategory quotgumquot tridentgrowMoreStale As we can see the member variables and functions are accessed using the dot operator It39s also perfect legal to set a member variable to equal the value of another member variable For example if we had another candy called Orbit we can set its category to gum Candy orbit orbitcategory tridentcategory So far so good But wait something s still missing right Running the above code doesn t compile because we haven39t de ned what trident growMoreStale accomplishes 12 De nition Member functions QSTN So let39s de ne that now Fill in the function definition void CandygrowMoreStale cout ltlt quotyo I grew more stalelnquot age In general member functions are of the format ltclass namegtltfunction name with parameter listgt QSTN What does the main function print out Class Candy i public string category int a e void growMoreStale l void CandygrowMoreStale cout ltlt quotyo I grew more stalelnquot age int main Candy trident tridentcategory quotgumquot tridentage 1 Candy mystery mysterycategory quotblobquot mysteryage 41 cout ltlt tridentage ltlt endl mysterygrowMoreStale cout ltlt mysteryage ltlt endl tridentage mysteryage tridentage cout ltlt tridentage ltlt endl Answer 1 yo I grew more stale 43 Notes 0 By convention the rst letter of the class name is capitalized Candy Date 0 Every instance of a class in this example 39trident39 and 39mystery39 has its own set of members 0 Member functions and variables by themselves aren39t useful in main They must be associated with speci c instances 0 We can have multiple class de nitions in a le each with their own variables and functions 13 Public and Private Members Ok back to the quotpublicquot statement which means that the following variables and functions are quotpublicquot to others and can be accessed in main for example This might sound convenient but it can be dangerous For example look at this statement trident age 50 According to C this is perfectly legal However does it make sense Another classic example is the BankAccount class containing a variable for balance Obviously some problems may arise from such a setup Therefore we can also set member variables and functions to be quotprivatequot using the private statement so that only code within the class39s functions can modify that variable M Complete the following code by setting the 39age and 39category to private and the function to 39public39 Also state what happens when run the main function Class Candy public void growMoreStale private int age string category l void CandygrowMoreStale cout ltlt quotyo I grew more stalelnquot age int main Candy trident tridentage l cout ltlt tridentage ltlt endl As we noted earlier only functions within the Candy class can change the age QSTN Keeping the current variables private how can we change the age from the main class We39ll can add either mutator functions or constructors First of all an accessor function allows us to read data We39ll write a function named getAge that returns he current age Mutator functions allow us to change the data We ll also write a function named setAge that takes in an int newAge as a parameter setAge sets the age to newAge ifthe age is at least 0 otherwise it prints Invalid age setting to age 0n and sets the age to 0 The following code should return 4 int main Candy trident trident setAge 4 cout ltlt tridentgetAge ltlt endl QSTN Complete the above task of writing getAge and setAge making changes to the class definition as necessary class Candy i public void growMoreStale private int a e string category i int CandygetAge i return age l void CandysetAgeint newAge i if age gt m age newAge else cout ltlt quotInvalid age setting to age 0nquot age 14 Constructors Even with the above accessors and mutators we still have to manually call the accessors and mutators after declaring the variable of type Class end sectionl notes 2 Misc Topics in Preparation for the Final I encourage you to see your own midterm exam during the professor39s OH if that s allowed Q Do we always have to declare functions other than main before using them Chasing a blind snail through a random maze Random walk on random graphs Marek Biskup Based on joint work with Noam Berger Random maze Bond percolation on 22 ruzzuqr quot 39 THE FL E39E 35 by ngzx E I 1 Bond percolation gt Keep edge with probability p gt Remove it with probability 1 p 1 539 Th39kf1 1 39 r 39 In 0 p ltlt 1 rm Random maze Bond percolation on 22 Bond percolation gt Keep edge with probability p gt Remove it with probability 1 p Thinkof1 pltlt1 Snail performs random walk ruzzuqr quot 39 THE FL E39E 35 by ngzx E I 1 I 3 5 i 3 12 m 3 E lhtl Random maze Bond percolation on 22 Bond percolation gt Keep edge with probability p gt Remove it with probability 1 p Thinkof1 pltlt1 Snail performs random walk ruzzuqr quot 39 THE FL E39E 35 by ngzx E I 1 I 3 5 i 3 12 m 3 E lhtl Random maze Snail s random walk 5 i r 1 hr x 15 hr 391L 39 Main questions gt Exit point distribution Random maze Snail s random walk supra quot31 Main questions gt Exit point distribution gt Time needed to exit Random maze Snail s random walk Main questions gt Exit point distribution gt Time needed to exit gt Snail s path is there a scaling limit Hitting probability Walk exits through the top side Electrostatic version Potential at the origin Discrete harmonic analysis Definition 1 Let G V E be a graph A function 1 V gt Rd is called discrete harmonic if VX e V AW quot2 Z wylJX0 y MIKE Discrete harmonic analysis Definition 1 Let G V E be a graph A function if V gt Rd is called discrete harmonic if VX e V AW quot2 Z wylJX0 y MIKE Automatic properties no conditions 1 Maximum principle Subtle properties depend on the graph 2 Lieouville s theorem 3 Harnack inequality Connections with random walk Let X1 X2 successive positions of the random walk on V Walk started at X Probability distribution PX expectation EX Connections with random walk Let X1 X2 successive positions of the random walk on V Walk started at X Probability distribution PX expectation EX Theorem 2 Dirichlet problem Let V0 C V be nite Let 1 V0 gt R be harmonic on V0 With boundary conditions 11 on 6V0 Then W ExlXT VX 6 V0 Where T rst time the walk leaves V0 Connections with random walk Let X1 X2 successive positions of the random walk on V Walk started at X Probability distribution PX expectation EX Theorem 2 Dirichlet problem Let V0 C V be nite Let 1 V0 gt R be harmonic on V0 With boundary conditions 11 on 6V0 Then 00 ExVXT VX 6 Va Where T rst time the walk leaves V0 Proof X gt EX1XT is discrete harmonic on V0 with be 11 Maximum principle gt 3 at most one such function 1 Electrostatic problem revisited Geometric embedding Electrostatic problem revisited Geometric embedding The position pX X is not discreteharmonic Electrostatic problem revisited Harmonic embedding The position pX X XX is discreteharmonic Electrostatic problem solved Notations 1 Infinite slab Xy y 5 N 2 Potential 1 on top bar 1 on bottom bar 3 X XX new position of site X Electrostatic problem solved Notations 1 Infinite slab Xy y 5 N 2 Potential 1 on top bar 1 on bottom bar 3 X XX new position of site X Theorem 3 If p X is the potential at X then M XXXez Martingales Definition 4 A sequence of random variables M0 M1 is a martingale if EM 1M09quot39JMI7MI79 n20 Eg game with zero expected profit Martingales Definition 4 A sequence of random variables M0 M1 is a martingale if EMI71M09quot399MI7MI79 ngt0 Eg game with zero expected profit Theorem 5 Harmonic RW martingale Let G V E be a graph and let 1 V gt Rd be harmonic Let X0 X1 be the random walk on V De ne MnzpX n2 0 Then M0 M1 is a martingale Hitting probability Martingale calculation Let M7 X7 XX Then M7 is gt random walk on deformed graph gt martingale A classic martingale calculation 92 39 M0 92 E0M0 92 E0MT 2P0top hit before bottom 1 From here we get P0wak exits thru top glt1 Snail s slimy trail w 31 z39 39r 39 quotEin 39 3913 Martingale functional CLT Diffusive scaling Scale space by J5 and time by n Explicitly 1 570 UV Ltnj tn LthMLtnj1 Mm Note t gt Bt is a continuous path Martingale functional CLT Diffusive scaling Scale space by J5 and time by n Explicitly 1 570 owitnj tn LthMLtnj1 Mimi Note t gt Bt is a continuous path Theorem 6 Martingale functional CLT folk version If margingale Mn has stationary squareintegrable increments then as n gt 00 the laW of Bnt t 2 O converges to that of Brownian motion Precise conditions of this theorem hold for M7 X7 XX on almostevery percolation configuration Correction on deformation All those thing under the rug Previous slide Deformed walk gt Brownian motion Need to correct on deformation We show that NO 0IX I Proof quite nontrivial see BBO5 for details Some pictures Percolation cluster and its deformation 50 x 50 box p 095 Some pictures Percolation cluster and its deformation 50 x 50 box p 085 Some pictures Percolation cluster and its deformation 50 x 50 box p 075 Some pictures Percolation cluster and its deformation 50 x 50 box p 065 39 a l9quot 4 3 Some pictures Percolation cluster and its deformation 50 x 50 box p 055 Lecture 15 Last time We cornpieted our discussion or a yery powerrui analytical approach to roiiowing he work or eosset and Fisher we rnotiyated the standardized or studentized testa istic and the accompanying trdiStHbutiOh The approach taken by eosset and FisherWas to anaiyticaiiy deriye the exact distribution for hohehorrnai popuia ions we need iarge sarnpies iarge enough Whit iiii iui rain 7 u hetedisthbutioh we contrasted this approach with our reerahdomiza ion and re 39 39 the 9 toois tend to agree Fihahy we repeated Fisher s anaiysis or DarWih S Zea Mays data ending with a rerandornization test Fishery you wiii recaii was an eariy proponent or 1 r4 1 been arrived at by this elementary method quot Today We are going to start to douoie pack and introduce sorne toois that aiiow us to dig a bit deeper into the structure or a data set sorne or he ideas We H discuss are a rnikture or rnodeiing and data anaiysis We wiii start with a procedure caiied Muitidirnensionai Scahhg it is a tooi that wiii heip us see reiationships in data sets wi h rnany yariaoies We wiii then rnoye to a kind or treeebased rnodei that heips us identity irnportant predictor or ekpianatory yariaoies in an inputerespohse rrarnework tor iack or a bet er errn Con dence intervals and tests Under the ciassicai approach we saw hatthe diSthbutioh or hetestatis ic did not depend on an unknown pararneters e or 0 i out instead on the number or suoiectsiterns in our study n or neiyto be precise We can a t N ihteNai With the formuia ii t s Wherel is the 0 957 quahtiie ofthe tr distribu ioh With hri degrees offreedorh in that exampiey data were paired dirrerences and we wanted to know it the orrspring or cross or Seifefertihzed piants wouid be taiier Con dence intervals and tests Con dence intervals and testing lfthat is the case then the value ofthe test statis ic we observed is not large it f Suppose hat 0 was in the 95 con dence interval hen we would knowthat Is not out In he 5 tails and hence we would not reject the test 0 hypothesis at the 005 level xitquotsh S 0 S Wax Recall that Fisher39s actual confidence interval is or rewritten a bit xi 214s 262i 214 472v1 00045229 Reasoning as in the last two slides the value of is outside the 95 linkt 39 rur large values u quotypluu39 u ofthe tstatistic for example under the null hypothesis hat there is no dilference in offspring heights he statis ic above would have a tdistribu ion with n1 degrees offreedom Your midterm con dence intervals and teS ng Your midterm concerns data from socalled DNA microarray experiments r u r a experiments on thousands of genes at one time r y evidence 39 quot we can The result is a kind of largedata problem but transposed rather than struggle r u k quot A the wil a 39 39 these studies elfect were initially at least expensive and there were a small number of subjects but many many variables Similar quot y u lu our 39 randomiza ion approaches to the same problem The data for your midterm comes from a classic paper by Golub et al that 39 39 39 1999 he goal on r agenetic emia tersrt 39 leuk AML and acute lymphoblastic leukemia ALL 139 a mum s quotmomentquot III mash Irhhmmtunh to Your midterm The robterhs posed by thts kmd ofwork ts ChaHengmg th short you are searching fordifferences among a large number orcandidates run mm mratatmmm There are how prtyate corhpaht es that are atterhptthg to up the sarhpte size of hese Studtes 23a dme forexampte wut run a strhttar ahatysts for 399 thetr hawmu phce tust dropped from the thtttat 1000 a yearago 39NIt thhtmwm m muf n VWh a targe hurhber of subscrtbers thts groust proposmg a new kind of h u genetic research based on surveys ra her hah the typtcat approach that tooks more We a chmcat that here s what they say about the prospects 23adee ts a new more etbteht way odomg gerrettc research that we belteve WW produce resuts where tradtttoha methods haye tated o t cu omers as cotaborators adytsers ahd parttctpahts by cohductthg research that correates thetr re oh s to We surye s wtth thetr gehettc data The tdea ts to mate possbe V hat wou be theasbe ushg cu Hquot m u mosty thyohe recruttthg pattehts through phystctahs pradtces and mm other rheahs g 5 mum m an quotwankRIMth ten mustmm nustuszu law Nmnaltlxd Expvwsunu mgh 4 475 2 45 I 415 a mum turmn w 7 maternal line 17 um ht WNWr WW Wynmsroobgmum warm Sonatva mmt M 1 I r tamttyamanus quot 395 39 1 5 IT395155 19 935 kuAumnnupmg mom Ammstm mm m an 5 rs V CE u powmmmmm tquot v mm mm m um um and uman yum m YourFmlly and marm rmu39vj 1 v A Hnwn mdn Albmnscnm nuu tr mm STMHHMD who to Flow Namwu amt communt msmhsas tuscnoarrou mmmNnbm Amman Mmogmunsmalmyssed mm Nam r WW M u Amnrumn man 397 uunyunuagnwm Stavnn wmm mmnnnu o39AiIm tommamq FlmuI Plnpll homaneusmots ur Mwmem account Human Pnhtstary vmea E J mymu 39nr 15391th m Wannnsd39mu Mumn pramnavy PMOWI L mquot mm welcome to 23andWe new approach to genetic research Z endwelsanawwayaidnln nell vas chlh m smepmenualmWDdWE valuablelnslghummequlcklyandlnexpenswelymahua lllunalmemn s wlmme lumaie nal oitvslier Malaysianan ourselves and canvlbuung m lmpmve health care n a Participating is as easy as taking an online survey DurstNaysdun39tluslaskynuquesunn mm M m lveyuuans l y Hnlsmngoneyu mlul mmuuihawiaslyuurvenexesave urwhemevynulvavepe eol n x Ham w um arem verynnel a m A wuacumpunanmalavmwenwuesaamn mum Collective action is the key to 23andWe Mobndyexuenslo uve cannerhyuarllcluaung lnacnavllywalknrmke nae Slmllarly nnslngle uersnnssurveywlll IeH useveryllvlng we nee m knuwt ailndgenes related in Favklnsonsnlseesearmlgralnas Euuusrasmausan soinamcmans nan ralse mum Didnllars we Delleve Ihe cnmhme lniurmallnnllvese surveys wllecl could lead m major lnslgmswhenwmmne wllh genellclninrmallanWummeZaanaMedatahase newan if my surveys 23andWe begins with you Learn about yourself while contributing to research Relaled Lopxm39 Ahaui Zaan we Feazuven Resaavch zsanaWe FAQ F03 ch Rosoamh Survey re You From Alxxu w smva uw mm suns Rm quotn m mm mm unrshamd prenlsiory zm sues 29m Gem 45 SNPs 53v Bases res um Ec mu 5w lek eases 37 Gene 1x 5N9 2 3 4 5 mm mm mm mm mm mm mm mm WW 3mm mmllwe v w Eases 135M Bases Mk Basus rm We mm 28 SNPs 7 31k SNP 39k SNPS l 7 l mm mesamslmmw QWW WW www mw mm WW mm am 7 r l l l t 22 X 52M Shela 46W 02 39 N B l I54 Bases nm gmm Wm WM MM mm M WM om Bases v 52 1 acres on 5w Hypothesis testing in a modern age lines of research new involve scanning data for structure 1 L I different 39om what we expect to see a null hypothesis of no difference is rejected it s typical to think ofthese problems in terms of testing rather than con dence intervals These problems are vas Iy diiferent 39om those in ei her Fisher s time the computer has made it easy to make thousands or millions of comparlsons Hypothesis testing in a modern age Hypothesis testing in a modern age Biosurveillance Microarray analysis Post 9711 tnere is an interest in identitying unusuai patterns in disease reporting a d iii i aiiaoi aii As you naye seen a DNA rnicroarray can hag We mace aiiow tor tne sirnuitaneous comparison ot tne dtterencei exp essi n ieyeistortnousands ot genes How do you identity tnat sornetning strange nas happened Many ot tne current rnetnods are based on nypotnesistests nuii nypotnesis ot no cnange in reporting in the previous tirne perio n comparison is a nypotnesis test Witn a U hypotheg g 0f quot0 were How do you identity tirnes ot eieyated repor ing Tnere can be trernendous onai ettects variation and seas iiii Eti WSW i990 ptym uit Outbreaks Wiii occur iocaiized in time and space events We an anthrax reiease may have a singie source from which the effects fan out computationai toois tnat iet you driii into tne data are extremeiy usefui iden nyrng iocai events and driiiing down to the actuai addresses of patients Hypothesis testing in a modern age MRI studies func ionai scan of their brains is taken the bran specificaiiy it measures increased biood ow to tneiocai vascuiature r idwiig pm To do tnrs ne brain is divided into a number orvoxeis and orn bioodfiowisc pared fortne nousands tensof thousands of voxeis nuii hypothesis of no difference mature Maven m Inna brie communications mmrnmn Sound of silence activates auditory cortex Haw m s nit m mi 01 human mg nmrrn Hm m a iumunml mtnmt momma quotMW in Memm md r mm u nuuui mama 1 Mpm unymmymi warm imagu Amt imJ u Th mular mt Muml pnnjwic t 1 unit at auditmvmwgu m min mni hm mmm r n mm g t mu m a may nimntduuiync nm39 itl mkl m A u by lming sanmng Vuhyux pamnnh39 ht m with in spit km anti Jr in Luing Hypothesis testing in a modern age MRI studies Patients are subiected to different stimuii wnne a functionai scan of their brains is taken Tne scan the brain n ifi aii the iocai vascuiature wnere neurai activity is taking piace To do tnrs orvoxeis and thousands of vofeis nuii hypothesis of no difference an hereutranvd dificimipmnhdur mg urt munduutL ma uplan with aim at mum mummicl mt mm mm mm naiirictidunngl m ihvr mm if rmnodc m wuppicmm m rr untrmwngs rs s u g mmngmuripmrm 4 mm mm mu n r prim vltuu l mm mun mt c dundvd mm m m mumn m m mniim mt mm m rmhcr iun war the mm m mg di urcm mun m n wninhngapwapnn um sung purdmvd gicam mmmm no a than an Luminr gt01gi u 14 Minn imtru lumuun in iamiiurn im mppimmntaryunoumuun n w 015 I lhhh nr hr 3 nmngn m g 01 4 Lumim nwludvinuncs E MS Imul V39lmmi prcvmm nwmugnumhh g cxr u dircucd sulxicdx l n imaging a FL H UL I ttiutaum wm L mu n mi ml aimply Jim 1 guys ntfamilur mum mxsul lnggvl mdllmv quotan a rmde um 1 pnn 1ch mmng 1 1 l0 muttm tlw obligatu quotmum um phr nnm m Lnrrulmmung ngtubgtuivmlmh lll numb nymlcd whitm 1mm w mm W 5 mi imx ul mt uimmilnl wup during um W m v I m mum 0139 v39 dctulimnud mt llliguhlic mum n ma imagined mpuiuuic A licn wm klmuludgc Hl a iymi nuld um mu m gunuulc the musing mlnrmnhnn rualn unlmnn rrminmcd m midilm nsm annn Mu 39 n m rueIng lama mute m mnulsi mm mmdcd m tsunami quot 39 mw nut llv i H m n lhquunanaudilunmkxlFlgthJl WHWMHM song mime quotHI mm 39mvlm am quotI39ll WM Mar mhllmr val139 w u r 7 Aimnmxi m mm mm m gm m um mung Ii m 3 quotnui Inn m F n 1 Int 39 ml lmmm u 1 mm mm tum m quotimmian lmmnh lih nlrm rd xmm r 5 H and immm at m m mmpmum u t Incl an um m a Iiiilll gtlrdn39rll raglanomnch Mal immmg un um mn mun Mun am mm mu mumd on quottum M mu m mum cuntL Mum bmm lmJging w um um w tn gap In rauan mugs wm command mm gaps in unknovm sang Jddlllunal mniiy w 5 uhscncd minintu m Jamilmmml pmmumi mm IEA a and m m sunnlxmunlar mnlurnrrm Supplemental infurmatiun tn Nature article March in mus Hypothesis testing in a modern age Data mining Huge 39 39 data now 39 ra her 39 39 39 about you and your activi ies much ofthis came up in our rst assignment Data mining is a term used to describe a set of algorithms that identify interesting structures in ata39 is is not morally different than searching for outbreaks or signi cant genes br active regions in the brain The problems tend to involve marketing 39aud detec ion and various prediction schemes Will you default on your mortgage Not pay back a credit card lBBE Cllrk mun Dlvld max rum preyu a Indral smyn swims may hm little to offer me Withinquot In a dam minquot t umh and in autumn m mun Statistical Inference and Data Mlnlng eruimxnminn nmmummAIIM f mm rm m a diubmm m mxmn m mm mum cums mots mmunumun mm unreturde mm m clurdrd Im a d u I am the 1 m w mm manhunstalx m mhruunlcr I n appllr um uni th lamb it I mam mnml 1mm rim mm m x in unm mum mmcmrnn39sampll39xnhnmnm dam m pm Inquot mmmwug rh 39rull wmnimmImmmblr inlrnnre pmhlrmx mzmmwi urx wmrm ll nude and nxhrnlmrxplnn rl arm Iqu quotMilt1 l mum mum Min mum mm minim Ingmnli mnnimwmmmmmugmim an 39wan llulupln Iulapplirlmzwnwnltnrulile u lmlmhinultrrmmulrnumllgvmrr nrluirnumthilt grml n mm and ln mammal um u m allti m mimum urlu pullmcx m H mu mung u mum at a d applnprldu mm um null Dealing with more than one variable yur1191uvv V more than one variable in fact it39s fair to say that aside from our simple twoby two tables associated with randomized controlled trials our data sets have been fairly complicated Our analv i hnwe Fr 39 1 1 1 r plays of a v 1 1 mosaic plots of pairs of variables we39ve plotted one smoothed histogram over another to compare the distribution of two variables Mu a a r r our rst new tool will build on the notion of distancesquot Let39s start with a data set that we can visualize easily Physical distances Boston New York DC Miami Chicago Denver Seattle San Francisco and L a V 0 an objec of Have a look at the entries and see that they make sense nos CHI DC DEN Multidimensional scaling Suppose we are given a set of objects say the 9 cities on the previous slide the starting point for this geometrical picturequot is a set of approximate 39 airs of 139 quot39 39 quot quot39 dissimilarities distances or proximities to refer to this kind of data 1 r he al orithm r nge 39 39 39 graph quotmatchquot your similarities that is two similar objects are represented 39 39 oge her 39 39 39 39 r quot r points that are far apart Let39s see what happens with our city data LA MIA my SEA 99 909 0 953 A29 19A9 2979 150A 205 2975 3095 CHI 953 0 571 995 205A 1329 902 2013 21A2 DC A29 571 0 1515 2531 1075 233 ZEEA 2799 DEN 19A9 995 1515 0 1059 2037 1771 1307 1235 LA 2979 205A 2531 1059 0 2597 2795 1131 379 MIA 150A 1329 1075 2037 2597 1309 3273 3053 my 205 902 233 1771 2795 1309 0 2915 293A SEA 2975 2013 ZEEA 1307 1131 3273 2915 0 909 99 3095 21A2 2799 1235 379 3053 293A 909 0 g Mill 0 7 LA 8 lt0 5 SF 0 e DEN DC CHl NV 8 EiOS 0 SEA l l l l l l 1000 500 o 500 1000 1500 Multidimensional scaling Tern r mm V remember hat MDS is concerned with he distances between points so we can flip the picture heck we can spit or rotate it any way we like and the distances between he points in the picture will be the same Of course this seems like a lot ofwork to recreate a map of he continental US the real power ofthis technique is when we consider data types hat don39t already exist on a map Returning to Fisher for a moment we can hink of each quant39tative variable in a a a s as a kin of axis in an n mensiona quot spac instead of latitude and longitude we would have height and weight Multidimensional scaling Let39s consider the complete set of data for your midterm in all there are 7129 genes on the GeneChip used by Golub and his coau hors for your midterm you are looking at a sample offour Suppose our data are in a matrix genes where genesik is he expression level of gene kfor tissue sample i we can then create a simple measure of distance between two tissue samples distance between tissue and J 7129 2 genesh k e genesLhk 2 k1 This calcula ion treats the expression of each gene as a new axisquot this is p 39 r 39 39 yuu sun I 39 39 just two genes then three and so on SEA 0 7 8 ElOS NV CH DC 0 DEN 3 SF 5 8 7 LA 8 9 7 MA l l l l l l 1500 1000 500 O 500 1000 cooid 1 Multidimensional scaling n L a r r fed them plotted r 45L 439 original map m ore or less With the gene expression data we start with points that do not live on a h 7 4 r r w r a distance between these points by summing up he squared differences and taking a square root 139 quot39 ama so that L r points on he map recreate as best as the can he original distances between the 7129long vectors And here39s what we get Multidimensional scaling Once we have a plot ofthis kind we have a View of the relationships 39 hslwssn lhs 39 n I I clustering places where points group together or other structures to give us a hint about how these data are distributed in ndimensionalquot space Here it doesn39t look like there is much going on well until you replace the points with the tissue type labels coom 1 8 a a o g o o o o g s o a 3 e o 3 8 a a a 5 e a a o 0 0 0 g a o a o o a a u a 5 q a o a a 539 o 8 one o a o o a o a a o o o o a a 8 a n a gt a d15tltgEnes 8 gt ids chidscalEd 8 a gt pluthuds 7 l l l 50000 0 50000 comm o O 8 l xML o g 7 all mm mi 8 g 7 NHL N M u A g g l L Main M 8 N mth llML ML mm mm wiur o e l k in 0 Li 8 n L L 8 m l 8 L 8 7 l l l 50000 0 50000 Multidimensional scaling r a y o wi h so many variables he cases AML and ALL cluster this is essentially the nding of Golub et al the idea that groups of genes could be used to classify e two kinds of cancers word of caution the distance measure we ve used for this problem is fairly simplistic if you are excited about his kind of idea especially as it relates to gene expression data there are other distance measures that are more appropriate The goal ofthis lit le tour is to simply show you hat there are various kinds of techni ue nu 39 a 39 39 u a big daia si 39 you to see structure a little more clearly MDS is just one ofa large number of data analytic tools that operate on a distance matrix of some kind next lecture we39ll examine socalled heirarchical uu leing mm 39 39 r r like 39 me bottom up um Inlmmk yumllll mu IhmIlyhlm n w minmm ww lt l l 39 Illll I all z u I l n hmpm Mu ml I 91 lamu mumm I yummy 7 t um i mum Prelude to modeling Let s be a little precise about the term classi cation Golub and his coau hors would like to find one or more genes that have different expression levels for he two kinds of issues so that they can be used to predict whether a new issue sample isALL or AML As a statis ical modeling exercise we consider he different genes or rather heir expression levels to be potential predictors and the true tissue type the response sometimes especially in the context of an experiment we refer to he former as independent variables and the latter as he dependent variable While his could be a useful exercise beyond cementing vocabulary let s have a look at another situa ion one which might be a little easier to reason with Decision Tree The ObamaClinton Divide n ma nummallng munly mmsis smns lunar cmmlllss ilnmlmamd by irts puunwmmrm lnr l mnre lemma win Awlllhvhlgh dwnl EllenIllquot m Mgr1 rm u plmmn m l I nigh uhuul Ellnlan wlnl 9mm n mm In mm mm 1 yumquot m r us n e m mum 39 E Anuw I Mm Mme 39 5 as m ml were many cum win mum Dunn In mullI 1 m In 79 5 quot mm aw pupal ml 52 on 25 quotWWquot u 71mm mm m Km mm w mw an u narzrlaue mm V a Decision trees The figure on the previous slide was made by Amanda Cox a statistician working in he graphics department at the New York Times and yes the analysis was done in R The itle Decision Tree is a pun as he data refer to countywide decisions for Obama or Clinton in the Democratic Primary the object is also a kin f statistical model known as a decision tree Coded in this elegant structure is a pretty serious computational algorithm a relatively recent addi ion to the statistician s toolbox The Democratic Primary m 510 Think backto Apni ot iast yean wnen it seemed Hiiiaiy Ciinton s momentum 6 3 was picking up again tne decision tree was buiit tiom a data set concerning 422 all the counties in states where primaries had already been held 5l3 56 NW N 5 demographic measures age and etnnic makeup education ievei ieiigious breakdown poiiticai measures didtne county go to Busn or ilteny in 04 and 39 H1 5 t ot county and so on Whiie it feeis very much We a data Set We ve Seen before what s different mnnet ennnty nene 155 Oh de 133 tonne antennente 157 3b 3 Blplne 19 E11 5 w 153 mm mm 131 ehntnn ann 3etnen31ne 159 chnt3 e 192 1 nt San D an 150 chnt3n Calaveras 193 hams San Fran 151 mm n 53 19A e11 ten ann Jeenntn 152 11nt3n Contra C3ste 195 obs a San Luis obxspu The Democratic primary 153 chnt3n e1 Harte 195 Elm 0 S at D 1 mm 1 D an 137 tonne enntn 3ennnte 155 1 m F 133 ehntn en tn 1 n H H 155 mm 61 133 n Santa cm Now giyen aii nese poten iai piedictoiyaiiabieswhich ones explain 157 chm mm zoo ehntnn sneetn countyievei voting patterns Suppose tor exampie We start Simpiy and 153 mm Imperial 2o 3n n at consideiwnetneioi not a maionty ot ne county yotedtoiBusn in 04 159 mm 2o 3n n ateittynn 1 mm B 203 ehntnn 5 non m 171 c1111 n Kings 20A abstie Somme cllnton 171 1039 172 llntun Lake 205 Clixmun Stanislaus obama 302 72E 1 chm Lam 205 ehntnn an t 7 mm L 99195 1 2 21 Therefore Obama Won about 64 ot tnose counties not yotingtoi Busn in 04 j if 209 ch n Mm wniie Ciinton Won about 59 ottnose counties tnat didyote tor Busn in 04 177 ehnte n ttneen 13 1mm Tumme 1 E Dbama MexAdmer intun ventura 179 mm m 212 tonne Me Now consideitne simpie prediction rule it a county yoted tor Busn in 04 13 nhnt n 1113333 213 31mm vube We H say tnat ney WiHvOte tor Ciinton in tne primary Wniieitacounty was 13 n 3 Mann mostiy in tayoi ot ilteny in 04 We H assign tne Win to Obama 132 ehntnn iiinntetey 13a ehntnn nan 13 tonne evenin 13s ehntnn en 1E5 ehntnn meet 137 tonne 31 es 133 ehntnn mvetetee The Democratic Primary Of course this rule isn t perfect by applying it we would make 171 728 899 mistakes out of 2240 counties or about 40 error we refer to these mistakes as having been misciassi ed by our simple ru e So the question becomes can we do any better There might be better of A indi s them consider he top of he tree the rootquot The Democratic Primary ueci iun nee 4 ly split ing 39 r he root or rst splitquot39 39 a 39 39 39 39 the lowest misclas cation error To investigate this a little fur her let39s consider he predictor hat represents the percen age of a county hat is African American we now choose a breakpoint a h quot39 quot h 39 39 h 39 39 percemage r 1 ofAfrican Americans and those with a smaller percentage We then form a table as we did for coun ies hat went for Bush or Kerry in 2004 and count the misclassi cation rate The Democratic Primary W ao1 A r Pct Af Am gt 01 FALSE TRUE cllnton 959 2 2 1 Obama 4 76 fthe 697 counties that are more than 10 African American 221 went for Clinton and 476 wen or Obama 68 in favor of Obama since Obama won more counties in this group we label he node Obama and we would be making 221 errors In the remaining 1543 counties 898 went for Clinton and 554 for Obama 64 in favor of Clinton hen we will label his group Clinton and we woul make 554 errors The Democratic Primary 139 quot quot h r 39 g magic point Mr hat minimizes the misclassi cation errors In fact Once this node has been chosen we work our way down the tree conduc ing t e same search but on the speci ed subsets ofthe data at each s ep attempting to minimize our errors M267 March 2003 Lecture 1 Eddy De Robertis Page 1 GENETIC CONTROL OF EMBRYONIC DEVELOPMENT ESTABLISHING CELL ASYMMETRIES Lecture 1 1 Characteristics of Cell Differentiation The differentiated state is stable eg neurons vs the lac operon in bacteria Pattern formation the difference between an arm and a leg is not in the ingredients it s in how they are mixed The same genes that control development control body form and evolution Cells are the true miracle of evolution Once the basic building block the eukaryotic cell became available the form of metazoans evolved by changing the arrangement of cells with respect to each other Differences between cells first arise as a result of two broad mechanisms 1 cytoplasmic determinants which are molecules asymmetrically localized in the cytoplasm of the egg or of somatic cells see below which become unequally distributed among cells after cell division and then affect the activity of genes 2 cellcell interactions in which cells induce new fates on their neighbors Both mechanisms are used over and over in the course of development 2 Early development is so rapid that many important molecules must be made during oogenesis and stored in the egg Eggs stockpile materials required for early development In Xenopus the rapid rate of division during early development allows little time for new synthesis A female lays 1500 eggs and each is 12 mm in diameter The first 12 cleavages take place synchronously every 30 minutes compared to 12 days for eukaryotic cells and 20 min for Ecoli This rapid pace of division is achieved through an increased number of replication origins and eliminating the G1 and G2 phases of the cell cycle At the 4000 cell stage or mid blastula transition lVlBT the cells start to divide slower and asynchronously and start synthesizing RNA There is no RNA synthesis until MBT but the initial differences between cells are already laid down by this stage These decisions are made using maternal molecules determinants already present in the egg The marginal zone equatorial region gives rise to mesoderm The mesoderm involutes through the circular blastopore starting on the dorsal side The blastopore gives rise to the anus in deuterostomes and to the mouth in protostomes The involuted mesoderm and possibly endoderm too induces the ectoderm to form the central nervous system in the overlying ectoderm By the end of these morphogenetic movements the body plan is outlined and the places of future organs determined e g muscle kidney The neural plate forms into a tube Neurulation Remarkably the general outlines of a molecular a FERTILIZED Ecr b TwoCELL STAGE C MIDBLASTULA d EARLY mumquot e LATE CAiTRUl A 90 MINUTES moon CELLS 7 HUURs 25000 CELLS a mm H1 HOURS ANIMAL POLE BLASTQEOLL CAvmr NEURAL PLATE EUDDERM MBODERM vEcmL POLE amsmwu summan LAT NEURULA h 511000 CELLS f 2 quotcum 139 Nncnwc TADPOLE j Arqu Won mu ugunum um NEUKULA aosmc 500000 cms 3 Luv 12 MCNTHS NEURAL mum PLAYE 5 VLATE cw CAVITV nu induces pan at nm unndnnn m m became mum The mesoderm ls Ihv d2 induced formation and movemenlnfcojl lawmas secnln nu cynical layer ma chnnlncs me Embryn39s anlcrupuslcrkn hemmnlha nLImll plulelLAs scan m dmlopmem aIXenamrIacvsa SuulhAfrkan Imu rhmugh I n qu39IEY I39M one 39 ball L39 L a 39 5 on itself and becomes a neural lube r m I p y 39 memnsu mespinalmmlmhc mmmanimalm Scienti c American July 1990 De Robertis Oliver and Wright 931ng 01 muu1f 9q 9112 uouemnse 01 uouezqum u101 1u9u1d019A9p esmp smeln zu 12q1 Aemqled sweqoa ea Appa 002 mm 19m z 95 1 amwaq M267 March 2003 Lecture 1 Eddy De Robertis Page 3 3 Gem cell determinants In 1875 Hertvvig found that sperm contributed a nucleus In 1833 Van Beneden and Boven argued that each parent contribrutes a set of chromosomes Ascans was a very favorable material E Fig 23 7 Fertilization of the egg of Ascaris megalocephala var bivalens illustrating Van Beneden s demonstration that each parent contributes an equal number of chromosomes A the sperm has entered the egg and its nucleus has swollen d The female nucleus is completing the second meiotic division and is eliminating the second polar body pb Each nucleus contains two chromosomes the diploid number for this variety of Ascaris is four B both pronuclei 9 6 have swollen the centrosphere 3 contains the dividing centrioles C start I 39 D two L clearly visible in each nucleus E first division the nuclear membranes dissolve and the chromosomes align in a common metaphase plate F first cleavage anaphase only three chromosomes shown on this section In Ascaris as in frogs the pronuclei do not fuse before first cleavage In other species such as sea urchins the membranes of the 3 and 9 pronuclei fuse before the first cleavage Drawing by T Boveri 1888 M267 March 2003 Lecture 1 Eddy De Robem39s Page 4 Boveri showed that Ascaris which has only two chromosomes has a process of chromosome fragmentation called chromatin diminution in all somatic cells except for the germ cell precursors However if the eggs are centrifuged granules of the germ plasm are redistributed and cells do not fragment the DNA This suggested that the type of cytoplasm inherited by cells is important Weissman proposed an in uential at that time theory of heredity stressing that the germ cells and the soma body are separate lineages But his theory of development was incorrect 4 4 Fig23 13 C 39 quot 39 39 39 l39 39 quot iquot 39 39 L 39 p5 primordial somatic cell yet to undergo diminution 5m somatic cell which has undergone diminution arrow germline stem pg cell a Second cleavage in progress in the primordial somatic cell chromosome diminution is in progress h Later stage Eliminationchromatin at equator at upper spindle c 4ceil stage showing olrminated chromatin in the cytoplasm of the upper two cells 0 Third cleavage in progress second diminution at p5 tei iOcell eml 0 showing mitosis of somatic cells with diminished nuclei each containing any small chromosome I IZ cell embryo g About 32 cells fourth diminution in progress leaving primordial germ Cell pg in prophase h Gastrula templeth with two primordial germ cells From the sludies niT Boveri i899 M267 March 2003 Lecture 1 Eddy De Robem39s Page 5 Boveri reasoned that there had to be a cytoplasmic substance being segregated into the germ cell that protects it from chromatin diminution In 1910 he centrifuged Ascaris and showed that multiple germ cells were formed x r 2 399 95 5 sq Sam 9 I 59ml 3quot 0 o 2 rr A m 12 3 Am do quot Gun 1 cum ll 0K7 Ccll fig 23 m Wuivmnnlheorv o1 hereillly rn each generahun rm germ ccllgweukennlhr nne hand to me body or sum and on he other 10 man germ Lull whlrh Carry rm Imu of hemrluy From EB rlun I697 4 Cytoplasmic cnnlml of hromomrvw limp Ascarls normal eggs I B39 cenlrifrlged Fig 23 1 nuliun in egg The shaded arru urdicalos rho rlrsrriburiun ol hy purlu 39 Iu39al Lvtuplmnur maronal In 1 1 n e fhl nmrerial the two large hromummes remall m a I blusromvrcs lacking ll llurv Chromosomes undergo dumnmiun m1 up rell l liq m res runranung l in M267 March 2003 Lecture 1 Eddy De Robertis Page 6 4 Nuclear transplantation and therapeutic cloning Weissman s idea that DNA might be lost during differentiation of the soma was disposed of by ligature experiments and by nuclear transplantation inXenopus which showed that somatic cells can be totipotent Fig 23 9 Spemann s demonstration that embryonic nu clei are totipotent until the l6cell stage Original explant Foot web explant removed Adult frog of l rm strain Outgrowth of as nuclear donor epidermal cells UV Parent of lst transfer Donor cells for Enucleation of nuclear transfer recipient eggs E recipient eggs l Wander transfer Cells trypsinized Foot web outgrowth and washed prove frog was Znu O UncleavedP V H l gimpletely cleaved 70 7 artia y c eave So o 25 Dissociated cells for serial transfer UV39 Parent of serial transfer recipient eggs Enucleation of recipient eggs k gal nuclear transfer Uncleaved 39 Completely cleaved 40 Partially cleaved 30 30 1 Foot web outgrowth prove frog was 24m Nuclear transplant tadpole Inu diploid from nucleolus and chromosome counts present in 36 of serial clones Fig 23 10 Serial nuclear transplantation in Xenopus Skin cells from the foot web of one nucleolu5 frogs the number of nucleoli provides a genetic marker were cultured and shown by immunofluorescence to produce keratin which is a characteristic of differentiated skin cells Nuclei from these cells were transplanted into eggs whose own nucleus had been killed with ultraviolet light Nuclei from partial blastulae tsee text for details were then transplanted into new recipient eggs and swimming tadpoles were obtained Thus adult skin cells contain all the genes necessary to build a tadpole or the stage indicated Courtesy of LB Curdon M267 March 2003 Lecture 1 Eddy De Robem39s Page 7 You must have heard about Dolly the sheep Who passed away in 2003 Alan Colman s company has now used cloning to inactivate the 113 galactosyl transferase gene in pigs for xenotransplantation They circumvented the need for pig ES cells Stock values went up 44 in one day Jan 2002 V i w yum ngre 1 Sltps involvrd in uhmpmm rim1 a An exylnnl in lake rm m nduk lune donor h Tm mm mmpl I5 pmwn up In nlmrc dl proliferallon c Nude rm he oulgruwlh tells Ire KmuApIMud Inlo elmthaan Km 25 from which he and mumI has been moms uxnm nudturlrnnsplnnl aquot quottauum m m I u n r Tinted re on ormmmi 3 Selected can bcg n m m I a dam spedquot71M u rypea i 11 repluazrnmt Kiss is Imlnplnnlud imo lhe onng donor Frum Gnrdun and Colman Nillllrl 402 743446 1999 If all nuclei had identical information then the initial differences must reside in the cytoplasm of the egg M267 March 2003 Lecture 1 Eddy De Robenis Page 8 4 A cytoplasmic determinant in Ascidians mosaic eggs Some eggs have what is called mosaic development When individual blastomeres are separated from each other they will adopt fixed fates such as muscle In other words mosaic eggs develop cell autonomously with little in uence from signls from their neighbors Some ascidian eggs have regions of different pigmentation and it is possible to visualize that after fertilization these regions undergo extensive movements and eventually become included in the cells that give rise to certain tissues For example in his classic 1905 paper Edwin Conklin showed that the ascidian Slyela has a region of yellow cytoplasm rich in mitochondria which eventually gives rise to mesoderm and muscle If the embryos are compressed so that the yellow cytoplasm is distributed into more cells than usual the cells that acquire it will give rise to muscle cells suggesting that this yellow cytoplasm contains determinants for muscle tissue lIx I mud1mmLizxuli ulhm mm m MW mm m mm rum quotMauird mum w n w Hmuvivmumnln r r r mm r u q u r in mi lumr mmukllm r um um mmm m an r H mm H y ML um mm a Mm ulr I Wm m mm x r in m Mmmu Wumm W x w m Prumrm 111 The cosegregation of an hypothetical muscle cell determinant with the yellow cytoplasm of Sgela was revealed by another classic experiment by Whittaker in 1979 He used an experimental trick The enzyme acetylcholinesterase is a good marker of muscle di erentiation but does not normally appear until the embryo is 9 hours old and has several hundred cells When developing embryos are placed in sea water containing cytochalasin B an inhibitor of actin microfrlaments the cells no longer ivide The nuclei however continue to multiply and the acetylcholinesterase activity appears at the normal time if these cleavagearrested embryos are incubated for 9 hours M267 March 2003 Lecture 1 Eddy De Robertis Page 9 He found that the potential to produce acetylcholineterase which was present in the unfertilized egg became progressively segregated into subsets of cells during cleavage Fig 23 12 Acetylcholinesterase AchEJ devel opment in cleavagearrested embryos To stop Cy tokinesis embryos were placed in cytochalasin B at following stages A lcell B 2cell C 4cell D 8ccll F 16cell F 32cell C Gitcell The embryos were kept for 15 to 16 hours after ter tilization at quot3 C before being stained histo chcmically for the muscle enzyme AchE H AchE in a 9hour nonarrested control embryo The enzyme stains the region of tho tail muscles of the future swimming larval In 2001 Nishida and Sawada Nature 409 724729 isolated a muscle cell determinant from the ascidian egg They isolated an RNA enriched in the vegetal half of the fertilized egg Although they work in Tokyo they called the new gene macho no reason given as to why The gene encodes a transcription factor with five C2H2 zinc fingers The localization of macho mRNA follows the movements through the emblryo of the hypothetical muscle determinant proposed by Conklin and Whittaker 1 c d 2 First phase of Second phase at Unlerttlized egg segregation segregation Xenoptts vnopus qiobin qtntim UTR Wid type Xenonnx igneous r m co tlobr 5 um 3 UTR Ml ant AAM Figuret Structure and lowlization ot machot transcript b Structure oi machot protein and constructs lot in vrtro mRNA synthesis Machoi has itve zinc lingers tn the central nan 39 cit ORE the wild type 39 ling domain p e 39 39 39 and enmzttterthehrst d and second tel phase at ooplasmit segregation Animal pole ts top and posterior is to the right llt Distribution ot maternal machol mRNA shown by in situ hybridization in eggs 39 39 at stages corresponding to ce A anterior P posterior im Localization at machot From NiShlda and salvada rnRNA during embryogenesrs Anterior ts to the left i Eightcell stage lateral view in situ Emil VOL 4quotquot l quot9quot quot quotl l quot quotquot 39 39 39 39 39 39 39 39 thiaclnmame 5 I mquot stage a and 110ccquot stage lit embryos shown in vegelal View um M267 March 2003 Lecture 1 Eddy De Robertis Page 10 Depletion of machoI mRNA with antisense oligonucleotides leads to the lack of expression of muscle actin in primary muscle cells Injection of myogenic cytoplasm of macho I depleted eggs lost its ability to promote muscle formation Injection of macho I mRNA led to rrescue of the loss of function phenotypes and to ectopic muscle expression arm rr urerexuressrbn Causes ecluprc musrlc rurmabbrr n r Embryos m We TWOrceH stage mad and m rn srru hybrrdrzed wrm the muscru abrm HMM probe n Ummecterr Comm Actrn expressron rs ubserueu m ten precursor rer b nu nthnrd maohr mRNA was subsequently rmected Adm rs ecmp Dally expressed rrr enduderm Figure 2 Depreubn br maremar mRNA by anlrsense urrgbnubreubdb a c Anusbnse segregacrbrr h and erthell embryos c prbbea rur mabrrbr mRNA a e Sense r g Antrsense brrgbnucreubue complementary m machor was rmemed men me Fmquot ishida and S wnda V n nnl 39 v machurl mRNA Scale Dem 100 um Conclusion almost 100 years after Conklin we how have a molecule required and suf cient to act as a muscle cytoplasmic determinant 5 Somatic cells have cytoplasmic determinants We are now learning that cells of adult tissues are very much like embryos They have stem cells eg intestinal villi skin blood that divide asymmetrically one of the daughters remains as a stem ce 1 and the other one marches through a program of cell differentiation that ends in apoptosis without leaving progeny Mouse bone marrow stem cells can contribute to skeletal muscle Adult Neural stem cells propagated from CNS in the presence of high FGF can give rise to hematopoietic cells The principles we learn in embryos apply to the maintenance of adult tissues Experiments in the 1950 s had indicated that the type of cytoplasm of neuroblasts in insects grasshoppers determines cell fate for one could rotate the chromosomes 180 with a needle without affecting the outcome cell differentiation M267 March 2003 Eddy De Robem39s 235351 quotrzmimtquot 39 A 5 w an thlisl cell ln ll itliftlxt39mm n m d If Fl a naugnnn Daughmr ganglmn cell neurohlul cell e pc 39 ed ncuroblhn normal dcclopmcnl with he daughlc39 chromoxomcs only 2 pairs are illuslrulcd which are Untitled to enter he Flilllre nclu39ubhisl c nnmhlc Milges of mitosis ofneuroblual cells in n nnngingdxup cullllre A nctdlc n pm oulsitla arm uividin ruutlimi M the mcumhasc hpll l gt mu chrammomes cmcr m daughter ganglion chL The divergent differenlia M is therefore delermined no chromosomes or spindle but by the asymmetricle an rangcmem ufsume cytoplasmic components ofll39u pur cm cell After Carlson J G I952 Chronmxtmm llJvrIJ S I NZ39IOJ Lecture 1 Page 11 lzlsmic determination ofgttngliunccll ion in grunhuppcr ncul39obllmls in normal dcvclopmcnl ol the glunhnpper Churnquotme m l V or t e t rccllbcmnmn divides no timer and di dcx l urlhcr A 0 F in olnrG lo I show gomv mulnluincd l d t m c 5 is g ncuruhhm st us to cause 1 mo 39 39 ul A rcwll llu col Iitm of c t duughlcl cells y properlics of the In Drosophila RNAbinding proteins such as Staufen have been shown to migrate to the daughter cell at metaphase homeobox gene that determines neuron ganglion cell fate Staufen binds prospero mRNA a In mammals homologues of Staufen and of the other proteins have been found as we Mllar ua Stanlen 4 p105 mRNA 39 3quot GMC Prospero a l Anapnnse a lnlelphase Melaphase Illscmeahla Mllanda coordlnates the asymmetrll prospero mR he complex muves mm the aplCal bottam to lhe basal top Si 9 ol the cell and is segregated ln lhe datlghtel cell GMC at wloklnesls Conclusion the asymmetric distribution of RNA is important in later development in addition to eggs The cleavage plane can determine cell fate 4 D enmmmnlm Avl39ul FthllE t masonquotla llevmhlasls lllmle ln an ans ommamula n mt o n untilunoblasl mvlsio M267 March 2003 Lecture 1 Eddy De Robem39s Page 12 6 Cytoplasmic determinants in the Xenopus egg Vgl was isolated by D Melton in 1987 as an RNA localized in the vegetal pole of the egg In the oocyte it is tightly localized to the vegetal cortex in the egg and blastula it is distributed more uniformly in vegetal cells Isolation of CDNA clones 390 localized Egg RNAS Aman 9 A G g r Vgrl in oocyte lsome poly A FINA Vg l m blastula 39al cm libnly Vgl encodes a growth factor of the TGFB superfamily Mcroinjection of synthetic mRNA made for example by using SP6 polymerase encoding a processable form of Vgl causes endoderm and mesoderm differentiation Other TGFBS of the Nodal and Activin subfamilies have similar activities Because Vgl is a maternal mRNA asymmetrically localized in the egg with biological activity it is a very good candidate for an egg cytoplasmic determinant VegT was isolated more recently as a cytoskeletonassociated egg mRNA and found to have a distribution almost identical to Vgl in the oocyte and early cleavage VegT encodes a Tbox family transcription factor the T stands for the shorttail mouse mutant Brachyury aka T VegT has the same early distribution as Vgl VegT is a cytoplasmic determinant required for endoderm and mesoderm formation The third important Xenopus egg determinant is B calenin which is stabilized in the dorsal side of the embryo VegT is a transcription factor expressed in the vegetal part of the embryo Its activity was depleted by DNA oligonucleotide depletion Oocytes are removed surgically from the abdomen of a frog and injected with specific DNA oligonucleotides for the targeted maternal mRNAs Endogenous RNAse H digests the RNADNA hybrids and the mRNA is degraded The injected oocytes are placed in solutions containing vital dyes such as Nile Blue Sulphate and phenol red matured with progesterone for 12 hours this induces meiotic division I and transferred to the peritoneum ofa host female that is laying eggs The egg acquires a cover ofjelly and becomes fertilizable by sperm In the case of VegT this has been very successful When VegT is depleted in this way and embryos dissected M267 March 2003 Lecture 1 Eddy De Robertis Page 13 into animal equatorial middle and vegetal base thirds markers of epidermis and mesoderm are now found in the base fragment that normally only contains endodermal markers This indicates that the VegT transcription factor is required for endoderm formation Subsequent experiments using higher doses of oligonucleotides showed that neither mesoderm nor endoderm are formed VegT depleted base vegetal fragments are unable to induce animal caps to form mesoderm in Nieuwkoop recombinants Thus VegT is required for the transcription of zygotic signals that induce mesoderm F I Uninj ang sng A Eq Bs Eq 35 Eq 55 Epidermal keratin nrp1 NCAM chordln 39wi Xbra Figure 6 In egT Depleted Embryos Meso xmyon derm Forms In the Vegetal Mass and Not In the Equatorial Zone in u w a n EFLaIDha A shows the dissection carried out on com F Oocytes were uninjected or injected with trol uninjected and VegT depleted mldblas 3 or 5 ng oligo and equatorial zones Eq tulae to explanl the three regions the animal and vegetal masses 83 were dissected at caps equatorial zones and vegetal masses the midblastula stage The explants were cul tured until the early tailbud stage and then analyzed by RTPCR for ectodemral and me sodermal markers UNTRFATED VegT DEPLFI39E D Animal