INTRO POL SCIENCE
INTRO POL SCIENCE POLS 2000
Popular in Course
Popular in Political Science
This 27 page Class Notes was uploaded by Estelle Prosacco on Saturday September 12, 2015. The Class Notes belongs to POLS 2000 at University of Georgia taught by Madonna in Fall. Since its upload, it has received 30 views. For similar materials see /class/202238/pols-2000-university-of-georgia in Political Science at University of Georgia.
Reviews for INTRO POL SCIENCE
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: 09/12/15
mgmwwmmmw mm m w wasnd l elp mxulsiwnm nunqrnm mmmwwmm mw immmundm mm gwmsmq WM mung mm 3n mmmwmwmm mm msxamvmbvlmmp wmmu m m Warnway p mn39m nxvuzm 3n mmmmmmwamAWW WWWAmmwsmas sd lrmamsanjww xulsl msnmanawaswd mmrmwW Vlcx39rlmvasnsn my mum mu intrubnim snim mammals mm mm mrl mumn Man a a 5 jfk wm sfm aw mn awn mmwmmgmwpww sq MW WWW dd um um a n H mm mm 4 NWWWV mm Wm MW Ja AWNRM mm put WWW mamas m mm y WWW um immdn w mm 3 mm H m 5 gt7 VSE WW The Politics ofSapreme Court Nominations A Theory of Institutional Constraints and Choices Bryon J Moraski University of Iowa Charles R Shipan University of Iowa When a vacancy occurs on the Supreme Court the president can attempt to use his power of nomination strategically in order to bring the Court in line with his own policy prefer ences However the president faces two constraints when attempting to do so First he may be constrained by the presence of continuing justices and the existing Court median Second he may be constrained by the Senate which must approve his nominee In this paper we develop and test a theory that examines the conditions under which a president is constrained in his choice of a nominee Our results show that presidents can and do behave strategically with respect to Supreme Court nominations 1 INTRODUCTION In recent decades vacancies on the Supreme Court have appeared ap proximately every two years Because the Court plays such a central role in politics and policymaking presidents place a great deal of importance on filling these vacancies with nominees who will produce a Court that looks more favorably on their agendas Surprisingly however while we have learned a great deal in recent years about Senate voting on Supreme Court nominations very little systematic analysis has been done on presi dents choices of nominees Given the involvement of the judiciary in nearly every important area of policymaking and the centrality of the nomination process to an understanding of the checks and balances in our political system it is imperative to understand the process by which presi dents select nominees Here we analyze presidential choices of Supreme Court nominees Our starting point is the observation that the Senate almost always approves Su preme Court nominees Since the turn of the century only four nominees have been rejected by the Senate while fifty five have been approved a Earlier versions of this paper were presented at the 1996 meeting of the Southern Political Science Association and the 1997 meeting of the American Political Science Association We would like to thank Chris Achen Matt Gabel Michael Giles John Huber Eric Lawrence Ed Schwartz Jerry Sorokin Katie Tenpas and especially Kathy Bawn Cary Covington and Becky Morton for helpful comments and discussions We also would like to thank Tim Groseclose Greg Adams and George Krause for providing us with ideology scores American Journal of Political Science Vol 43 No 4 October 1999 Pp 1069 1095 1999 by the Midwest Political Science Association 1070 Bryan J Moraski and Charles R Shipan success rate of greater than 93 percent1 Given this tremendous rate of suc cess we believe the important question to examine is how the president de cides what kind of nominee he should choose Our primary interest is in determining how the political context in u ences nominations In particular we look to see whether the existence of other political institutions the preferences of those other institutions and the constitutionally prescribed sequence of the process affect the president s choice We also explore whether other factors such as presidential popular ity and years remaining in of ce affect nominations Our goal is to deter mine which of these factors in uence the president s choice 2 THE SUPREME COURT AND THE SEPARATION or POWERS The selection of a Supreme Court justice begins with an opening on the Court due to either the death or retirement of a justice When a vacancy oc curs the president nominates a candidate and the candidate is con rmed or rejected by a majority vote of the Senate If the position of chief justice be comes vacant the president may either nominate both a sitting justice for the position and a new associate justice or he may nominate a new chief justice from outside the Court2 Surprisingly even though the president plays a major role in this pro cess most recent systematic analyses of Supreme Court nominations have focused mainly on the Senate s con rmation vote3 