Animal Eranquotmg f5 Epidermis CNS Eyer Fpidtermis 139 Mandelm V 1 od Endudcrm budWm CNS eg al Vegclal Model From Zhang et al Cell 94 515524 2000 Other methods to achieve loss of function in embryos in which targeted gene knockouts are not available will be mentioned Phosphorothioate oligos are more stable Morpholino antisense oligonucleotides instead of the normal phosphodiester bonds are very stable and provide a way of blocking translation VOYAGER FUND GLOSSARY Fiie Fund Yaais Heip Save 5 ngvixi Hin LedsgjandFundS mum HAEWSSPNUS emu WW SDWZ VVdLE EWW Aiiacated V FundTyue EantvactLEvant V a i a enaosiel Dcauzuua Expend Dn y Nalesl Plavgvhes 1 a muomaiAuacahan saw 55 Total 38355 97 Total 25236 59 incleates sznunn nu Pending 33 85 Pending man as Decleates s szal an i 7 New Fund E 37 AM Hi 2EIE5 uner 2 terms mnz at muster t 4 LBS Mi VZEIEIS 8 37 AM use Muihen l meay austere graces Emma g g y gauge 5 mm Original Allocation Initial allocation that ms input at the beginning ofthe Fiscal Year or at the time the fund was rst created Increases amp Decreases 39 39 39 cumulative total amount oftransfers made by LBS To determine ividual transfer amounts that make up this total look under the Transactions 1 Net Original allocation plus increases minus decreases Hence this is the total allocation forthis fun Commitments Total Indicates total amount ofmoney committed on this fund This gure c u c h this particular fund Commitments Pending Indicates total amount ofmoney committed on this fund This gure com s from the rm orders input but notyet approved by acquisitions stajf onto this particular fund than acquisitions staff 113007 Week 9 Notes Section 1C Classes Structs and classes textbook reference Ch 6 Ch 7 up to half of page 291 most of Ch 103 1 Classes Introduction So far we have seen different types of data such as those oftype int doubl e char etc What if we want a type that contains other types because we want to associate multiple types of data with a certain type Perhaps we want a Date type that has a month day and year We can do this with an array of ints However keeping track of Date 0 Datel and Date2 can be cumbersome This is where classesstructs come to the rescue We39ll use the term 39class39 in this discussion We can create types called classes that keep track of these data elds in a more structured and exible way and classes let us de ne our custom types These classes can also have functions as well and we can add functions that give us the date in various forms Without further ado let39s start with a simple Candy class and add functionality as we go along 11 De nitions Member Variables Member Functions A class is a type that group together related variables and functions known as member variables and member functions Let s create a 39Candy39 class containing the variables 39category string type and 39age int type and the function growMoreStale of void type QSTN For this problem use either the struct or the class keyword One difference by default the member variables and functions of a struct are public while those for a class are private More on public and private later Having the above code let39s say we have a Candy variable called 39trident Next we want to set the member variable 39category to be the string quotgumquot and call the growMoreStale function QSTN Complete int main Candy trident fill in below As we can see the member variables and functions are accessed using the dot operator It39s also perfect legal to set a member variable to equal the value of another member variable For example if we had another candy called Orbit we can set its category to gum Candy orbit orbit category trident category So far so good But wait something s still missing right Running the above code doesn t compile because we haven39t de ned what trident growMoreStaleO accomplishes 12 De nition Member functions QSTN So let39s de ne that now Fill in the function de nition ll in here cout ltlt quotyo I grew more stalelnquot age In general member functions are of the format ltclass namegtltfunction name with parameter listgt QSTN What does the main function print out Class Candy public string category int age void growMoreStale l void CandygrowMoreStale cout ltlt quotyo I grew more stalelnquot age int main Candy trident tridentcategory quotgumquot tridentage 1 Candy mystery mysterycategory quotblobquot mysteryage 41 cout ltlt tridentage ltlt endl mysterygrowMoreStale cout ltlt mysteryage ltlt endl tridentage mysteryage tridentage cout ltlt tridentage ltlt endl Notes 0 By convention the rst letter of the class name is capitalized Candy Date 0 Every instance of a class in this example 39trident39 and 39mystery39 has its own set of members 0 Member functions and variables by themselves aren39t useful in main They must be associated with speci c instances 0 We can have multiple class de nitions in a le each with their own variables and functions 13 Public and Private Members Ok back to the quotpublicquot statement which means that the following variables and functions are quotpublicquot to others and can be accessed in main for example This might sound convenient but it can be dangerous For example look at this statement trident age 50 According to C this is perfectly legal However does it make sense Another classic example is the BankAccount class containing a variable for balance Obviously some problems may arise from such a setup Therefore we can also set member variables and functions to be quotprivatequot using the private statement so that only code within the class39s functions can modify that variable QSTN Complete the following code by setting the 39age and 39category to private and the function to 39public39 Also state what happens when run the main function class Candy void CandygrowMoreStale cout ltlt quotyo I grew more 5talelnquot age int main Candy trident tridentage cout ltlt tridentage ltlt endl As we noted earlier only functions within the Candy class can change the age QSTN Keeping the current variables private how can we change the age from the main class UNIVERSITY OF CALIFORNIA Los ANGELES Department ofEconomlcs c ameon Ecmumics 134 e Envimnmmml Ecunumics Lecture 9 storable Renewable Resources Forests Bale Forest Blology Efflclmt Management wltlnout nonrtlmber values Sources oflnetflclency posltlve ertennalltles e Deforestatlon e ebtrforrNaLure Swaps 1mplennentlng Efflclmt Managennent Readlngs Tletenbeg 6 chapter 11 Some matmal drawn from Hartwlck and olewlle 4 r tauonpenodquot for Mr PM at quot and pap e e pulp productlon but Wlldllfe habltat recreatlonal opp ortumtles ecosystenn support carbon sequestrattonwatershed enhancement re cycles etc Relevant Farest Biology stylized case As a stand oftrees grows the amount ofw stand volume vtt 0mm 5 e E d W 3 E 5 E E the trees become u ova maturequot and begln to decay frcrn old age dlsease lnsectpredatlon re or Wlnd and eventuall collapse Wood volume develops slowly m the beglnnlng and lncreases untll leetquot then slows toward date t when the maxlmum volume 15 achleved The volumerage relatlonshlp can be ln uenced by thnnlng fettltztng andrepresslng pests age ntstenn nammw 2 economlst Blologlstsused to focusonaconcept called mean annual lncrennentwlleI Theblologlcal M Tl ma lml d the l V cm called tn CAI as themarglnal product oftlme and the MAI as the average product oftlme stone volume vtt W r at eweer W vttt maxlmum MAl lncleasmg MAl Nb age at stone Simplest case une single harvest cycle strrctly blologlcal consrderatronswould lead to a decrsron to harvest when the age ofthe stand maxlmlzesLheMAI Thrs blologlcal crrterron for decldlngupon aharvestt sever Cmsldaatlon to account To begln wrth we mu worry about the costs ofharvestlng and plantrng E costs be sold 4 r WD stone volume vtt age at stand I T c then Rlttgtpvtgt Let dc srgnrry the set of current values at each ornt rn trrne w rch have a common present 0 when drscounted back to the present pro le dc represents the value onev qogt when rt rs caprtalrzed forward for each number ofyears Flndlng the harvest trrne T whrch maxlmlzes the present value of e harvest rneans ndlng the hlghest lsorpresmtrvalue lrne that Just torches the ha est revenue functron R t We can see that harvest trrne T maxlmlzes the present value ofthe harvest There rs only one drscount rate rn tlnrs dlagram The hrgner lsorpresmtrvalue curves rrse rnore qulckly because they start at a hrgner base level 1 lnturtrvely the trrne rnteval T wrll depend upon the drscount rate atTt ofthe former rs dqdtrq In drscretetrrne when oneunrt oftrrnepasses q rncreases by rq 1 gt as 1 Substrtutrng we can wrrte The slope Thuswe 7RRr A utilnl orer error 7 Back in 1930 American economist Irving Fisher noticed that this meant that trees should be cut when the proportional rate of growth of their net sales value equals the interest rate The should be cut when their value grows as fast on the stump as in the bank Living trees are natural assets that compete with nancial assets in terms of rate of return At time less than T the rate of return on the natural asset exceeds the interest rate on financial assets Beyond time T the rate of return on the natural asset falls short of the return on financial assets Therefore time T is the optimal point at which to convert your holdings from the natural asset to the nancial asset When tree growth slows so that the rate of return is higher on financial assets its is time to liquidate your natural assets time for portfolio adjustment Comparative statics ie what happens to the optimal harvest time T when we change any of the features of the problem a Higher interest rates This will make the family of isopresentvalue curves steeper meaning that the tangency with the current net value function for the harvest will occur at an earlier point in time Lower interest rates will have the opposite effect b Higher planting costs Planting costs occur at time 0 and will have no effect on the optimal harvest time unless they exceed the present value of the net harvest revenues in which case it would not make any economic sense to plant such a crop of trees As long as planting costs are less than the present value of net harvest revenues they will have no effect on the optimal harvest time c Higher lumber prices or higher harvest costs Recall that Rt p 7 c Vt If p or c changes each is assumed to remain constant over time then Rt will be scaled up or down proportionally However the condition for T is determined by RR r Any change in scale for R will factor out of the le handside leaving no effect on the optimal harvest time T Result c leads to some interesting implications for taxation policy Any tax levied on a perunitofwood harvested basis would be equivalent to an increase in harvest costs or a decrease in harvest price This would decrease the present value of revenues from the forestry operation without affecting the optimal harvest age Tietenberg shows this numerically Repeated harvests some material here is adapted from Hartwick and Olewiler 4 p 316317 Note here I had to make a decision There is a rather elegant model of optimal forest rotation periods that has been in existence since the mid 1800 s Basic textbooks like Tietenberg avoid showing students the model39 sophisticated textbooks like Hartwick and Olewiler devote considerable space to it Either we do the whole model in gory detail or gloss over its main results in words Since modern forestry economics has gotten away from simply the pro t 7m aximizing behavior of a forestexploitation rm some would argue that the time is better spent on institutional details However since this is an upper division course I have elected to outline for you how the model is developed and the standard technique that is used to solve for the optimal rotation period Butl will not march you through proofs of all the comparative static results that the model provides 1 will review the main results so you have the avor of how early natural resource economists approached this problem Why this morecomplicated model The single harvest model above is unrealistic for most developed countries In some developing countries we still see forestry practices involving a single harvest of virgin timber a er which the operation moves on to new stands without planting When land areas are nite from the point of view of the forestry operator however plans for replanting and subsequent harvests are the norm When subsequent planting and harvest cycles are planned there is an additional cost to be considered in determining the optimal harvest interval Delaying the harvest one more period in the current year means you have also delayed the start of the next planting cycle For the clearest treatment in this case it is convenient to suppose that interest is paid and compounded continuously In this case qO exprT qT where the notation exp means e raised to the power of whatever is in the parentheses In the continuous case r is interpreted as the instantaneous rate of interest and exprT is the discount factor comparable to ll r Let s assume it costs D to plant a unit of land say and acre and c per cubic foot of wood to harvest the trees The cost in present value of the first round in the infinite cycle is D 0 var i To eXP39139T1 Tel Where To is the date of planting and T1 is the date of harvesting The term exp rT1 7 T0 transfers costs at harvest time back to the beginning of the cycle which is planting date To Let p be the revenue per cubic foot sold upon harvesting assuming perfect competition The present value of profit forestowner s bene t is P 0 VT1 To eXP39139T1 Tel D After harvesting the land is replanted at cost D and a new round is undertaken The same formula would discount the harvest in this second round back to time T1 To get it back to T0 we would have to discount it over the interval from T0 to T1 For the third round amounts discounted back to T2 would have to be discounted over the interval from T0 to T2 to bring them back to T0 The complete present value from an in nite sequence of planting and harvesting cycles would be given by W P 0 VT1 To eXP39139T1 Tel D eXP39139T1 Tel P 9 VT2 T1 eXP39139T2 TID D eXP39139T2 Tel P 9 VT3 T2 eXP39139T3 Tzl D eXP39139T3 Tel P 9 VT4 T3 CXP39139T4 Tal D Now if the interest rate remains unchanged and nothing else differs across periods each harvest interval will be the same so Tj Tjil l for allj This means we can write W more compactly as W p 7 c Vl exp rl 7 D t egtltPr1l P i 0 VG egtltPr1 D t egtltPr21l P i 0 V16XPrfll D t egtltPr31l P i 0 V16XPrfll D If we factor the term exp rl out of the second and all subsequent terms we get W p 7 c Vl exp rl 7 D t egtltPr1 P i 0 VG egtltPr1 D t egtltPr1 P i 0 VG egtltPr1 D t egtltPr21l P i 0 VG egtltPr1 D The stuff in the curly brackets is now the same as the original expression for W written in terms of I This means we can write W p 7 c Vl exprl 7 D exprl W Solving for W we get W pc Vl exprl 7 D l 7exprl The forest manager then maximizes W by choosing rotation interval 1 ie set dWdl 0 This exercise yields the condition This means we can write W p 7 c Vl exprl 7 D exprl W Solving for W we get W pc Vl exprl 7 D l 7exprl ttttt tt ttt att ntet pelt tt Ovalhmnpm d L ll A L 41 ll lL V mt mu wh A lee The tenn rprc n Wthe TherWtenn ts called them rent w 39r h tau A followlngdlagam l L L LL A A a A at A a t hutnmtn tall also assoetated wtth auntquesttevalue mgaemntenestratesmean Iowa stte values Secondrorda aremlmrmzmg pro ts wovememal Ens s Recall that LHs ts tntapreted as the marglnal mmememal bene ts RHS PM bene ts funotton an RES ts emarglnal a costs funotton The opttrnalrotattonpenod RHS FM happens where LHs RHS Contrast ths wtth what happens unda oLha wan maxlmlzeLhe undlsocun e annualtzed stream ofrevenues prlantlng cosLsD were also 250 the emetent solutton would be to harvest at the oulmtnatton of mean annual tncrenent Lhe blologlst s answa 8 Fave l Fave mutton wtewol age at stand atttttt tt ptatttut u tn tatttttt oflandlnlt tt quot ttuttt t and convatto a dlffamt land use altogether P0116165 ta mamputme apnmm tlmber ramtlan parted nt nseq of tntewenttons but you should h 4 utt t tt tttet ttt p ttt that would not eonpletely eltmtnate the lndush39y a Tax per ton harvested royalty or severance tax equivalent to an increase in harvesting costs Will cause longer rotation periods More wood will be harvested because efficient rotation interval is typically less than the time for maximum mean annual increment so longer rotations will increase average annual harvests Siteuse tax tax per acre equivalent to an increase in D Will increase rotation intervals more wood harvested overall c Tax on pro ts tax on residual income accruing to land in forest production cannot be shi ed by changing rotation interval License fee per year rather than per rotation Optimal rotation period unaffected Property tax based on value in current use Encourages keeping land in forestry rather than converting to highervalued uses especially if rollback clause is included where change in land use incurrs cumulative back taxes for differential in value 7 V 09 VL Sources of Inef ciency The discussion above applies to the case where the only value derived from a forest is its value in commercial timber production Nontimber values N TV values of forests have surged to the forefront of modern forest management problems a Forests as reserves of biodiversity Gene splicing now allows researchers to transplant desirable genes from one species into another creating new species with more valuable characteristics Declining forest habitat contributes to the escalating rate of extinctions among species which may prove to be valuable Option demand for biodiversity Global climate change Trees absorb carbon dioxide one of the major greenhouse gases suspected of contributing to an apparent global warming trend Trees sequester store carbon Reforestation would store carbon burning of forests or natural decay releases carbon The role of forests in carbon sequestration has been receiving considerable attention Some call it pickling of carbon Habitat Ecosystem values Watershed maintenance prevention of soil erosion Population pressures Squatters with insecure tenancy on forest lands have no incentive for proper management of the resource ln poorer countries fuelwood is a signi cant source of energy for cooking population pressures can contribute to deforestation See Virtual Handout 12 U 1 p 39 39 of forest 39 quot issue formal permits 7 forest concessionsto individuals or private corporations to manage certain areas of public forest for timber production Concessionforest utilization contract Governments can capture economic rents via forest charges including taxes charges per volume harvested area charges and exploration fees Evidence that forest concessions have been widely and seriously underpriced lf government does not appropriate rents they are earned by the rm as profits Many countries are now reforming their forest concession policies One consideration is that concession time period should be long enough to foster ef cient dynamic optimization by firms ST V cm LV Towards more Ef cient Management recognition of the global scale of some of the uncompensated positive externalities associated with good forest managem ent39 explaining and reversing trend towards deforestation especially in developing countries Deforestation issues Classic article is Deacon 2 Additional discussion and more empirical work in Kahn and McDonald 5 and in Capistrano and Kiker 1 two articles identi ed in this year s SSCl submissions Debtfornature swaps Deacon and Murphy 3 The structure and occurrence of debtfornature swaps are examined empirically Contracts executing debtfornature swaps are studied to assess the role 0 transaction costs in determining how these agreements are structured The emerging contract form is a product of weak enforcement of legal claims to environmental resources in developing countries high costs for delineating and monitoring environmental outcomes and nominal government ownership 0 t e resources involved The occurrence of swaps in individual countries is significantly related to host country attributes including the presence of tropical land and threatened species democratic political institutions and large debt burdens Empirical finding in Virtual Handout 12 A Theory of Media Politics How the Interests of Politicians Journalists and Citizens Shape the News By John Zaller Draft October 24 1999 Under contract to University of C hicago Press A version of the book was given as the inaugural MillerConverse Lecture University of Michigan April 14 1997 Versions have also been given at the Annenberg School for Communication at Penn and at the University of British Columbia and at seminars at UCLA UCLA School of Law UC Riverside Harvard Princeton UCLA Program in Communication Studies and Chicago What audiences at these places have liked and disliked has been immensely valuable to me in developing my argument though perhaps not always in the ways they might have expected I am also grateful to Michael Alvarez Kathy Bawn Bill Bianco Lara Brown Jim DeNardo John Geer Shanto lyengar Taeku Lee Dan Lowenstein Jonathan Nagel John PetrocikTom Schwartz Jim Sidanius Warren Miller and especially to Larry Bartels Barbara Geddes and George Tsebelis for helpful comments on earlier drafts Chapter 1 The New Game in Town Introduction A few years after he left office in 1969 President Lyndon Johnson was asked by a TV news producer what had changed in American politics since the 1930s when he came to Washington as a young Texas Congressman quotYou guysquot Johnson replied without even reflecting quotAll you guys in the media All of politics has changed because of you You39ve broken all the party machines and the ties between us in the Congress and the city machines You39ve given us a new kind of peoplequot A certain disdain passed over his face quotTeddy Tunney1 They39re your creations your puppets No machine could ever create a Teddy Kennedy Only you guys They39re all yours Your productquot Halberstam 1979 pp 1516 In the old days political disagreements were settled in backroom deals among party big shots As majority leader of the Senate in the 1950s Johnson achieved national fame as master of this brand of insider politics But in the new environment disagreements are fought out in the mass media and settled in the court of public opinion The weapons of combat are press conferences photo opportunities news releases leaks to the press and quotspinquot When the stakes are especially high TV and radio advertisements may be used Politicians still make backroom deals but only after their relative strength has been established in the public game of quotmedia politicsquot By media politics I mean a system of politics in which individual politicians seekto gain office and to conduct politics while in office through communication that reaches citizens through the mass media Thus defined media politics stands in contrast to the older system of quotparty politicsquot in which by conventional definition politicians seek to win elections and to govern as members of party teams Although party politics is by no means defunct it now shares the political stage with media politics an emerging system whose properties are only beginning to be understood 1 The references were to Ted Kennedy widely considered at the time to be a likely future president and to John Tunney a photogenic media savvy Senator from California 11 When I say that media politics is a system ofpoitics I mean to compare it to such other systems as legislative politics bureaucratic politics judicial politics and as already suggested party politics Within each ofthese domains one can identify key roles diverse interests routine rules of behavior and stable patterns of interaction that taken altogether define a distinctive form of political struggle In my account of media politics there will be three principal actors politicians journalists and whom is animated by a distinctive motive For politicians the goal of media politics is to use mass communication to mobilize the public support they need to win elections and to get their programs enacted while in office Forjournalists the goal of media politics is to produce stories that attract big audiences and that emphasize the quotIndependent and Significant Voice of Journalists For citizens the goal is to monitor politics and hold politicians accountable on the basis of minimal effort These goals are a source of constant tension among the three actors Politicians would like journalists to act as a neutral conveyor belt for their statements and press releases Yet journalists do not want to be anybody s handmaiden they wish rather to make a distinctive journalistic contribution to the news which they can better accomplish by means of scoops investigations and news analyses all of which politicians detest In my account of media politics journalists value journalistic voice at least as much as big audiences2 and they care nothing at all about helping politicians to get their story out to the public lfjournalists always reported the news just the way politicians wanted them to or gave audiences only the political news they really wanted journalism would be a much less lucrative and satisfying profession for its practitioners than it presently is In fact it would scarcely be a profession at all The public wants as indicated to monitor politics and hold politicians accountable with minimal effort And because there is a surfeit of politicians and journalists vying for public attention in a competitive market the public tends to get the kind of political communication it wants But not entirely The politicians39 inherent interest in controlling the content of political news in combination with journalists39 2 Journalists may be compared in this regard to professors at research universities who typically care about undergraduate ratings oftheir courses only because and to the extent that they have to but care deeply about expressing voice through research The difference is that professors are much more insulated from market pressure inherent interest in making an independent contribution to the news create a farreaching set of tensions and distortions The argument of the monograph simply put is that the form and content of media politics are largely determined by the disparate interests of politicians journalists and citizens as each groupjostles to get what it wants out of politics and the political communication that makes politics possible Although media politics is pervasive in American national life this book focuses on presidential selection The reason is methodological Presidential elections have a fixed structure and recur at regular intervals thereby making it possible to observe patterns and to test generalizations across multiple cases Even though media politics probably has the same basic properties in nonelectoral settings eg the struggle to pass the Voting Rights Act of 1965 or the infamous Federal Government Shutdowns of 1995 and 1996 it is harder to demonstrate these properties or sometimes to perceive any sort of regularity at all in nonelectoral settings Why Because systematic political analysis depends upon a delicate balance of similarity and difference a stable common background against which to observe meaningful differences To a greater degree than almost any other kind of media event presidential elections have that balance politicians reporters and voters going through the same basic routines over and over but under somewhat different conditions And because presidential elections are so important to our democratic life the differences in conditions are closely studied and painstakingly recorded in the form of polls news and books Little of importance goes unnoticed As a result of all this it is easier to discern and measure the dynamics of media politics in this setting than in others But to reiterate This book aspires to be more than a study of the role of media politics in presidential elections it aims to be a study of media politics in a context in which the dynamics of media politics happen to be relatively easy to observe and study As I shall argue there are good reasons to believe that the forces that animate media politics are essentially similar in both electoral and nonelectoral contexts The approach to studying media politics in this book is distinctive in two respects The first as already suggested is that it focuses on the diverse selfinterests of the participants and how they shape the nature of media politics This is a departure from most studies of media politics which tend to see 13 media politics through different theoretical prisms One major strand of media research focuses on the values and conventions ofjournalists such as their delight in covering the political quothorseracequot Patterson 1993 Lichter Rothman and Lichter 1986 or the routines by which reporters organize their work Cohen 1962 Sigal 1973 Epstein 1973 Gans 1980 Another major strand of media research emphasizes the symbolic side of media politics especially its creation of illusions images and spectacles that masquerade as a depiction of reality Edelman 1988 Bennett 1996 Without challenging the validity of insights in previous studies this book offers as a corrective the view that media politics is like other forms of politics driven most fundamentally by conflicts in the goals and selfinterests of the key participants And in an even stronger corrective to existing research it maintains that media politics is driven by the selfinterest of the public at least as much as by the selfinterests of other actors3 The other distinctive aspect of this study is that it is organized deductively rather than inductively In the inductive mode of analysis one begins by describing a set of facts and then draws or induces from them a theoretical explanation In the deductive mode one begins by positing a handful oftheoretical claims and then logically derives or deduces from them specific hypotheses which are tested against a set of facts In keeping with the latter mode of analysis I shall make a point of deriving all of my hypotheses from clearly stated premises and referring ostentatiously to each deductive inference by number as in D1 D2 and so forth For the type of study undertaken here that is a heavily empirical study that employs no strictly formal analysis the difference between the deductive mode of analysis and the more familiar inductive mode is largely stylistic Yet I believe the stylistic difference has important practical value First in beginning with theory rather than data the deductive mode tends to focus the reader39s attention where I think it belongs on the general processes at work rather than on the particular and sometimes distractingly colorful facts that are at the base oftheories Second in focusing attention on theory per se the deductive mode makes it easier to see how the various elements of one39s theory logically relate to one another This in turn makes errors of analysis on the part of the researcher me and failures of 3 Perhaps the only study of media politics to emphasize the importance of mass interests in determining media content is that of Bovitz Druckman and Lupia 1997 14 comprehension on the part of readers you both less likely though of course far from impossible in either case Chapter2 The Players in the Game The theory of media politics I propose is in effect an extension of Anthony Downs39 study An Economic Theory of Democracy In this 1957 classic Downs showed how party competition for the support of rational voters could explain many ofthe most salient features of democratic politics1 But Downs39 theory hardly mentioned journalists and gave them no independent role in politics In the present study I create a theoretical role forjournalists within Downs democratic system and trace out certain effects ofthis change Specifically I require officeseeking politicians to communicate with voters at least some of the time through a journalistic profession whose interests are quotvoicequot and audience share Because both Downs39 theory and my extension of it are rooted in basic political forces it is plausible to believe that my theory of media politics applies to political news in the US generally and not merely to presidential elections I shall later offer some modest evidence for this view WHY INVOKE RATIONAL CHOICE ln following Downs my theory of media politics takes a loosely rational choice approach to its subject That is it treats media politics as the product of goaloriented behavior on the part of key actors in the political system namely politicians journalists and citizens The fact that the goals ofthese actors as specified below often conflict is what makes politics and as I hope readers will conclude my theory of A straightforward implication of rational choice is that individuals take account ofthe goaloriented behavior of others with whom they interact It is extremely hard in my view to overestimate the importance ofthis point for the understanding of media politics or other forms of political struggle for that matter Everyone in politics does what he or she does in significant part because of what others are doing or expected to do Thus to take a commonplace example candidates create the kinds of 1 Downs39 book is in important respects an incisive digest of prior theoretical and empiricial work notably that of Schattschneider 1942 Schumpeter 1942 Key 19xx Black 1958 and Arrow 19XX campaign events they do because of their beliefs about how journalists are likely to cover the events Or to take an example that I will develop more fully below journalists facing multicandidate fields in presidential primaries routinely limit their coverage to the two or three contenders that they think voters are most likely to favor When candidates do what they do because of how they thinkjournalists will respond and when candidates are covered or ignored because of how they are expected to fare with voters one cannot provide a satisfactory explanation by focusing on any single actor in isolation from the others Rather one must take at least theoretical account ofthe full set of actors and in particular how the actions and anticipated actions of one set of actors affect the actions of others This is not easy to do but it is more natural to attempt it within a rational choice framework than any other for this reason Whereas psychological theories tend to focus on the effects of internal drives and perceptions on individual behavior and whereas sociological theories tend to stress the effects of external structure on behavior the notion of strategic behavior that is inherent in rational choice posits that individual behavior is shaped by both external forces what other individuals are trying to do in a particular situation and internal drives personal interests and goals Although taking a rational choice approach I by no means assume that everyone39s mind works like a computer calculating all possible contingencies at each decision point and making the move with the best expected return I make a much milder set of assumptions That individuals at all levels of politics m to behave in ways that advance goals that are important to them that individuals are embedded in groups such as classes or professions whose rules and values help them to achieve their goals and that thus assisted individuals establish 1 of behavior that do generally reflect their goals Politicians are probably the only political actors who regularly and consciously calculate the expected gain for every important action But voters who for example somewhat mindlessly support the party of their social class or journalists who are equally mindless in their distrust of authority may also be rational in the sense that their basic patterns of behavior may have initially developed and continue to exist primarily because they serve individual goals Thus in my use of rational choice individual choices need not be calculated or even selfconscious in order to represent interestoriented behavior and hence qualify as rational A danger in this brand of quotsoft rational choicequot is that anything anyone does might be sloppily described as rational But I am quite aware ofthis danger and do not believe that my theory will suffer greatly from this form of indiscipline My basic theoretical posture then is that politicians journalists and citizens behave in ways that generally reflect individual goals and interests that in pursuing their various goals individuals take account of the goaloriented behavior of other individuals with whom they interact and that the essential features of media politics can be usefully analyzed as the outcome of all this goaloriented and strategic behavior Rational choice is often seen as a controversial perspective especially when it invades new intellectual terrain It is however hard for me to see what general objection there can be to the theoretical posture outlined in the preceding paragraph THE DOWNSIAN FRAMEWORK Downs theory of democracy is based on a handful of theoretical postulates The most important are that politicians are organized into party teams that care about winning office and nothing else that voters wish to elect politicians who give them as much as possible of what they want out of government and that both politicians and voters are coldbloodedly rational in the pursuit ofthese goals What voters want out of government is anything that happens to give them utility whether in the form of individual benefits eg social security low taxes a prosperous national economy or social justice for others There is no requirement in Downs model that voters be selfish the only requirement is that voters support parties that deliver what they as voters want From these simple assumptions Downs deduces many theoretical expectations that most observers regard as true For example Downs argues that in a twoparty system both parties will converge to the position ofthe median voter that is the voter who occupies the dead center ofthe ideological spectrum This is because if either party moves left or right of center the other party will then capture ofthe votes of centrist voters and thereby win the election The actual tendency of the Democratic and Republican party to stay near the middle of the road in most elections seems wellexplained by this argument Another of Downs arguments is that producers are more likely to organize to get what they want out of government than consumers Consider for example the case of dairy farmers For such people government policy toward milk is extremely important since their whole livelihood depends on it For consumers on the other hand milk is only one of hundreds or thousands ofthings that they purchase At the same time there are relatively few milk producers which makes it easy for them to know one another and organize Milk consumers on the other hand are more numerous and therefore harder to organize For these reasons milk producers are more likely to form effective lobbying organizations This argument which generalizes to businesses of all kinds seems a plausible explanation for the advantages that many special interests have in getting their way with government Actually these and other arguments in Downs book were originally proposed by scholars other than Economic Mg Democracy is to pull many such arguments into a cohesive theory about how democracy works if everyone is rational in the pursuit oftheir political goals One of Downs most intriguing arguments is that it is rational for voters to pay little attention to politics and to rely on simple heuristics such as party attachment and ideological labels to decide how to vote This argument will be especially important to my theory of media politics and so will be considered below in more detail Some four decades after it appeared in print Downs39 study still captures some ofthe most important features of our political system Politicians who cling to the middle of the road and voters who rely on party attachment remain as they were in the 1950s among the most salient features ofthe American political system Notably however Downs specifies an entirely passive role for the journalistic profession in his theory The assumption seems to be that reporters reflect the political biases oftheir publishers but do not otherwise affect the political process In the 1950s this may still have been a plausible assumption The partisan press of the 19th century in which newspapers functioned as virtual adjuncts of the parties had become far more neutral but some vestiges ofthe old way remained For the most part the mass media seemed unaccountably unassertive perhaps less assertive than at any other time in American history In the nomination phase of presidential selection reporters basically just stood around outside the quotsmokefilled roomsquot at which the real decisions were made hoping for crumbs of information And they were scarcely more intrusive in general election campaigns Even in Time and Newsweek magazines known for their interpretive style typical campaign news consisted of chronological accounts of what the candidates were doing laced with lengthy verbatim quotes from their speeches There were to be sure some publishers who played an active role in politics but they were acting as agents of their quotparty teamquot rather than as members of an independentjournalistic profession In these circumstances there was no need for Downs to posit an independent role for the mass media in the process of elections and governance Circumstances however have now changed The old partisan press is fully defunct and so for the most part is the quotlapdogquot press of the 1940s and 1950s this apt term is from Sabato 1993 Journalists no longer stand idly by while party nominations are made or mechanically relay candidate information to the voters in elections They are key intermediaries in the process by which competing politicians attempt to mobilize public support in both the nomination and general phases of presidential elections The change in the role ofthe mass media is part of a much larger change in American national politics a transition away from Party Politics as the predominant form of political organization and toward a new system in which media politics is also important Elaborating on earlier definitions suggest that these terms be understood as follows The de ning characteristic of Party Politics is that politicians compete as members of organized teams In strong forms of Party Politics party leaders choose candidates for party nominations conduct the campaigns for of ce and coordinate their activities in of ce Voters recognizing that the parties compete as teams cast quotstraightparty ballots for one of the teams The de ning feature of media politics as the term is commonly used is that politicians seek to gain of ce and to conduct politics while in of ce through communication that reaches citizens through the mass media Parties and interest groups formerly unchallenged kingpins of mass politics are often left on the sidelines as independent politicians do battle by means of speeches press conferences advertisements photdops and various other quotpublic relations events The basic dynamics of Party Politics have been wellunderstood for some decades through the work of E E Schattschneider V 0 Key Jr Joseph Schumpeter Kenneth Arrow and Downs but media politics is a relatively new form of organization and hence less well understood My aim in this book is to develop a theory of the new form and to accommodate it to traditional understandings of American politics as encapsulated in the work of Downs The first step in developing the theory is to specify the general goals of each of the key actors candidates voters and journalists I begin with candidates the group whose behavior is easiest to fathom From the interests of all three types of actors the dynamics of media politics will later be deduced THE GOALS OF CANDIDATES Downs theory focused on parties and assumed that their only political goal was to capture and hold political office formulating policies as necessary to achieve this goal I make the same assumption except that my focus will be on individual politicians rather than on party teams Going beyond Downs I shall also deal with the process by which politicians communicate their policy proposals to voters which is the defining feature of media politics Let me begin with some historical background In the heyday of 19th century party politics communication with voters was not something that presidential candidates worried about As titular head of the Democratic or Republican party they relied on their fellow partisans to conduct their campaigns for them McGerr 1986 For the most part this meant turning the campaign over to city and state units which canvassed doortodoor for the party ticket and offered public entertainment in the form of torchlight parades and family picnics as a means of mobilizing support Most newspapers in the 19th century had an informal party affiliation and openly boosted its party s candidates To the likes of Joseph Pulitzer Robert McCormick and Otis Chandler fiercely partisan coverage was more a scared duty than a cause for embarrassment and it was not a duty that was shirked any more in the news columns than on the editorial pages Thus in a study of partisan bias in of the Chicago Tribune between 1900 and 1992 Burgos 1996 found that headlines attacking Democratic candidates at the turn of the century were ten times more frequent than ones attacking Republicans For example following a dinner gathering of GOP luminaries during the election of 1900 the paper proclaimed on its front page Hosts Gather At Great Feat President39s Position Correct McKinley Was Right Bryan Denounced As Demagogue The person denouncing McKinley in the Tribune headline was Robert B McArthur the pastor of the local Baptist church If there were other local pastors who felt that the Republican candidate was the demagogue they were not given access to the Trib39s news pages As Burgos goes on to show the Tribune39s blatant and onesided partisanship declined gradually over the course ofthis century As a result the paper had become essentially balanced in its presidential campaign coverage and at much lower levels of negativity by the 1970s Despite its glorious past the tradition of unabashedly partisan journalism has been in decline since about 1870 the point at which a group of dissident journalists founded a reform movement dedicated to the ideal of nonpartisan and objective coverage McGerr 1986 The transformation in the 1960s of such partisan holdouts as the Q Angeles Times the Chicago Tribune and Time magazine marked the final triumph ofthis movement The capacity of local party organizations to mobilize support for candidates has also declined greatly since the 19th century The upshot is that candidates must now make their own way both in presidential primaries and in the general election That is they must get out on the campaign trail and try to create events that a nonpartisan press will see fit to report as news The new situation is wellcharacterized by Ansolabehere Behr and lyengar Today political leaders communicate with the public primarily through news media that they Q not control The news media now stand between politicians and their constituents Politicians speak to the media the media then speak to the voters 1993 p 1 Paid advertising helps presidential candidates out of this bind Jamieson 1996 but does not eliminate the great need to achieve favorable notice in the quotfree mediaquot How politicians go about trying to create favorable news is fairly well understood On the one hand they attempt to take actions and create events that promote their campaign agenda and that are so compelling that reporters will feel obligated to report them as news and on the other hand they attempt to avoid situations such as news conferences that make it difficult for them to control what gets reported as news The kind of coverage that politicians want is also fairly obvious They seek to be associated with honesty competence likability and popular policies Candidates however may not always be completely clear about the policies they favor As Downs argued a degree of ambiguity may increase their appeal to voters who might otherwise feel distant from them As Downs put it Ambiguity increases the number of voters to whom a party may appeal This fact encourages parties in a twoparty system to be as equivocal as possible about their stands on each controversial issue And since both parties find it rational to be ambiguous neither is forced by the other39s clarity to take a more precise stand Thus political rationality leads parties in a twoparty system to becloud their policies in a fog of ambiguity Chapters 7 and 8 Subsequent scholars have not always agreed with Downs on this point Shepsle 1972 Bartels 1988 Alvarez 1997 but also Page 1978 Jamieson 1992 Chapter 9 But whether or not it is rational for candidates to be deliberately ambiguous it certainly is rational if they can get away with it for them to do something rather similar To take different positions in front of different audiences For example during the 1968 presidential campaign Nixon told northern audiences that he strongly supported the Supreme Court39s 1954 desegregation ruling but in a TV broadcast beamed to southern audiences he carefully suggested otherwise Witcover 1970 p 38586 Often because they are in danger of losing candidates also sometimes change positions during campaigns make extravagant or unrealistic promises or distort the records of their opponents Jamieson 1992 When for whatever reason candidates do any ofthese things they wantjournalists to report their statements as quotstraightquot news without any hint of challenge Also most politicians like most nonpoliticians have done things in the past that they find embarrassing to admit in public and that they therefore try to keep secret For politicians then the new goal of media politics is to get certain helpful kinds of campaign information reported as news and to keep other unhelpful kinds of information out of the news Put more simply the goal of politicians is to Use journalists to quotGet Our Story Out quot As we shall see this goal tends to bring candidates into more or less continuous conflict with journalists who have no interest in running the kind of news that politicians would most like and some considerable interest in running stories that politicians typically do not like THE GOALS OF CITIZENS I shall assume that citizens have the same basic outlook in the age of media politics that they did in the earlier age of party politics as theorized by Downs That is citizens want to elect politicians who will do what they as individual citizens want to have done Yet as Downs also argued citizens are busy people and they are sensible enough to appreciate that as individual voters their chances to affect election outcomes are minuscule Hence they instinctively minimize their electoral involvement hoping for a good result but refusing to put significant effort into it including the effort necessary to study the issues and candidates in the election The payoff is simply not there Voters are more likely to be mugged on the way to the polls than to actually affect an election or other political outcome2 Thus as Downs reasoned for most citizens most of the time it is individually rational to be ignorant about politics Citizens will prefer to use their limited time for matters that provide a more direct and certain return for the effort such as playing with children working overtime or perhapsjust watching a comedy on TV The question that now arises is the attitude of rationally ignorant citizens toward political news The answer in broad outline is obvious They will mostly disdain it Yet the little attention voters do pay may be very important to politicians and journalists since their livelihoods depend on the response ofthe mass audience to political news I should add that there are