We have learned that the Senate vote is more likely to be con ictual when the nominee is less quali ed or is ideologically out of step with the Senate Cameron Cover and Segal 1990b We have also seen that the timing of the nomination Segal 1987 and the preferences of constituencies and interest groups Overby et a1 1992 Segal Cameron and Cover 1992 Caldeira and Wright 1998 in u ence the Senate s vote Additional evidence suggests that the ideology of the departing justice also may matter Ruckrnan 1993 1The four nominees rejected by the Senate were John J Parker in 1930 Clement Haynsworth Jr in 1969 G Harrold Carswell in 1970 and Robert H Bork in 1987 Despite being rejected these nominees all received a good deal of support when the Senate voted ranging from Bork s 42 percent to Parker s 49 percent In addition to these rejections two other nominees were not con rmed by the Senate Johnson withdrew his nomination of Abe Fortas for the position of chief justice before vot ing took place and the Senate took no action on Johnson s nomination of Homer Thomberry 2William Rehnquist for example was an associate justice when nominated and con rmed for the position of chief justice His predecessor Warren Burger was not on the Court when nominated and con rmed 3A wealth of descriptive material on how presidents choose from among potential nominees can be found in a number of studies For recent examples see Abraham 1991 Maltese 1995 and Silverstein 1994 For systematic evaluations of presidential nominations see Massaro 1978 1990 Cameron Cover and Segal 1990a Watson and Stookey 1995 and Nemacheck and Wahlbeck 1998 SUPREME COURT NOMINATIONS 1071 These studies of the con rmation vote treat the choice of a nominee as exogenous Put somewhat differently although many of these studies as sume that members of the Senate behave rationally in deciding how to vote they do not make the same assumption for the president Instead the strate gic manner in which a president might choose a nominee is ignored Yet most nominees receive a large majority of votes which suggests the need for a careful analysis of the nomination stage Lopsided con rmation votes are almost certainly due in part to the careful consideration given by presidents to the choice of nominees consideration given with an eye toward the con rmation vote How might the president behave strategically Given the Court s key role in setting public policy the president will want a Court that shares his ideology and thus will nominate someone who will bring the Court closer to his preferences At the same time however the president is constrained by institutional features of the nomination process First the Senate has the power to con rm or deny the president s choice which may force him to take the preferences of the Senate into account when nominating someone for a seat on the Court Second depending on the con guration of prefer ences of continuing justices and his own ideology the president may not be able to shift the Court so that it completely shares his own preferences In stead he may be able to move it only a short distance toward his ideal point 3 THE NOMINATION GAME The requirement that the Senate must approve a president s nominee places a certain structure on the nomination game In the rst stage a va cancy occurs in the second stage the president nominates someone to ll that vacancy and in the third stage the Senate votes on the president s nominee4 Because the president knows that the Senate will vote on his nominee when he makes his choice he will take the preferences of the Sen ate into account Given the constitutionally prescribed sequence of this game and relying on the assumption that actors are rational and forward looking we can model the nomination game using the equilibrium concept of subgame perfection5 4For reasons discussed later in this paper we omit the Judiciary Committee from our analysis 5Analyses using the same equilibrium concept to examine the president s appointment power include Nokken and Sala forthcoming Hammond and Hill 1993 and Snyder and Weingast 1994 Like Snyder and Weingast we model the appointment process as occurring along a single dimension Although some recent analyses most notably Nokken and Sala forthcoming indicate that a second dimension might be empirically relevant for Supreme Court con rmation voting pre vious studies have demonstrated that a single dimension provides a useful approximation of the more complicated multidimensional reality of the appointment process eg Segal Cameron and Cover 199239 Cameron Cover and Segal 1990b In addition singledimensional analyses are more 1072 Bryan J Moraski and Charles R Shipan If the Senate rejects the nominee the game ends In other words we treat this as a singleperiod game Several arguments provide justi cation for focusing on a singleperiod game To begin with as our analysis demon strates below there are a number of interesting and testable implications that derive from a singleperiod