many kinds of news besides political news These varieties include entertainment news consumer news sports news and medical news Most business advertising is also 2 My colleague Tom Schwartz claims credit for this formulation ofthe classic problem 14 a form of news namely product news My theory of media politics is concerned only with political news by which I mean news that is primarily about public policymaking and leadership selection So what do rationally ignorant citizens want out of the relatively small amount of political news they consume suggest several interests each following in a loosely deductive sense from the basic notion of rational ignorance Rational voters want to keep tabs on political events if only to know how their tax bills or benefit checks are likely to change They just don39t want to devote much energy to it Hence rational voters do not want to be immersed in details nor do they want large quantities of dense substantive information and analysis nor do they want news reports that attempt to be encyclopedic and comprehensive full of context and history about every aspect of the public affairs Stated negatively the overriding message of rational voters to their information providers is quotDon39t waste my timequot Stated affirmatively the message is quotTell me only what really need to knowquot Remember that this imperative concerns political news but not necessarily other kinds of news Indeed the contrast with other kinds of news is illuminating It is probably not rational for citizens to ignore or mostly ignore health news since it conveys information that can tangibly improve the length or quality of their lives Even if most health news were boring or irrelevant to one39s personal condition it could still be worth paying close attention to it since the individual benefits of even an occasional story that is personally relevant can be very great But the same cannot be said for political news A citizen can spend his entire waking life digesting political news and in consequence make extremely wise political choices and yet be no better offthan if he or she had done no studying at all What the rational voter wants then is help in focusing as efficiently as possible on those matters that absolutely require attention But what requires attention As indicated voters know or at least intuitively appreciate that it is not worth their time to give careful consideration to their vote choices because their power to affect events is tiny Yet despite this election outcomes can have quite large effects on individual voters Depending on who wins taxes may be cut or raised welfare or Medicare benefits may be expanded or slashed the government may draft young people to fight in overseas wars In light of this fundamental asymmetry elections can affect individual voters far more than individual voters can affect elections I reach the following conclusion The rational citizen will be more interested in information about how the election is likely to come out than in information that will help him to cast a wise vote To whatever modest extent rational voters seek information whose purpose is to help them form an informed opinion or cast a wise vote they will seek information about matters that are controversial When elites achieve a consensus on a policy the policy is likely to be adopted no matter who wins the election and if this is so there is no reason for each voter to try to figure out for herself or himself which side is best and which candidate favors it If on the other hand elites disagree the election outcome may determine what policy is adopted thus giving voters an incentive to pay some bit of attention By this reasoning I reach the conclusion that The rational voter is engaged by political conflict and bored by political consensus When elites do disagree each side works hard to articulate the best arguments for its position and to expose the weaknesses ofthe other side39s position And they have every incentive to state their arguments in terms that ordinary people can readily understand By monitoring such disagreements citizens can often get incisive information on the basis of little effort Of course even a little bit of effort may be more than most voters want to make Yet they know that some oftheir fellow citizens will be paying attention if only for the entertainment value of politics and they want this minority of politics junkies to be able to see what it going on And finally even if voters do not themselves want to pay attention to most conflicts they want to retain the option of paying attention in case some really important issue should come up For all these reasons rational voters do not want political conflict swept under the carpet away from public view nor do they want any elite group politicians or journalists to monopolize public discourse with its own point of view Rather When political elites disagree rational citizens want exposure to both sides of the argument and under no circumstances do they ever want to see one side monopolizing public discussion Nonetheless rational citizens are ambivalent toward elite conflict including conflict between politicians and journalists They are as indicated engaged by it and insofar as they pay any attention wish to know both sides But they also want to limit their attention to politics and if elites engage in too much fighting then paying attention to conflict loses its value as a heuristic Much like the harried parent who scolds bickering children to quotjust work it out among yourselvesquot citizens wish to avoid being called upon to arbitrate all of the numerous issues on which ideologically contentious and often selfinterested elites may get into fights Hence Rational citizens become impatient with elites who disagree too much withdrawing attention trust or votes as appropriate Synthesizing the last three of these points we may say that Rational citizens want to be exposed to some but not a great deal of elite con ict An important difficulty with this line of argument is that although I have claimed that citizens wish to focus on controversial matters because their vote or opinion is more likely to be consequential in such matters the possibility that an individual voter could E be pivotal is extremely remote even in a close election turning on a controversial issue Thus as my UCLA colleague Tom Schwartz has observed the claim that a voter is more likely to be pivotal in a close election is like the claim that a tall man is more likely to bump his head on the moon In light ofthis basic political reality it seems prudent to develop an alternate rationale for the propositions just offered Since the difference between news and entertainment is often a subtle one3 the most promising line of argument is that citizens watch political news in order to be entertained The question then arises What kinds of political news will citizens find most entertaining It is beyond my power to develop an original theory of entertainment so I will work from the conventional view of what citizens find enjoyable in nonpolitical domains of entertainment sex violence 3 As Neuman 1991 p 114 observes quotTheories of education and mass communication have been troubled by a naive distinction between information and entertainment Although in common parlance we all routinely make such distinctions in the practice of daytoday mass communications the two elements are inextricably intertwined Neither the communicator nor the audience can meaningfully determine which element of a message or which characteristic ofthe delivery medium is most successful in attracting attention or in amusing or informing the audiencequot 17 suspense humor and human drama Perhaps unfortunately politics offers relatively little sex or humor though it must be said thatjournalists are quick to exploit what there is of them but politics does offer an abundance of a nearequivalent to violence namely political conflict And where there is conflict there is often suspense and drama as to how it will be settled Hence one might argue that journalists would tend to focus on political conflict because their audience will find conflict more entertaining than consensus But how much conflict To judge from movies and sports the taste for conflict probably varies greatly across individuals Some people watch movies like The Texas ChainSaw Massacre and go to ice hockey games while others prefer The Sound of Music and golf Yet even in the most violent movies one rarely sees more than one episode of major violence every 15 minutes or so and the same may be true even for ice hockey Boxing is more violent but it has a small audience If we take something like Star Wars as the exemplar of a successful mass entertainment offering we might infer that the taste of the median entertainment consumer is for some but not a great deal of violence that is wellorganized and not too brutal From this reasoning I infer that entertainment audiences prefer political news having some but not a great deal of conflict4 As a separate matter I note the widespread but by no means universal popularity of sports broadcasting and sports news Most sports offers some sort of violence and all offer the distinctive element of organized competition From this one may infer that many citizens find competition per se to be entertaining and that by extension many will be attracted to political news that describes such competition To keep my parallel lines of argument clear let me recapitulate Reasoning from the notion of rational ignorance l infer that citizens want 1 to avoid wasting time on political news whose only purpose is help them develop informed opinions and cast wise votes and that insofar as citizens want any political 4 Violent entertainment nearly always includes stereotypically good guys and bad guys thus suggesting that having someone to cheer for and against is essential to the enjoyment of conflict lf today39s citizens fail to enjoy political conflict as much as my discussion suggests or as much as they seem to have enjoyed it in the 19th century it may be because the nonpartisan press unlike its 19th century counterpart does not frame domestic political conflict as a battle between good guys and bad guys See McGerr 1986 news at all they want news that 2 emphasizes what government is likely to do to citizens more than how citizens can affect what government will do and that 3 provides some but not a great deal of conflict Because the latter two or these inferences derive from the debatable assumption that rationally ignorant citizens want any political news at all I provided an auxiliary justification for them which is that citizens derive pure entertainment value from news that stresses competition and some but not too much conflict This analysis of mass preference for news is not based on the direct testimony ofthe citizenry as expressed in public opinion surveys Such testimony seems to suggest higher levels of public interest in politics than can be justified from the notion of rational ignorance For example 49 percent of respondents to a 1992 survey said that they were quotvery much interestedquot in that year39s political campaigns while 40 percent professed to being quotsomewhat interestedquot and only 11 percent said they were quotnot much interestedquot Another question found that 27 percent claim to follow what39s going on in government and public affairs quotmost of the timequot 41 percent follow it quotsome of the timequot and 32 percent follow it only quotnow and thenquot or quothardly at allquot These numbers though not extremely high nonetheless indicate more interest than my theory can comfortably accommodate but also more than probably really exists For there is a clear tendency of many citizens to attribute more interest to themselves in verbal statements than they exhibit by their actual political behavior Thus Doris Graber 1984 found in her study ofthe news consumption habits that ordinary citizens were often bored by the news but that they nonetheless grumbled frequently about the oversimplified treatment of all news including elections news on television Yet when the debates and other special news programs and newspaper features presented a small opportunity for more extensive exposure to issues they were unwilling to seize it For the most part the study subjects would not read and study carefully the more extensive versions of election and other news in newspapers and news magazines Masses of specific facts and statistics were uniformly characterized as dull confusing and unduly detailed Such attitudes present a catch22 situation If more detail and specificity is resented how else can the demand for greater depth be satisfied p 105 Over the years journalists have occasionally tried schemes to increase the attention citizens pay to news mostly without success But as Lance Bennett 1996 reports many editors and marketers think that the few noble experiments to improve election issue coverage and offer more indepth political reporting are up against a basic obstacle People really do not want more serious news even when they say they do p 2223 W Russell Neuman 1991 makes the same observation Those who call for publicaffairs programming on television do not tend to watch it when it is made available Those who claim to attend to the media for purposes of acquiring information do score slightly higher on tests of learning and recall but the differences are surprisingly small The key finding that must be dealt with candidly if we are to understand the nature of lowsalience learning in regard to politics and culture is simply that people are attracted to the path of least resistance For knowledge acquisition in general and for publicaffairs knowledge in particular people are not inclined to give such matters a great deal of effort p 95 103 Politicians seem to have arrived at a similar conclusion In the 1996 election the major party candidates were offered free TV time on an experimental basis by several networks provided they use it for a serious discussion of the issues But Dole claimed only about three quarters of the time allotted him and Clinton used his time for what seemed like boilerplate excerpts from his stump speeches In the last election in Britain neither party used the full two hours of free TV time they are guaranteed by law and in Israel there is a joke that when the candidates claim their free TV time water pressure throughout the country falls as viewers seize the opportunity for a bathroom break The little attention citizens pay to the serious news they currently get suggests that they may want even less As Bennett writes of newspapers in particular All over the country the trend is to hire market research firms to find out how to win more subscribers The main casualty of packaging the press has been the amount of space devoted to hard news whether local state national or international which has dropped sharply as publishers bend to popular tastes and business pressures p 20 Perhaps the clearest indication that many citizens are not as interested in politics as they claim to be is how few citizens possess even a rudimentary knowledge of the political system and its leading figures Only about a quarter can typically name the Speaker of the House of Representatives and only ten percent the Chief Justice ofthe Supreme Court information that is scarcely obscure In 1992 after nearly four decades of continuous Democratic control ofthe House of Representatives only about half 20 knew which party controlled the House5 It is easy to multiply such examples of citizen ignorance Delli Carpini and Keeter 1995 The fact that citizens know something about government and politics shows that many pay passing attention to public affairs But it is hard to make the case that more than a few more than say the mere 10 percent who can name the Chief Justice of the Supreme Court pay much more than that THE GOALS OF JOURNALISTS Journalists are a highly differentiated group They are spread across newspapers news magazines TV and radio and they vary in style from quothappy talkquot TV news anchors to the erudite Robert MacNeil of PBS My theory of media politics primarily concerns elite journalists by which I mean journalists who specialize in coverage of national politics or who work for a nationally prestigious organization such as the New York Times CBS News or Newsweek I focus on this group ofjournalists because by common observation they tend to set the news agenda for other media I shall sometimes refer to other journalists especially local journalists who do not specialize in national politics but do sometimes cover it because they may also affect media politics But unless I specifically say so all of my references to journalists should be understood as references to elite journalists What then do elite journalists want How if at all can their quotinterestsquot be generally characterized A simple answer to this question is that like politicians and just about everyone else journalists want career success In the case ofjournalists career success means producing stories that make it onto the front page or get lots of airtime on the evening news from whence flow fat salaries peer respect and sometimes a degree of celebrity status What we must then inquire gets onto the front page and the top ofthe evening news 5 Actually 59 percent named the Democrats in the 1992 survey while about 10 percent named the Republicans If as seems prudent we assume that the 10 percent who named the Republicans were guessing there must have been another 10 percent who guessed Democrats and happened to hit the right answer Subtracting the likely percentage of guessers from 59 percent yields 50 percent an impressively low number for such an obvious piece of information 21 Certainly one part of the answer is that in the competitive business ofjournalism the stories that make it onto the front page are the ones that the public is interested in From this it follows that the most successful journalists are the ones who are most adept at appealing to the tastes of the mass audience Yet this is scarcely the whole story For although the tastes and interests ofthe mass audience must certainly affect the kind of news thatjournalists provide it would be very dubious to assume that quotwhat elite journalists wantquot is to provide the mass audience with exactly what it wants Indeed the opposite assumption may be closer to the mark That what journalists want is to be freed from subservience to the mass audience so that they can provide the public with the kind of news that they as professional journalists feel the public needs quotToo many of us in hard newsquot as CBS news anchor Dan Rather has bluntly written quotare looking for that extra tenth of a ratings pointquot and thereby quotblurring the distinctions and standards between news and entertainmentquot6 In a similar vein NBC anchor Tom Brokaw has openly pined for the early days of TV news when journalists could dictate to a captive audience 39When I started out in the 196039squot he said in an interview quotthere were effectively two network news programs and at 630 PM people turned on either HuntleyBrinkley or Walter Cronkite and got their news for the day And I39d like to have that back againquot7 The ambivalent attitude of elite journalists toward their clientele that is wanting a large audience but not wanting to kowtow to its low brow preferences is I believe similar to that of many other professional groups including architects doctors lawyers and professors in research universities What professionals want is to sell their customers the most sophisticated product they can whether the imaginative structures of elite architects the heroic scientific medicine of top doctors the hypercomplex legal instruments of corporate lawyers or the scientific research of university professors By sophisticated I mean products that are complex nonroutine and dependent on the special skill of the provider The reason that as suggest professionals want to offer products that are sophisticated 5 quotLetter to the Editorquot New York Times March 8 1994 7 quotSimpson Case Gives Cable An Edge on the Networks quot by Lawrie Mifflin p D1 New York Times February 20 1995 in this sense is that they can charge more money for them find them more interesting to work on and can more readily use them as vehicles for showing off to their peers Consider architecture If an architect had a choice between designing a nofrills quotboxquot or a building or instead an irregularly shaped subtly shaded and elaborately styled quotstructurequot of her own design which would she choose The latter of course since architects can get higher fees more intellectual satisfaction and greater peer recognition for producing the latter type of building The major constraint on this professional impulse is the consumer who might want quotjust a boxquot or at least something that costs what a box costs A primary difference between professionals and others kinds of business people is that professionals are to some extent free of market constraints They achieve this freedom by developing standards of what good professional work consists of socializing fellow professionals into accepting and applying these standards and educating the public to accept the standards To whatever extent they can professionals also seek institutional support for their standards whether in the form of favorable government regulation monopolistic control over work in theirjurisdiction or private sweetheart arrangements These professional standards may of course serve the socially useful purpose of limiting charlatanism and quackery They may also result in higher quality service than would be produced in a purely competitive market though any such judgment needs to be made on a casebycase basis But they also help professionals to do more lucrative and interesting work than they otherwise could ln economic terms professional standards constitute an attempt to create cartels in restraint of trade Or as George Bernard Shaw more colorfully put it quotevery profession is a conspiracy against the laityquot There is a venerable tradition of studies in the sociology of the professions that emphasizes these unsavory aspects of professional life eg Larson 1978 Yet professional cartels by which I mean control of a workjurisdiction by an exclusive group are difficult to maintain particularly under conditions of rapid social or technological change and they are doubly hard to maintain in the presence of free market competition Abbott 1988 Consider briefly again the case of architecture Because architects like journalists and many other professionals must deal with clients having lamentably unsophisticated quottastequot and because even clients who have been socialized into accepting architects39 notion of good taste may lackthe money to pay for it there is always a market for architects willing to forsake elite values by putting up nofrills buildings at low cost Thus within architecture and many other workjurisdictions there can be intense competition between higher and lower status providers Frequently moreover new groups rise to challenge old ones Thus as Abbott has described social workers challenge psychiatrists for control ofthe mental health jurisdiction Likewise accountants have taken over a large part of the business formerly done by lawyers Information technologists are displacing traditional librarians Solo practice physicians have lost ground to numerous groups from nurses to anesthesiologists and most recently to accountants Throughout the professional world there is a continuous jostling among service providers and resultant reshuffling of both work jurisdictions and the rewards that go with them Abbot 1988 The constant challenge for high status or elite professionals then is to develop sophisticated services to fend off competition by lower status and nonprofessional providers and to get the consuming public to accept their high status product Acceptance may be achieved through open market competition but more often it is achieved by restraining competition through professional codes of conduct and where possible legal protection All of this applies in a straightforward manner to political journalism Elite reporters would like to produce a highly sophisticated news product which in their case means a product rich in journalistic interpretation and critical analysis They want to do this because for reasons of pay status peer recognition and intellectual interest it is more personally rewarding to do so Thus journalists have an interest in creating and selling a form ofjournalism that offers more than stenographic transcription of what others have said or one that appeals to the lowest common denominator ofthe mass market What elite journalists want is a profession that adds something to the news a profession that not only reports but also selects frames investigates interprets and regulates the flow of political communication Whatjournalists add should be in their ideal as arresting and manifestly important as possible if possible the most important part of each news report so as to call attention to journalists and to the importance oftheir work Commenting in this vein on the rise of interpretive reporting in recent years Patterson 1996b writes 24 The interpretive style empowers journalists by giving them more control over the news message Whereas descriptive reporting is driven by the facts the interpretive form is driven by the theme around which the story is built As Paul Weaver notes facts become quotthe materials from which the chosen theme is illustratedquot The descriptive style casts the journalist in the role of a reporter The interpretive style requires the journalist to act also as an analyst The journalist is thus positioned to give shape to the news in a way that the descriptive style does not allow The interpretive style elevates the journalist39s voice above that of the news maker As the narrator the journalist is always at the center of the story p 102 The extent to which journalists can in practice get away with elevating themselves above the newsmakers they cover is limited since the news consuming public tends to be more interested in the newsmaker than in the news reporter Yet as description of the journalistic ideal Patterson39s observation is exactly right Summarizing my general argument in a form specific to journalism I propose as a cornerstone ofthe theory of media politics that Journalists aspire individually and collectively to maximize their independent and distinctive quotvoicequot in the news By quotvoicequot I mean any sort of distinctivelyjournalistic contribution whether it be hidden information analytic perspective or simply personality It is not necessary for my model to workthat every journalist have a realistic chance to become Bob Woodward or George Will or Sam Donaldson whose voices are renowned throughout the land It is enough that ordinary journalists find it materially and psychologically rewarding to express as much voice in the news as they can persuade their audiences to accept The drive forjournalistic voice is far from innocuous ln ways I will describe more fully below it leads journalists to adopt an adversarial stance toward others most notably politicians who venture onto their turf and who as already noted also wish to control the content ofthe news it leads them to create and emphasize distinctive news products over which they can maintain control and which affirm their status as being quotin chargequot of political communication and because so much political conflict now consists of what are in effect propagandistic battles for public opinion the desire for voice leadsjournalists to contest political parties for quotthe organization of political conflictquot By the organization of political conflict I mean the selection of issues and candidates for voter attention the criteria for so selecting and the kinds of appeals that are made to voters As I argue below reporters often end up selecting the same candidates and issues that party professionals select or would select but they also make a distinctively journalistic contribution to the process Like other professionals journalists would not describe their motivations in such selfinterested terms They would instead stress their commitment to supplying the hidden information and analytic perspectives necessary for ordinary citizens to understand what is really happening In their eyes their aggressive and increasingly interpretive styles of reporting serve to quotprotec quot their news audiences quotwho cannot gather their own newsquot from politicians and others who have quotaxes to grindquot and are trying to mislead the public3 But while such motives can lead to the same type of news product as the motive of maximizing voice it is tempting to interpret them as simply an ideological justification of the role they would like to play This justification has more than a little validity most successful ideologies do but its validity is not the main point forjournalists The main point is the sophisticated conception of journalism that it tends to legitimate Yet reporters are constrained in their desire to produce sophisticated product by the need to sell the product to a consuming public that has as noted relatively little interest in political news They are further constrained by their inability to restrict competition from lowbrow providers such as tabloids quothappy talkquot anchor personalities and talk radio And as we will see in a moment journalists must also contend with the challenge of an extraprofessional group politicians who would also like to control the content of mass communication Ross Perot39s brilliant use of interview programs like the Larry King Show and Today is only one of many indications ofthis challenge Hence when elite journalists like Rather and Brokaw complain about the decline of news standards they are in effect complaining about their inability to maintain control over their workjurisdiction What they would like is to return to the days when 3 The general thrust ofthis paragraph along with the particular words in quotations are adapted from an insightful discussion in Gans 1979 186 on the importance journalists attach to objectivity l have altered Gans39 meaning by claiming that journalists see the background and analytic perspectives they supply as serving the same function as objective information namely quotprotectingquot the public from deception 26 owing to the limited number of news outlets they could do so on the basis of what was in effect a professional cartel in restraint of trade Although elite journalists project an air of great dignity and cool selfconfidence their most important mass outlets top newspapers national news magazines and network news shows are all losing audience share In contrast local TV news and other forms of soft news are gaining market share Writing of network TV news New York Times media critic Walter Goodman has written Television news as your local anchor might put it is under fire The target is not the violence that is agitating viewers and politicians but a creeping tabloidization not only of local news which serious observers have never considered of much account but of national news too pride ofthe networks9 What is true of network TV news is to a lesser but still significant extent true as well for other mass outlets Elite journalism is under fire moreorless continuous fire from a mass audience that isn39t much interested in politics lowerstatusjournalists willing to meet the mass audience on its own level and politicians vying to control their own communication and increasingly adept at doing so Elite journalists are no patsies in this struggle and they certainly do not appear to be in danger of going the way of homeopathic healers mediums and other once successful but now defunct professional groups At the very least they will survive as niche providers in a few big city newspapers offpeak television hours PBS and various cable and smallcirculation venues But elite journalists are in a more precarious position than many outsiders realize and they know it BASIC CONFLICTS IN MEDIA POLITICS Let us then assume the existence of a citizenry with an interest in holding politicians accountable on the basis of minimal political involvement or attention to the news a journalistic profession with interests in attracting large audiences and expressing journalistic voice and politicians with an interest in building political support via communication that reaches citizens through the news media What follows from these assumptions 9 quot39Tabloid Charge Rocks Network Newsquot The New York Times February 13 1994 section 2 p 29 1994 What follows generally speaking is a great deal of tension and sometimes open conflict among the players The key actors have quite different interests and they frequently jostle with one another in the pursuit of them The three most basic conflicts may be identified as follows Conflict between the interests of journalists and citizens Journalists would like to produce a more sophisticated news product than many citizens wish to consume Conflict between the interests of politicians and journalists Politicians and journalists both have an occupational interest in controlling the content of the news Conflict between the interests of politicians and citizens The basic interest of citizens is to hold politicians accountable on the basis ofwhat the politicians have accomplished while in office or say they will accomplish if elected to office Depending however on their accomplishments in office or ability to deliver on their promises some politicians may have an interest in bamboozling the public In these and other ways media politics is rife with actual and potential conflicts between the major actors But it does not follow that any problem necessarily exists Perhaps for example politicians have an interest in bamboozling the public but are unable because ofjournalists interest in exposing them to do so Or perhaps it would be good for democracy ifjournalists were able to sell the public a little more news than rationally ignorant citizens really want to consume Before we reach any conclusions about whether the conflicts l have identified are helpful harmful or merely innocuous for democratic politics it is necessary to know how they play out in practice In the course ofthis book I argue that these conflicts play out in the form ofthree patterns of recurring behavior which I describe as behavioral rules The rules are The Rule of the Market or the tendency of market competition to force journalists to lower the overall quality and amount of political news The Rule of Anticipated Importance or the tendency of journalists to devote attention to occurrences in proportion to their anticipated importance in American politics The Rule of Product Substitution or the tendency of journalists to substitute their voice for that of politicians in deciding what s news The next chapter takes up the Rule ofthe Market Chapters 4 and 5 then develop the theoretical and empirical groundwork necessary for testing rules of anticipated importance and product substitution This testing occurs in Chapters 6 and 7 Finally Chapter 8 assesses the big question of how media politics helps or harms or otherwise affects the operation of democracy Chapter 3 THE RULE OF THE MARKET We saw in the last chapter that journalists have an ambivalent attitude toward the news audience On the one hand they wish to maximize the audience for news This is because larger audiences mean fatter paychecks more prestige and a greater stroke to the ego Yet l have not maintained that elite journalists are the humble servants ofthe mass audience wishing only to provide the public exactly what it wants Indeed this would be a violation of another premise on my argument which is that journalists want to provide a sophisticated type of news one that permits them to express voice My theoretical argument therefore is thatjournalists seekto exercise their prized voice within limits set by audience tastes This sort of tension l have argued is universal within the professions Every professional group wishes if possible to have as much business as possible Yet they typically wish to offer products that are more sophisticated than what the clientele wants A nearly universally feature of professions therefore is the attempt to insulate the professions work from market pressures What professionals want is a captive public one that will pay top value for their product without exercising much control over the nature of that product The present chapter looks more carefully at the effect of professional insulation and its opposite market competition The argument is that from religion to medicine to journalism the effect of professional insulation is to strengthen professional values and the effect of these values on the product offered for sale Conversely the effect of market competition is to erode professional values and their effect on product quality For example British TV news which has until recently enjoyed a state monopoly and still has a subsidy offers higher quality news than TV news in the United States where numerous providers compete for the news audience The US produces some highquality TV journalism but it is mainly on PBS where it is shielded from competition by a subsidy Meanwhile the lowest quality American TV news is produced in the most competitive news sector namely local television Moreover the very worst TV news is produced as we shall see in the local TV markets that are most competitive A comparison of major British and American newspapers is also telling In this domain the America media which still typically enjoy monopolies in their local markets seem to have the quality edge over media in Britain where the most important papers compete against one another in a national market The method of this chapter is to make as many such comparisons as possible between more and less competitive sectors ofthe news business Some of these comparisons are as will become apparent extremely soft in the sense that they depend on little more than impressionistic evidence Most however involve some sort of quantitative indicators And all run in the same direction For every set of cases in which I am able to make plausible comparisons higher levels of market competition are associated with lower levels of news quality The chapter begins with a brief look at two wellknown professions to which journalists may be usefully compared the clergy and the university professorate The next step is to develop fruitful concepts of news quality and news quality Finally I present empirical evidence of the relationship between news quality and market competition A LOOK AT Two PROFESSIONS If insulation from market pressure is what every profession strives for the professorate at American research universities must be considered one ofthe most successful professions in the world This group has managed to convince the public or at least the public s representatives that highquality education requires lifetime job security for professors tenure the freedom of professors to teach whatever they want academic freedom and the opportunity to do research Research as university professors like to say is the most important product they offer But although research professors with their captive clientele of students are to a large extent outside the market there is a great deal of competition among professors Most of this competition involves entry to the profession rather than advancement within it and essentially all ofthe competition is on terms on which research professors wish to compete namely the provision of top quality research Much university research is in such areas as health science and engineering and obviously has great value to society Thus a strong argument can be made that society will be better off if its top researchers or at least some of its top researchers are sheltered from market forces while they conduct basic research But whether the insulated life of university professors can be justified or not professorial life would be quite different if professors were more directly exposed to market forces Without much doubt there would be more demand for highquality teaching and less opportunity for research among other large changes The clergy are an instructive contrasting case Like other professions it has attempted to use government to restrict competition but with strikingly mixed success In a handful countries eg Sweden lsrael the clergy have been able to obtain what all professions aspire to obtain a state subsidy for their services lifetime job security and restrictions on the right of competitors to enter the field In these countries religion tends to be highly intellectualized as suits the tastes ofthe highly educated persons who offer religious service lannaccone 1995 At the same time church attendance in these countries tends to be relatively low since the appeal of heavily intellectualized religious doctrine seems to be limited In many other countries however there are no state subsidies for religion notable lack ofjob security and few if any restrictions on entry to the field with the result that clergy must compete for their clientele In these countries religion has become both emotional and popular The United States with its rigid separation of church and state scores of highly emotional religious creeds and unusually high levels of church attendance exemplifies this type of case The thesis that market competition tends to erode clerical control thereby making religion more emotional and more popular is as would be expected controversial within its academic community And indeed the thesis is far from proven But the general argument fits a number of important cases well and it has recently been extended by a political scientist studying the Catholic Church s response to revolutionary movements in Latin America Gill 1998 has shown that whether the Catholic religious establishments of this area embrace liberation theology thereby siding with the impoverished masses of their countries against the economic elite is determined by the degree to which they face competition from Protestant missionaries This obvious case of capitulation to market competition is not of course described as such by Church officials Rather it is justified in terms ofthe teachings of Jesus Christ especially the Sermon on the Mount which argue for solicitude for the poor But the pattern of acceptance and rejection of liberation theology by the Catholic clergy as determined by the level of Protestant competition suggests that another type of logic is at work The situation of the clergy is suggest analogous in important ways to the situation faced by many professionals including journalists The work product they would most like to provide a product always justified in terms of high cultural values will sell well enough under conditions of restricted competition so that is what clergy provide when they are insulated from market pressures But when for any reason competition increases the effect is to undermine the ability of professionals to provide what they consider a quality product As applied to journalism this perspective leads to the following specific expectations D1 All else equal journalists will be best able to produce highquality news when they are most insulated from competitive market pressures D1 is short for deductive inference 1 D2 lncreases and decreases in competitive pressure should be associated with increases and decreases in the quality of news D3 When a new news program successfully enters a previously noncompetitive market it will locate itself to the downmarket side ofthe existing entrant since the existing entrant will have been providing higher quality news than market competition can sustain D4 As professionals journalists should be expected never to lead and always to resist efforts to lower the informational content of news It does not seem to me that my argument yields any clear prediction with respect to the entertainment content of news Journalists know that they are always to some degree in a struggle to maximize audience share and there is no reason that other things being equal they should resist bright and lively reporting that is news that is entertaining It is only when entertainment drives out highquality news content that they should object1 DEFINITION OF CONCEPTS 1 This case seems to me comparable to that of teaching in universities It is no violation of academic values to offer entertaining lectures and many professors do indeed try to be entertaining Only when entertainment displaces intellectual content is there a violation of academic values 33 The first step in evaluating these hypotheses is to measure two key concepts news quality and market competition Neither concept is easy to measure and the former is hard even to define Further complicating the problem is that wish to make comparisons across a range oftimes places and types of media sometimes using my own data and sometimes relying on data collected by others The need therefore is for concepts that are easy to operationalize and adapt I begin with news quality Many authors have found it useful to distinguish between an information model ofjournalism and an entertainment model Of course good journalism must both inform and entertain but the balance may vary One element ofthe conception of qualityjournalism to be used in this paper is that it is primarily intended to provide information about the larger world The other key element is the content of news The news media provide information about a great variety of topics from the activities of government to stock prices to tips on how to pick highquality Cabernets Some of this content refers to matters of general social or political significance and is implicitly intended to help citizens in their role as democratic decisionmakers other information is intended primarily for purposes of entertainment or personal consumption My notion of news quality stresses the former Thus I define highquality news as information about matters of general political or social significance Other definitions of quality are certainly possible For example journalists might make huge expenditures of energy and enterprise to find out and report how Princess Diana spent the last day of her life The same can be true of reporting on a more significant subject such as whether red wine has special powers to prevent heart attacks as 60 Minutes reported probably incorrectly Such stories might therefore be considered highquality journalism However it is precisely the rise of such news reporting including the several varieties of news you can use that many descry as evidence of the decline of news quality Conversely it is the kind of news l have described as quality news that the critics of contemporaryjournalism would like to see increase A more important reason for favoring my conception of news quality is that it describes the kind of journalism that commands the greatest prestige within the journalistic profession itself A perusal of journalism textbooks which stress public affairs reporting of the biographies of famous journalists which never fail to stress the extent to which the ego has done stories of great general significance and indeed the kinds ofjournalists who become famous all support the notion thatjournalists value the reporting of public affairs information more highly than anything else Why it is important thatjournalists value the kind of news l have defined as quality news will be explained in the next section I turn now to competition News is offered to the public in three main formats print radio and television Within each format there are different kinds of offerings In the domain of TV for example there are network news shows local news shows news magazines and morning magazine shows To some extent different kinds of news programs appeal to different market niches which means that they do not directly compete But from inspection of media as different as the New York Times and the New York Daily News it is apparent that the boundaries between market niches are vague and permeable since both papers aspire in their own ways to be fullservice news providers The Times did not for example fail to cover Princess Diana s funeral nor is it above putting sports news on its front page the Daily News for its part does not fail to cover wars elections and even certain acts of Congress One must therefore assume that aH news programs that offer their product within the same geographical market are to some degree in competition with one another such that gains by one tend to reduce the market share of others Competition becomes more intense as Two or more news providers focus on the same general type of news eg local news Two or more news providers compete in the same medium such as print or television Two or more news providers offer their product in the same time slot On the other hand competition becomes less intense when one or more news programs receives any sort of subsidy whose effect is to free it from the need to win audience share through market competition Altogether then news competition may be defined as the extent to which two or more news providers offer the same kind of news product to the same audience in the same format at the same time By this accounting competition is especially intense for American local TV news since several programs offer the same news product at the same time in the same medium Competition between local TV news and network TV news is somewhat less intense but still significant since both types of programs use the same medium at the same time Competition is perhaps least intense between newspapers and TV since they use different media and different time periods It may nonetheless be the case that the greatest competitive threat to newspapers comes from the morning news magazines which offer the same type of general news product at the same time as most newspapers are delivered EVIDENCE CONCERNING THE EFFECTS OF NEWS COMPETITION The method ofthis chapter is to make as many headtohead comparisons of news quality as possible between more and less competitive sectors ofthe news business Several kinds of comparisons will be made Overtime comparisons of the same type of news outlet with itself as competitive conditions change Comparisons of different types of news outlets with each other in the same market when competitive conditions differ Comparisons of British and American media The quality of this evidence as acknowledged earlier varies from rigorously quantitative to merely impressionistic Future versions ofthe chapter will I hope bring all of the evidence up to a common high standard 1 Local television news 1960 to the present The earliest local TV news shows were often staffed by experienced print journalists bearing the traditional news values of their profession The programs they produced were immediately popular and profitable and they grew rapidly Figure 1 illustrates the trend for three cities The data show trends for all news programs including network news and news magazines but the bulk ofthe overtime increase is due to local news broadcasts INSERT FIGURE 31 ABOUT HERE The quality of local news offered on local news shows declined in this period In the early 1970s these shows featured information about city government schools and state government but by the late 1980s and 1990s local TV news shows became heavily laden with stories about crime natural disasters and other episodic matters Many news departments no longer even employ city hall beat reporters or state capital bureaus McManus 1994 These trends have been widely noted in the popular press but I have found only one quantitative study It is however of high quality In 1976 a scholar did a content analysis of 10 local news programs in four Pennsylvania media markets Adams 1978 to determine the amount of