model6 Furthermore in order eventually to model the nomination process as a multiperiod game it is useful rst to un derstand the singleperiod game Most importantly a single period game is a very plausible representa tion of the nomination process For a number of reasons the president will not want the game to continue The president s public approval and standing with the Senate may suffer from such a rejection Thus there is a loss of po litical capital associated with putting forth a nominee who is rejected More over the president has to expend this capital in a losing cause whereas it might otherwise be spent more pro tably on other policy issues In addition there are timerelated costs If his term in office is ending the president might not get another chance to nominate someone Alternatively the Senate might change and become more inhospitable to his preferred nominees e g the president s party often loses Senate seats at the midterm election Fi nally the president will lose time in which he could have a Court that pro duces outcomes more in line with his preferences Because of these costs he has a strong incentive to nominate someone who will be approved We make the following assumptions in the game First all players have perfect and complete information about the preferences of other actors and the sequence of the game Second players have Euclidean preferences Third the actions of players are driven by their preferences over policy an assumption we elaborate on below Fourth recognizing that the size of the Court has remained at nine members for well over one hundred years we assume that after a justice retires or dies the Court will have eight members Fifth we assume that if the Senate does not approve a nominee the seat re mains open and the Court functions with eight members Finally in the spa tial models presented below we also assume without loss of generality that the president s ideal point is to the left of the Court median The game begins when a seat opens on the Court7 Prior to the creation of the vacancy the median of the ninemember Court is equal to the position tractable theoretically and empirically We agree with Krehbiel s 1996 assessment that multidi mensional institutional theories are more likely to yield examples than general propositions and as such it is dif cult to discern whether or how the general properties of multidimensional choice within interbranch institutional settings differ from the unidimensional results 1996 34 6A singleperiod approach also facilitates empirical testing by narrowing the potential equilib rium predictions Morton 1999 Allowing a game to be repeated often produces multiple equilibria which could prevent the model from yielding clear and testable implications 7We treat the existence of a vacancy on the Court exogenously In a more complete model va cancies could be endogenous as some openings occur due to strategic considerations Squire 198839 Hagle 1993 SUPREME COURT NOMINATIONS 1073 Figure 1 Supreme Court with a Vacancy Ideology of the fth justice Once the vacancy occurs however this changes In Fig ure 1 each of the eight remaining justices J 1 through J8 is located along a scale according to his or her ideology Because there are now only eight members on the Court no single justice occupies the median position In stead the policy outcome will be a lottery over the interval J 4 J5 and we therefore allow the midpoint of this interval J J 4 J 52 to be considered the Court s median after a vacancy occurs Once a seat opens the president has an opportunity to nominate a poten tial new justice We assume that the president whose ideal point we denote by P is motivated by policy concerns Thus he wants to move the median of the Court as close as possible to his own ideal point In other words the president wants to minimize the distance between P and J where J is de fined as the location of the new median if the president s nominee is ap proved by the Senate8 Consider rst what the president would do in the absence of the require ment that the Senate confirm his nominee In such a case of course the president still cannot simply choose a new median for the Court Rather he is constrained by the presence of eight sitting justices Because of these continuing justices the new median will remain in the interval J 4 J5 re gardless of the ideological position of the nominee More speci cally if the president chooses a nominee N such that N lt J4 then the new median J 8We do not explicitly model the process through which the Supreme Court reaches decisions instead we rely on the median voter on the Court to provide a proxy for its decision making out comes Somewhat surprisingly very little attention has been paid to the accuracy of using the me dian voter to represent the Court s policy positions We believe that the relevance of the median voter to Court decision making deserves further attention but here use three arguments to support its use First and most simply we rely on