hard news Sixteen years later Slattery and Hakanen 1994 returned to the same stations and applied the same coding scheme to Figure 31 Hours ofNews Broadcasting in Chicago New Orleans and St Louis 1960 1970 1980 1990 2000 the same programs They found that the percentage of news about government education and politics had fallen from 54 percent in 1976 to 15 percent in 1992 At the same time the percentage of news rated as sensationalistic or human interest rose from 25 percent to 48 percent By my definition of news quality this is evidence of decline and hence support for D2 Moreover Slatter and Hakanen argued that embedded sensationalism within stories categorized as government news had led to an underestimate of the amount of sensationalism in current news and hence an underestimate of the amount of change that had occurred in the 16year gap between studies Why the change By all accounts the pressure for change has emanated not from reporters but from upper managements that sought higher audience ratings and profits In his study of local TV news in MarketDriven Journalism McManus 1994 interviewed one news director who told him bluntly that he had learned to think with a cash register in my head As McManus continued this station manager refused to permit my access to the station arguing that he did not want his reporters to think about news values or journalism while gathering stories Instead he wanted them to think about ratings He instructed his reporters to imagine that he was placing a certain number of viewers in their hands at the beginning of their story he explained and he wanted them back at the end 168 Reporters and even some managers disliked doing this sort of marketdriven journalism but felt they had no choice When McManus challenged his interview subjects for abandoning journalistic values The most common response of those interviewed was something like I know you re right But I hate to think of myself that way 168 In a study of a Florida TV station that sought to offer a fare of more serious news journalist Michael Winerip reported that when the policy of highquality news was announced the newsroom erupted in applause Kathy Marsh a reporter who under the previous regime was assigned to do a special report on penis and bust enhancers leaped out of her seat clapping Dan Billow who covers the space program at nearby Cape Canaveral said It s like we re having honor restored to our occupation Wineripi 1997 p 31 Winerip found however that the policy of highquality news was losing out in the ratings war to two stations that emphasized crime violence and bizarre occurrences The thing is people like it Winerip wrote making it clear that the new station manager had an uphill fight to turn around his ratings before losing his job This evidence even if accepted as applicable to the universe of local TV news programs which it probably is is only equivocal support for my claims It is after all possible that owners might seek to maximize ratings and profits even in the absence of competition and that it is only a coincidence that they discovered a news formula for doing so at the same time that competitive pressures were increasing I shall address this issue later on For the moment my claim is only that the evidence concerning local TV news which involves tandem overtime changes in competitiveness and news quality is consistent with my general argument as embodied in D2 I note that the qualitative evidence of lack of enthusiasm for low quality news among journalists is also consistent with my argument 2 Newspapers in comparison with local TV news Although as noted above most local TV news programs face heavy competition typically several shows offering the same product in the same time slots in the same medium most newspapers have it much easier The large majority of American newspapers have enjoyed monopolies in their local markets for decades Except in New York where there are three citywide daily papers no city in America has more than two citywide papers and a large majority has only one These days the most serious competition for newspapers comes not from other newspapers but from television It is often observed probably correctly that evening TV news destroyed afternoon newspapers and it now appears that morning TV news programs which offer a mixture of local and national news including local traffic and weather reports are harming newspapers which occupy the same time slot Still since newspapers and TV news are different media rate the competition between them as moderate rather than intense What then is the quality of American newspapers compared to local TV news In the article cited above Winerip writes Most anyone in the press and academia who has given much thought has concluded that while there are exceptions local television news is atrocious p 33 I believe that this is an accurate statement and further that no one would make such a sweeping statement about American newspapers There is some evidence to be reviewed immediately below that the quality of newspapers has slipped but I have found no blanket claims that newspapers are anything like atrocious My impression for which there is evidence below is that most newspapers produce the highest quality journalism they can under the resource constraints they face I take this as tending to support D1 The largely impressionistic evidence that local newspapers typically offer a higher quality form of journalism than local TV programs is consistent with but again not strong evidence for my claim that higher levels of competition are associated with lower levels of news quality 3 American newspapers 1950 to the present The increase in local TV news programming counts by my conception of competition as a moderate increase in the competitive pressure on newspapers At the same time newspaper circulation has begun to lose readership most likely because ofthe competition from television From 1970 to 1988 total circulation of American newspapers increased but at a rate less than the general increase in the population Bogart 1992 p 87 In a sample of 67 newspapers gathered for another purpose see below I found that circulation has recently begun to decline in absolute terms Between 1990 and 1998 daily circulation of my newspaper sample fell about 6 percent in the market area of sampled papers Newspapers that ceased to exist between 1990 and 1998 within the markets of sampled papers are counted within the denominator on which this change in circulation has been calculated if failed newspapers are omitted from the calculation the decline is only 2 percent l have found no studies oftrends in newspaper quality during the period 1950 to the present There are however two indications of what appears to be a small but significant declines in newspaper quality Bogart 1992 p 89 comments without presenting any data that there has been more emphasis on features relative to news in newspapers in the 1980s In the same vein Diamond 1994 in his book on the New York Times describes how the gray lady of the 1950s and 60s was forced by competitive pressure to brighten its writing add more human interest features and even big color pictures to its news pages In pursuit of circulation as persons interviewed by Diamond said the Max Frankelled Times was willing to get down and scratch for the same kind of dirt that in the past it left to the city s rude tabloids p 9 In his chapter on Soft Times Diamond writes It was the summer of 1978 The Times was introducing its new sections covering such topics as food furniture and design These daily magazines one for each weekday represented the Times39 prime editorial initiative of the 1970s They signaled a major investment of both money and staff the centerpiece of the effort to attract new readers Americans were spending an ever increasing amount of time in front of their television sets They were getting the first hard reports of developments in Washington Wall Street or the Middle East from network news on the nights before their morning papers were delivered Attracted by nightly television and the early morning shows like quotTodayquot on NBCand later by quotnew mediaquot networks like the twentyfour hour CNNthe traditional audiences for news seemed to be drifting away from their newspapers Around the country editors tried new formats to lure readers The Washington Post the Los Angees Times and the Miami Herald had all taken the lead in developing sections devoted to quotlife stylequot features p 84 l undertook a few small test bores to see if Diamond s general observations could be supported by quantitative measurement In particular I asked a research assistant to rate a sample of front page stories on a fivepoint scale where 5 represented information useful readers in their roles as citizen decisionmakers and 1 represented information that was useful to readers purely as private individuals such a information about health or entertainment The stories were from one week of Times front pages in 1970 1980 and 1999 The results showed a decline that was highly statistically significant but substantively rather modest The mean level of news quality on the fivepoint scale fell from 496 in 1970 to 484 in 1980 to 444 in 1999 However it was perhaps notable that the number of frontpage stories in the sample weeks fell from 79 in 1970 to 49 in 1980 to 45 in 1999 In expanding fontsize and increasing picture content the Times also cut down on the number of stories it could place on the front page It would be quite easy to extend these measurements to a larger sample of papers and I expect to do so In the meantime I take the scattering of available evidence in conjunction with my own strong impression that newspapers have in fact gone somewhat soft as additional support for the general argument ofthe effect of competition as embodied in D2 4 Market size and competitive pressure Consider the case of TV news in which production costs are nearly constant with respect to audience size once a big antenna has been purchased but advertising revenue increases with audience size If lowquality news is no more expensive to produce than high quality news and has more audience appeal then every incentive is for owners to move downmarket except one Downmarket news is less prestigious than high brow news and some owners value their reputations as producers of quality news Now imagine two stations one in a market of 500000 viewers and one in a market of 18 million viewers which is about the size of the New York city TV market Suppose that by watering down news quality a program could attract an additional one percent of viewers in each market In the small market this would amount to 5000 additional viewers whereas in the big market it would mean 180000 additional viewers On the assumption that advertising revenues are a linear function of audience size the pressure to go after the additional one percent of audience share would be vastly greater in the big market The expectation therefore is that pressures to abandon journalistic values in pursuit of larger audience share would be greater in bigger markets There is however an opposing logic Bigger markets can mean more advertising revenue per hour of news programming which could translate into more resources for the production of each hour of news which could lead to more reporters more indepth reporting and generally higher quality news The question then is whether the additional revenue generated in larger markets will be captured by owners in the form of higher profits or byjournalists in the form of support for higher quality news Figure 2 showing the relationship between audience size and news quality for newspapers and TV news attempts to answer this question The measure of newspaper quality in Figure 2 is the number of Pulitzer prizes won in the period 1977 to 1998 The measure of TV news quality is the news quality index produced by The Project for Excellence in Journalism an affiliate of the Columbia School of Journalism The TV news index seeks to measure depth of reporting balance of viewpoints within stories number of story sources the expertise ofthe sources and the degree of local relevance INSERT FIGURE 32 ABOUT HERE As can be seen the data seem to support both arguments and hence neither Bigger markets tend to produce better newspapers but worse TV news Why might this be I suggest that the differences in competitive pressures may explain difference in quality As Figure 3 shows there is vastly more competition in the provision of local TV news than of local printjournalism Moreover this differential in competitive pressure increases with market size The effect of this may be as follows In the absence of competition no one knows what might sell best in a market or what kinds of profits can be made Owners may come to feel at the urging oftheirjournalistic staff that they are getting most of the profit that can be squeezed out of the market by offering highquality journalism since they will never confront any painful evidence to the contrary Put somewhat differently newspapers can afford to offer a noncompetitive product even in large markets because there is no competitor to punish them for doing so But in a large Figure 32 News quality and market size for newspapers and TVnews Average news quality scoresfor local TV news markets by media market size in 20 selected markets 420 7 Emsvllle o Tallahasee a 0 Louisville 380 7 340 7 300 7 260 7 o Pittsburg 220 500000 1 million 3 million 10 million 20 million Media market size in logged millions Note News quality was mm sured by a resmrch tmm spon sored by Project for Excellence in Journalism an affiliate of the Columbia University School of Journalism Data in gure are average of scores for network stations in each of 20 media markets The news quality scores are published in Rosenstiel et al 1998 Number afrecent Pulitzerprizes by circulation for a non sample ofdaily newspapers 507 Washington Post 0 0 200000 400000 600000 800000 Daily circulation 1000000 1200000 Note The non random sample of newspapers was drawn from those available on LexisNexis The sampling criterion was to pick all of the major papers in LexusNexus eg The New York Times Los Angeles Times plus an arbitrary selection of the rest Total number of cases is 67 Circulation gures are for daily editions and come from the 1998 edition ofEditor and Publisher Pulitizer prizes are for the period 1977 to 1998 market that does have competition the less successful programs will become aware of their deficiencies in the form of low competitive ratings and will be sorely tempted to break ranks with journalistic orthodoxy When as has tended to happen in TV news this leads to higher audience share it sets off a downmarket spiral that eventually forces even highminded owners to abandon their journalistic scruples INSERT FIGURE 33 ABOUT HERE Yet as Figure 3 shows there are some large markets in which two or in the case of New York three newspapers compete lfthe previous TVdirected argument is correct it ought to be the case that some big city newspapers in particular those with competitors are induced by competitive pressure to go downmarket l have used a sample of daily newspapers on the LexisNexis information service to test this proposition To measure news quality I calculated a Lewinsky quotient that is the ratio of front page stories about the Lewinksy scandal to front page stories on other serious news topics namely stories about Bosnia the federal budget deficit and problems in Social Security and Medicare Although coverage of Lewinsky was often serious a high Lewinsky quotient all Monica all the time would be indicative of commitment to entertainment rather than to highquality journalism lnitial examination of the data supported my theoretical expectation On a simple difference of means test the Lewinsky ratio was 59 percent larger in cities in which there were competing dailies p 02 onetailed on assumption of unequal variance However an examination of a scatterplot ofthe data as shown in Figure 4 undermined this result In this figure solid black dots indicate newspapers with a city wide competitor in the same market and clear white dots indicate a newspaper monopoly As can be seen the only cases in which competition led to markedly more Lewinsky coverage were cases in which one of the competitors was a tabloid For the majority of cases in which the competitors were not tabloids there was no effect of competition Thus in a regression which controlled for circulation and the presence of a tabloid the tabloid dummy had a large and statistically significant coeffieint but the competition dummy had essentially no effect INSERT FIGURE 34 ABOUT HERE Although the regression results do not support my argument they do not really damage it either After all the highest Lewinsky ratio occurs as expected in competitive markets What appears to Figure 33 The effect of market size on level of competition in selected newspaper and TVnews markets Number ofnewspapers in 20 selected media markets Number afTVnews programs andpragram hours in 20 selected media markets 50 50 Number ofhours of daily 45 45 programming 40 40 g o 35 35 0 30 30 25 25 20 20 15 15 0 Numbers of 0 0 news programs 10 10 5 LA 1 NY 3 5 o 1 1 1 1 1 o 1 0 5000000 10000000 15000000 20000000 0 5000000 10000000 15000000 20000000 Media market size Media market size Figure 34 The quotLewinsky quotient for a sample of US newspapers Ratio of Lewinsky story count to story count for stories about Serbian crisis federal bud get de cit and social security issues 600 7 500 7 400 7 300 7 200 7 100 7 000 0 NY Daily News Boston Herald 0 Chicago NYPost SunTimes ArizonaRepublic o oO oo o 0 Chicago Trib USA Today I 0 Q9 g d sash LA Times 650 g Boston P051 Q39Ny39rimes O m Globe 0 250000 500000 750000 1000000 1250000 1500000 19 9 8 Circulation happen is that in smalltomedium sized newspaper markets newspaper competition has no effect The papers probably do compete but they compete within the standard journalistic paradigm trying to outdo each other on hard hitting reporting rather than on lurid crime and sex news But as market size becomes large there is an increase in the chance that one ofthe competitors will adapt a tabloid strategy2 Thus exactly as in the case of TVjournalism competition alone is not associated with the abandonment ofjournalistic standards rather it is the combination of large market size and competition that has the critical effect The main difference between TV and print appears to be that due to lower overall levels of competition in newspaper markets the necessary combination of market and competitive pressure occurs less often for newspapers This argument has two notable implications Highquality big city newspapers such as the New York Times Los Angees Times and Chicago Tribune could probably increase circulation by repositioning themselves down market Doing so would cost them dearly in terms of national prestige but would probably increase profitability That these and other highquality papers generally resist downmarket pressure is probably best explained by an idealistic commitment either to prestige or to quality journalism or to some combination of both The opportunity for profit does not translate into low quality journalism except in the presence of intense competition as in the case of local television and a few large newspaper markets In the absence of competition journalists seem to be able to persuade owners to cast their fates with respectable highquality news In a general way these results are consistent with the basic Rule of the Market That increases in market competition lead to lower levels of news quality I note in passing that another interesting result of this analysis Whether a newspaper was owned by a chain in general or by the oftenvilified Gannett chain in particular had no effect on the Lewinsky quotient It is also interesting that smaller papers did not in general have a higher Lewinsky quotient I take this as evidence that apart from the resource constraints that prevent small papers from competing effectively for Pulitzer prizes they aspire to produce highquality news 2 Total newspaper circulation has a substantively large and highly statistically significant effect on the probability 5 American newspapers 1900 to 1950 l have already presented evidence that news quality declines as competitive pressures increase The argument that the latter causes the former would be stronger if I could run the argument in the opposite direction if that is I could find a case in which news quality increased as competitive pressure decreased I believe there is one such case Between about 1900 and 1950 competitive pressure within the newspaper business fell markedly Although the evidence is less compelling as regards quality there appears to have been an improvement in news quality during this same period Let us lookfirst at the evidence of competitive pressure According to data compiled by Emery and Emery 1996 the percentage of daily newspapers having no competitor was nearly flat from 1880 to 1920 but shot up suddenly in the decade of the 1920s from about 40 percent to about 80 percent This is a remarkably sudden consolidation and all the more so since it occurred in a time of general prosperity and generally high newspaper profits Mott 1962 p 593 Moreover although the Emery and Emery data do not continue beyond 1930 newspaper consolidation obviously continued such that by 1950 the vast majority of newspapers had no competitors or at most one Different authors give different reasons for the great newspaper shakeout My hunch supported by some initial compilation of data is that newspapers were in effect redefined in the two decades prior to consolidation In 1880 the typical big city newspaper was four to eight pages long mixed different kinds of news in haphazard fashion and had three or four and sometimes as many as 10 or 12 competitors But about that time newspapers began to grow in size and depth adding more pages separate sections on sports business culture and fashion and in general becoming more like the multifaceted cultural fountainheads of today But cities that could support three or four ofthe old fourpagers could not for long support so many multipurpose papers This led eventually to the somewhat odd situation in which many newspapers were being driven out of business but those that remained were larger and more profitable than ever In any case the low point for competitive newspaper pressure was probably the 1920s Profits were high consolidation was proceeding rapidly and a key point there was not yet significant competition from radio In most cities one or two newspapers ruled the news market that a tabloid competitor will emerge The 1920s were also the time of one ofthe most important reform movements in the history of journalism The decades in which the pressure for consolidation was building and in which therefore competition was probably at its peak were the time of yellow muckraking and generally sensationalist journalism But beginning in about 1920 a reaction set in in the form of a movement for quotobjectivequot news reporting In its original sense writes Streckfess objectivity meant finding the truth through the rigorous method of the scientist 1990 p 975 In Schudson s 1978 account objectivity was likewise a more rigorous reporting method According to Walter Lippmann who is credited by Streckfess and Schudson with leading the movement for objectivity a central purpose ofthe new creed was to make journalism less a romantic art and more an application of trained intelligence cited in Streckfess p 981 Although these conceptions of objectivity are different from my notion of highquality news they are clearly an attempt to increase the information content of news in relation to its entertainment value The key question then is how the objectivity movement affected actual news content According to a careful content analysis of six bigcity newspapers over the period 1865 to 1955 Stensaas 1986 certain elements of objective journalism pre dated the objectivity movement ofthe 1920s Stensaas measures objectivity as the extent to which stories make assertions that are strictly observable or verifiable p 13 Thus to be counted as objective Stensaas required on one hand that stories avoid statements ofthe writer s opinion and on the other that stories link assertions of fact either to concrete events or to statements by particular sources Stensaas s findings are shown in Figure 5 along with the Emery and Emery data on decline in newspaper competitiveness As can be seen the decline in competitiveness is not welltimed for explaining the increase in objectivity as measured by Stensaas Stensaas data thus fail to support my theoretical expectations INSERT FIGURE 35 ABOUT HERE Yet this is not the end of the story For one thing Stensaas data refer to six newspapers that survived well into the 20m century and may for that reason alone be atypical ofthe newspaper universe by contrast the Emery and Emery data refer to the universe of American daily newspapers most of which were small The two time series thus refer to rather different populations For another contemporary Figure 35 Certain trends in news competition and news quality 1865 1955 Percent of all daily newspapers having no daily competitor 20 0 i 1860 1880 1900 1920 1940 1960 From The American Press An Interpretive History Michael Emery and Edwin Emery 1996 p 293 Percent of sample of Qty daily newspaper stories rated as objective 100 186574 188594 190514 192534 194554 From The Objective News Report A Content Analysis 0 Selected Us Daily Newspapers for 1865 to 1954 Harlan S Stensms unpub lished doctoral dis ertation 1986 University of Southern Missisippi p 57 observers felt that the objectivity movement did made a difference Writing in the Yale Review in 1931 Lippmann said it had brought about a revolution in news writing over the previous decade The most impressive event of the last decade in the history of newspapers has been the demonstration that the objective orderly and comprehensive presentation of news is a far more successful type of journalism today than the dramatic disorderly episodic type3 This assertion merits further study which I hope soon to supply In the meantime it remains clear that sometime between 1900 a time of famously sensational journalism and about 1950 American newspapers toned down in terms of sensationalism and toned up in terms of quality During this same interval the level of competition within the newspaper business also fell dramatically This could be a coincidence but it seems more likely that the decline in competition created the breathing space in which the new duller style of newspapers could establish itself 6 Network TV news 1969 to 1997 We have seen in Figure 1 that local TV news much of which goes headtohead with network news in the same time slots has grown dramatically and that the most dramatic growth has come in the last 10 years My theoretical expectation therefore is that network news will have declined in quality in the last decade In a small test bore into the data asked a research assistant to use the Vanderbilt TV News Abstracts to assess the percentage of the network news broadcasts devoted to serious coverage of national government and foreign affairs The results based on the analysis ofthree months of programming in each ofthree years are shown below 3 Cited in Streckfess p 981 Percent of Network News Devoted to Stories about Government and Foreign Affairs 1 69 1 81 1997 ABC 62 57 42 CBS 55 60 40 NBC 58 59 27 Average 58 59 36 As can be seen there has been a marked decline in news quality with all ofthe decline occurring after 1981 The timing of the change thus accords well with the increase in competitive pressure though obviously more empirical work is needed to tie down the timing question Ken Auletta s book on the Big Three networks in the 1980s Three Blind Mice points out that more than competitive pressure affected changes in network news content Each of the networks was taken over by a new owner in the 1980s who was more concerned with profit than previous ones Thus the drive for profits can be considered an alternative explanation for the decline in news quality in this case Yet as Auletta also shows the rise of new TV networks including CNN and Fox and cable programming occurred at the same general time as the ownership changes In the absence of other evidence we might therefore have to throw up our hands and say that we have no way of telling which factor is important But inasmuch as news quality seems to have fallen wherever competition has intensified and even in media such as the New York Times in which no ownership change has occurred we can feel reasonably confident in ascribing causal importance to competitive pressure The key factor has been welldescribed by NBC Anchor Tom Brokaw quotWhen I started out in the 196039squot he said in an interview quotthere were effectively two network news programs and at 630 PM people turned on either Huntley Brinkley or Walter Cronkite and got their news for the day And I39d like to have that back againquot4 Perhaps the most interesting feature of Auletta s book was his description of how network news journalists had managed to convince owners that their news shows should not have to make a profit as they generally failed to do in the 1970s Their argument was that news had such special importance in a democracy that it ought to be exempted from the need to produce profits They added however that their news programs were flagship operations whose prestige value did help the profitability of other network operations These are exactly the kind of arguments that professionals of all kinds make when they want to get themselves exempted from market pressures But when the new owners of the 1980s refused any longer to accept them the journalists at first resisted then squealed then capitulated as my general argument would suggest I take this as evidence consistent with D4 7 Network news magazines 1970 to present The news magazine 60 Minutes began broadcasting in the late 1960s and for some 15 years maintained a welldeserved reputation for highquality television journalism From its prime time spot on Sunday night it was also able to attract consistently large audiences But about 1980 its success began to attract competitors ABC s 2020 and NBC Magazine The competition moved up another notch around 1990 when Primetime Live 48 Hours Turning Point Fox Files and EyetoEye began to appear Thus 60 Minutes has had at least two competitors since about 1980 and a halfdozen since 1990 It is possible to estimate the news quality ofthese various shows by examining TV Guide listings in newspapers which often carry a description of story content For example Discussion of the Shah of Iran s Secret Police and President Nixon s Vietnam Troop Withdrawal Plan were listed as stories and rated as 5 on the news quality scale described earlier 5 represents information useful to readers in their roles as citizen decisionmakers while 1 represents information that is useful to readers purely as private individuals An interview with Elizabeth Taylor was rated as The results obtained from this exercise are shown in Figure 6 and support my argument in two ways 1 The new entrants to a previously noncompetitive field came in at distinctly lower levels of news quality consistent with D3 and 2 60 Minutes initially resisted but was eventually forced to go downmarket to meet the competition consistent with D2 Note that the biggest decline in 60 Minutes news quality occurs only after 1994 at which point the level of competition had grown quite heavy INSERT FIGURE 36 ABOUT HERE These results which are based on the coding of 173 stories from 60 Minutes and 229 stories by its various competitors cast the original 60 Minutes show in a very favorable light In the decade or so in 4 quot Simpson Case Gives Cable An Edge on the Networks quot by Lawrie Mif in p D1 New Y ark Times February 20 Figure 36 Trends in news quality among network news magazine shows 5 4 News Quality Scores 3 7 2 60 Minutes n173 Other news magazines n229 1 1968 1978 1988 1998 which 60 Minutes had no direct competition the average quality score for its stories was about 38 on the same scale on which The New York Times front page stories now average 44 This is a very respectable rating for a TV show attempting to reach a mass audience but also very difficult to sustain in the face of heavy competition I note that all ofthese numbers are preliminary and that further coding may produce changes 8 Comparison of British and American TV News Since the BBC had a monopoly on national TV news coverage for some two decades and still retains a substantial subsidy the expectation is that it would produce higher quality news than American network news l have found no relevant data comparing British and American media and have not yet had time to collect any However Semetko et al 1991 do make a systematic comparison of NBC coverage ofthe 1984 election with BBC coverage of the 1987 parliamentary election Most oftheir evidence much of it quantitative has no bearing on my argument but their report makes it clear that the BBC provided more news with more information about the candidates positions on issues than their American counterpart which makes the BBC news higher Thin as this evidence is there can be little doubt that the BBC provides higherquality news than its American counterparts as expected by D1 9 Comparison of British and American Newspapers The Semetko study also compares American and British newspaper coverage of the two elections but again most ofthe evidence is irrelevant to the concerns of my chapter However two pieces are relevant The study examines the length of the average campaign story which can be taken as an indicator of depth of coverage And it also provides counts of the number of straight news versus feature stories which can be taken as an indicator of concern for information versus entertainment The comparisons are made for three classes of papers 1 Typical American newspapers Indianapolis Star and Louisville Courier Journal 2 British broadsheets Times and Guardian and British tabs Sun Daily Mirror Daily Mail The two American newspapers compete in local markets 1995 which they monopolize while all British papers compete in a frenzied national market The expectation therefore is that the American newspapers will provide higherquality news in the sense I have used this term At first glance they do seem to The two American papers ran stories that were slightly longer at 203 inches than the British broadsheets 194 inches and much longer than those in the tabs 89 inches5 And the American papers ran a much higher ratio of straight election news to features news than either of the British groups see Table 83 It would be foolish to maintain that two small American papers provide coverage that is in any general way better than that of the elite British press At the same time the data in the previous paragraph may well indicate that the British papers are being forced to work harder to hold market share than the monopolist American papers Again D1 seems to be supported 10 Competition for the BBC After two decades of BBC monopoly two commercial news programs have been introduced in Britain one in the 1980s and one in the 1990s This leads to two clear theoretical expectations That the new entrants should provide lower quality news than the BBC and that the presence of downmarket competition should pressure the BBC to go downmarket as well In a series of publications Jay Blumler and Michael Gurevitch make clear that the first expectation is supported fn The two new stations one of which has a state subsidy provide less election coverage and more horse race entertainment than the BBC though as Blumler and Gurevitch emphasize the new programs are still well above the quality level of American network news As they comment Britain s approach to its 1997 campaign was still largely sheltered from those building commercial and competitive pressures that were so much more rampant in the US system p 19 Americanization Reconsidered UKUS Campaign Communication Comparisons Across Time ND The second expectation is more difficult to evaluate On the one hand the two new stations have certainly brought pressure on the BBC to change But on the other hand the BBC has not changed much if at all Blumler and Gurevitch quote a news reader who says There has been a growing perception fed by market research that many people are not digesting political reporting from the BBC as well as 5 See Table 82 p 152 they might It was not as accessible as it should be Change in the Air Campaign Journalism at the BBC in 1997 p 20 ND Yet BBC programming remained as lengthy and issueoriented as ever Blumler and Gurevitch write An editor even spoke of the fantastic luxury he enjoyed since Nobody would say that was a jolly interesting programme but the ratings went down And when evidence of a significant drop in the audience became available the newspeople tended to rationalize arguing that they were not going to be deterred from doing their duty as public service broadcasters In a producer s words We are relaxed about it We have been told from the top that the BBC has a duty to do this ibid p 11 The attitude seemed to be the same in the 1993 election when one producer said that the BBC was prepared to test viewers boredom thresholds to do justice to the campaign 1995 Crisis of Public Communication p 169 l interpret these remarks as evidence that the BBC is in fact under pressure to go downmarket to meet its new competition but has so far been successful in resisting the pressure In the classic manner of professionals BBCjournalists are prepared to be as boring as they have to be to uphold the standards of their profession and class so long as the subsidy holds out Thus the evidence from the effect of competition on the BBC supports my D4 that journalists will resist competitive pressure to dumb down the news but not my D2 that competition will lower news quality However it seems too early to take the disconfirmation of D2 in this case very seriously CONCLUDING REMARKS A central tension of media politics is between journalists who wish to produce a sophisticated news product and ordinary citizens who want something much less sophisticated The best evidence ofthe kind of productjournalists would like to produce comes from markets in which they face relatively little competitive pressure to cater to mass tastes In these markets modern American newspapers TV network TV news in the 1960s and British television we find a relatively highquality news product and a determination to keep it so But from markets in which competition is greater especially local TV news we find a lower quality news product that is one must assume closer to what mass tastes in news actually are The most significant shortcoming of this analysis is its failure to consider how news organizations may attempt to build niche markets that afford a degree of insulation from unfettered competition The essential idea in nichemarketing is to develop a product that appeals to a particular wellbounded segment ofthe market in appealing to such a clientele one is protected from competition with other more general service providers An example of successful niche markets is Spanishlanguage news broadcasting which has developed in many American cities The difficulty in developing niche markets for news is that there are few market boundaries that are as impermeable as the language boundary that protects Spanishlanguage broadcasting What seems to have happened in the American news business is that news organizations that enjoy any degree of insulation from market pressure such as monopolist newspapers still enjoys or network TV news for a brief time enjoyed have become a de facto market niche for relatively highquality news What impresses me however is how very small the potential niche for highquality news in the US seems to be The US has two highquality national newspapers the Wall Street Journal and the New York Times each widely available in the US Together their circulation is about 3 million The audience for Lehrer News Hour widely though not universally available is currently about 1 million and falling And the audience for National Public Radio which is almost universally available is 77 million daily Given this total of roughly 12 million consumers of elite news given that the adult population of the US is about 230 million and given that elite news services are available in a very large fraction of the nation the niche audience for serious news in the US is trivially small probably only five or so percent of the potential market The rest ofthe audience is up for grabs under conditions of moreorless general competition conditions that as l have argued prevent journalists from offering as much serious news as they would like It is interesting to ponder exactly how much freedom mostjournalists enjoy to depart from what a perfectly competitive market might force them to offer the public My guess is that the typical print journalist enjoys more discretion than most car manufacturers who would like to load their products down with expensive gizmos but are greatly constrained by component pricing and cutthroat competition but less than what the most successful professions such as research professors and lawyers have been able to manage6 I would guess however that TVjournalists have little if any freedom to depart from what market pressures demand Whatever exactly journalists do manage do get away with it is a good bet that like other professional groups they are constantly trying to upgrade the sophistication of their professional product for the basic selfinterested reason that sophisticated products return greater pay social status and intellectual satisfaction 6 Many members of my own professional group political scientists are acutely aware that the public supports their research activities through its taxes and tuition payments at higher levels than it would if given a direct choice 53 Chapter 4 Struggle to Control the News Except when they enjoy a state monopoly on the news journalists are under constant audience pressure to keep the news short simple and dumbeddown The technologyinduced explosion of competition within the news business in recent decades has as we saw in the last chapter intensified that pressure Highquality professional journalism though not yet an endangered species is on the defensive across Europe and the United States as businessminded news programs increasingly cater to audience tastes Nor is the market the only source of pressure on journalists Politicians are increasingly adept and aggressive at managing the news that is staging news events that constrain journalists to report the words and pictures that the politicians wish to have conveyed to the mass audience Candidate George Bush s visit to a New Jersey flag factory in the 1988 campaign is a good example of how such news management works From a flagbedecked podium at the factory the candidate declared quotFlag sales are doing well and America is doing well and we should understand that and we should appreciate thatquot The statement was vacuous but the visual imagery was spectacular and since the candidate did little else that day the media had no choice but to report the event Journalists are thus fighting a twofront war to control their professional turf On one side they must fend off market competition that forces them to dilute the news values that are their professional bread andbutter And on the other side they must struggle with politicians to maintain control oftheir work product This chapter focuses on how journalists manage this dual struggle It begins with a theoretical sketch of the basic tension between journalists and politicians It then develops two rules of behavior that serve journalists interests in the ongoing struggle THE BASIC CONFLICT Important politicians such as presidents and presidential candidates feel they have a right to give a speech or take an action and have it straightforwardly reported as news Butjournalists jealous of their autonomy and voice do not cede this right They have their own ideas about what news consists of Hence politicians often find themselves jumping through journalistic hoops in order to get their story out and they resent having to do so Moreover politicians sometimes refuse to jump through the hoops instead creating events so compelling in terms of visual images symbols or drama thatjournalists have no choice but to report the events as news even if they don39t thinkthey really are news In this way politicians force journalists to jump through their hoops Bush s appearance at a flag factory was an example of this Politicians also resent the criticism that journalists routinely heap on them It is one thing to be attacked by one39s partisan opponents but another to be attacked as regularly occurs by the supposedly nonpartisan press They reporters love to destroy people President Clinton has been quoted as saying That s how they get their rocks off 1 In the ideal world of politicians therefore campaigns would generate information and journalists would dutifully pass all of this information and onlythis information on to the public News reports of campaign activities would differ little from paid advertisements except that they would be free and would run under the byline of a reporter Politicians have important resources in the struggle to control news content Most importantly they determine in both a positive and negative sense the daytoday content of campaigns On one hand they take actions and stage events that promote their campaign agenda and that with the advice oftheir media advisors are often so compelling that reporters feel obliged to report them as news On the other hand they attempt to avoid situations such as news conferences that make it difficult for them to control the kind of news that gets made Both elements were present in the flag rally discussed above The Bush campaign calculated that journalists would be unable to resist the visual appeal ofthe patriotic setting even if what the President said was somewhat vacuous Also campaign managers kept Bush physically separated from reporters on that day so as to prevent journalists from asking questions that would force Bush to address questions that would distract from his primary message of patriotism If as campaign managers try to 1 From Behind the Oval Of ce Dick Morris p99 ensure journalists can find nothing more interesting to report they are constrained to report what the candidates offer up All this smacks of manipulation But politicians have little choice but to try to influence news content in this manner With the demise of parties as campaign organizers political news is one of the most important means and through long stretches of time the only means of mobilizing the public support presidents need to get elected to office and to be effective while in office A president or presidential candidate who always spoke with complete candor without any strategic thought to the kind of news his words and actions would make would be as remiss in his duties and as foolish as a lawyer who always told the jury everything he thought about his client And he would succeed at hisjobjust about as well When politics is conducted by means of mass communication politicians must approach communication strategically rather than sincerey Candidates who fail to be strategic will be beaten by that is judged by voters to be inferior to candidates who do behave strategically But if candidates are constrained to approach communication strategically journalists are not constrained to like it and most do not But why exactly not Couldn t journalists sell as many newspapers or get as many audience rating points by providing straight reports of lavishlystaged campaign events Why do journalists so often feel compelled to make sarcastic or other negative comments when as in the case ofthe flag factory rally the candidates do such a good job of appealing to the production values ofjournalists The answer to these questions as l have suggested is that reporters would cease to be professionals and hence cease to enjoy the social status pay and selfsatisfaction that go along with being a professional ifthey were forced into the role of news readers for politicians Hence journalists insist that it is their professional prerogative to determine what counts as news and they resent what they see as the incessant efforts of politicians to manipulate them They have no objection to reporting the speeches and other campaign events staged by candidates But in the ideal world ofjournalists candidateinitiated information would be only the bare starting point rather than as candidates prefer the totality of news reports As journalist Eleanor Randolph has written Journalists who cover politics bridle at any suggestion that their job is simply to transmit campaign speeches like a conveyor belt from the campaign trail to the reader or viewer They say their job is to give some idea of what kind of presidents these candidates would be which means looking beyond the portrait presented by the campaign managers quotThe premise we have to challenge as journalists is that the candidates have the exclusive rights to control the dialoguequot said David S Broder of The Washington F ost2 As indicated earlier journalists always want to add their own information and analyses to stories and they want the material they add to be as important as possible When in contravention of this ideal journalists can find no important information or perspective of their own to add to a news story they feel frustrated Either they have failed to do what as professionals they expect of themselves or someone usually a politician has prevented them from doing theirjobs as they feel they have a right to do them If the latter is the case they take countermeasures One ofthe simplest ofthese countermeasures is acid commentary what Mark Levy 1981 has 3 Thus in the case ofthe flag factory visit journalists did run the visual images that the Bush campaign wanted but they framed them in terms that were anything but helpful to the vicepresident On NBC Tom Brokaw announced that quotThe vice president wrapped himself in the flag againquot Dan Rather said on CBS that quotGeorge Bush gives his 39my patriotism is better than yours39 the hard sell ABC s Brit Hume reminded viewers of an event a week earlier in which the VicePresident had used the word America 31 times in 15 minutes for an average of twice a minute4 NBC s Lisa Myers added that Bush s use of national symbols lead some to quote Samuel Johnson that patriotism is the last 2 quotCandidates Limit Media Access Aides Fear Reporters Will 39Step on39 Messagequot Eleanor Randolph Washington Post Sept 21 1988 A1 3 The elements of Levy s insightful argument are different than mine but still as it seems to me essentially similar In his view journalists disdain news when competitive pressures force them to report information they believe to be tainted By tainted Levy means information that is not genuine news according to standard journalistic criteria by competitive pressures he means the expectation that most journalists will be reporting the story thus forcing all others to go along by disdaining news Levy means role distancing behavior by which the journalist signals the audience in such a way as o maintain credibility despite reporting tainted news Translating my argument into Levy s categories I would say that when politicians manipulate reporters into reporting information that reporters believe to be vacuous or phony the information is tainted The competitive pressures to which Levy refers are in my model market pressures to provide the mass audience what it wants Thus I think individual journalists would report stories like the flag factory visit even ifthey knew others were not simply because they recognize that the