Black s median voter theorem which holds that in a voting body whose members can be arrayed along a line with singlepeaked preferences the outcome will be the median voter Second although some studies have demonstrated that the chief justice can act strate gically in assigning opinions Maltzman and Wahlbeck 1996 it remains true that both the chief jus tice and the opinion author must take care to satisfy the median member of the Court for without the support of that member they cannot sustain a majority Thus changes in the median lead monotoni cally to changes in the Court s output Finally and most impressionistically close observers recog nize the importance of the median member of the Court When National Journal recently published a list of the most important political actors in Washington Sandra Day O Connor and Anthony Kennedy usually at or near the median on the Court made the list but Justices Rehnquist and Scalia did not See The Washington 100 National Journal June 14 1997 1074 Bryon J Moraski and Charles R Shipan will be equal to J4 Similarly if N gt J 5 then J J5 and if N is located in the interval J 4 J5 then J N We maintain that the president s choice is motivated by the effect the nominee will have on the Court s median The president will choose a nomi nee who if approved will bring the Court s median closer to his own ideal point He can do this by choosing a nominee whose ideal point is close to his own ie N P Thus what is key from our perspective is that the president can achieve both of these goals minimizing both IP Jl and IP N simultaneously Consider Figure 1 keeping in mind that for the moment we are still ig noring the role played by the Senate If the president is located to the left of J 4 he will want to move the median to J4 and can do so by choosing a nomi nee such that N P Similarly if P gt J 5 any nominee such that N gt J5 will move the median to J5 and the president again will choose N P Finally when the president is located in the interval J 4 J5 choosing a nominee at his own ideal point will cause the new median to be located at his ideal point ie N P will lead to J P In other words regardless of his location relative to the existing Court median the president can move the median toward his ideal point by choos ing N P9 In some cases he is constrained by the presence of the continu ing justices in the sense that he can only move the median a portion of the distance to his own ideal point But in all cases by choosing N P he ob tains the best possible new median What happens when we introduce the Senate into the model Obviously the president now must take into account the preferences of the Senate be fore deciding on a nominee His goal will be to choose the nominee who will produce the best new median and who also will be approved by the Senate10 Whether the Senate constrains the president however depends on the con guration of institutional preferences As the model we develop in the next section demonstrates there are three distinct regimes and which vari ables affect the position of the nominee depends on which regime exists In the rst two regimes discussed below the Senate and president agree on the direction of change for the median although they may disagree on the amount In the third regime they disagree about even the direction of change 9Choosing N P therefore is a weakly dominant strategy for the president The president could of course choose other values of N However by choosing N P he not only moves the me dian toward his ideal point he also increases the likelihood that future medians will be located close to his ideal point 10We assume that the Senate also is motivated primarily by policy concerns and is most con cerned with how the new nominee will if con rmed affect the median on the Court We base this assumption on the demonstrably strong relationship between senators policy preferences and their votes on Supreme Court nominees see eg Cameron Cover and Segal 1990b Ruckman 1993 SUPREME COURT NOMINATIONS 1075 31 Regime 1 Unconstrained Presidential Power First consider the case where S lt P lt J as illustrated in Figure 2a The president knows that if he nominates someone who shares his ideology ie N P and that person is con rmed then the new median will be equal to J4 In addition he knows that the Senate prefers any new median 1 such that J lt J Since a nominee located at P produces J J 4 the president knows that the Senate which prefers J 4 to J will approve such a nominee Of course the Senate might prefer a nominee located closer to S than the one the president is willing to put forward Since it prefers J 4 to J however the Senate s hand is forced by rejecting the president s nominee it would decrease its own util ity The president by virtue of having the rst move can present the Senate with an option that while not perfect is one it prefers to the status quo More generally in this regime the president is unconstrained by the Senate by choosing a nominee at his