Bush campaign had created the elements of an appealing story What Levy calls disdaining news is in my model simple one kind of media negativity as described in the next chapter 4 Goldwater Quip about Bush Reflects Changing Dynamics of Campaign on TV Lloyd Grove Washington Post September 23 1988 A16 refuge of scoundrelsquot5 Thus although all three networks carried Bush s rally at the flag factory the coverage may not have won many votes for Bush quotThatquot as campaign manager Lee Atwater said afterward quotwas one flag factory too many 6 In many cases journalists go beyond acid commentary They may openly challenge the president s or other politician s account of events give space in their stories to critics or in rare cases conduct full fledged investigations As we shall see in Chapter 7 media negativity of all kinds is correlated with the attempts by campaigns to dominate the campaign agenda Conflict between politicians and journalists for control of the news is often quite literally a contest to see who can get more news space or air time Candidates exhaust themselves flying around the country to create campaign events that are so compelling in a staged Hollywoodish sort of way that the journalists have no choice but to report them as news Journalists for their parts can become almost openly jealous of what many seem to regard as quottheirquot air time and newspaper space Thus the length of the average candidate sound bite on the evening news has fallen from about 42 seconds in 1968 to less than 10 seconds in 1988 and 1992 Adatto 1990 Hallen 1991 Patterson 1993 The length of candidate statements quoted on the front page of the New York Times has fallen from an average of 14 lines to an average of six lines in the same period Patterson 1993 p 76 The struggle is especially fierce on live TV where journalists surrender their usual right to edit what politicians say in exchange for the right to do a live interview of an important personage Once when ABC anchor Peter Jennings observed a politician being interviewed and apparently hogging time on another news program he exclaimed quotI hope I never interview a politician on the nightly news that way because when they39re on live they just own the fucking timequot Rosenstiel 1993 p 29 This sort ofjealous reaction led to a famous incident on the evening of the 1996 election As is customary all of the big network news teams had gathered their star correspondents onto special sets for live broadcast of the election results These shows are an American ritual and one ofthe top opportunities for correspondents to strut their stuff As part ofthe ritual President Clinton was permitted by the networks to make a victory statement on live TV But the President went on much longer than expected relegating numerous star reporters to 5 Cited in A Campaign Dominated by Images Not Issues by Ed Siegel Boston Globe September 24 1988 p 10 offscreen silence In the ABC studio alone those crowded out of the limelight by the longwinded president included Sam Donaldson George Will Cokie Roberts Peter Jennings Jeff Greenfield and David Brinkley plus other correspondents in the field Also wasting was ABC39s flashy hightech graphics display set which had been built in order to exhibit large quantities of statistical data The longer the President spoke the more reporters39 segments had to be cut back or canceled as each ofthe ABC stars was no doubt well aware as he or she listened to the President go on All this apparently became too much for Brinkley After getting back on camera though seemingly unaware that he was he blasted Clinton39s victory speech as quotgoddamn nonsensequot and quotone of the worst things I39ve ever heard quot and quottotally unnecessary Everything in there he39s already saidquot Brinkley concluded ofthe newly victorious President quotHe39s a bore and will always be a borequot7 What is most notable about this event however is not Brinkley39s lapse of professionalism but the fact that ABC as well as the other networks ran Clinton39s halfhour speech at the length they did Obviously they would have preferred to reduce it to a minute or two perhapsjust a soundbite so as to save the time for their own correspondents But journalism is as my theory holds constrained by the public39s wish to have original exposure to what politicians are saying Thus once the public realized that Clinton was making a victory statement it was hard for the networks to cut him off The same constraint is at work in campaign reporting generally Even though a fastpaced exchange between Sam Donaldson and George Will or a witty commentary by Cokie Roberts would probably be more interesting than a politician39s speech the public wants to see the speech anyway or thinks it does And certainly the public has no natural sympathy for the occupational aspiration ofjournalists to express voice These attitudes strengthen the hand of politicians in their turf war with journalists constraining journalists to run stories they might prefer to ignore Yet the public also remains both easily bored by politicians and suspicious of them and so is willing to cede journalists considerable leeway to make the news more interesting and to entertain allegations that politicians have shaded truth or engaged in other subterfuges or shenanigans This of course strengthens the hand ofjournalists against politicians3 6 Germond and Witcover 1989 p 408 7 Washington Post November 7 1996 quotBrinkley39s Parting Shots at Clintonquot by John Carmody p E1 3 It should be noted that the goals of neither politicians norjournalists imply any tendency to lie or make false statements Their competing goals imply only that from the infinite variety of true facts that might 59 This threecornered conflict politicians and journalists struggling to control news content within constraints set by the mass audience is as I claim at the heart of media politics Yet as described so far this conflict explains little about the actual dynamics of media politics What is needed is a more specific account of how the conflict plays out It is to such an account that now turn THE RULE OF ANTICIPATED IMPORTANCE Reporters must peddle their product to an audience that as maintained earlier wants exposure to all important politicians and points of view but doesn39t want to waste its time on unimportant ones But who determines what is important and what isn39t At times the answer to this question is obvious as in the case of President Clinton39s victory speech But in many other cases the answer is not obvious Uncertainty about what is important is an opportunity for journalists to express voice by using their own judgment to determine what is important and every indication is that reporters eagerly seize upon this opportunity Meanwhile politicians who fail to meet the journalistic criteria of importance whatever exactly they are have difficulty getting any coverage at all much less controlling the nature of the coverage they get Consider the case of presidential primaries Except when an incumbent president is seeking re election each party fields 5 to 10 experienced and reasonably wellfunded candidates who would like to be president For a mass audience that doesn39t want to waste its time on politics this is too many candidates to learn about Even the C SPAN crowd doesn39t want to study this many platforms and delve into this many political backgrounds Recognizing this journalists are highly selective in their coverage ignoring most ofthe field and covering only those two or three who seem most likely to succeed be reported as news each side wishes to focus selectively on what serves its own goals Each side that is wishes to tell a different partial truth Of course politicians orjournalists may in practice lie or make false statements but nothing in the logic of media politics impels them to do so Nothing moreover quickens the passion for truth in one side more than a demonstrably false statement by the other As a result it is normally a serious error to make false statements since this simply gives the other side ammunition in the contest for control of news content It is an interesting question whether it is easier for politicians orjournalists to get away with untruths and misleading statements My impression is that each side thinks it is easier for the other side to lie easier for politicians to lie because they control the government and can speak faster than reporters can investigate and easier forjournalists to lie because politicians have no ready means of rebuttal Whichever is the case however neither politicians norjournalists can achieve longterm success in their professions by a strategy of consistently distorting the truth 60 Politicians who are blessed with media coverage in these conditions have at least a chance to succeed others bear a considerable handicap Likewise in general elections There are normally four or five third party candidates on the ballot in most or all of the 50 states but the public doesn39t want to waste its time on them Many voters hardly care about the major party candidates let alone the minor party candidates Hence journalists ignore minor party candidates unless there is some indication that they will be unusually consequential This problem is entirely general There are always more ambitious politicians more controversial issues and more serious national problems demanding attention than the public cares to know about Hence the need for selectivity arises in every domain of public life And in each of these domains the public39s demand for selectivity is the journalist39s opportunity to exercise the discretion that as professionals they relish In deciding exactly how to use this discretion journalists follow a rule that is very close to the public39s rational interest in wanting to be told only what it really needs to know That rule which I call the Rule of Anticipated lmportance9 may be initially stated as follows Coverage of candidates and issues should be allocated in proportion to its marginal value for shedding light on future developments in American politics Thus in the case of multicandidate fields the press will cover candidates it expects to do well and to ignore candidates it expects to do poorly Similarly in other domains of politics reporters will concentrate their energy and attention on issues and problems that they expect to be most important to future events The reference to the quotmarginal valuequot of coverage indicates that journalists will concentrate on candidates and issues that are relatively new and unknown for whom the value of any new information will therefore be especially high The Rule of Anticipated Importance can also be justified from the auxiliary assumption that citizens value political news primarily for its entertainment value The justification is as follows Citizens who turn 9 Although it may have an older pedigree I encountered this general idea in Entman and Page 1995 They noted that in coverage of Congressional hearings on the Gulf War reporters seemed to be interested in the statements of policymakers in proportion to the importance of the policymaker rather than the inherent importance or novelty of the statements 61 to political news for entertainment must do so for some special reason since they would otherwise consume the more conventionally quotpurequot entertainment offered up by Hollywood and the professional sports business That special reason has to do with the charisma of power What makes political news distinctively entertaining is that it involves powerful and important people Notjust any politician or political program is fascinating only those who possess or are likely in the future to possess power have appeal In allocating coverage according to the Rule of Anticipated Importance journalists are assured of covering individuals and topics having the special charisma that only political power bestows It is often said the only thing thatjournalists really care about is 39What sells newspapersquot commands high ratings or serves other commercial needs But what is it that sells newspapers or gets high ratings In grounding my argument in specific claims about what a rational public will find interesting or entertaining I have tried to flesh out an otherwise vague argument about how journalists pursue commercial values I have maintained that journalists devote coverage to candidates whom they anticipate will be important But this is not all Journalists also devote more of their enterprise time and talent to candidates whom they expect to do well hoping to be able to quotscorequot on them To be blunt journalists are most harsh when dealing with the most powerful politicians What after all is the point of launching a majorjournalistic investigation of a candidate who is unimportant and going nowhere The public won39t care and so will little notice whatjournalists might report Journalists therefore want to make their mark on bigger fish so as to be more noticeable Hence a more general form of the Rule of Anticipated Importance can be stated as follows Journalistic resources of all kinds coverage talent effort are allocated to candidates in proportion to their marginal value for shedding light on future developments in American politics The Rule of Anticipated Importance is a powerful one since it serves the audience interest in conserving its time the interests ofjournalists in expressing discretion and voice and the commercial interest of quotselling newspapersquot The only actors not served by it are politicians who are judged by reporters to be unimportant But politicians39 only real opportunity for redress is to do something that will make them seem more important that is to play to reporters39 sense of what is newsworthy I should add that in saying that politicians must play to reporters sense of what is newsworthy I do not mean to imply that reporters make these judgments arbitrarily To the contrary Chapter 6 will offer considerable evidence that reporters sense of anticipated importance is grounded in a very plausible sense of actual importance THE RULE OF PRODUCT SUBSTITUTION Once politicians establish themselves as important by winning an important nomination or office the power equation between them and reporters shifts radically Politicians are no longer in the position of hoping for crumbs of coverage from journalists they know that journalist must cover them and try to determine the nature ofthat coverage Thus one of the first moves of a newly powerful politician is to beef up his or her staff of campaign consultants and public relations specialists Their purpose is to create words and images that make their candidate look good and thatjournalists will convey to the public as quotnewsquot The elite journalists who are on the receiving end of an important politician39s strategic communication do not relish this position They feel constantly pressured into providing what might be called quotpass throughquot coverage of candidates39 public relations events but they do not find it professionally gratifying to do this But there is as l have been arguing a dilemma here Public relations events having been created by highly skilled professional campaign staff often have real mass appeal Thus however much reporters may wish to ignore professionally unrewarding campaign communications they cannot entirely ignore the tastes ofthe mass audience This dilemma is rooted in the basic tenets ofthe model of media politics On one side is a public that wants in general to see what important political figures are doing that does not want in this particular case to cede to journalists the right to censor the kinds of images that presidents and presidential candidates can present to the public and that therefore wants some degree of original exposure to what the candidates are doing especially so if the visuals are striking10 On the other side are professionals who hate reporting quotpress releasesquot broadly construed but nonetheless feel constrained to do so The response ofjournalists to this dilemma is captured by what I call the Rule of Product Substitution as follows The more effectively reporters are challenged for control of a news jurisdiction the more assiduously they will seek to develop new and distinctive types of information that they can plausibly substitute for what politicians are providing and that af rm overall journalistic control of mass communication The claim here is that if politicians are so thorough and effective in staging news events that journalists have no opportunity to express voice journalists will quotfight backquot by substituting information and perspectives into the news that are distinctively their own What journalists substitute must however meet two constraints First it must permit politicians some opportunity to speak directly to the mass audience This is because as explained the public dislikes having any one group dominate communication Thus journalists cannot offer general commentaries on the election in place of stories that show what the candidates are doing Second like a detergent company that wants to get consumers to buy liquid gel instead of soap bars journalists must offer something that is the functional equivalent of the product they replace ie something that provides information about the campaign Much horserace coverage in which for example journalists let the candidate deliver his sound bite of the day but then explain how everything he has said is really just an appeal for votes meets both ofthese constraints Clever editing of sound bites to show how the candidate has contradicted himself also does so Thus much of the information that journalists substitute for candidatesupplied information cynical commentary investigations and so forth will reflect unfavorably on the president or presidential candidate But this need not always be the case If a journalist were to dig out positive information that a powerful official was keeping secret I believe it would have just as much prestige value as negative 10 The ability of ordinary citizens to act on this preference is obviously limited but it is plausible to suppose that citizens would be upset ifjournalists simply stopped covering presidential events deemed by journalists to be exercises in public relations and that journalists would be sensitive to such dissatisfaction information Thus if to take a hypothetical example a bachelor president were to secretly wed while in the White House or if a billionaire president were to secretly give most of his wealth away to charity l conjecture that reporters would be as eager to report this type of secret information as comparable negative information For it is not the negativity per se that is valued it is the opportunity for a journalistic contribution to the news Clever politicians find ways to turn journalists occupational interest in the expression of voice to their own benefit Perhaps the easiest way to do this is to offer journalists a leak or even better an exclusive interview Journalists are nearly always happy to accept such stories and to report them in moreorless the same terms that the politician wishes to have them reported because they are able to call attention in these stories to their own role in the creation ofthe news Leaks and exclusive interviews are by definition limited in their effectiveness since one cannot leak information or give exclusive interviews to the whole press corps at once Candidates do however have the option of press conferences press availabilities and other forms of direct access to the candidate by reporters Reporters appreciate these events since they can by the questions they force the candidate to confront seize control of the campaign agenda from the candidate Yet the fact that media access involves the potential loss of agenda control for the candidate makes candidates reluctant to grant such access Generally speaking they are most inclined to grant media access when they know what the reporters are likely to ask about and wish to be questioned on that subject And this in turn lessens the value of access to reporters making them feel manipulated by the campaign rather than genuinely in charge oftheir professional turf THE SOURCES OF MEDIA NEGATIVITY The game of media politics consists in my theoretical argument of virtually continuous struggle between politicians and journalists to control the content of news Given the occupational stakes involved struggle follows and given the existence of struggle mutual dislike among politicians and journalists would seem to follow as well Finally given mutual dislike it is no surprise that much media coverage of politics and politicians is highly negative High levels of media negativity are thus a direct implication of my theory of media politics But not uniformly high levels of negativity I shall argue later that increases in media negativity since the 1950s and more recent increases in media negativity in parts of the European press are a journalistic response to more aggressive attempts by politicians to control the news Beyond this the Rule of Anticipated Importance holds that journalists are more apt to launch serious investigations of which is to say dig up dirt on candidates who have importance in American politics Also the Rule of Product Substitution holds that reporters will be more negative in their coverage of candidates who attempt to manipulate the news And since only the more important politicians will be in positions to make heavyhanded attempts at news management they are the ones most likely to be heavily criticized by the media Thus both the Rule of Anticipated Importance and the Rule of Product Substitution have direct implications for the conditions under which journalists will be most likely to provide negative coverage of candidates Chapters 6 and 7 will further develop and test these implications But in order to conduct the tests it is necessary to develop a workable measure of media negativity This task which is more difficult than it initially appears is first taken up in Chapter 5 Chapter 5 The Nabobs of Negativism quotToo much attitude is the main problem ofthe press todayquot 1 said a White House correspondent quoted by James Fallows in his book on the press Breaking the News Fallows himself writes To judge by the coverage everything in public life is a sham Conflicts are built up and they blow over and no one is sincere As onlookers we can laugh at and look down at the participants because everyone knows it39s all done for effect p 179 181 Presidential campaign coverage has become so negative that it is now a barrier between the candidates and the voters rather than a bridge between them says Harvard scholar Thomas Patterson Election after election the press tells the public the candidates are not worthy p 25 These views are widely echoed in the writings of other academics in public opinion polls in self criticism by journalists and in the speeches of the politicians who are the target of the media onslaught When VicePresident Spiro Agnew complained in 1970 that the media had become nattering nabobs of negativism it was widely seen as a partisan attack Today the view that media negativity has gotten out of hand is almost universally accepted From many points of view this development is a puzzle Why would reporters who typically profess love of public affairs and fascination with politics insist on tearing down so much of what they cover Why has a profession struggling with the pressures of market competition dished out so much more negativity than by all indications its mass audience wants Why don t reporters and politicians strike a mutually beneficial bargain Easy access to information forjournalists in exchange for positive coverage for the politicians Given that both journalism and public relations are now thoroughly professionalized why haven t media relations become smoother rather than more conflictual The answer as I will argue is a generalized competition between politicians and journalists for control of the news Politicians have a constant need to mobilize mass support and except for mass advertising news is the principal means by which they do so Their wish to control the content ofthe 1 Emphasis in the original news is therefore a deep occupational interest But ifjournalists were to concede control of the news to politicians ifthey were to become people who simply read or reprinted the press releases of politicians their professional status would fall to zero Their occupational interest is to make some independent contribution to the news and criticism of politicians is one important means by which they do so How exactly the competition between politicians and journalists plays out will be analyzed in Chapters 5 and 6 The task for this chapter is to develop and validate a measure of media negativity that will carry the weight ofthis analysis In particular this chapter will develop measures of media negativity for three separate media national news magazines Time and Newsweek the New York Times and TV network news The task is not straightforward Everyone agrees that the news media must report a certain amount of negative news It is only when reporters cross over some invisible line and begin to express bad attitude that they become in the eyes of some nabobs of negativism But determining when journalists have crossed this line presents daunting problems FROM PARTISAN TO NONPARTISAN NEGATIVITY Media negativity is nothing new in American politics Consider the following paragraph from the Chicago Tribune s coverage of the 1896 election William J Bryan Democratic candidate for President was denounced as worthy only of contempt a dangerous man a teacher of Anarchy an advocate of the Gospel of Hate of wallowing at the feet of the Tammany King and the foe of law and order by the Rev Robert B McArthur this evening2 The Tribune s scathing coverage of Bryan is quite easy to explain It was a Republican newspaper and Bryan was a Democrat To be sure the Tribune was unusually partisan even in the heyday of the partisan press But Michael McGerr describing Joseph Pulitzer39s relatively independent New York World writes 2 Cited in Burgos 1997 Eager for Democratic victory Pulitzer used most of the weapons from the arsenal of party journalism News stories in the M attacked James J Blaine portrayed a dispirited Republican party and proclaimed the certainty of Cleveland39s election The paper like most New York journals did not run the party ticket on the editorial masthead because as an editorial explained quotit is not necessary Every column of our paper tells the story of our devotion to the principles of the Democratic partyquot But the World did celebrate Cleveland39s triumph with a traditional display three roosters crowded quotVICTORYquot on the front page This conventional partisanship was a basic element of Pulitizer39s journalism He used the World to sell his politics and he believed his politics sold the World In the late 19th and early 20th centuries then most American newspapers had a reliable partisan bias and this bias explained most of the negativity in their coverage And some of this negative coverage was indeed very negative Although there has been no systematic examination ofthe decline ofthis sort of partisan negativity it is obvious from any cursory examination of newspaper archives that such a decline has occurred A few papers including the Chicago Tribune remained blatantly partisan through the 1930s But for many newspapers the blatant partisanship of the 19th century died relatively early in the 20th century Why this happened is an open question There are many conjectures and essentially no published evidence But by the 1950s most American newspapers had become for whatever reason mostly if not entirely neutral3 Not only did American newspapers cease by the 1950s to be overtly partisan they also ceased to be very negative Lapdog journalism is the way that Larry Sabato has described the journalism ofthe mid twentieth century But beginning sometime in the 1960s journalism took a turn toward the negative The new negativity was not simply a revival of the old partisan negativity in which each paper supported one party and lambasted or ignored the other Rather the new negativity was heaped on all sides without fear or favor Some including VicePresident Agnew have claimed that the new negativity was a mostly antiRepublican negativity and there is some evidence to support this view But Democratic Presidents Jimmy Carter and Bill Clinton would deny that journalists ever cut them any slack and as we shall see in 3 As late as 1948 however the Tribune carried a prominent front page story on Soviet Dictator Joseph Stalin s endorsement of Harry Truman in that year s presidential election Burgos 1997 69 Chapter 6 much of what seems to be media bias against Republicans is better explained as journalistic reaction against the aggressive news management practiced by many Republicans The new media negativity seems to have escalated steadily since the 1960s Our task now is to measure the new negativity so that we can first delineate its upward course and then explain it CONCEPTIONS OF NEGATIVITY If for example one candidate attacks another and journalists report the attack no one can blame journalists for negativity Nor if a candidate is behind in the race or loses a debate can the media be blamed for reporting that the candidate is losing the election What reporters can be and often are blamed for is negative coverage that they themselves have initiated Thomas Patterson provides an extended example ofthis type of negativity in a quotReality Checkquot report by CBS correspondent Eric Engberg In that report Engberg questions candidate Clinton39s 1992 statement that he could not yet offer a firm opinion on NAFTA because of its complexity As Engberg exclaims Time Out Clinton has a reputation as a committed policy wonk who soaks up details like a sponge but on an issue which will likely cost him votes no matter what side he takes the onetime Rhodes scholar is a conveniently slow learner Commenting on this passage Patterson contends that Clinton was not as Engberg charged dodging the NAFTA issue Rather Clinton was making truthful statements of his own actual ambivalence toward what was after all 1078 pages of densely worded text As Patterson also points out Clinton maintained his ambivalent attitude during his presidency Patterson writes The Engberg news story is a case study in journalistic halftruths that pass for incisive analysis If Clinton the candidate was circumspect in his support for the North American Free Trade Agreement so is Clinton the president In a meeting with Mexican president Salinas Clinton said quotI reaffirm my support forthe NAFTA And restate my belief that some trade issues between our nations still need to be addressedquot But if Patterson s case is a prime example of what many people regard as excessively negative media coverage it is also an excellent example ofthe difficulty of measuring media negativity For no one including Patterson would argue that reporters should ignore politicians lies or halftruths Yet it is hard except in unusual cases to say when a politician has committed some actual offense and when a reporter merely thinks the politician has Thus in a published review of Patterson39s book Washington Postjournalist E J Dionne 1994 argues that Engberg39s rap on Clinton was quite fair Dionne writes Engberg had it right as any honest Clinton adviser would readily admit NAFTA was a terrible issue for Clinton because it divided his constituency It did take him an eternity to take a stand But Patterson doesn39t go into the politics of the issue He simply dismisses Engberg39s comments as quotfatuousquot He takes a similar approach throughout the book blaming the press for the things candidates themselves do for their own political purposes P 50 So who is right Patterson who stresses the true complexity of the issue as the reason for Clinton39s hesitation or Dionne who stresses the equally indisputable political bind that NAFTA presented for Clinton This type of question is one that cannot be resolved to the satisfaction of fairminded and intelligent observers as the disagreement between Patterson and Dionne attests It is moreover the kind of question that comes up over and over if one attempts to distinguish appropriate media negativityquot from I shall therefore take another approach To measure the sheer amount of mediainitiated negative coverage regardless of the appropriateness ofthe negativity tone For while fairminded people may disagree over whether a particular criticism is merited they may nonetheless agree on whether a negative comment has been offered or not The key element in this approach is mediainitiation of the negativity Hence reports of attacks by an opposition candidate will not be counted as media negativity since reporters do not initiate them An exception will also be made for reports that one of the candidates is doing badly in the electoral horse race since a losing performance by at least one candidate is the inevitable product of political competition4 What will count is any coverage in which reporters themselves initiate coverage that reflects badly on a given candidate Negativity could involve anything from a snide remark to a barrage of 4 If however a reporter reaches large conclusions about a candidate39s general competence or lack thereof on the basis his performance as a campaigner it is counted as mediainitiated criticism 71 critical commentary to a majorjournalistic investigation The amount of mediainitiated negative coverage taken as a fraction of all coverage will be the measure of media negativity It should be noted that measuring negativity in terms of sheer quantity does have drawbacks The most important is that there is no necessary relationship between quantity and normative appropriateness No matter how much or little negativity exists it may still be too little or too much depending on what is actually merited But as the PattersonDionne disagreement indicates analysts are unlikely to be able to reach agreement on the merits anyway Meanwhile a strength ofthe quantity approach is that it enables us to get on with the systematic analysis of what the reporters are doing Having now decided on a conception of medianegativity ie any negative information except information about the political horse race that reporters themselves initiate I can proceed to a more detailed discussion of how to count it As will become apparent there are still some difficult issues to be decided MUST NEGATIVTY INVOLVE AN EXPLICITLY NEGATIVE STATEMENT The sources of mediainitiated criticism are often individual reporters who take it upon themselves to offer critical commentary on some subject as in the Engberg example above But the source of negativity can also be collective as when the press corps as a whole becomes obsessed with covering a matter that one of the candidates finds embarrassing Media coverage in 1992 of Bill Clinton s alleged affairwith Ginniger Flowers is an example Sabato 1993 documents many other such feeding frenzies Much of what I shall consider negative coverage contains no explicitly negative evaluation Again the Flowers story is an example since most of this coverage simply reported information suggesting an affair without offering any sort of evaluation But since the reporting involved a matter that Clinton had denied and that was obviously embarrassing to his candidacy it should in my view be counted as negative coverage The alternative approach used by some scholars is to count only explicitly negative evaluations as media negativity An example ofthis was Engberg s comment that Clinton was a conveniently slow learner on NAFTA The advantage of this approach is that it involves a very clear standard of negativity which then makes the coding of news stories easy and more reliable The disadvantage however is that it misses a great deal of what almost anyone would recognize as negative coverage lndeed cases in which the reporters create major negative stories on the basis of straight often scrupulousy straight reporting of embarrassing facts or alleged facts can be found throughout the period of this study and are among the most memorable stories ofthe campaigns Examples include In 1948 Thomas Dewey39s contemptuous remark about a railroad engineer who caused his campaign train to lurch in 1952 Richard Nixon39s quotsecret fundquot which was heavily covered by the press without much direct comment except on the editorial pages in 1956 Eisenhower39s health which the press also reported in excruciating detail in 1964 the Walter Jenkins affair in 1972 the Thomas Eagleton affair in 1976 Jimmy Carter39s Playboy interview and in 1992 the Gennifer Flowers story Most of whatl have classi ed as mediainitiated criticism consists as in these examples of ostensibly straight news about topics that candidates do not wish to have reported and discussed WHAT TO COUNT It is useful to compare my concept of mediainitiated criticism with the notion of quotbad pressquot developed by Robinson and Sheehan 1983 in their highly regarded study of media coverage of the 1980 presidential campaign Over the Wire and On TV These researchers count material as bad press if it involves quotnegative informationquot that has arisen from any source mpg the following noncampaign events eg economic news attacks by partisan opponents criminals or antiAmericans or any aspect of horserace competition Given this definition quotbad pressquot for Robinson and Sheehan is to some degree a residual category In practice however their concept and mine are likely to identify much the same material Negative information that has originated in neither noncampaign news events nor horserace coverage nor partisan attacks is likely to be information that the press itself has raised5 5 The only point of difference is over statements that arise from nonpartisan sources which Robinson and Sheehan always include and that I include only if it appears that reporters have sought them out If for example an environmental group holds a press conference to denounce a candidate the rules used by Robinson and Sheehan would count coverage ofthis event as bad press whereas I would not because the press did not initiate the story even though it did use its discretion to report it If the same denunciations were carried in a feature story on the environment and in the form of statements by individual environmental experts Robinson and Sheehan would continue to count the remarks as bad 73 My approach to tallying or counting criticism is however markedly different from that of Robinson and Sheehan They seek to measure bad press at the level ofthe news report as a whole characterizing each story as bad press good press or something else I measure negativity at the level of each individual story element typically the sentence but sometimes at the level of a clause or even phrase In addition I have developed a somewhat elaborate coding schemes to capture the content of each sentence For example for coding campaign stories in Time and Newsweek I developed the following codes for specific forms of media negativity Candidate uses unfair sleiy campaign tactic Candidate is inconsistent fugy issues Candidate takes wrong madvised position on issue Candidate is immoral person Candidate lacks competence ability Candidate is M person Candidate is dangerous crazy Other press criticism For example press attention to Dewey39s contemptuous remarktoward the engineer on his campaign train was coded as indicating that Dewey was a cold person the Eisenhower health story was coded in terms of personal competence and ability since the suggestion was that the General might not be fit to serve as president Gennifer Flowers39 allegation was coded as a mediainitiated suggestion that Clinton is personally immoral The coding scheme contains numerous other codes including for example positive and negative codes for partisan attacks thus a candidate who can get the press to report his attacks gets a positive quotown messagequot code and the target ofthe attack gets a nonmediainitiated negative code Most stories end up getting a mix of positive negative and neutral codes with only some ofthe negative codes reflecting mediainitiated negativity Robinson and Sheehan acknowledge that stories may contain a mix of positive and negative elements but they nonetheless code only the story as a whole Their rule is that if a story contains three press and here I would go along on the presumption that the press had actively sought out the comment In practice this difference in counting rules is not very consequential in my opinion 74 times more negative information than positive the story taken Q a whole is bad press Thus isolated negative evaluations or negative comments that are offset by equal amounts of good press within the same story are not counted as bad press by Robinson and Sheehan Using the same threetoone rule Robinson and Sheehan also measure quotgood pressquot at the level of the story as a whole I can best illustrate both the approach I have taken and how it contrasts with the standard work of Robinson and Sheehan by means of an extended example In their book Robinson and Sheehan provide the following partial transcript of a UPI story By their coding scheme this story is as they say quotneither good nor bad press essentially colorlessquot p 113 Ronald Reagan today declined to characterize his handpicked running mate Geroge Bush as a second choice to Gerald Ford Bush appeared at a nationally broadcast news conference and said quotWe39re delighted I39m very pleased to have been selected quot Bush reminded by several questioners how he had differed on many issues with Reagan during a sometimes bitter primary campaign reacted sharply saying he did not intend to stand on the podium and 39be nickled and dimed to death39 over differences with the former California Governor Bush who clashed with Reagan in the primaries on abortion and the Equal Rights Amendment said the big issues this fall will be unemployment in the economy and foreign affairs By my coding scheme these four sentences would get seven separate codes The first sentence would be coded as a mediainitiated negative reference to Bush because it principally consists of the suggestion which has almost certainly been raised by reporters rather than Reagan that Bush is a secondchoice selection for vicepresident The second sentence would be coded as a positive reference to Bush because the reporter permits him to assert his positive feelings about his selection albeit in a somewhat defensive context The third sentence would get a positive and a negative code for Bush the negative code for the presssupplied reminder that Bush differed from Reagan on many issues during the primaries and the positive code for the report of Bush39s response to the press Finally the fourth sentence would get a positive code for Bush for his assertion of what the real issues in the campaign will be and a negative but not mediainitiated negative code to the Democrats for Bush39s reminder about the weak state of the economy There would also be a mediainitiated negative code for the reminder that Bush and Reagan had been at odds about important policies in the past even though Bush seemed no longer to consider them important6 By my coding scheme then this passage contains about equal amounts of good and bad news four mediainitiated critical references to Bush but also three instances in which Bush was able to get out his own message On balance therefore the story appears neutral and in this sense my coding agrees with that of Robinson and Sheehan Yet by my scheme of counting mediainitiated criticism this story is 56 percent negative which is a large departure from the Robinson and Sheehan coding which simply counts the story as a whole as neutral Yet despite this difference our approaches might well produce similar results across a wider range of cases I suspect for example that the rest of the UPI story which Robinson and Sheehan do not reprint is more neutral than the passages they do reprint so that if my scheme were applied to the whole story the result would probably not be 50 percent mediainitiated criticism but more like 20 percent which is closer to Robinson and Sheehan39s score of colorless neutrality Likewise stories that contain enough negative material to be called quotbad pressquot by Robinson and Sheehan counting it in effect as 100 percent negative would probably get less than 100 percent bad press by my coding Hence over many stories our approaches could well produce fairly similar results The point of this extended example as I hope has been clear has been neither to disparage the Robinson and Sheehan approach nor to acclaim my own It has been rather to explain as clearly as possible what I have done and how it compares with a standard piece of work so that readers will know what lies beneath the statistical summaries report WHAT S IN A FRACTION 5 In my calculation of scores I make separate tallies for each party so that if a candidate talks only about himself his remarks affect only his party39s scores but if he talks about both parties his remarks affect the scores of both parties Thus in my scheme a candidate quoted as saying quotI39m greatquot gets a positive code for having his message reported If he is quoted as saying quotThe other guy is a bumquot two codes are assigned a positive code for the party making the assertion and a negative code for the target ofthe remark There is one additional methodological issue to face In a sense it is simply a technical question of how best to combine the content data into an estimate of negativity But it is also a substantive question in that how one counts the data determines the estimate of how much negativity exists and of course one39sjudgment of whether the amount of negativity that exists is too high I can best illustrate the importance of this issue by comparing my approach to that of Patterson 1993 He describes his measure of negativity as follows It is based on favorable and unfavorable references to the majorparty nominees in 4263 Time and Newsweek paragraphs during the 19601992 period quotHorseracequot references are excluded all other evaluative references are included emphasis added Thus Patterson conceptualizes media negativity to include reports of candidates39 attacks on each other as well as reporters39 own evaluations It also includes evaluational remarks from voters quotsoccer momsquot quotangry white malesquot and so forth explaining why they like or dislike a particular candidate My coding of Time and Newsweek was not designed to measure Patterson s concept However my coding scheme does include codes for candidate and other partisan attacks partisan praise and self puffery and instances of both mediainitiated praise and criticism7 Hence it is possible to construct a measure from my data that ought to capture much or perhaps most of what Patterson39s measure does Patterson39s measure of media negativity may be expressed as follows All negative evaluations Bad News All negative evaluations All positive evaluations Despite the differences in coding and sampling periods when I calculate media negativity by this formula I get results very close to what Patterson got as shown in the top two lines in Figure 1 The r squares between the upper trend lines is 92 INSERT FIGURE 51 ABOUT HERE 7 My coding scheme includes voter appraisals but as part of quothorse racequot coverage since they almost always appear in the context of such stories Unfortunately I cannot extract them from this larger category Figure 5 Three estimates of trends in negativity of press coverage of presidential elections Percent of all coverage that is rated quotpressintiated criticismquot 100 80 HuntZaller I 60 Patterson 40 20 HuntZaller II 1 0 l l l l l l l l l l l 48 52 56 60 64 68 72 76 80 84 88 92 Middle line shows Patterson 1993 20 estimates of trends in quotbad newsquot coverage of presidential candidates in Time and Newsweek Upper line shows shows attempt to replicate Patterson results from HuntZaller coding Bottom line shows trends in quotpressinitiated criticismquot as a percentage of all coverage The two sets of estimates are not however identical Despite the substantial covariation in yearto year trends indicated by this statistic my estimate of negativity is consistently a bit higher than Patterson39s The difference however is quite likely due to an artifact rather than to real differences in coding standards Patterson39s estimate covers the entire election year whereas my estimate is based on data from only the final six weeks of the fall election campaign As Patterson also shows Figure 61 p 221 overall levels of media negativity were about ten percentage points higher in the fall campaign than in the spring and summer phases of presidential selection Hence my estimate of media negativity based on the fall campaign alone ought to show higher levels of negativity than Patterson39s estimate based on the whole year And this is exactly the difference that appears in the data These results indicate if nothing else that Patterson39s coders and mine though using different coding rules made very comparable judgments about the amount of negativity that existed in a given election year Yet reliable data cannot by themselves settle the question of how much media negativity exists Given the same set of data there are several different methods of calculating negativity Patterson made his calculation by the method described immediately above I propose however that a more valid estimate of media negativity can be obtained by calculating media negativity as a percentage of all campaign coverage as follows Pressinitiated negativity only Media negativny All campaign coverage period This method of calculating negativity differs from Patterson39s in two key respects lts numerator is moderately smaller in that it omits candidate and other partisan attacks and its denominator is much larger in that it includes all campaign coverage in the given year The estimates of media negativity obtained from the new method are shown as the lowest of the three trend lines in Figure 1 As can be seen the new method yields estimates of media negativity that are consistently lower than Patterson39s This is to be expected Shrinking the numerator and increasing the denominator of a fraction as l have done will necessarily make the fraction smaller But does the new method yield more valid estimates Both of my departures from Patterson39s method can be disputed Omission of candidate and other partisan attacks from the denominator leaves out material that the public might find alienating whether the reporters are responsible for it or not And journalists do after all decide how much of the candidates39 attacks on one another to report as news Yet on the other hand some candidates run campaigns that are genuinely more aggressive and negative and it could be misleading to assign responsibility for this negativity to the press For example if as may be the case underdog candidates offer more partisan attacks on the frontrunner than the frontrunner reciprocates Patterson39s method would make it seem that the press was picking on frontrunners by running more negative information about them Since the press already stands accused by Patterson and others of being especially tough on frontrunners it is important to create a measure that does not artificially create such a finding by a spurious method The danger of putting all coverage in the denominator as l have done is that the denominator could be changing over time in ways that distort the trend in negativity In fact the total amount of campaign coverage does vary somewhat across years but not I believe in ways that distort the actual level of mediainitiated criticism First the size of the Time and Newsweek quotnewsholesquot for campaign coverage has increased since 1948 Yet the change was mostly complete by 1968 so that total