ideal point he also moves the median closer to the Senate s ideal point Put somewhat differently any nominee that the president likes will create a new Court median that falls within the Senate s win set This holds true whenever S lt P lt J and also usually when Figure 2 Presidential Nominating Regimes 2a Unconstrained President Regime 1 I I I I I j I I s P J4 J 2b SemiConstrained President Regime 2 I J 20 Fully Constrained President Regime 3 1076 Bryon J Moraski and Charles R Shipan P lt S lt J with the exception of the condition discussed in the next section11 In effect when the president and the Senate are on the same side of the Court median the president is almost always unconstrained by the Senate position and N will be a function of P but it will not be a function of S or J 32 Regime 2 SemiConstrained President If the Senate is located closer to the median of the Supreme Court than is the president ie J SI lt J PI the president may not have as much power as he does in Regime 1 More speci cally if P lt S lt J S gt J 4 J 2 and P lt 2S J the president will no longer be able to present the Senate with a takeitorleaveit offer and be con dent that the Senate will take it12 To see this consider Figure 2b If the president chooses a nominee such that N P then the new median would be J J 4 The Senate however prefers J to J and therefore would reject the nominee In fact the Senate prefers any point to the right of IS which is de ned as IS 2S J and is the Senate s in difference point with respect to the Court median to J As opposed to Regime 1 where the president is constrained by neither the Senate s ideal point nor the median of the Court here the president is constrained by a combination of the two What nominee can the president choose in order to obtain the best possible new median Working backward we can see that the Senate will approve any nominee that yields a median that is closer to S than is J The president then knows that the best he can do is to choose a nominee who will produce a new median at IS He can do so by choosing a nominee such that N IS Hence in this regime the president has to nominate someone whose position is determined by the combination of the ideal points of the Senate and the Court He is semiconstrained that is he can choose a nominee who will bring the Court s median closer to his own ideal point but he cannot do so by choosing N P and he cannot move the median all the way to J 4 33 Regime 3 Fully Constrained President Finally the president and the Senate might be located on opposite sides of the Court s median as depicted in Figure 2c In this regime the president and the Senate disagree completely about the ideological direction the Court should take Indeed any attempt by the president to move the median closer to his ideal point will be rebuffed by the Senate and the Senate will never llWhen S lt P lt J P is always in the Senate s win set and the president can always choose N P 12In other words if the Senate is closer to J than to 1 and the president is farther from S than S is from J then the president is semiconstrained SUPREME COURT NOMINATIONS 1077 see a presidential nominee who is likely to move the Court s median toward S There is a standoff and the only outcome that would please both the president and the Senate would be a nominee located at J Regardless of whether the distance between J and S is small or large as long as P and S are on opposite sides of J the president will be fully constrained in his ability to move the Court median closer to his ideal point Equilibrium Outcomes The preceding discussion demonstrates that while P J and S each are important determinants of the location of N their importance is a function of the specific con guration of preferences More explicitly the model demon strates that there are three distinct theoretical regimes one in which N is a function of P one in which N is a function of IS which in turn is a function of S and J and nally one in which N is a function of J Figure 3 depicts the equilibrium outcomes as a function of the Senate median Holding P and J constant we see that when S lt P the outcome will be equal to P regardless of the location of S Similarly when S gt J the out come will be J regardless of the location of S Only when S is located to the right of J 4 J 2 ie the midpoint of the interval J 4 J does the location of the Senate median have a direct in uence on the outcome Similar gures could be drawn in which P is allowed to vary while S and J are held constant or where J is allowed to vary while P and S are held constant In each gure we would see that within one regime the variable of interest would have an effect on the outcome while in the other two regimes it would not Figure 3 Nominee s Ideology as a Function of the Senate s Location Nominee s Position 1 Regime 3 Regime 1 Reg 2 l 1 Location of S 1078 Bryon J Moraski and Charles R Shipan This theoretical nding that different variables are in uential in dif ferent regimes