amount of coverage has been roughly flat in the period in which the increase in mediainitiated negativity has occurred Other aspects of campaign coverage have also changed since 1948 but except for negativity there have been no important time trends in the composition of campaign coverage8 Hence the 3 In particular there has been no increase in the amount of horserace coverage in Time and Newsweek Though there has been considerable yeartoyear variation the average has been about 25 percent of codable references and less than 20 percent in the last two elections The major difference since 1948 has been how horserace information has been handled In the early period the news magazines published statebystate summaries of the race that were segregated from other campaign coverage and that were very dry by current standards In the current period coverage is more thematic and more integrated into regular coverage thus giving a horserace flavor to much ofthe overall reporting Patterson 1993 in the discussion associated with his Figure 21 notes the greater integration of horserace coverage as a change from use of quotpolicy schemaquot in campaign coverage to use of quotgame schemaquot Though Patterson reports no data pertaining to horserace coverage in the news magazines a large amount of change along the lines he describes seems to have occurred Yet the integration of horserace coverage into general campaign coverage though important does not eliminate the possibility of other kinds of content within stories framed by a game schema most notably mediainitiated criticism of the candidates for matters not directly related to the horserace 79 appearance of change in negativity in Time and Newsweek is not an artifact of change in any important aspect ofthe denominator it is due rather to change in the numerator which is to say an increase in the actual amount of negativity Meanwhile Patterson39s approach has a denominator problem of its own one that is in my opinion far worse than mine For in calculating negativity as a fraction of all evaluational statements rather than all coverage his measure is subject to distortion due to changes in the overall amount of evaluational material or due to small changes within a small volume of coverage Consider the following example Suppose that in some campaign the media offer one negative evaluation and two positive ones By Patterson s counting scheme that comes out to 33 percent negative 11 2 33 If in the next campaign the media offer two negative evaluations and one positive evaluation Patterson s counting method will show that coverage has become 67 percent negative 21 2 67 which is a doubling of negative coverage Yet because the sheet quantity of negative coverage is still low only two negative comments this would be a very misleading finding Although I do not claim that the number of negative comments in Patterson s measure is actually as low as in this example the example does show the kind of problem to which the Patterson measure is vulnerable Too MUCH NEGATIVITY Comparison ofthe two methods of calculating negativity as summarized in Figure 1 make an interesting point about the amount of negativity that exists in the media By Patterson s measure 70 Horserace coverage in the New York Times has by my coding scheme increased only moderately from about 15 percent of all codable references in 1948 to about 28 percent in 1992 This is a much smaller change than Patterson finds for changes in the use ofthe game schema though direct comparison is difficult because of differences in concepts and in our denominators By Patterson39s data it appears that about 70 percent of frontpage Times stories used a game schema in 1992 compared to about 40 percent in 1960 l have attempted to read these estimates off his Figure 21 eg 70 82 X 85 see the note to Figure 21 My quotstory element by story elementquot estimate for horserace coverage in the Times in 1960 is just 18 percent Horserace coverage on TV has varied greatly from year to year but has shown no time By my coding scheme 285 percent of air time was given to horserace material in 1988 and 344 percent in 1992 These estimates agree fairly closely with those ofthe Center for Media and Public Affairs which estimated network news horserace coverage as 27 percent of coverage in 1988 and 35 percent in 1992 cited in Patterson p 73 percent of all evaluationa references are now negative while by my method about 20 percent of campaign coverage consists of mediainitiated negativity The two estimates are not as l have taken pains to show mutually inconsistent but they give very different impressions From the estimate of negativity presented in his book as well as some more qualitative evidence Patterson contends that media negativity has gone beyond the point of responsible journalism and become quota barrier between the candidates and the votersquot But that conclusion is much less compelling if my estimate which is that only about 20 percent of media coverage is negative is accepted as correct So which is correct For the reasons given I believe that my measure which calculates negativity as a fraction of all coverage rather than as a fraction of a small subset of coverage is preferable But no single number whether 70 percent or 20 percent or something in between will answer the question of how much negativity is too much For the key question is how much negative coverage politicians get in m to two Wm they m and that is a question that depends on a substantive reading of each story in relation to what the given politician actually did I have not as explained earlier attempted to make such judgments I can however give readers a sense of the kinds of media reports that have been classified as mediainitiated negativity Data from the New York Times will work best for this purpose so I will briefly introduce readers to how Times stories were coded My coding of the New York Times is based on abstracts of stories appearing in this newspapers rather than the original stories These abstracts are regularly published in The Index of the Times Following is a random sample of abstracts that have been coded as instances of mediainitiated 9 9 Each element of each story abstract has been coded with a story element defined as the text between periods or semicolons All coverage has been included except editorials oped pieces picture captions speech excerpts and any explicitly partisan commentary such as cartoons The procedure was first to code material under the rubric of presidential election to proceed next to coverage under the names ofthe principal candidates and finally to examine US politics and policy for campaign references not included in previous sections A strenuous and cumbersome effort was made both to capture all relevant references and at the same time to avoid double counting This effort was made necessary by the fact that the organizing rubrics ofthe Times Index were not consistent across the period of study 81 1 Gov Clinton stung by recurring questions about his credibility gives television interview from his mansion in Little Rock Ark in effort to control campaign coverage M S 11 A3121 2 Clinton and Bush campaigns have started using paidfor airtime to fling mud ushering in season of negative advertising 3 Gov Bill Clinton and his running mate Sen Al Gore have campaigned together on 20 of 52 days since their nomination in July subtext is to draw comparisons between Democratic package and Republican one since Pres Bush almost never shares stage with Vice Pres Quayle also positive code for Clinton 4 Newsmen from at least 6 natl pubs and Dem Natl Com agents have been at work for wks searching through data for material on Agnew probes focus on old charges of conflicts of interest 5 Nixon lr to securities indus leaders pledging to ease Govt regulatory policies disclosed aide A Greenspan says it was not made pub because it covers 39narrow policy area39 6 Sen Brooke flies to Cleveland to rejoin Nixon campaign says he is bewildered about Nixon39s remarks on school desegregation but stresses he is not leaving campaign10 7 Pres Reagan during televised briefing says he will meet on September 28 at White House with Soviet Foreign Min Andrei Gromyko denies he has been motivated by election campaign and by criticism that SovietAmerican relations have worsened under his Administration and that he has not met with any Soviet leader 8 Pres Reagan assails suggestions from some Democrats and news commentators that he showed signs of age in debate with Walter Mondale 9 Mayor A Starke Taylor of Dallas insists that Republican National Convention is still the 39free enterprise39 convention city leaders said it would be even though city39s taxpayers will end up paying from 1 million to 15 million for conventionrelated expenses I stress the element of random selection in these examples There were some 420 examples of mediainitiated negativity in the coverage of the campaigns of Nixon in 1972 Reagan in 1984 and Clinton in 1992 From each ofthese campaigns I randomly selected three examples as shown above This method of selection makes it likely that the stories will be reasonably representative ofthe entire set of mediainitiated negative stories 10 In a broadcast on southern regional television that was initially missed by much of the northern press Nixon suggested reservations about the Supreme Court39s 1954 decision in Brown v Board of Education Reporters in subsequently seeking the reaction of Senator Edward Brooke a black Republican Senator who was a member of Nixon39s campaign found a way to provide information about the incident to northern audiences The criticisms in these examples obviously vary greatly in their seriousness with subtexts that range from an allegation of garden variety hypocrisy item7 to fitness for office item 8 to possibly illegal corruption items 4 5 Stories like these constitute roughly 10 percent of all New York Times campaign stories in the period 1968 to 1996 As l have said I cannot say whether this figure represents more or less negativity than was actually warranted But I hope that this exercise has given the reader a reasonably clear idea of what typical media negativity looks like and how much of it there is in relation to other kinds of coverage OVERVIEW OF MEDIA NEGATIVITY DATA As indicated this study is based on content coding of the national news magazines Time and Newsweek the New York Times and network TV news Figure 2 gives a visual overview of the data from each ofthese media Additional details of the coding are contained in the Appendix to this chapter Two points stand out The first concerns the pattern of partisan bias and the second concerns the amount of media negativity Prior to about 1968 there was relatively little mediainitiated negativity and what there was had a slight antiDemocratic slant But from 1968 onwards there is notably more media initiated negativity and it mostly runs against Republican side The big exception to the pattern of apparent antiRepublican bias is Democratic incumbent Bill Clinton who across all four media got as much or more criticism as any Republican including Richard Nixon The Clinton case should stand as a warning against concluding prematurely that the media are guilty of antiRepublican bias later evidence will indicate that there is little if any party bias in media coverage of presidential campaigns Nonetheless large changes in media behavior seem to have occurred in 1968 and taken altogether they suggest a sort of regime change in media behavior Prior to 1968 the media seem to have been a somewhat timid and slightly proRepublican institution After 1968 the media seems to be a much more assertive and often proDemocratic institution The apparent regime change in 1968 is most striking in the case of Time magazine Up to and including 1964 it was reliably Republican But its behavior changed sharply in 1968 when for the first time in the postwar period it was much more critical of the Republican nominee Richard Nixon The change at Time was almost certainly part of another even larger regime change at the magazine A change of leadership In 1964 the unabashedly Republican publisher Harry Luce retired and in 1968 Luce39s managing editor and alter ego Otto Fuerberger followed suit In both cases the replacements were professionals who gave greater freedom to workaday journalists Halberstam 197911 INSERT FIGURE 52 ABOUT HERE Meanwhile the New York Times and Newsweek were switching in the 1960s from rough neutrality to often but not always proDemocratic leanings There is a clear suggestion in the data that Newsweek and the New York Times may have begun moving left in 1964 when for the first time in the postWorld War II period both showed a noticeable tilt toward a Democratic candidate But however this may be the big change came as it did for Time in 1968 As we shall see later on 1968 also seems to have been a turning point in the way candidates behaved toward the media CONCLUDING REMARKS The main point of this chapter has been to develop a plausible measure of mediainitiated negativity and to familiarize the reader with what it measures Chapters 5 and especially 6 will put this measure to work in analyzing patterns of interactions between politicians and the media 11 As Halberstam 1979 has recounted a similar regime change was occurring at about the same time at another traditionally Republican publication the Los Angees Times and with the same result By 1968 the venerable Republican mouthpiece was leaning left Also at about the same time at the traditionally Republican Chicago Tribune public incorporation led to the appointment of a professional editor and a decline in rightleaning partisanship on the news pages Burgos 1996 84 Figure 52 Trends in Pressinitiated criticism in elite news media 19481996 Percent of each candidate s coverage that is rated quotpress initiatedquot criticism Newsweek Democrat U I I I I I I I I I I I I 48 52 56 60 64 68 72 76 80 84 88 92 96 New York Times Republican Democrat 25quot Percent of each 20quot candidate s coverage that is 15 rated quotpress initiatedquot criticism 10 5 i i i 48 52 56 60 64 68 72 76 80 84 88 92 96 Democrat Time magazine o 48 52 56 60 64 68 72 76 80 84 88 92 96 25 Network TV news 20 Democrat Republican o I l l l l l l l l l l l l 48 52 56 60 64 68 72 76 80 84 88 92 96 Note Data are based on last six issues of the fall campaign in Time and Newsweek and October and November in the New York Times and on network TV news APPENDIX This book reports content analyses of presidential election coverage in Time and Newsweek magazines from 1948 to 1996 TV network news from 1972 to 1996 and New York Times coverage over the period 1948 to 1996 Only for the news magazines has it been possible to code the actual stories For the other media abstracts of media content have been used as described below All coding has been done by Mark Hunt a recent graduate of UCLA Technically Hunt is a professional research assistant to this project but he has done more work than I have been able to pay him for simply because he finds it interesting On two occasions I have made a careful evaluation of his coding and found the quality very high 12 Comparison of his coding with findings from other studies as well as the internal structure ofthe data he has created indicate that the reliability and validity of his work on this project have been high Note for example that when Hunt s coding is compared with results published in Patterson s study as in Figure 1 the correlation is 96 Coding of newsmagazines was based on an elaborate 39code scheme as briefly described above and available from the author upon request Because this monograph focuses on negativity it makes little use of data from the full coding scheme as all codes are usually collapsed to either quotmediainitiated criticismquot or quototherquot Some use is made ofthe horserace codes however The coding scheme was designed to capture overall coverage atthe level of each individual candidate rather than at the level ofthe campaign as a whole Hence any story that referred to both candidates has been counted as part of each39s coverage This has led to a considerable amount of what 12 Near the beginning ofthis research Hunt and I both coded 30 weeks of news magazine coverage of the presidential primaries The correlation between our codes aggregated at the level of weekly summaries was 96 In a later project unrelated to this one I wrote out coding instructions and on the basis of this written communication only asked Hunt and two other people to code some news magazine data One of the other coders had extensive coding experience and the other had none Aggregated at the level of weekly summaries their scores correlated at the level of 70 with no coder standing out as good or bad However Hunt39s coding proved more valid in practice for his coding the correlation with the dependent variable in the analysis was 71 while for the other coders it was around 50 See Zaller and Chiu 1996 especially note 9 p 404 and associated text Another evaluation project based on a sampling of Hunt39s coding for this project is in process and will be included as part of the published report of this research amounts to doublecoding at the level of the campaign as a whole even though it is singlecoding at the level of individual candidates For example the statement that quotNixon calls McGovern too radicalquot would be counted in Nixon39s coverage as positive partisan coverage because the press has reported Nixon39s campaign message and counted negatively in McGovern39s coverage as negative partisan coverage because he has been the target of a partisan attack though not a mediainitiated attack 13 The main part of my analysis is based on coverage at the level of individual candidates especially Figure 2 for which these counting rules are wellsuited But in few places sum the individual candidate scores in order to make statements about coverage at the level ofthe campaign as a whole The counting rules are not ideally suited for such statements Yet because I deal with rates rather than absolute amounts of coverage and since the major party candidates get fairly equal amounts of coverage it seems unlikely that these campaignlevel statements are misleading The 39code scheme used for the newsmagazines could not be reliably applied to the abstracts of New York Times stories Hence an abbreviated sixcode scheme was developed one of which codes was quotmediainitiated negativequot as described above The other codes were quotpositive horse racequot negative horse racequot quotother positivequot which included most of candidates39 own statements including partisan attacks quotother partisan negativequot including cases in which a candidate has been the target of an attack and quototherquot As with news magazines most stories and some story elements within a story have been assigned multiple codes For example a statement that quotNixon ahead of McGovern in Times Pollquot would be positive horse race for Nixon and negative horse race for McGovern A statement that quotNixon aide denies Washington Post story of Watergate breakinquot would be coded as quotother positivequot to credit Nixon39s denial and quotmediainitiated negativequot to indicate that the story originated with a charge by journalists Randomly chosen examples of these abstracts are given above in the text 13 But if Nixon said quotI39m the right candidate for Americaquot it would count in his coverage only Likewise if a story said quotNixon ahead of McGovern in critical statesquot it would count as positive horserace for Nixon and negative horserace for McGovern But ifthe story said quotNixon excites crowds on campaign trailquot it would be positive horserace coverage for him only Coding of networktelevision news on ABC CBS and NBC was based on news abstracts published by the Vanderbilt Television News Archive The abstracts provide less detail than the Times abstracts so data obtained from them may be less reliable as noted below The coding scheme distinguished ten categories 1 all references that the candidate would consider positive including quotes of his own statements and those of supporters 2 positive horserace for candidate 3 negative horserace for candidate 4 neutral horserace 5 horserace but direction if any unclear from abstract 6 neutral background or general campaign relevant news 7 candidate is target of partisan attack 8 event news that is bad for candidate eg report of faltering economy 9 mediainitiated criticism 10 uncodable The Vanderbilt abstracts have become more sparse in recent years as indicated by the fact that the amount of uncodable material rose from about two percent in the early period to about 12 percent in 1992 The uncodable material may have had an actual slant but its direction was not evident from the abstract In the analyses reported in this monograph uncodable material has been excluded from the denominator on which rates of negativity have been calculated This exclusion is tantamount to assuming that the uncodable material contained media negativity at about the same rate at the material that could be coded The Vanderbilt abstracts supply the total time for each story but not the amount oftime taken up by each part ofthe story The general rule was to treat each sentence that is text separated by periods as a distinct story segment and to assume that each segment received an equal amount ofthe total time for the story This procedure though a rough one was constant across all story types and so probably would not tend to create bias 14 The following example illustrates the counting procedure All punctuation is original but the division into numbered story elements reflects the coding rule 1 Studio Garrick Utley The candidacy of Ross Perot featured President Bush39s attack on Democratic presidential candidate Bill Clinton this week recalled 2 On Today PEROT comments ie a quote from Perot 14 Some exceptions were made when they seemed appropriate In cases for example in which a report was exclusively positive or negative the anchor39s introduction was coded as having the same tone as the report on the assumption that it would be setting up the material that followed 87 3 Perot39s ad on the American economy shown economic solutions in his bookM We Stand outlined on screen 4 Perot spokesman Orson SWINDLE says the people wanted to see Perot39s ad show again 5 New shorter Perot ad shown The first element which seems to be the anchor39s introduction to a story on Perot was considered neutral for Perot The remaining four elements which appear to present generally proPerot material without any suggestion of criticism were coded positive for Perot According to the Vanderbilt abstracts the whole segment ran 140 seconds so each ofthe five segments was assumed to be 28 seconds Since four of the five segments were rated positive for Perot 80 percent of story a total of 112 seconds was rated as quotother positivequot coverage of Perot Perot also got 14 seconds of neutral code for Utley39s introduction In addition Bush receives 14 seconds of quotother positivequot positive code for his attack on Clinton and Clinton as the target ofthe attack gets 14 seconds of nonpress initiated negative code Estimates of media negativity across these several media are moderately comparable for the years 1968 to 1996 as the following intercorrelation matrix indicates New York Times Newsweek Time magazine Network TV news 62 46 62 New York Times 83 75 Newsweek magazine 86 The average ofthese correlations is 59 The average of comparable correlations forthe period 1948 to 1964 is much smaller about 28 Chapter 6 The Rule of Anticipated Importance In the 1996 election Bill Clinton and Bob Dole were both vulnerable to media slams on a similar set of allegations Each candidate raised millions of dollars in soft and arguably illegal campaign money with substantial amounts of it coming from foreign sources Clinton s foreign contributions came principally from Asia Dole s from the Caribbean Also each was rumored to have cheated on his spouse For Clinton these affairs were mostly old news except for the one with Lewinsky which the media didn t yet know about For Dole the affair was old but not yet old news and the Washington Post s Bob Woodward made a major investment in investigating it including it must be assumed the rumor that Dole had paid for an abortion for his 1970s mistress One might suspect that with each candidate vulnerable to roughly the same extent for the roughly the same presumed offenses the mass media would make a point of being evenhanded in the investigation of both sets of allegations But this was not the case Despite the media s reputation for antiRepublican bias reporters paid almost no attention to Dole s alleged transgressions Even after the Washington Post had Woodward s meticulously researched account of Dole s affair in hand and ready for print it declined to run it Instead reporters focused their attention on Bill Clinton most notably the allegation that he had engaged in illegal fundraising which became the subject of intense media investigation in the last month of the campaign Because the news media did not seriously follow up on the Dole allegations it is impossible to know how comparable they were to the Clinton allegations But there is reason to believe that the allegations had some degree of substance1 This pattern of highly selective media negativity perfectly illustrates the Rule of Anticipated Importance Since Clinton maintained a commanding lead over Dole throughout 1 On Dole s questionable fundraising see Politics The contributors Foreign GOP Donor Raised Dole Funds Leslie Wayne New York Times October 21 1996 B8 Concerning the allegation that Dole secured an abortion for his mistress see Press Clips James Ledbetter Village Voice November 5 1996 p 20 1996 there was never a point at which Dole s anticipated importance warranted a heavy investment of resources in attacking him I should perhaps add that the Rule of Anticipated Importance does not focus on media negativity per se The claim rather is thatjournalistic resources of all kinds including space and air time depend on the anticipated importance ofthe object of coverage The present chapter develops a series of empirically observable implications of Rule of Anticipated Importance on resource allocation and presents the evidence necessary to evaluate them Most refer to highly specific empirically measurable features of presidential elections There is also a special analysis ofthe candidacy of Ross Perot No single piece of evidence is definitive and many ofthe empirical regularities that I cite could be explained as well by other theories I contend however that no other theory can as effectively explain the range and number of empirical regularities as the Rule of Anticipated Importance PRESIDENTIAL PRIMARIES As discussed earlier ordinary voters do not want to study the records of every senator and governor and billionaire who may decide to run for president They want to know only about the two or perhaps three who have a realistic chance of winning As also explained earlier journalists are delighted to provide this kind of screening service since it gives them an opportunity to exercise journalistic voice Yet nothing in my theory of media politics implies thatjournalists will make these screening decisions arbitrarily On the contrary their incentives are to accurately anticipate what will happen and to report it in a timely fashion These incentives derive from two sources one internal to journalism and the other external The external incentive is the quotcry wolfquot syndrome Ifjournalists were to regularly pick out and boost weak candidates while ignoring strong ones too many ofthe favored picks would fall flat and too many of the ignored ones would do well thus eventually embarrassing the profession2 So journalists have a collective incentive to get the story right Reinforcing this collective incentive and perhaps the key to it is the cutthroat competition that exists among individual journalists Although journalists like 2 The sensitivity of a minimally attentive and not very selfassured public to failures of press prediction should not be overestimated as the novelist George Orwell drives home in his book 1984 it is likely however that some sensitivity exists stock market investors often run in packs each individual reporter like each individual investor has an incentive to find undervalued candidates and invest in them Thus poor choices by existing pack leaders create opportunities for wouldbe pack leaders and journalism is full of such ambitious individuals The collective interest ofjournalists in making good picks in combination with individual interests in pointing up weak picks creates a strong incentive for journalists to pay attention to the evidence as they decide which candidates to cover and which to ignore This leads to my first deductive inference from the Rule of Anticipated Importance which concerns the earliest stage of presidential elections the period of the socalled quotInvisible Primaryquot Buell 1996 This is the period prior to the Iowa and New Hampshire contests when candidates have begun to campaign but primary election balloting has not actually begun The inference labeled D5 for quotDeductive Inference Number Fivequot is as follows D5 The amount of coverage allocated to candidates in the Invisible Primary will be roughly proportional to standard indicators of political strength because stronger candidates have greater anticipated future importance The most obvious indicator of a candidate39s strength in the Invisible Primary is his or her standing in the polls which regularly query citizens about which candidate they favor for their party39s nomination One would therefore expect to observe a strong relationship between poll standing and amount of press coverage Buell 1996 and Mayer 1996 have presented a good deal of evidence tending to support this inference and Figure 61 presents some more For all candidates who have run in the Invisible Primary between 1980 and 1996 and received any degree of support in the polls Figure 61 shows the relationship between Gallup Poll standing in the last preDecember poll in the year prior to the election and amount of coverage in the New York Times in that December Poll support and amount of coverage are measured as fractions of all poll support and all coverage within a given party respectively INSERT FIGURE 61 ABOUT HERE A handful of points in Figure 61 are labeled as aids to interpretation In 1991 for example President Bush was favored by 78 of Republicans in the last preDecember Gallup Poll and received 70 percent Figure 61 Poll standing and percent of coverage in New York Times 100 39 80 Bush 92 o Percent ofParty Coverage in 60 quot Clmton 92 Carter 80 December New York Times 40 O quotquot 20 u 339 0 0 Jackson 88 x 0 go 0 Brown 92 0 I I I I I I 0 20 40 60 80 100 Percent of party members supporting candidate for nomination in latest preDecember Gallup Poll Note Estimate of coverage based on number of references i an York Time sabstracts poll coverage omits Don39t Know in calculation of support rates of the campaign coverage allocated to Republican candidates in the New York Times The simple correlation between December coverage and preDecember poll standing is 81 The corresponding correlation between December coverage and January poll standing is even higher 91 This suggests thatjournalists are as my theory would suggest doing a betterjob of anticipating the future than reflecting the present There is however an alternative interpretation for these simple correlational data which is that New York Times stories along with similar coverage in other media cause January support for candidates rather than anticipate it Since I cannot rule out this possibility on the basis of the present evidence I do not claim that these data prove the Rule of Anticipated Importance but only that they are consistent with it This evidence however is only the initial piece of a larger pattern of theory and data Let me therefore continue with my argument Although reporters have strong incentives to pay attention to the polls as indicators of a candidate39s future importance there is no expectation that they will be slaves of the polls for two reasons First reporters understand that polls are highly fallible indicators of political strength and that particularly in the early stages of presidential contests they may measure a candidate39s name recognition rather than any real support Second reporters do not want to be overly dependent on any single source since this would limit their own ability to decide what to cover and ignore Hence in making coverage decisions reporters give weight to other qualities especially intangible ones that require journalistic judgment such as a candidate39s capacity to give a good speech to attract a good professional staff to win support among party activists and to perform well in front of television cameras Here for example is how veteran reporter Jules Witcover explains his decisions about which presidential contenders to cover and ignore during the period of the Invisible Primary If a guy is a bomb it39s our job to ignore him If I have decided that a guy doesn39t deserve any more attention than I give him it39s not because of the polls It39s because I39ve been out there I39ve heard what people say and I39ve heard what the candidate is doing and I39ve made a judgment that this guy is just not cutting it3 3 From quotThe Campaign for Page Onequot PBS Frontline documentary report 1984 92 In light of these considerations the Rule of Anticipated Importance leads to a second deductive inference D6 Some candidates will get considerably more press attention than standard indicators of political strength alone would seem to indicate while others will get considerably less and further these departures will be intelligible in terms of clear strengths and weaknesses of the affected candidates A close look at Figure 61 reveals three points that are far enough from the trend line to be considered outliers Jerry Brown and Jesse Jackson in 1992 who got little coverage in the Times despite relatively strong poll results and Bill Clinton who got heavy coverage despite poor poll numbers All three outliers are readily intelligible in terms of the Rule of Anticipated Importance Brown and Jackson though riding high in the early polls had shown in previous presidential races that there were sharp limits to their appeal Brown because of his image as the flaky quotGovernor Moonbeamquot and Jackson because too few whites would support an AfricanAmerican minister with strongly liberal credentials Meanwhile Clinton though unknown to most ordinary citizens was wellknown to the national press which judged that he had the political skills and ideological flavor necessary to go all the way I do not claim that these or otherjournalistic judgments were necessarily correct but I do contend that most informed political observers would grant that they were at least plausible just the kinds of judgments according to my theory that rationally ignorant citizens want reporters to make and that reporters relish making Another test of D6 is that the candidates who go on to win a party nomination ought if reporters are doing a good job of anticipating future events to get somewhat more coverage in December than their poll standings alone would justify4 Candidates with strong financial backing ought also to get more coverage since money helps win votes once balloting begins Both expectations are met Controlling for poll standing Table 61 shows that candidates who went on to win their party nomination in the period 1980 to 1996 got 123 percentage points more December coverage in the New York Times than 4 I thank George Tsebelis for suggesting this test candidates who did not win a party nomination Fundraising in the year prior to the election as reported to the Federal Election Commission also predicts heavier press coverage5 INSERT TABLE 61 ABOUT HERE The Invisible Primary ends with the primary contests in Iowa and New Hampshire whereupon the Visible Primary commences This phase ofthe nomination process runs for several months in a drawn out sequence of statelevel contests Since 1976 it has been conventional wisdom that the earliest contests winnow down the field of candidates to a relative handful and may in addition create sufficient momentum to propel an early winner all the way to nomination Matthews 1978 Polsby 1983 Bartels 1988 In the most rigorous analysis of momentum Bartels 1988 has been able to quantify the boost that winning the New Hampshire primary gives to overall chances of winning the nomination Given that early primary contests break many campaigns while making a select few the Rule of Anticipated Importance leads us to expect that D7 Journalists will allocate more coverage to early primaries than to late ones since early primaries contribute more marginal information to who is likely to win nomination There is much evidence consistent with this expectation especially as regards the Iowa and New Hampshire primaries which almost always get highly disproportionate treatment see Adams 1987 But quotalmost alwaysquot indicates that exceptions do occur and these exceptions are as in the analysis ofthe residuals in Figure 61 quite informative In 1992 Tom Harkin was the clear winner of the Iowa caucuses but got little media attention and no momentum out of it The reason was that Harkin was a Senator from Iowa and this led reporters to interpret the result as the victory of a favorite son rather than a rising star The media as this indicates are not impressed by early wins per se they are impressed by early wins only insofar as they augur genuine political strength which the victory of a favorite son does not 5 Something like the Rule of Anticipated Importance seems also to be operating in congressional races where local reporters try to cover candidates who have a chance to win and to ignore others See Westlye 1991 Table 61 The effect of poll standing and subsequent capture of party nomination on December coverage in New York Times 19801996 E Him Poll standing in 044 58 471 Later Wins Party Nomination 0 1 91 22 179 Fundraising Year Prior to Election 018 20 235 in millions Intercept 50 N 46 Adiusted rsquare 71 Note Dependent variable is percent of party coverage i New York Time sgoing to each candidate Poll standing and funding data are based on latest pr e December Gallup poll quotdon t knowquot responses omitted in calculation of candidate support rates as reported in Mayer 1996 Table 21 Table 23 and Political Hotline service New York Time scoverage was estimated by the number of references i Wew York Times Abstract swhere semicolons delineated references The 1992 New Hampshire primary offers another example of the same lesson With Bill Clinton wounded by press allegations about extramarital affairs and draft evasion Paul Tsongas won the Democratic contest But the national press corps was unimpressed They considered the former Massachusetts Senator a merely regional candidate who was too lacking in political charisma to win a major party nomination Hence they refused to give Tsongas who was noted in several stories as resembling TV39s Mr Rogers the bonus coverage that winners ofthe New Hampshire primary traditionally get quotI just don39t see Paul as the real storyquot said one reporter in asking his superior to assign him to cover someone else quotI don39t know ten reporters in one hundred who think he can be the nomineequot said another Rosenstiel 1993 p 1356 I take the examples of Harkin and Tsongas like those of Jerry Brown and Jesse Jackson as prime cases of the Rule of Anticipated Importance in action Reporters pay attention to objective indicators of likely success notably poll standings key victories and fundraising prowess but not in a mechanical fashion They also take account of subjective judgments about who is in Witcover39s terms quota bombquot and quotnot cutting itquot My claim thus is that both the general tendency of reporters to allocate heavy coverage to the winners of early primaries and reporters39 departures from this tendency in the particular cases of Harkin and Tsongas tend to support the Rule of Anticipated Importance As l have laid out this argument the cases of Harkin and Tsongas are classic examples of exceptions that do tend to prove a general rule The general rule in D7 is that reporters allocate extra coverage to early contests because they believe they contain more information about future events than other contests But when as in the case of Harkin in Iowa and Tsongas in New Hampshire reporters have particular reason to believe that early contests are poor portents of future events they withhold the usual bonus coverage thereby revealing that the motivation underlying the general rule Another wellrecognized regularity ofthe primary election process is that poorly rated candidates who suddenly do quotbetter than expectedquot receive large amounts of coverage in the short term Thus as Polsby 1983 has observed the most important candidate to beat in any election is always that elusive fellow named quotexpectedquot The Rule of Anticipated Importance provides a ready explanation for this phenomenon D8 Since candidates tend to be covered in proportion to media estimates of their future success candidates who suddenly do 39better than expected will have been underestimated and hence undercovered in the past and so will merit a ration of unusualy heavy coverage The first four deductions from the model attempt to explain which candidates get covered and how much I turn now to the nature of this coverage Which candidates get covered favorably and which unfavorably A central claim of my model is that reporters seekto make a distinctive journalistic contribution to the information that gets reported as news Reporters can on certain occasions expressjournalistic voice by injecting positive information about candidates into news stories Explaining why a quotdark horsequot candidate is likely to do especially well in the future or why a candidate has done quotbetter than expectedquot in the New Hampshire primary are examples of occasions on which journalists will eagerly report positive information about presidential candidates But such occasions are uncommon especially for candidates who have established themselves as pack leaders Journalists count on such candidates to provide morethanadequate amounts of positive information about themselves and therefore concentrate their own energies on finding negative information about them And in keeping with the Rule of Anticipated Importance they may be expected to concentrate especially on information about little known candidates because the marginal value of their contribution is larger and stronger candidates because stronger candidate have greater future importance This leads to two further deductions D9 D10 For top tier primary candidates reporters will offer pressinitiated negative information a in positive proportion to how well the candidate is doing in the horse race and b in negative proportion to how much is already know about the candidate For primary elections from 1976 to 1992 Zaller and Hunt 1995 examined Time and Newsweek coverage of all candidates who made it into the quottop tierquot that is the elite circle of candidates who won or came close to winning a major primary and were generally considered potential nominees5 The 5 A total of 19 candidates in both parties made our cutoff The two weakest candidates who nonetheless made it into our study were Jerry Brown in the period after he won several primaries late in the 1976 contests and Pat Robertson in the period after finishing second in Iowa and winning the Michigan 96 results were consistent with D9 and D10 Within the top tier stronger and less known candidates got the highest rates of pressinitiated criticism7 PRESIDENTIAL ELECTIONS Similar considerations come into play in coverage of general elections but lead to different expectations As in the primaries stronger candidates ought to attract more pressinitiated criticism The argument here can best be stated as an interrogative Why spend time flogging a dead or dying horse Why bother to criticize someone who39s going to lose Reporters ought rather to concentrate their fire on candidates who have a future If it sounds strained to suggest that the press is especially tough on front runners and incumbents because it cares about their power to influence future events perhaps it will help to think about the matter in terms of resource allocation Pressinitiated criticism is often a form of enterprise reporting It utilizes the best journalistic talent for long periods oftime with no certain returns Why squander such resources on hopeless challengers nonincumbents and third parties with no capacity to affect outcomes Who cares what their shortcomings are Here is yet another version of the same argument From the crassly commercial point of view of quotselling newspapersquot why would anyone waste time exposing the shortcomings of losers No one would care The inference therefore is D11 Pressinitiated criticism of candidates in general elections will be positively associated with political strength Because as we have seen the Rule of Anticipated Importance implies thatjournalists should provide more information about less known figures it might be further argued that reporters should concentrate their attention on the lesser known of the two general election candidates namely the candidate ofthe nonincumbent party Every election in the last 48 years has been contested by either an incumbent caucuses in 1988 Candidates remained in the study only during the period in which they remained viable 7 Hagen 1996 shows that candidates are more likely to attacktheir rivals when the rivals become more successful That phenomenon is separate from my empirical result which concerns pressinitiated criticism only president or in a few cases an incumbent vicepresident This argument is sound as far as it goes but there is an important countervailing argument Incumbent candidates are newsworthy not because they are less known but because they are the current occupants of very powerful offices thus giving them automatic quotanticipated importancequot Incumbent presidents and also incumbent vicepresidents when they run as presidential candidates also often carry baggage from their term in office that attracts press scrutiny even when the incumbents are wellknown For example Richard Nixon was suspected in 1972 of Watergate crimes Jimmy Carter bore responsibility for handling the lran hostage crisis in 1980 and George Bush had to live with press suspicions over the IranContra scandal in 1988 all of which matters attracted press criticism Our expectation then is that one ofthe candidates will attract more scrutiny because of the power and baggage that come with incumbency and the other will attract more scrutiny because he is the lesser known candidate thus yielding no clear prediction beyond D11 These data on media negativity as described in Chapter 4 will constitute the dependent variable in the test of D11 that follows8 The main independent variable will be quotpolitical strengthquot which can be measured in a variety of ways from poll and electoral data Rather than pick the one measurement that quotworks bestquot which tends to capitalize on chance variation in the data l have used an average of the most plausible measures9 As control variables I use party year and incumbency status None of 3 Actually the data in Figure 2 cover different time periods for different media For TV and the New York Times they cover the entire fall campaign from September 1 to election data But owing to resource constraints the Time and Newsweek data include only the final six campaign issues One reason for example that Time and Newsweek were so much more critical of Clinton in 1996 than the other media is that the pressinitiated controversy over illegal fundraising broke in the last weeks ofthe campaign and so constitutes a larger fraction of coverage in the newsmagazines than in the other media Both to eliminate this sort of problem and to facilitate another test to be reported below my analysis of press criticism in this section will use data from October 1 to election day 9 The dependent variable in the analysis will be press criticism during October as explained in the previous note Presumably the measure of political strength should refer to the same period To get such a measure I used the average of Gallup39s early October poll and the actual election returns It was unclear however whetherl should use the average of the candidate39s support in the two indicators or the average of a candidate39s twoparty support thus excluding undecided survey respondents and supporters of third party candidates from the calculation of the political strength of the two major party candidates My solution was to average the two averages or more clearly put to average each candidate39s 1 percent support in the October poll 2 percent of the twoparty vote in the October poll 3 percent support in the election and 4 percent of twoparty vote in the election In a recent paper Bartels 1997 suggests averaging coefficients obtained from running separate regressions for each potential measure The implicit assumption in this recommendation is that each potential measure defines a unique model In the present case however I am undecided between alternative indicators rather than alternative models which is much less serious than uncertainty over 98 these variables is required by theory but the test of D11 will be more credible if it controls for these possibly confounding factors The appropriate measure of incumbency however is uncertain It could count presidential incumbents incumbent presidents and vicepresidents or something in between Given this uncertainty I again resort to the average specifically I have created a measure that gives one point to incumbent presidents a half point to incumbent vicepresidents running for the top job and zero to nonincumbents The results divided into two time periods to reflect the visual break in the data in 1968 are shown in columns 1 to 4 of Table 62 for each ofthe four media outlets l have used logit in this analysis to take into account the floor effects that are apparent in Figure 2 For the earlier period the results show a time trend toward more negativity in two ofthe media and a party bias in favor of Republicans in Time magazine but no consistent tendency for stronger candidates to get more criticism The time trend moreover is almost wholly due to Time magazine39s attacks on Johnson in 1964 The one medium with a coefficient for Political Strength that approaches statistical significance has the quotwrongquot sign That is in Time stronger candidates get lei criticism Hence D11 fails for the earlier time period In the second time period however Political Strength has the expected effect in all four media and achieves statistical significance in three of the four This supports D11 Also note however that all four media now register a significant proDemocratic bias I shall return to this disturbing indication of media bias in the next chapter10 INSERT TABLE 62 ABOUT HERE To test the overall statistical significance ofthese findings l transformed the criticism scores in each media to a common mean averaged them and made a logit transformation Results of a test based on these overall scores are shown in the fifth column of Table 62 As can be seen Political Strength has a moderate overall effect with a standardized coefficient of 40 The tratio for this coefficient implies a onesided pvalue of 015 models so my form of averaging seems preferable and not only because it avoids the Draconian penalties built into Bartels39 procedure 10 Patterson 1993 Chapter 3 reports several figures that relate press negativity to a candidate39s standing in the polls But owing to differences in his