guides our empirical approach In particular the theory points to the following speci cation of the empirical model where D1 D2 and D3 are dummy variables indicating the nature of the re gime13 As the theoretical model and the empirical speci cation in Equation 1 make clear the nominee s position is affected by either the president s ide ology or the Senate s indifference point or the Court s median It is not however a function of all of these variables at the same time Whether each variable matters depends on the location of S relative to J and P In effect then this adopts the approach of switching regressions where different in dependent variables matter under different conditions14 4 MEASURING INFLUENCES ON THE NOMINATION To examine the choice of Supreme Court nominees we consider all jus tices nominated to the Supreme Court between 1949 and 1994 ie from Tom Clark through Stephen Breyer This sample is relatively small with only twentyeight observations and it is limited in time15 However since these were the only nominees for whom we could obtain data for all the vari ables this problem could not be avoided16 Our dependent variable which follows from our theoretical model is the ideological position of the nominee To operationalize this variable we use the ideology scores developed for Supreme Court nominees by Segal and Cover 1989 Segal and Cover derived these scores which are meant to be predictive of a justice s future behavior from a content analysis of pre con rmation editorials about the civil liberties and civil rights tendencies of the nominees Like ADA scores these predicted scores range from zero to 13More formally D1 1 if S lt P lt N or if P lt S lt N S lt J4 J2 and P lt IS otherwise D10D21ifPltSltNSltJ4J2andPltISotherwiseD2OandD31ifPltJltS otherwise D3 O 14The switching regimes regression approach is especially valuable when we do not know which regime exists eg Lee and Porter 1984 In our analysis the location of the various actors spells out which regime exists See also footnote 29 15Another potential problem is that we cannot observe all the people that a president chooses not to nominate Instead we make the reasonable assumption that the president is allowed to choose from a set of potential justices whose preferences range across the ideological spectrum For a per ceptive analysis that includes people who were considered but not nominated see Nemacheck and Wahlbeck 1998 16The limiting variable is the ideology of elected of cials ADA scores extend back to only 1947 and WNOMINATE scores for the president extend back to roughly the same time SUPREME COURT NOMINATIONS 1079 one with zero being the most conservative and one being the most liberal ideology Table 1 presents the descriptive statistics for this variable and for other variables we discuss below To measure the ideology of the Senate we use the yearly ADA scores of the median member of the Senate We adjust these scores according to the methodology recently developed by Groseclose Levitt and Snyder 1999 which controls for the possibility that the ADA scale shifts and compresses or expands over time Measurement of presidential ideology is somewhat less straightforward One option would be to use a dummy variable to designate a president s par tisanship We prefer to use a more differentiated measure however and thus use presidential ADA scores which are computed by examining each presi dent s positions on the votes ADA uses to calculate scores for members of Congress Zupan 1992 As with the ADA scores for the Senate the presi dential ideology scores also range from zero to one with zero being most conservative and one most liberal And as with the Senate ADA scores the presidential ADA scores are computed yearly and are adjusted While ADA scores and the nominees ideology scores both can be aligned along the same zerotoone scale a potential problem exists ADA scores are general measures of ideology computed in order to demonstrate the overall liberal or conservative tendencies of each member of Congress The nominees scores on the other hand focus on civil liberties and civil rights and thus are more issuespeci c than ADA scores Yet there are rea sons to use them together First justices cast far more votes in civil liberties and civil rights cases than in any other policy area Thus a good portion of their overall ideology is comprised of these more speci c scores Second at the congressional level ADA scores and ACLU scores are highly correlated Third both types of scores represent attempts by third parties to locate po litical actors along a similar zerotoone conservativetoliberal dimension In the end we maintain that it remains the best option The only precon rmation measures of nominees ideologies are the civil liberties scores and ADA scores for the Senate and president are relevant and extend farther back in time than do other issue speci c scores To measure the median position of the Court we look at the median voting score on civil