measurement of press negativity as described in Appendix A and his use of time categories I cannot be sure ofthe applicability of his results to D7 99 Table 62M0dels of pressinitiated criticism 19481996 Newsweek Time NY Times TV All media 1948 to 1964 Year 5 070 020 066 na 056 0t0 4 b 019 014 017 016 t 2 53 102 234 238 Incumbent 009 021 039 030 0 5 1 008 046 032 027 0 27 090 1 14 1 05 Political strengtha 030 038 030 002 percent vote share 001 005 001 044 092 1 64 091 1 61 Democratic Candidat e 021 079 032 050 0 0r 1 016 150 023 039 074 386 111 205 Intercept 309 188 378 362 Adjusted r2 31 64 28 50 Number ofcases 10 10 10 10 1968 to 1996 Year 5 045 048 034 001 036 5 to 13 b 018 013 010 000 010 t 242 274 229 008 258 Incumbent 012 041 006 064 031 0 5 1 024 056 009 098 043 058 210 036 479 201 Political strengtha 035 018 049 034 040 percent vote share 005 002 005 004 004 168 090 298 250 252 Democratic Candidat e035 025 054 020 036 0 0r 1 064 031 072 028 046 176 131 348 157 243 Intercept 647 470 543 451 523 Adjusted r2 47 53 68 78 70 Number ofcases 16 16 16 16 16 Note The dependent variable is the log odds of the proportion of each candidate39s October coverage that has been coded pressnegative The rst entry in each cell is the standardized coef cient the second is the unstandardized logit coef cient and the third is the tratio a An average of early October poll strength and nal vote share as described in text Although this analysis has grouped 1964 among in the earlier time period it might make sense to regard it as a transitional election On the one hand the press was more critical than in any previous election in this dataset which was as it turned out an omen of a big change to come But on the other hand the increased negativity was confined to one outlet Time which was still a Republican organ and therefore seems better understood as Harry Luce39s last partisan attack on a liberal Democrat than as an instance of Johnson39s anticipated importance11 Meanwhile Newsweek and the New York Times were actually more critical of Goldwater the landslide loser than of Johnson a clear indication that they were not yet paying attention to future importance On the whole then 1964 seems a better fit with the earlier press regime than with the later one12 The notion that a candidate39s standing in the political horserace affects the inclination ofjournalists toward criticism fits quite well with qualitative evidence During the 1996 campaign Margaret Warner of the NewsHour asked the press secretaries of recent losing candidates whether reporters had treated losers differently The following exchange ensued with Marlin Fitzwater who was President Bush39s press secretary in the losing effort of 1992 and Maxine lsaacs who played the same role for Walter Mondale in 198413 MS WARNER Does the press start to treat you differently as it looks worse MR FITZWATER Well they treat you better I think when they see disaster ahead MS WARNER A sudden wave of sympathy comes over MR FITZWATER They39re at their worst when you39re on top winning and they39re going for you and uh so at the end you know reporters are coming up and saying you know thehe has really done a great job what a great guyand he has this dignity and inner strengthyeahwel let39s see that in print MS WARNER Do you find the same thing surpassed the partisan venom of Republican outlets during Franklin Roosevelt39s time 12 These results in Table 62 are indeed sensitive to whether 1964 is counted among the early or late cases That is if 1964 is counted among the early ones there is no time trend toward greater negativity in the early period And in the later cases the pvalue on Political Strength coefficient in the overall model falls from 037 to 029 though remaining statistically significant at exactly the same level This regression test omits TV from the comparison since TV data are not available in 1964 and also restandardizes the criticism scores to include 1964 a wholly technical adjustment 13 November1 1996 11 Although Time39s attack on LBJ produced the most negativity to date in these data I doubt that it 100 MS ISAACS Absolutely THIRD PARTY CANDIDATES The Rule of Anticipated Importance has implications for Third Party candidates as well as Major Party ones Most Third Party candidates receive scant attention in the mass media The reason as Rosenstone Behr and Lazarus 1996 p 35 write is that broadcasters and publishers do not thinkthey warrant attention As James M Perry of the Wall Street Journal put it quotWe base our decision on the simple proposition that readers don39t want to waste their time on someone who won39t have a role in the campaign We39re not going to run a pageone spread on a fringe candidate We don39t have a multiparty system Until we do nobody39s going to cover these candidatesquot The presence of numerous potential quotthird partyquot candidates in each election cycle in 1980 some 150 presidential aspirants filed preliminary statements with the Fair Election Practices Commission virtually forces reporters to ignore most of them Yet some Third Party contenders manage to get the press to take them seriously How do they do it The straightforward implication of the Rule of Anticipated Importance is that reporters want to cover candidates who have strong potential support and are therefore destined to do reasonably well if covered and to ignore movements that have no realistic chance of gaining support no matter how much they are covered The press in other words wishes to report and if possible to anticipate the news but not to make it This argument leads to the inference that reporters will cover Third Party candidates in direct proportion to how well they expect them to do in the election If therefore we imagine a scatterplot in which quotpercent public supportquot is on the Xaxis and quotpercent share of coverage is on the Yaxisquot we should observe that all points cluster along a line with intercept 0 and slope 1 Thus if a candidate had 10 percent of public support he would get 10 percent of coverage if he had 20 percent of public support he would get 20 percent of coverage and so on The line defined by such a series of points might be called the quotequal share linequot But it is not quite so simple In many cases Third Party candidates who have no chance to win may nonetheless affect the election either by tipping the balance toward one ofthe major party candidates or 101 throwing the outcome into the Electoral College In addition Third Party candidates tend to be new to national politics and hence relatively unknown This means that the public39s interest and therefore reporters39 interest in new information about them will be high From the Rule of Anticipated Importance both ofthese auxiliary considerations lead to the expectation that Third Party candidates may get somewhat more coverage than their level of political support alone would suggest Hence the inference I reach is D12 Third Party candidates will be covered at a level somewhat above what would be expected on the basis of their public support alone if they are new to national politics or capable of affecting the outcome of the major party contest Data on the allocation of New York Times coverage to major Third Party candidates from 1912 to 1992 are shown in Figure 62 The results except for very minor Third Party candidates are fully consistent with D12 INSERT FIGURE 62 ABOUT HERE Other data confirm that this pattern of Third Party coverage is not unique to the New York Times In the last five weeks of the 1948 campaign Henry Wallace and Strom Thurmond each got seven percent or more of total campaign coverage in Time and Newsweek magazines but only two percent of the vote each in 1968 George Wallace got 28 percent of the coverage in the two news magazines but 135 percent ofthe vote in 1980 John Anderson got 10 percent of coverage but 66 percent of the vote and in 1992 Ross Perot averaged 23 percent of newsmagazine coverage but 19 percent of the vote For network TV news the percent of all October coverage was 26 percent for Wallace in 1968 11 percent for Anderson in 1980 and 22 percent for Perot in 1992 Thus in each ofthese cases in all four media Third Party candidates were covered at rates that were proportional to political strength yet somewhat above the quotfair sharequot coverage line These departures from a rule of strict quotequal sharequot coverage are so consistent as to make it unlikely that reporters were actually trying to achieve it Reporters obviously pay considerable attention to viability in allocating coverage but equally obviously it was not the only thing they were paying attention to 102 Figure 6 2 Press coverage of major Third Party candidates Percent of all New York Time 3 coverage devoted to Third Party candidate 10 T Roosevelt Equal 30quot LaFollette Share LIne 25 George Wallace P633039 20quot Henry Wallace 15quot C Anderson no 0 Perot 96 0 5 10 15 20 25 30 Percent of Vote Won by Third Party Candidate Lemke Thurmond ln striking contrast to these results however all four elite media gave Perot substantially less coverage in 1996 than his percentage share of the vote which was 9 percent This result also fits D12 ln Perot39s second run for the presidency he was no longer novel and in contrast to the 1992 race the 1996 race was so much a runaway that no Third Party candidate could affect the outcome of the Major Party contest There is of course a chickenandegg problem here Media coverage could as l have suggested reflect reporters39 anticipation of election results But it could also be a cause of election results in the sense that more coverage could lead to more support rather than as I maintain vice versa Hence I claim the data in Figure 62 as consistent with my model rather than as tending to prove it Third Party candidates like their major party brethren often suffer pressinitiated criticism The incidence of such criticism ought to be consistent with the Rule of Anticipated Importance as follows D13 Pressinitiated criticism of Third Party candidates will be positively associated with political strength This expectation holds Within each of the four media outlets separately and all ofthem together Political Strength in the early October Gallup poll and vote share in the November election were significant predictors of Media Negativity14 On average each additional percentage point ofthe Political Strength led to an additional 8 percentage points of Media Negativity15 Note that the dependent variable in this analysis is the proportion of all coverage that is pressinitiated criticism Hence the result is more than a mere repetition ofthe previous finding that amount of coverage is proportional to political strength Stronger candidates get more coverage and a larger proportion of that coverage is negative This result is important in two respects First it suffers no chickenandegg problem That is it is unlikely that high levels of press criticism are the cause of political strength rather than vice versa And second the finding goes against the grain of a plausible alternative hypothesis namely that press criticism of stronger candidates is motivated by a desire to make the race closer and therefore more 14 In the allmedia test used a fixed effects model that controls for correlated errors within each media and within election years For both poll strength and vote share the effect of Political Strength was significant at about p035 onetailed The fixed effects model was an OLS regression that included dummy variables for each election year and each medium 15 The intercept in both models was about 7 percentage points 103 exciting rather than as l have claimed by anticipations of future importance For ifjournalists were motivated only to make the horse race closer as a counterargument would assert increases in Third Party strength would lead journalists to boost Third Party candidates in general elections and to withhold negative coverage up to the point at which they take the lead in the race Yet as the data indicate this is not how journalists behave I stress that both of these points are clean findings of considerable theoretical interest THE SPECIAL CASE OF Ross PEROT On February 23 1992 Ross Perot appeared on CNN39s Larry King show and said that he would become a candidate for President Unsurprisingly the national media failed to take notice Perot had no prior electoral experience no mass following and no clear plan for capturing the White House beyond a promise to finance a firstclass campaign from his private fortune By any reasonable standard his quotanticipated future importancequot was negligible Nonetheless his candidacy caught on A month later two national polls found that about 20 percent of the public preferred Perot in a threeway race with Bill Clinton and thenPresident George Bush At that point the national media began to pay attention Over the few weeks virtually every important news outlet profiled the candidate including his bold plan for a 50cent gasoline tax to help balance the federal budget Riding this wave of mostly positive publicity Perot pulled ahead of Bush and Clinton in the polls in late May and through about midJune his lead continued to edge upward But media coverage then turned sour massively so According to Zaller and Hunt 1995 Perot got far more pressinitiated criticism than any presidential candidate in the history of the new nominating system Perot39s support in the polls now plummeted and in midJuly he withdrew from the race What is most notable about the events that launched the Perot candidacy in 1992 is how disconnected they were from the traditional bases of power in American politics or any base of power whatsoever As an Independent Perot had no ties to the political parties and hence no builtin partisan support As the selfproclaimed enemy of special interests he got none ofthe group endorsements and organizational support that often sway voters Nor did any nationally prominent politician embrace his 104 candidacy The whole spring campaign in both its ascent and descent phases was driven by words and images carried in the mass media It was in other words an unusually pure case of media politics Given this my theory of media politics ought to have something to say about the Perot candidacy And indeed it does All but one of the major turns in media coverage of both Perot39s 1992 run for the presidency and his 1996 campaign are as I will now seek to show wellexplained by my theory of media politics in general and the Rule of Anticipated Importance in specific The one important feature of the Perot phenomenon that the theory cannot explain is what happened in the first weeks after Perot39s announcement on the King show An obscure but energetic political activist and organizer Jack Gargan had been urging Perot to run for president for some months In November 1991 Gargan had induced Perot to speakto members of his group and in a newsletter mailed to 100000 group members at about the time of Perot39s appearance on the King show Gargan had urged group members to contact Perot and ask him to run Also producers ofthe morning talk shows on CBS NBC and ABC invited Perot to appear on their shows after seeing him on CNN Perot39s performance on these and other news shows was by all accounts extremely persuasive and it launched his campaign In particular the combination ofthe Gargan newsletter and the TV show appearances rapidly generated 11 million phone calls to Perot39s Dallas headquarters from people wishing to volunteer in his campaign It also seems to have netted Perot about 7 percentage points of the roughly 20 percent support he had in the earliest polls Hunt and Zaller 1995 The reason my theory cannot explain this development is that the primary actors were not the national news media but a grass roots organizer and the producers of morning quotInfotainmentquot programs But the theory can explain what came afterward The Rise ofRoss Perot in spring 1992 According to the Rule of Anticipated Importance reporters cover candidates who appear likely to be successful The three indicators of likely success that I discussed earlier were poll standing financial viability and the demonstration of raw charisma By late March Perot had abundantly succeeded on all three indicators whereupon coverage began D5 D6 15 The group was THRO Throw the Hypocritical Rascals Out which urged the defeat of all incumbent members of Congress 105 The great volume of this coverage may be explained by D6 See Zaller with Hunt 1994 for further discussion ofthe emergence of Perot Although theories based on the notion that the press pays attention only to candidates ofthe two major parties or quotresponsiblyquot screens out unsuitable candidates might be surprised or embarrassed by the coverage given to Perot my theory is not since it makes no reference to such factors After Perot39s candidacy takes off media coverage turns extremey negative According to D9 and D10 candidates who emerge as viable presidential candidates are criticized by the press in positive proportion to political strength and in negative proportion to their prior experience In May and June Perot was leading in the threeway race for president and had no prior political experience Thus the extremely heavy criticism is again consistent with expectations from the model Perot got generous amounts of coverage in the 1992 general election but stingy coverage in 1996 In particular he got more than his quotfair sharequot of coverage during the 1992 fall campaign when he was a new force in American politics and seemed capable of affecting the race between Bush and Clinton and less than quotfair sharequot coverage in 1996 As explained earlier this is consistent with D12 The press was tougher on Perot in 1992 when he was stronger than in 1996 On the network news 85 percent of Perot39s coverage was pressinitiated negative in 1992 compared to 37 percent in 1996 For Time these figures are 27 percent and 11 percent for the New York Times they are 14 percent of all coverage and 6 percent These data are consistent with D13 which holds that the rate of press initiated criticism is proportional to political strength Newsweek however fails to show this pattern 30 percent of Perot39s 1992 coverage was pressinitiated negative compared to 40 percent in 1996 This failure of expectation does not seem very serious however in light of drastic reduction in Perot39s coverage in 1996 when it was only 37 percent of election coverage compared to 30 percent of all election coverage in 1992 And overall evidence suggests D13 holds strongly as shown in Table 63 The October 23 1992 edition of the MacNeilLehrer NewsHour provides an interesting insight into journalistic criticism of Perot Jim Lehrer opened the usual Friday night political discussion with Mark Shields and David Gergen by noting that quotthe big news ofthe week is the surge in the the Perot surge There is some indication now that it may have stoppedquot But Shields and Gergen disagreed that the 106 surge has stopped offering several reasons to believe that Perot remains a force in the election and in American politics generally This part ofthe discussion ends with Shields observing quotBut he39s had a free ride He Perot hasn39t had the kind of media scrutiny that 17 Here Lehrer interrupts to describe pressure from the Bush campaign on the media to be tougher on Perot The interesting point is Shields39 perception that the press was giving Perot a quotfree ridequot Although this was scarcely true in a literal sense it may well have been true in another and more important sense Given that quotthe big news ofthe weekquot was Perot39s continuing quotsurgequot Perot may have been getting less quotscrutinyquot than by the Rule of Anticipated Importance he merited But the scrutiny deficit did not last long The next day at a Saturday taping of an appearance for the 60 Minutes show reporter Leslie Stahl induced Perot apparently through a ruse Goldman et al 1993 p 595 to say on camera that he had quit the race in July because of fear that the Republicans planned to sabotage his daughter39s wedding His comments were released as a wire story that evening critically discussed on the Sunday morning talk shows and became the basis of extremely critical news reports on Monday on all three networks programs Coverage of Perot39s remarks that Monday became in fact the single most negative night of TV news for any candidate in October Zaller and Hunt 1995 thereby causing a fivepoint drop in public support for Perot over the next few days and ending his surge The upshot then is that both the amount of Perot39s criticism in 1992 and its timing reflected Perot39s political strength or as I prefer anticipated importance THE SPECIAL CASE OF HORSE RACE COVERAGE One ofthe first deductions from my model was that a rational citizen would know or at least be able to intuitively sense that elections are likely to have more effect on him than he is likely to have on them That is the chances that an election may affect individual welfare taxes benefits military obligation are quite real whereas the chances that an individual can affect an election outcome are quite negligible In light of this asymmetry citizens should be more interested in consuming coverage about 17 The second part of this comment referring to press scrutiny does not appear on the Nexis transcript of the show but is clearly audible on my VCR tape of it 107 who is likely to win than issue coverage helping to inform them on how to cast a wise vote I therefore infer that D14 Reporters will provide more horserace coverage than issue coverage My data collection was not designed to test this proposition Although it did measure the frequency of horserace references in some media it made no effort to measure issue coverage However Thomas Patterson39s study of media behavior Out of Order attempts to measure both Patterson39s 1993 conclusion is that The dominant schema for the reporter is structured around the notion that politics is a strategic game When journalists encounter new information during an election they tend to interpret it within a schematic framework according to which candidates compete for advantage p 56 Patterson39s claim here is not that the bulk of information conveyed in the news concerns the political horse race it is rather that news tends to be framed or organized by concern about the strategic game between the candidates Thus issues may be discussed but discussion is typically framed in terms of the impact of the issue on the election outcome rather than as a guide for citizens in choosing between the candidates or choosing the best policy Patterson finds that in recent years about 70 percent of all New York Times coverage of presidential campaigns is framed in terms of what Patterson calls the quotgame schemaquot as against about 15 percent of what he calls a quotpolicy schemaquot and about 15 percent of coverage in other schemas13 I take this as empirical support for D14 The Rule of Anticipated Importance has a further implication for the kind of coverage candidates get The closer a candidate gets to actually winning the election and assuming office the greater the anticipated importance of general information about him especially information about the policies he has promised to implemented For candidates on the verge of losing by contrast this sort of information becomes completely uninteresting From the view of the Rule of Anticipated Importance the only thing interesting thing about losing candidates is whether can get themselves back into a competitive race 13 l have attempted to read these estimates off his Figure 21 eg 70 82 X 85 see the note to Figure 21 108 To test this implication which was suggested to me by Kathleen Hall Jamieson l divided coverage of candidates into two exhaustive categories horserace coverage and everything else The expectation is that D15 In the nal month of the campaign the greater a candidate39s political strength the larger the proportion of coverage that will be devoted to substantive information and the lower the proportion devoted to horserace matters To make the test ofthis inference more stringent I set aside pressinitiated criticism since we already know that winning candidates get more ofthis nonhorserace form of coverage The data strongly support D15 Within each of the four national media there is a negative correlation between Political Strength and the proportion of a candidate39s coverage that was devoted to horserace matters In an overall test in which the dependent variable is the average of criticism across all four media the correlation was 54 CONCLUDING REMARK I noted in the opening chapter that there is an important methodological advantage in detecting general patterns of media behavior in the context of presidential elections It is that presidential elections have a fixed structure and recur at regular intervals with politicians journalists and voters going through the same basic routines over and over except under somewhat different conditions This common structure has made it possible to see the effects of differences in conditions thereby confirming or disconfirming various arguments that might be made Nowhere has this methodological advantage been on greater display than in this chapter For a set of eight general elections 16 major party candidates and dozens of lesser politicians it has been possible to obtain comparable measures of political strength and test their effect on media coverage Given that the interests of politicians journalists and citizens are the same in other news domains it is quite likely that the Rule of Anticipated Importance holds in these other domains as well domains such as foreign policy news Congressional politics and economic policymaking Indeed the basic idea for the rule was initial formulated by Entman and Page 1997 in the context of a study ofthe Gulf War 109 But because of lack of comparability of cases within these domains it will be much harder if not impossible to marshal the sort of systematic data presented in this chapter 110 Chapter 7 The Rule of Product Substitution Although a central tenet of my theory of media politics is that politicians and journalists struggle with one another for control ofthe content of news that struggle has not been much in evidence in my analysis so far I have to be sure depicted journalists covering some candidates while ignoring others and heaping extra criticism on frontrunners while letting alsorans off lightly But these patterns of coverage have had by my account little to do with a jealous professional desire to control news content Reporters39 primary motivation has been to provide a rationally ignorant public with the kind of information it wants viz information about politicians having future political importance The analysis in this chapter brings the struggle between politicians and journalists to the forefront Its subject is the effort of quotimportantquot political candidates to control their news message to the public and the response ofjournalists attempting to keep control over their professional turf The central theoretical claim for which I offer several discrete pieces of evidence is that the harder candidates work to constrain what journalists report about them the harderjournalists workto find something else they can report instead I refer to this effort as product substitution The nature ofthis substitution is by now familiar instead of long sound bites from highminded speeches images of enthusiastic crowds and litanies of group endorsements the mass audience is treated to investigations of possible illegalities quotreality checksquot and other mediainitiated negativity The specific expectation is D16 Mediainitiated criticism will be positively associated with campaigns39 level of news management To test this claim it is necessary to measure the attempts by politicians to manage the news and to do so for each of the 16 major party campaigns since the new press regime began in 1968 Mainly because of the need to get comparable information from so many campaigns this is no easy matter One possible source of information about news management is the testimony of campaign officials But their statements are too general and too infrequent to provide a basis for systematic comparison among campaigns This leaves journalists as the only viable source of comparative information For every 111 recent election they have provided daybyday and sometimes hourbyhour accounts of the activities of the major candidates in the fall phase of the campaign These accounts as I shall argue provide a reasonable basis for measurement of candidates39 efforts at news management But journalistic accounts of campaigns must obviously be approached with caution As we saw in Chapter 5 reporters are more apt to criticize Republican than Democratic candidates which creates the appearance of bias and quite possibly more than mere appearance And if reporters39 stories are biased how can we trust them to provide accurate descriptions of the daytoday activities of candidates This concern though real cannot be permitted to block investigation For it remains possible that reporters are not biased against Republicans and that they more often criticize them only because Republicans are on average more likely to practice the kind of news management that reporters dislike and that as l have emphasized reporters dislike on an entirely nonpartisan basis At this point in my argument there is no reason to believe that Republican candidates campaign differently than Democrats but there soon will be In effect then we have competing hypotheses One is thatjournalists are biased against Republicans the other is that Republicans are more inclined to practice the type of news management that attracts press criticism Either or both hypotheses could be true And there is a third even more challenging possibility That Republicans are more prone to aggressive news management than Democrats but only because reporters are more apt to criticize them The challenge of this analysis is to devise an empirical test that will shed impartial light on these various possibilities My approach will be as follows I will begin by using press accounts to reconstruct as fully as possible the daytoday activities of presidential campaigns As described below all specific campaign activities of any importance and only m activities will be noted From these basic descriptions will be gleaned quantitative measures ofwhat I shall call News Management For example one measure of news management will be the physical exclusion of reporters from campaign events Though this rarely happens at important campaign events it does happen at minor ones and more frequently in some campaigns than others When it does happen I will assume that reporters will note it and critical to my analysis they will note Democratic exclusions as reliably as Republican ones 112 I will also note more subtle indicators of campaign style such as frequency of speeches whether they occur in friendly or unfriendly territory and number of press conferences Even if the reporters were biased against Republican candidates it is unlikely that they would be so biased as to fail to report their speeches and press conferences Having collected many such indicators my main analysis will take the form of a multiple regression in which the dependent variable is mediainitiated criticism of each candidate in October and in which the independent variables are news management style as measured in September mediainitiated criticism as measured in September the party ofthe candidate political viability incumbency and year ofthe election The control variable for September press criticism is especially important As noted it can reasonably be argued that candidates attempt to manage press coverage only because the press is attacking them Press criticism from the first half of the campaign will be used to control for any such defensive tendency on the part of candidates Since as would be expected September criticism of candidates is strongly correlated with October criticism r80 this is a very strong control If an aggressive news management style in September predicts mediainitiated criticism in October even after controlling for September criticism and other variables including party I will take it as evidence that a candidate39s style of news management is a causal determinant of mediainitiated criticism To set the stage for this analysis I begin with a brief look at the nature of presidential campaigns in the US especially the new style of electioneering that developed around 1968 This examination lays the groundwork for developing a measure of news management style from the observable activities of presidential campaigns THE GENERAL STRATEGY OF NEWS MANAGEMENT Since presidential candidates first began to campaign actively for office at the end ofthe 19th century they have sought to control the image they projected to the public The attempt to exercise such control is inherent to political campaigning But the rise of electronic communication especially TV dramatically affected the calculus on how to go about it Before television candidates sought to visit as big a part of 113 the vast nation as possible partly it seems to invigorate local party organizations and the partisans they could reach and partly to create a show of energy and enthusiasm for the national audience Thus as late as 1960 Republican candidate Richard Nixon made a point of visiting all 50 states including remote Alaska and Hawaii Such a strategy makes no sense in the TV age Already in 1952 aides to Democratic candidate Adlai Stevenson urged him to reduce the number of purely local campaign events and to concentrate on a handful whose real target would be an electronic audience As one journalist wrote Governor Stevenson is concentrating on the formal speech carefully prepared ahead of time for presentation primarily by radio and television His advisers are telling him that the whistle stop technique is out of date that such stops can be used from time to time primarily as a backdrop for a television or radio audience scattered over an entire state or region 1 Stevenson appeared to take this advice holding even fewer campaign events than Dewey had in his playitsafe 1948 campaign But this was only the beginning The trend toward fewer campaign events is shown in Figure 4 These data based on campaign activities in the last two weeks of September of each election year have been compiled from accounts of presidential campaigns in the New York Times and Washington Post The data represent an attempt to count of all rallies photo opportunities radio speeches motorcades fundraisers meetings with dignitaries and any other formal events that were mentioned in newspaper accounts ofthe campaign2 As the Figure shows the number of such events varied from 1948 and 1960 with some candidates running extremely energetic campaigns and others doing little Truman39s legendary whistle stop campaign in 1948 marks a highpoint the fact that he roared back from an initial deficit to win the race while his opponent was hardly campaigning at all is an interesting suggestion that presidential campaigns really can make a difference at least they can if only 1 The New York Times September 18 1952 p 26 2 Even if as is certainly the case some events were missed the trend across time ought still to be roughly valid The major threat to validity of the trend data is that they might reflect changes in reporting conventions rather than changes in candidate behavior But this seems unlikely News accounts in all periods routinely describe candidate behavior over the entire campaign day including what the candidates do when they are not doing anything Data through 1964 are based primarily on the Times and after that primarily on the Post The coding period was September 16 to September 30 As described below I subsequently expanded the coding period to September 10 and increased the number of media that were examined but only for the period beginning in 1968 114 one side is campaigning The 1956 campaign in which neither candidate did much campaigning in September was an early lowpoint The most probable reason for this dip is the combination of Eisenhower39s 1955 heart attack and Stevenson39s forwardlooking campaign style Forjoint intensity of both candidates 1960 was the peak After that the rate of campaign activity in both parties fell off rapidly first on the Republican side and then on the Democratic INSERT FIGURE 71 ABOUT HERE Once candidates began to concentrate on a smaller number of events whose real purpose was to reach a regional or national audience the imperative to manage the news created by those events became acute Nixon39s 1968 campaign was probably the first to fully recognize and systematically exploit the new strategic context quotNixon and campaign manager John Mitchell had decided that a good deal of what goes on in the usual political campaign was wasted effortquot Chester et al 676 Hence they cut out whistlestop tours in which the candidate made 10 or more speeches every day instead they focused each day on a few or often only one carefully orchestrated madeforTV event The Nixon campaign also virtually abandoned the ancient custom of night rallies reasoning that because they occurred after the TV news deadlines they were not worth the effort The rallies and other events that Nixon did hold were carefully staged so as to give reporters only one thing to report the message the candidate wanted to get out and campaign officials were relentless in their efforts to focus the attention of reporters on that message At one point for example reporters sought but were denied permission to come into a broadcast studio to watch Nixon read a speech into a microphone Why not let us watch the reporters pleaded Because a campaign official named Frank Shakespeare told them quotlfthat happens you39re going to write about the lights the cameras and that sort of thing and you39re not going to understand what happens in the living rooms across America quot3 Shakespeare39s instinct was sound Journalists like other professionals value the opportunity to exercise initiative and they dislike anyone else telling them what to do So ifjournalists had been permitted to watch the speech they would certainly have been hoping for some little event or telling detail 3 Frank Shakespeare as cited in Jamieson 1996 p 260 The quote is Shakespeare39s recollection ofthe occurrence 115 Figure 71 Frequency of campaign events 1948 to 1996 1 20 Democratic candidate 1 00 Count of all 8 0 public events during the second half of 6 0 September Republican 40 candidate 2 0 0lllll 48 52 56 60 64 68 72 76 80 84 88 92 96 Source Washington Post and New York Times reports perhaps the way the studio was lighted but more likely a stutter sweat on the brow or last minute coaching from an aide that would enable them to put their own stamp on the story And precisely to preventjournalists from doing this the campaign shut them out ofthe event leaving them with nothing to report but the substance of the speech itself Most of the techniques by which campaigns attempt to control what journalists report are more subtle than physical exclusion from campaign events In general campaign officials follow a twopronged strategy The first is to create primary events that the public will find interesting and that journalists will therefore feel compelled to report as news whether they want to or not The second is to avoid any sort of activity that reporters might choose to cover instead ofthe main event Thus from the campaign39s point of view an ideal campaign day is one in which the candidate stays in his hotel room all morning comes out to deliver a snappy sound bite in front of a cheering crowd and a striking backdrop and then retreats into the hotel for private meetings the rest ofthe day Journalists are then left nothing else to report except the snappy line the cheering crowd and the striking backdrop see Arterton 1981 Bennett 1996 Chapter 3 Although the demands of electioneering make this simple scenario unachievable in practice the basic strategic ideal to control what reporters can report by serving up a sharply limited number of carefully crafted events is at the heart of modern media campaigning Nothing in the theory of media politics implies that reporters resent the first prong of this strategy the creation of wellcrafted campaign events Compelling political theater sells newspapers lures TV news audiences and entertains even jaded reporters What journalists resent is the second prong Their opportunities to express voice are greatest when there are numerous diverse events to provide the raw materials for stories when unexpected and serendipitous episodes occasionally intrude on the campaign and when reporters themselves can help set the campaign agenda by raising questions and issues to which the candidates respond ln snuffing out these opportunities campaigns are in effect challenging journalists for control ofthe news In light ofthis my attempts to measure news management will concentrate on the second of the strategic prongs namely the concerted effort to eliminate serendipity and othenvise avoid giving anything for journalists to report except what the campaign itself wishes to have reported 116 It is often suggested that there is something illegitimate in efforts by candidates at news management but I do not make that claim Efforts to avoid activities that are likely to produce unfavorable media images are as l have indicated inherent to campaigning For professional reasons journalists do not like such efforts but that is another matter Modern campaign practice represents nothing more than an attempt to rationalize and control what any good campaign does anyway The rationalization however conflicts with the interests ofjournalists who would prefer to see more free wheeling campaigns My view ofthe resulting contest between politicians and journalists is similar to that of Nixon who after he had left office wrote ofthe new style of politics he helped to pioneer Public officials devote enormous energy to trying to rig the news to be reported their way When two sawy insiders reporter and official are in the ring together each trying to bamboozle the other neither should complain4 It should be added however that the effort to control media images through news management has costs as well as benefits to campaigns especially if it becomes heavyhanded One cost is that it angers journalists who as l have argued find ways to even up the score Another is that control requires attention and resources that could be used for other purposes And finally control tends to sap campaigns of drama and spontaneity thus undermining their appeal PARTICULAR FORMS OF NEWS MANAGEMENT The preceding section discussed the general strategy of news management Let me now proceed to more specific forms The aim ofthis section will be provide an overview of the particular behaviors that I will attempt to capture in my measure of News Management In extreme form news management involves as mentioned earlier the physical exclusion of reporters from events But the cost for this form of news management is extreme as well It is that reporters become enraged and typically respond by turning out lurid stories about quotisolated secretivequot and quotreclusivequot officials Even members ofthe public who dislike the media have no rational interest in opposing its efforts to open up quotsecretquot activities and may resonate to the kind of criticism journalists heap on reclusive candidates 4 Nixon 1990 p 299 117 In consequence campaigns lookfor more subtle ways of controlling what journalists can report One centers on choice of campaign venue Candidates may confine their campaigning to friendly partisan strongholds where they will encounter only cheering crowds but also little drama Or they may campaign in neutral settings where the risks of disruption that reporters could seize upon are somewhat higher but the number of votes to be won and the interest generated may also be higher Or if candidates feel especially daring they may take the campaign to opposition territory where they increase the risk but also probably also the potential gain from their efforts On one such venture Ronald Reagan used a black ghetto as a backdrop for dramatizing incumbent President Carter39s failure to solve the problem of urban poverty Reporter Lou Cannon describes the scene as follows Soon local residents formed a shouting crowd which jeered at Reagan and alarmed the Secret Service Reagan could not hold a press conference because the crowd shouted quotTalk to the people not to the pressquot When Reagan tried to talk to the people he was heckled unmercifully Finally in his most effective burst of emotion since Nashua Reagan shouted back at a heckler quotI can39t do a damn thing for you if I don39t get electedquot The crowd quieted down enough for Reagan to finish his presentation though a few still jeered when he left Reagan39s commanding presence had once more dramatically saved the day The evening television news showed an angry but controlled candidate forcefully putting down a hostile black crowd in a manner which won the respect of the crowd itself It was the perfect image for a candidate campaigning on the theme that his opponent was a failed leader but it was a nearrun triumph which had narrowly courted disaster Cannon p 270271 This excerpt shows a campaign taking a risk and getting good press out of it but just barely Campaigns normally seek more control than this and certainly Ronald Reagan39s campaigns usually did By the time of his 1984 reelection effort Reagan39s advance team was routinely screening members ofthe crowds that would see Reagan keeping potential hecklers as far away from the rally site as possible avoiding unfriendly territory and preventing the candidate from getting into unscripted exchanges with friend or foe The two candidates in the 1968 election were a study in contrasts with respect to unscripted exchanges Hubert Humphrey frequently held rallies on college campuses where he knew hecklers were likely to be present and when he encountered them he tried to engage them in dialogue sometimes setting aside his prepared speech in order to do so Nixon on the other hand once canceled a rally and 118 simply sat in his hotel room rather than speak before a crowd that would have some protesters in it He liked to answer questions from members of the public but only under the most controlled conditions Reporter Jules Witcover describes one such Nixon question session as follows The telethon was too extremely important for the campaign to be left to random phone calls So the Nixon media boys devised a shrewd system for preserving the appearance of authenticity without the substance Questions were written by the staff on subjects and in language that would be most helpful to the candidate Then when questions in the same general area were called in the ersatz questions were substituted using the names or the original callers 1970 p 445 An important part of news management is careful rationing ofthe access of reporters to the candidate Press conferences and press availabilities can sometimes be useful for getting the campaign39s message out but only when the candidate wants for some reason to be asked questions that reporters wish to ask For example candidates often hold press conferences or press availabilities when their opponent has made a gaffe on which the press is seeking comment Otherwise press conferences are at best a distraction They enable reporters to force the candidate to address issues of their own choosing thereby seizing control of the agenda from the campaign itself Lynn Nofziger a former press secretary to Ronald Reagan lays out the basic logic Let39s look at a hypothetical case This week the candidate wants to emphasize national defense At every stop he will talk about national defense The schedule has been set up carefully He visits a naval base an air base a shipyard a missile base He makes a speech to the American Legion and another to the Reserve Officers Association After a couple of days of this the press grows bored And the questions start coming about the candidate39s health or what he thinks of something his opposition has said or about anything else that is irrelevant to the topic ofthe week reporters would have the candidate answer those questions even though to do so would detract from the point the candidate has been trying all week to makeBut I am not going to let him answer their questions if I can help it As Nofziger explains quotIt was my job to keep the candidate and the campaign on track Otherwise the other guy winsquot p 18 Many campaign officials share this view and so routinely deny reporters the opportunity to question the candidate except under the special conditions just noted Thus on a day on which the Reagan campaign theme was wooing ethnic voters the following rather typical scene unfolded 119 Reagan ends up a walking tour of a Lithuanian neighborhood at Ramune39s Restaurant and Delicatessen where he is scheduled to stop for coffeeThis too will make fine pictures and Reagan39s staff shuttles the national and local press through the delicatessen so that they can record the sceneWhile Nofziger and others herd the press through shooing off anyone who tries to ask Reagan a question Reagan chats somewhat absently with the people on either side of him When Reagan begins to answer a question from a local radio reporter quotWhat did you think of your welcome herequot Nofziger and another aide almost have apoplexy quotNo interviewsquot Nofziger shouts waving his arms and diving for the reporter quotJust let me reply to this onequot Reagan says to Nofziger calmly and then he says quotMost heartwarming Anyone who wouldn39t be thrilled is unconsciousquot Drew 1981 p 272 Nofziger39s evident assumption is that if reporters have any opportunity to question the candidate they will quickly use it to change the subject so that even friendly questions must be prevented My contention in the Rule of Product Substitution is that Nofziger is exactly right Constrained or as the reporters themselves would put it manipulated by the campaign to produce stories about Reagan39s walk in a Lithuanian neighborhood reporters will look for something else they can do stories about instead Of course another reason that campaigns try to keep their candidates from speaking to reporters is that the candidates may misspeak Reagan and Bush were particularly prone to footinmouth disease as Elizabeth Drew 1989 has written the campaign team feared letting the candidate out very much on his own speaking for himself With Reagan the worry was that he would engage in rambling detours as he did in the first debate with Mondale or display his tentative grip on the facts With Bush the concern was that what came to be called his quotsillyquot factor his propensity for saying odd things would be on display5 In a moment of candor Bush seemed to accept this view of himself and in the same breath to illustrate why the view of him as a bungler was correct When asked at a rare news conference whether his campaign chairman James Baker had been keeping him away from reporters to maintain tight control over his campaign39s message the president replied quotIt wasn39t Jimmy lt wasjust some lowlevel handwringers who think I39m going to screw it upquot6 5 Page 338 For an example of how the press responded when cut off from a candidate see Chester et al 1969 p 689 5 quotBush Talks of Lasers and Bombersquot Maureen Dowd New York Times September 17 1988 A8 120 Consider also the following excerpt from a Q Angeles Times story on the 1996 campaign of Robert Dole Reporters following Dole who were interviewed in recent days noted that they seldom get to talk with the candidate He has not had a press conference for the national media since March they complained and traveling reporters rarely get background papers or briefings on what the campaign is trying to accomplish quotIt39s really hard to say whether Dole39s aides don39t know how to deal with the press or they don39t want to do anything that might help us or whether they just hold us in such contempt that they39re not going to go out of the way for usquot said Jodi Enda who has been covering Dole since the primaries for KnightRidder newspapers 39Whatever the reason I think it only hurts themquot quotGenerally reporters like Dolequot she added quotHe39s nice He39s funny He39s easy to talk to when you talk to him but the staff worries about every little thing he says getting into print quot Of course earlier in the campaign quotlittle thingsquot that Dole said particularly his offhand comments about tobacco perhaps not being addictive which became a running campaign story in 1996 did get into print causing huge difficulties and making Dole39s aides understandably gunshy But Enda said one reason that such comments made for big stories was that quotwe didn39t have much ready access to him so we wrote about everything we gotquot7 I noted earlier that candidates may limit press access in part because they fear the press is out to get them But when candidates fear the press the reason is often not so much that the press is out to get the candidate as that the candidate can39t deal effectively with the sort of pressure the press puts on everyone Perhaps no one made a greater effort to isolate himself from the press than Nixon and probably for this reason no one earned greater enmity from the press than he did The following passage reveals the state to which journalists sink when deprived of access to the candidate Reporters and columnists who had covered Presidential campaigns for many years eventually realized even accepted that it was not going to be possible to put questions to the candidate They fell into a state of what one can only call torpor As a mark ofthe state that intelligent men and women could be reduced to by this organized tedium there was the controversy which broke out in South Dakota over whether or not Nixon shaved his nose There were two schools ofthought on this some said that bristles on the nose 7 Eleanor Randolph quotSqueeze on Media Coverage May Be Bad News For Dolequot October 3 1996 p A1 121 are unheard of others that Nixon did have them After much craning and peering the matter was settled by a distinguished columnist who declared that at a certain angle he could see the cut hairs glinting in the sunlight Chester et al p 689 In Nixon39s case reduced contact with the press was indeed motivated by beliefthat the press was out to get him8 though to judge from the data reported in Figure 2 of Chapter 5 Nixon had little to complain about the amount of press criticism he got in the 1960 election Neither candidate got much and Nixon got even less than Kennedy But however this may be other candidates especially Republicans have often followed Nixon39s example of avoiding the press in order to limit the ability of reporters to seize control ofthe campaign agenda The more candidates isolate themselves from the press the more they must rely on press secretaries or other surrogates to respond to press queries about the issues that arise in any campaign Reliance on surrogates for such matters has several advantages but this is perhaps the most important Anything a press secretary says will be less newsworthy than it would be if the candidate himself had said it Hence having a press secretary handle sensitive matters is a way of downplaying the importance reporters can attach to them and thereby controlling what reporters can write For this reason reporters dislike being forced to deal exclusively with surrogates on sensitive issues as many campaigns force them to do One ofthe arts of news management is to find ways to get the media to carry one39s message while keeping the candidate39s direct exposure to reporters at bare minimum Ronald Reagan39s campaign team came up with an unusually clever idea in 1984 when it had the candidate visit the home of an eightyear old black child who had written the president a letter Accompanied by his wife Nancy the president had dinner with the boy and his family where he offered the child a gift ofjelly beans the president39s favorite candy Reporters were not present at the visit but afterwards the parents spoke graciously to the cameras about the president39s gesture Thus the president got highly positive coverage on network TV news as an apolitical father figure who concerned about all Americans and all without a single second of direct exposure to reporters39 prying eyes and shouted questions 3 Wicker 43940 Nixon was especially angered by an incident in the 1960 campaign when reporters suggested at a news conference and in stories that Nixon was using the anticommunism issue in demagogic fashion See New York Times xxx This was however one of the few instances of media initiated criticism of Nixon in that campaign 122 Although this event was unusually successful it is a general type of activity that is fairly common in elections Mark Hunt who first recognized the type gave it a fitting name quotClosed Photo Opportunityquot that is a photo opportunity at which the media are blocked from full access to the candidate The distinguishing feature of a closed photo opportunity is that candidates meet with groups or individuals as part of their official campaign schedules thereby focusing news coverage on what is usually a direct or indirect endorsement but fail to appear in public themselves In this way candidates campaign through the news media without having to directly expose themselves to reporters Sometimes of course candidates do appear in public to talk about an endorsement they have received but press avoidance often makes more sense Thus in 1972 the Democratic major of Philadelphia Frank Rizzo made news by endorsing Nixon for reelection and was rewarded with a private meeting with Nixon at the White House Afterward Rizzo made himself available to reporters telling them about the 52 million in revenue sharing that Nixon had promised to Philadelphia Meanwhile Nixon remained in the White House lf Nixon had appeared with Rizzo reporters would probably have asked the president about something of greater national significance than a revenue sharing grant to Philadelphia and this could only have undermined the message that the campaign wanted to get out to the voters that a Democratic mayor had endorsed Nixon and been rewarded for it Hence there is little mystery in why Nixon left Rizzo to speak to reporters on his own Even when candidates do meet with reporters they seek to do so under conditions that minimize the ability of reporters to gain the upper hand For example candidates may agree to answer a few press questions but only on a noisy airport tarmac that permits the candidate to pretend he hasn39t heard any question he doesn39t wish to answer Even fullblown press conferences may be staged by campaigns in ways that make it hard for reporters to pursue their own agendas as the following excerpt shows On one occasion when the traveling press started to complain midflight to Dulles Reagan aide Stu Spencer announced that there would be a press conference upon landing quotThey couldn39t talk to their editors they couldn39t look at the wires they couldn39t preparequot Spencer says quite pleased with himself quotThey were so goddamm mad at me We had a great press conference I thoughtquot 9131980 p 01 123 Each ofthese forms of news management involves concretely observable behavior and is therefore susceptible to direct measurement from press reports One must as l have noted be concerned about press bias but so long as it is possible as I believe it is to rely on the press to give unbiased reports on such matters as the number and location of rallies whether attendance at rallies is open or controlled by the campaign whether protesters who sometimes show up at rallies are permitted to remain or physically removed whether citizens who are permitted to ask questions at town hall meetings have been selected by the campaign or not the occurrence of press conferences and press availabilities whether reporters have been permitted to attend particular events or not then it is possible to gain at least a rough idea of the news management styles of different campaigns A MEASURE OF NEWS MANAGEMENT On the basis of these ideas and observations I developed a set of 48 codes each denoting either highpositive or lownegative concern for news management For example excluding reporters from a fund raising event is coded as a positive indicator of news management while taking reporters questions at an informal press availability is coded as a negative indicator Similarly screening attendance at rallies is counted as a positive instance of news management while taking unrehearsed questions from crowd members is counted as a negative instance A sample of positive and negative codes grouped into six subscales is shown in Table 71 INSERT TABLE 71 ABOUT HERE Working with these codes Mark Hunt my research assistant sifted through written accounts of each presidential campaign from 1948 to 1996 Specifically he examined campaign stories covering the period September 10 to September 30 in the New York Times Washington Post Q Angeles Times Vanderbilt TV News abstracts and for elections since 1980 the Associated Press wire9 Each time a candidate gave a speech spoke to reporters or took any of the other codable behaviors Hunt recorded the behavior Altogether he noted about 1100 behaviors or roughly four per active campaign day per 9 In some years one or both ofthe candidates tooktime off from campaigning for a special reason such as debate preparation When this occurred replacement days were added to make up for the time off If 124 Table 71 Sample Codes for News Scale Message control Candidate cancels major rally or event in order to avoid demonstrators positive Nt I Candidate refuses to debate major party opponent positive E Candidate responds to speci c opponent attacks excluding debates negative 4 Candidate takes questions from group or individual where questioners 1 been screened or selected by the candidate himself Includes friendly talk show positive 5 Candidate engages in exchange that is backandforth discussion with demonstrators or hecklers in crowd negative Crowd exposure 6 Rally or speech in unfriendly territory eg Clinton addresses VFW Convention during draft controversy negative 7 Rally in controlled setting audience screened or selected by campaign positive Willingness to debate 8 Candidate refuses to debate with major party opponent Positive Interview access 9 Press conference for national press negative 10 quotPress availabilityquot ie candidate meets informally with group of reporters negative 11 On his own initiative candidate engages in light nonsubstantive banter with reporters negative Interview restrictions 12 No one in the campaign will respond to queries about sensitive issue including press secretary positive 13 In response to queries from reporters about sensitive issue the candidate or press secretary issues statement but no one will verbally respond to questions positive 14 Candidate has interview with selected print joumalists restrictions on content positive 15 Candidate refuses request from traveling journalists for press conference positive Media exclusion 16 Any public or quasipublic event from which reporters are excluded e g fundraisers positive 17 Campaign creates impediments to reporting of news e g party workers hold up signs to block picturetaking positive candidate in the period 1968 to 1996 The average of four per day is however somewhat misleading Some candidates did almost no campaigning such as Richard Nixon in 1972 while others did a great deal such as George Bush in 1988 As part ofthe coding task Hunt copied the newspaper text that he relied upon in assigning codes into electronic files and these files have been put on my webpage Thus each of the 1100 behaviors coded for the News Management scale has a publicly available justification 10 Further information on coding of candidate behavior is available upon request Converting 48 codes and l100 candidate behaviors into a usable measure of news management is not a straightforward task To do so I grouped the 48 codes into six subsets as indicated in Table 1 and gave each candidate a score on each subset of items by adding up positive and negative points Note from Table 71 that three of the subscales refer to behavior of candidates toward reporters and three refer to the management of campaign events independent of reporters As Table 72 shows scores on these subscales are correlated with negative coverage within each of the four media When I performed a principal components analysis on the six subscales all loaded reasonably well on a common factor as shown by the in the last column of Table 72 INSERT TABLE 72 ABOUT HERE An obvious concern in measuring candidate behavior from media reports as l have done is that the reports may be biased in some way This concern however is greater for some subscales than others For example one can be confident that when candidates give ontherecord interviews or press conferences some reference to it will appear in print eg speaking with reporters on Air Force One the President said Similarly one can be confident that when candidates exclude reporters from events reporters will usually note it often in the form of a complaint On the other hand one cannot be confident that every case in which a campaign screens access to its rallies will be noted the most one can hope is that there will be more frequent references to such screening for candidates who screen more In light of this concern it is reassuring that all six subscales of the news management scale have zeroorder relationships with media negativity as shown in Table 72 however candidates refrained from campaigning without special cause no replacement days were added 10 See wwwsscnet urla min I quot quot quot 39 quot39 39g iconductfiles There is a separate file for each election year The files are in Word format 125 Table 72 Correlations between News Management Subscales and Media Negativity Correlation with media negativitv subscale loadings Newsweek Time New York Network Row on general news magazine magazine Times TV news average management factor Limited interview access 18 25 41 44 Q 56 Unwillingness to debate 23 29 54 28 57 Interview restrictions 39 39 47 44 Q 39 Media exclusion from events 41 29 57 50 66 Limited crowd exposure 18 22 37 67 E 54 Message control 23 30 49 39 46 Column average 2 NOTE Cell entries are correlation coefficients based on scores of 16 major party candidates from 1968 to 1996 Underlined entries are averages of correlations in the indicated row or column Figure 72 shows the scores of each of the major party candidates on the overall News Management scale over the period from 1968 to 1996 Table 73 shows the mean scores of the two parties on each of the six subscales of the overall measure From both visual inspection and statistical analysis of these data it is clear that the only important trend is a large average difference between Democratic and Republican candidates This difference appears on every subscale and in almost every election Table 73 missing INSERT TABLE 73 AND FIGURE 72 ABOUT HERE Why such large party differences exist is tangential to my analysis but is an interesting question An obvious possibility is that Republicans have a rational basis for fearing press criticism and therefore take steps to contain it But Republicans are not only more restrictive in press relations they also tend to be more controlling with respect to campaign events that have nothing to do with the press For example when Democratic candidates take questions from ordinary citizens they are apt to randomly select people from open rallies whereas when Republican candidates take questions from citizens the questioners are more likely to have been screened by campaign officials beforehand Republicans are also more likely to control access to their rallies than Democrats My hunch is that party differences in campaign style are rooted in the enduring and wellknown philosophical differences that distinguish the parties Republicans have traditionally been the party of law order and stability whereas Democrats have been the party of freewheeling reform When freedom of the press becomes controversial Republicans are also more apt than Democrats to favor restrictions In a study of personality differences between delegates to state party conventions two psychologists write that The Republican style appears to reflect a steady dependable personality particularly attuned to focused attention and purposeful effort but the Democratic style is expressive of a strong restless personality given to immediate responsiveness and robust initiative To some extent the former style embodies the abiding concern for proper conduct that makes social life possible the later style conveys the verve and forceful social presence that makes it fascinating The prototypical Republican leader tends to be thoughtful in both senses of that term considerate of others and deliberate in action the prototypical Democratic 126 Figure 72 Standardized News Management scores for major party candidates 1968 to 1996 Standardized News Managem ent Republicans Sc ores Dem ocrats leader when intensely impelled to autonomous action may be thoughtless and headstrong It does not seem farfetched to extend this observation to a suggestion that deeply rooted ideological differences between the parties may explain why the more conservative party favors a more controlled campaign style and the more liberal party favors a more freewheeling one The reasons for party differences in campaign style are beside the point of my main argument What is important is whether party differences in campaign style exist and whether they may affect how reporters cover candidates ofthe two parties Let us therefore look more closely at the relationship between Media Negativity and News Management MAIN EMPIRICAL RESULTS Table 72 has shown levels of News Management as measured in September are correlated with higher levels of media negativity as measured in October This is the main empirical result so far and the one needed to confirm D16 But in order to be certain that this relationship is a causal rather than a spurious one it is necessary to control for potentially confounding variables as follows Inspection ofthe data indicate that beginning with Nixon in 1968 Republican candidates tended to make more aggressive efforts at News Management than Democrats This makes it necessary to control for the party of the candidate Absent such a control any effect of News Management could be a spurious indicator of media bias against Republicans As Patterson 1993 in particular has shown media negativity has increased in recent decades To control for this general trend it is necessary to control for year ofthe election We saw in Chapter 6 that reporters are more inclined to dig up negative information about candidates who are politically strong This reflects the Rule of Anticipated Importance To control for a candidate s anticipated importance I use the average of his share of the twoparty vote in the early October Gallup poll and the final election results Candidates may resort to aggressive management as a response to media negativity toward them To control for this possibility I control for the September level of media criticism Since September criticism is correlated with October criticism at the level of r 77 this is a strong control 11 Constantini and Craik 1980 127 o The data indicate that reporters are more critical of incumbents Hence I add an incumbency control variable Using the five control variables just described in a regression having only 16 observations makes it difficult to show the effect ofthe variable of interest news management Compounding this difficulty is the fact that three of the five controls are correlated with news management at the level of r 50 or greater Nonetheless Table 4 shows that news management has a significant effect INSERT TABLE 74 ABOUT HERE The dependent variable in Table 4 is mediainitiated negativity which has been formed by combining negativity scores from all four media Time Newsweek New York Times television network news The key independent variable is news management which is a linear combination of the six subscales in Table 72 as weighted by the factor scores from a principal components analysis Look first at column 1 of Table 4 where the effect of news management on criticism is both statistically significant p 03 one tailed and substantively large The standardized coefficient of 51 means that a change of l SD on the news management scale is associated with a change of 51 SDs in pressinitiated criticism Column 2 of Table 4 breaks the news management scale into two subscales a threeitem subscale that I shall call event management crowd exposure message control willingness to debate and a threeitem subscale that I shall call reporter management media exclusion interview restrictions lack of interview access In the regression in column 2 reporter management has a very large and significant effect while event management has almost none Column 3 shows that when reporter management is taken out of the model event management has a moderate but statistically marginal effect on media negativity The latter results suggest that how a candidate treats reporters has a big effect but that little else matters Yet it would be mistaken to accept to this conclusion The two subscales of news management are to begin with correlated at 80 In light ofthe measurement error that no doubt exists in both subscales this is a high correlation one strongly suggesting that event management and reporter management tap a common syndrome It is quite possible that the vagaries of measurement error in combination with multicollinearity in a small dataset have made it artificially easier to show effects for one 128 Table 74 Effects of News Management on Media Negativity 1968 to 1996 1 2 3 4 News management 3 51 Mean 0 SD 1 53 one sided pvalue 03 Reporter management 3 55 Mean0 SD 1 b 57 pvalue 05 Event management 3 04 26 34 Mean 0 SD 1 b 04 27 36 pvalue 44 15 03 September media criticism 3 18 17 20 Mean 0 SD 87 b 22 20 12 pvalue 30 28 29 Political strengt 3 38 39 37 44 range 36 to 62 b 06 06 05 06 pvalue 03 03 05 01 Year 3 34 38 26 28 0 to 7 15 17 11 06 pvalue 02 01 07 04 Democratic candidate 3 14 18 02 0 or 1 b 29 35 04 pvalue 28 34 47 Incumbent 3 15 12 24 32 0 50 1 b 32 25 52 68 pvalue 25 30 17 04 Intercept 347 363 3 16 379 Rsquare 84 86 79 78 Adjusted rsquare 73 73 65 69 Note Estimation is by means of Ordinary Least Squares Number of cases is 16 All pvalues are one tailed The dependent variable is a weighted average of the October media negativity scores of Newsweek Time the New York Times and television network news To create this variable 1 standardized all four variables averaged them so as to give equal weight to each type of media that is onesixth weights to each news magazine onethird weight to the newspaper and onethird weight to television news which was already an average across the three major networks and restandardized the final variable to mean 0 and SD 1 September criticism scores from the New York Times and television news were standardized and combined September scores for Time and Newsweek are unavailable An average of a candidate s support in the early October Gallup poll and in the final vote The nonincumbent nominee of the incumbent party receives a score of 50 part of the syndrome than for the other even though the overall syndrome rather than either part alone is what matters A partial test ofthis supposition is possible Note that the coefficient for party is small and statistically insignificant in columns 1 2 and 3 and further that it has the wrong sign in two of the tests The positive sign indicates that contrary to the usual expectation the media appear to be slightly more critical of Democrats than Republicans These results suggest that party is a superfluous control variable with no real effect at all Note also that September media criticism was included only to control for the possibility of a reciprocal relationship between media negativity and candidate behavior toward the reporters and that there is little reason to worry about such reciprocity if we are testing the effect of event management by itself Given this it is reasonable to omit party and September media criticism as control variables when testing the effect of event management separately from the effects of media control The results of such a test as reported in column 4 of Table 4 show that event management is both statistically and substantively significant when freed of the need to compete with a set of highly collinear and arguably superfluous rivals The conclusion I draw from these results is that attempts by candidates to manage journalists and campaign events are part of a common syndrome and have common effects on media negativity It may be as the evidence in column 2 suggests that reporters are more sensitive to attempts to manage them than to attempts to manage campaign events But it is quite possible that attempts to measure the former seem more important simply because they are easier to measure accurately Other issues raised by these data are as follows Although this analysis is based on 16 candidates they competed against each other in only eight elections This raises the possibility of correlated errors across pairs of observations in the same year To evaluate this possibility I calculated the withinyear correlation of residuals for the model in column 1 of Table 4 A positive correlation would indicate that reporters are more negative toward both candidates in some years and more positive toward both in others A negative correlation would indicate that reporters pick a favorite within each election such that if they are harsher toward one they tend to go easier on the other The observed correlation turned out to be positive but it was neither large r17 nor statistically significant p69 twotailed n8 paired observations This 129 result suggests that correlated errors are not a problem in these data nonetheless estimated a fixed effects regression model to control for withinyear correlated errors More specifically I ran a regression in that included a dummy term for all ofthe elections years but one thus controlling for any yearspecific disturbances The results indicated that correlations among error terms was minimal since the joint F test on them had a significance level of 48 The coefficient estimates for the key variables Political Strength and News Management were hardly affected by the seven dummies but their standard errors were greatly enlarged As a result the significance levels ofthe coefficients for News Management and Political Strength fell in each case to plt15 onetailed If however the arguably extraneous control variables incumbency party and September media 1 2 are dropped both News Management and Political Strength are statistically significant at about p01 onetailed ln fixed effects models in which the only independent variables were Political Strength and Press Management the two variables had effects on Media Negativity at about the 01 level of significance onetrailed In a similar model testing Event Management and Political Strength both variables were significant at the 025 level onetailed The News Management variable may have somewhat larger effects for Republican candidates compared to Democratic ones The same is true for its component parts Press Management and Event Management But these differences do not approach statistical significance in these small samples and are in my opinion best understood as the product of chance variation 13 This last set of results especially the results for the eight Republican candidates alone are quite notable For if News Management explains differences in Medianegativity both across parties and within parties it must be more than simply an alternative way of measuring the party affiliation of the candidate This in turn exonerates the media ofthe taint of an antiRepublican bias to their coverage Reporters 12 The Ftest for the joint significance of these three variables in the fixed effects model was p58 13 In a model including News Management Political Strength Year Party and Party X News Management for all 16 cases the interaction term does not approach statistical significance p77 In a standard fixed effects model the pvalue is 85 130 do of course have biases but they appear to be biases in favor of campaign openness rather than in favor of one of the parties Because the key variables in this analysis have no natural metric it is hard to say much more about the sizes ofthe effects observed in these data than the standardized Beta coefficients say But it is perhaps worth noting that the two biggest campaign scandals in American politics Watergate in 1972 and Campaign Finance in 1996 were visited upon candidates whose scores on my predictor variables put them well within the danger zone Nixon in 1972 had the highest score of any candidate since 1968 on Political Strength and News Management a prescription for highly negative coverage and Clinton in 1996 was tied with Dole for the highest score on the Year variable had the highest score of any Democrat on Political Strength but only moderate overall and also had second highest score of any Democrat on News Management but average overall 14 Hence the high Media Negativity scores of these candidates are reasonably well accounted for by my model The implication ofthese results is that if Nixon had been in a close race with McGovern in 1972 and had been open to the press besides reporters might have made little fuss about Watergate For 1996 the implication is that if Clinton had been in a closer race and had been nicer to the press the Democratic party could have raised as much quotsoft moneyquot as it wanted without stirring up much criticism among reporters Are these implications plausible This is hard to say The overall results do however suggest that media criticism of candidates is heavily situational in the sense that it is determined as much byjournalistic interest in voice as captured by the News Management variable and public interest in powerful figures as captured by the Political Strength variable as by what the candidates have done to merit criticism This does not imply that journalists concoct the negative information they report about candidates it implies rather that of the abundance of negative information that might be reported journalists invest resources in digging out information the pertains to candidates who are politically strong or take an aggressive approach to news 14 lfthe suspicion lingers that high scores on News Management are caused by press criticism rather than vice versa l observe that in Nixon39s case his score on News Management was almost as high in 1968 when he faced no scandal as it was in 1972 In fact Nixon has the two highest scores on News Management in the sample In the case of Clinton recall that News Management is measured in September while the scandal on campaign finance broke in midOctober 131 management 15 Candidates get some criticism strictly on the quotmeritsquot the base level of criticism for an average candidate in 1996 is about 12 percent but the high rsquares on the models reported in Table 4 indicate that much of the variance in criticism from one candidate to another is situationally determined Reagan39s 1984 coverage stands as counterpoint to the argument that situational factors drive media negativity According to his scores on News Management and Political Strength which were both very high Reagan should have received a mountain of press criticism Yet as Figure 2 shows he received only a moderate amount more than his hapless opponent but less than half what would have been expected on the basis ofthe prediction equation in Table 4 16 One ofthe negative stories Reagan did get in 1984 is worth noting It is a backpage nearly vacuous item in the m mm about a Reagan TV commercial that featured a large bear stalking through the woods as a metaphor for the Soviet Union The point of the story was that the bear used in the commercial was not actually Russian but American an indication as the story insinuated of dissembling by the Reagan campaign No reporter would have bothered to do such a story about Mondale in 1984 Who would pay any attention to it But Reagan an incumbent president with excellent prospects for reelection was a target worth going after especially since the ads seemed to be working very well What makes this story notable in my opinion is that it shows how far reporters were willing to stretch in order to criticize Reagan quotThe Russian bear that wasn39tquot is thus an example of the situationally determined criticism that reporters heap upon frontrunning candidates who adopt a strategy of heavyhanded news management It is an interesting question why if my analysis is correct candidates persist in news management techniques that offend the media and tend to result in press criticism Why not simply ease up and get better coverage in return 15 When asked at a roundtable discussion why his newspaper failed to investigate reports that Bob Dole39s alleged affair in the 1960s had resulted in an abortion an editor for a nationally prominent paper replied in part that he faced a tradeoff between expending scarce resources on Clinton or Dole and felt that Dole as a hopeless candidate was not worth the effort Roundtable on Media Coverage of the 1996 Presidential Election August 29 1997 Meetings ofthe American Political Science Association Washington DC 15 In fact Reagan s 1984 coverage contributes the single biggest residual that is missed prediction in the dataset From Figure 2 it can be seen that only network news was especially tough on Reagan in 1984 according to the model all media should have criticized him at about the level that TV did 132 Much of the reason seems to be the belief of campaign consultants especially on the Republican side that candidates gain more from the controlled images they are able to get the media especially TV to carry than they lose from the criticism much of it petty and strained which journalists visit upon them in return This belief is on open display in an ofttold tale from Ronald Reagan39s 1984 campaign Frustrated that Reagan campaign consisted of vacuous hoopla CBS News reporter Leslie Stahl assembled a repetitive montage of campaign scenes that seemed to her especially vacuous cheering crowds colorful balloons rising into the sky Reagan smiling and waving and used it as visual backdrop for an acid commentary about Reagan39s supposedly empty campaign But as related by Schudson a White House official called Stahl soon after the piece aired and said he39d loved it quotHow could youquot she responded He said quotHaven39t you figured it out yet The public doesn39t pay any attention to what you say They just look at the picturesquot Stahl on reflection came to believe that the White House was probably right all she had done was to assemble free of charge a Republican campaign film a wonderful montage of Reagan appearing in upbeat scenes 1995 p 115 It is hard to believe that media stories about Watergate and Clinton fundraising were quite as harmless as Stahl39s attack on Reagan39s campaign style but there is as we shall see below no systematic evidence that press criticism of candidates during the final phase of the election campaign has any negative effect whatsoever and a slight suggestion that it might even help at least in the short run Consistent with this possibility I spoke to another Republican adviser who said that the Bush campaign knew in 1988 that it might be criticized byjournalists for visiting flag factories but felt that this sort of criticism from nonetoopopularjournalists could actually be helpful among swing voters No doubt this kind of thinking is the biggest part of the reason that campaigns are as willing to anger the media as some obviously are 133 Chapter8 Has All of Politics Changed quotAll of politics has changed because of youquot That was Lyndon Johnson s assessment of the accomplishments ofthe newly aggressive journalistic establishment ofthe 1960s Was he right In one sense American national politics has indeed changed over the last 50 years or so Much more ofthe nation s business is now conducted publicly through press conferences political road shows paid advertising and other forms of mass communication1 A century ago politicians spoke to voters mainly through political parties now they address them directly in the mass media including the paid media This is as true for presidents who wish to pressure Congress into passing their legislative program as it is for presidential candidates who would like to win their party s nomination But is this a deep change or a surface change Does it mean that different kinds of political groups are winning political battles for different kinds of benefits or that the same old groups are using new means to old ends ls American politics really different or is itjust conducted by different means Another question centers on the quality of the communication by which much politics is conducted Can media politicians easily bamboozle the public through false promises and demagoguery or are they basically truthful and honest in what they say ls news better or worse in the era of media politics than it was in the 19th century heyday of party politics These are obviously large questions and I am not so foolish as to think I can definitively answer them I do however have evidence and arguments that bear on them leading to the following general conclusions All of politics has not changed in the era of media politics The leaders of political parties are for the most part masters of media politics rather than its victims At least as regards presidential nominations parties are still tremendously important for organizing both elite and mass politics As regards general presidential elections party accountability for performance in office greatly affects electoral outcomes The degree of continuity in presidential politics over the past 50 to 75 134 years is all told surprisingly high The quality of mass communication in media politics is certainly not high This is because the large majority of citizens is not interested in and probably never has been interested in high quality political communication But mass communication seems to be mostly fair and honest at least insofar as it is in the interest of majority opinion for communication to be honest MEDIA POLITICS VS PARTY POLITICS Since the birth of mass political parties in the early 19th century the United States has had a two party system and since the replacement ofthe Whig party by the Republican party in the 1850s the two major parties have been the Democratic party and the Republican party The rise of media politics has not challenged or even importantly disturbed the dominance ofthese two political organizations In this fundamental sense continuity reigns in American national politics A widely accepted view is that the rise of media politics has been associated with the rise of candidatecentered politics which has in turn meant greater political instability and less political accountability ln President George Bush39s description ofthe 1992 presidential election politics has simply become more quotwackyquot In an impressive early analysis of the new mediadriven system of presidential nominations Nelson Polsby 1983 suggested that it would be prone to fads and contagions This view is I believe widely shared more citations needed Instability is actually fairly easy to measure In the period from 1900 to 1964 the average inter election swing in the presidential vote was 77 points including any effects ofthird party candidates on the vote In the period from 1968 to 1996 the comparable interelection swing was essentially the same 812 points2 1 See especially Kernell 1997 2 These figures were calculated as follows Starting with the 1900 election the percent of the national vote won by the Democrat in the given election was subtracted from the percent won in the previous election and the average absolute value was calculated For example William Jennings Bryan won 455 percent ofthe vote in 1900 and 467 percent in 1896 for an absolute difference of 12 percentage points The average ofthese interelection swings in the Democratic vote from 1900 to 1964 is 667 percent for the period 1968 to 1996 it is 753 percent The corresponding figures for Republican candidates are 772 percent and 818 percent The average of the Democratic and Republican figures are those reported in the text 135 For the period since the rise of public opinion polling it is also possible to measure instability within election campaigns Starting in 1948 the Gallop organization has regularly measured public support for the major party candidates in January ofthe election year just after the party conventions and in the first week of October In the period from 1948 to 1964 the average swing from January to the election was 148 points percent in the period from 1968 to 1996 this figure was 149 The comparable figures for the swing from the postconvention period to election day are 47 percent for the early period and 51 percent for the recent period And finally the figures for average swing from early October to election day are 40 percent and 43 percent3 It would be very hard to read these data as evidence that American presidential politics had become less stable in the new age of media politics But if politics are not any more fluid perhaps they are more quotwackyquot Perhaps that is Americans are about as changeable as they used to be but are changing in ways that are strange difficult to predict or somehow irrational This does not seem to be happening either In fact the trend is in the other direction The outcomes of presidential elections are if anything more intelligible and easier to predict in recent decades They also appear to involve more party accountability than in the past To see this requires a brief digression A major theme in recent research on presidential elections has been quotretrospective votingquot The idea is that voters use elections to cast a retrospective verdict on the performance ofthe incumbent president and his party If the country has been prosperous remained at peace and avoided serious political scandal voters tend to reelect the incumbent or the candidate of the incumbent party But if the country has suffered four years of bad times the incumbent party tends to get tossed out As V O Key Jr put it voters tend to act the role of quotrational god of vengeance and rewardquot Advantaged candidates are only too eager to encourage the electorate in this role Nonincumbent Richard Nixon running for president in 1968 against a backdrop of several hundred combat deaths a week in Vietnam racial disturbances in the cities and a youthful counterculture that many voters found shocking declared throughout his campaign that 3 For 1948 to 1992 these data are from Stanley and Niemi 1995 p 96 For 1996 data are from Gallup polls published on Hotline on January 10 August 30 and October 4 Calculations were made in the same manner as in the previous note 136 When you39re in trouble you don39t turn to the men who got you in trouble to get you out of it I say we can39t be led in the 3970s by the men who stumbled in the 3960s4 ln perhaps the most famous political exhortation ofthe last 20 years Ronald Reagan who in 1980 had the good political fortune to be the nonincumbent candidate for president in the midst of an economic depression and the lran hostage crisis declared to the audience watching the presidential debate Next Tuesday all of you will go to the polls you39ll stand there in the polling place and make a decision I think when you make that decision it might be well if you would ask yourself Are you better off than you were four years ago Is there more or less unemployment in the country than four years ago Is American as respected throughout the world as it was And if you answer all of those questions yes why then I think your choice is very obvious as to who you39ll vote for If you don39t agree then I could suggest another choice that you have This country doesn39t have to be in the shape it is in5 Four years later when Reagan sought reelection in the midst of an economic boom he again asked voters whether they were better off than four year ago and got the answer he hoped and expected to get a landslide reelection to office Figure 81 provides some systematic evidence of the extent of retrospective voting in US presidential elections for two time periods The upper graph is for presidential elections from 1932 to 1964 and the lower graph is for the elections from 1968 to 1996 For each graph the horizontal axis is a measure of national prosperity namely the percentage change in quotdisposable incomequot for the average American in the year ofthe election6 For example the upper figure shows that in the year of the 1932 election average disposable income in the US fell about 15 percent In 1936 by contrast average spending rose by about 10 percent The vertical axis shows the percentage of the vote won bythe incumbent presidential party Thus in 4 Newsweek November 4 1968 p 28 5 Bartels 1992 p 271 5 More specifically the data show the average growth rate in inflationadjusted disposable personal income per capita in the four quarters of each election year The economic data are from the Survey of Current Business Table 3 p 164 August 1997 The original data were not population adjusted to make the per capita adjustment population data were taken from the following sources For the period to 1988 National Income and Product Accounts of the United States 1929 to 1988 Table 21 of vol 1 and 2 Bureau of Economic Analysis Department of Commerce Washington DC US Government Printing Office February 1993 for the period 1989 to1991 Survey of Current Business July 94 Table 21 p 62 for the period 19921996 Survey of Current Business July 94 Table 21 p 62 137 1932 for example it can be seen that the candidate ofthe incumbent party Hoover got about 40 percent ofthe vote But in 1936 when the economy was stronger the candidate ofthe incumbent party Democrat Franklin Roosevelt got more than 60 percent ofthe vote INSERT FIGURE 81 ABOUT HERE Each ofthe points in Figure 81 many of which are labeled refer to a particular election between 1932 and 1996 Looking at the overall pattern of these points one can see that they tend to form a line running from the lower left to the upper right as summarized by the actual solid lines What these solid lines show is that as average personal income rises the incumbent presidential party tends to get a larger share of the vote7 Note moreover that the relationship between growth in personal income and vote for the incumbent party seems to be about as strong in the era of media politics as in the preceding period8 Thus politics in the current era does not seem to be either whacky or strange It seems as in the previous period to reflect a healthy concern for the actual performance of the incumbent party The electionyear economy is not however the only determinant of voting in presidential elections Another dimension of party performance concerns war and peace Parties that lead the nation into costly wars fare less well at the polls than do parties that maintain peace Thus the Democratic party which led the nation into World War II the Korean War and the Vietnam war appeared as we shall see in a moment to pay a price in the elections of 1944 1952 and 1968 The positions of the candidates may also to affect election outcomes Candidates who are closer to the center of public opinion seem to do better than candidates who adopt quotfar outquot or quotideologically extremequot positions The reason I say that ideological extremity quotseemsquot to affect voting is that this is a very difficult point to prove This difficulty stems from measurement From 1932 to 1996 there have been 17 elections and 7 The regression line for the earlier period omits 1932 and 1936 as outliers including them would flatten the line in the upper graph thereby making it appear that the presidential vote was more structured and leg erratic in the period of media politics 3 The relationship between the economy and the vote seems however to be notably weaker in the period before 1928 In fact to judge from official reports of the economy in the period 1892 to 1928 economic voting is nearly nil in this period This is based on GNP data published in Historical Statistics of the United States Colonial Times to 1970 Bureau of the Census 1975 Part 1 series F4 p 224 138 Figure 81 The effect of peace and prosperity on the presidential vote Twoparty share of presidential vote for incumbent Party Twoparty share of presidential vote for incumbent Party 1932 to 1964 6 5 quot 6 O 39 5 5 39 5 O quot y 498 153econ 348war adj r2 37 45 52 94 32 regression omits 1932 and 1936 as outliers O 4 O I I I I I I I 1 6 1 2 8 4 O 4 8 1 2 Election year percent change in real Personal Disposable Income 1968 to 1996 6 5 quot 6 O 39 5 5 39 5 O quot 45 80 y 471 223econ 531war adj r2 35 96 53 4 O I I I I I I I 16 1 2 8 4 0 4 8 1 2 Election year percent change in real Personal Disposable Income The first scholar to tackle this problem in the context of presidential elections was Steven Rosenstone 1983 who studied elections from 1948 to 1980 He obtained his measure of ideological location by asking some 40 scholars of presidential politics to rate each major party nominee on two ideological dimensions which he called New Deal social welfare liberalism and racial liberalism As he explains Raters were instructed not to judge how the public perceived the candidates or to recall the results of public opinion polls Rather I asked the scholars to score the candidates39 actual positions on these dimensions quotthe way an insightful political observer of the day would have evaluated the actions and positions of the candidate priorto the election p 174 Since I need evaluations of candidates from 1932 through 1996 I could not simply use Rosenstone39s measurements Hence I asked my Research Assistant Mark Hunt to replicate them That is asked him to rate each of the 34 major party candidates since 1932 on a simple leftright scale Hunt was unaware ofthe specific hypothesis l was testing and having read extensively about all recent elections as party of this study he was well familiar with the positions of the candidates on the major issues of the day To his task simple I asked him to use only a single ideological dimension My instructions were Please rate the ideological position of each presidential candidate listed below on a leftright ideology scale There are of course many dimensions of ideology social welfare race foreign policy In making your ratings use whatever dimension or dimensions were most important in public debate at the time ofthe election Do not rate the candidates on the basis of your impressions of what the public thought their positions were or what public opinion polls indicate that the public thought Rather rate the candidates on the basis of the actual positions and actions they would be expected to take in office given what was publicly known about them at the time of the election Candidates who run in more than one election may be rated differently if their positions change or better information has become available at the time ofthe second election Use a rating scale running from 3 most conservative to 3 most liberal where 0 is
Are you sure you want to buy this material for
You're already Subscribed!
Looks like you've already subscribed to StudySoup, you won't need to purchase another subscription to get this material. To access this material simply click 'View Full Document'