liberties from the Court s previous term We compute this by looking at the voting scores of the eight continuing members of the Court and using the midpoint of the interval 14 J 5 as the median For ex ample in 1994 Stephen Breyer was nominated to ll the seat vacated by the retirement of Harry Blackmun Of the eight justices from the 1993 term who were retaining their seats Justice O Connor had the fourth highest voting score and Justice Kennedy the fth Since their scores were 362 and Table 1 Descriptive Statistics Standard Skewness Variable Maximum Minimum Mean Median Mode Deviation Statistic Nominee s Ideology Scores 100 000 045 045 050a 033 026 President s Ideology 089 006 035 034 034a 031 026 Senate s Indifference Point 060 029 021 021 021a 024 026 Court Median 076 026 052 050 050 012 030 President s Ideology in Regime 1 085 000 022 011 000 027 096 Senate s Indifference Point in Regime 2 059 000 001 000 000 017 274 Court Median in Regime 3 065 000 011 000 000 021 159 Presidential Approval 079 040 059 062 065 01 1 0 15 Years Remaining in Of ce 100 025 073 075 075a 022 018 Quali cations of Nominee 100 011 077 099 100 027 111 aindicates one of multiple modes N 28 Note Scores for the president and the Senate are adjusted ADA scores These adjusted scores can take on values greater than one or less than zero We used the full range of values in our analysis but the results were essentially the same if we bounded these scores at zero and one SUPREME COURT NOMINATIONS 1081 447 respectively the median we use for Breyer s appointment is 405 the average of these two scores17 5 EMPIRICAL ANALYSIS Table 2 presents the distribution of nominations by regime type Of the twentyeight observations the majority eighteen to be speci c fall into Regime 1 Three nominations Stevens Bork and Kennedy fall into Re gime 2 The remaining seven nominations belong to Regime 3 The exist ence of nominations in each regime will allow us fully to test our theory but the small number of observations in the second and third regimes will de crease the likelihood of obtaining statistically signi cant results As our theoretical model indicates the location of the nominee N should be a function of the president s ideal point the Senate s indifference point and the existing median on the Court once a vacancy occurs How ever the type of regime determines which of these variables should have an effect on N In Regime 1 we expect P to have a signi cant effect in Regime 2 we expect IS to have a signi cant effect and in Regime 3 we expect J to have a signi cant effect Bivariate analysis provides initial support for our model In Regime 1 there is a strong relationship between the president s ideology and the nomi nee s ideology r 60 p lt 01 Similarly in Regime 3 we nd a strong re lationship between the Court median and the nominee s ideology r 76 p lt 05 In Regime 2 we do not nd support for our theory r 008 this is unsurprising given the small number of observations in this regime Table 3 presents the results of our switching regressions Column 1 in this table presents the most basic and straightforward test of our model All three of the variables have positive coefficients as the theory predicts Both the president s ideology in Regime 1 and the Court s median in Regime 3 differ from zero at p lt 0118 Only the Senate s indifference point in Regime 2 fails to reach standard levels of signi cance although here and in a range of other specifications its coefficient is in the expected direction On the whole the model does a good job of predicting the ideology of nominees to the Court Before proceeding to a discussion of alternative hypotheses and speci cations we wish to make two general points about our empirical analysis 17In some cases there were only seven continuing members on the Court when a nominee was chosen In such cases we use the voting score of the median or fourth justice In other cases there were seven sitting justices and also one justice who had been con rmed but who had not yet estab lished a voting record on the Court For these cases we used the voting scores of the continuing jus tice along with the predicted vote of the newly con rmed justice 18All signi cance levels we report for the theoretical variables of interest are based on one tailed tests 1082 Bryon J Moraski and Charles R Shipan Table 2 Nominees Classi ed by Regime Type Regime 1 Regime 2 Regime 3 Unconstrained SemiConstrained Fully Constrained President President President Warren Stevens Clark Brennan Bork Minton Whittaker Kennedy Harlan Stewart Fortas associate White Fortas chief justice Goldberg Souter Marshall Thomas Burger Haynsworth Carswell Blackmun Powell Rehnquist associate O Connor Rehnquist chief justice Scalia Ginsburg Breyer Note Regime 1 occurs when S lt P lt J Regime 2 occurs when P lt S lt J S gt J4 J2 and P lt ZS J and Regime 3 occurs when P lt J lt S The same regimes occur for the mirrorimages of these preference con gurations ie J lt P lt S also is classi ed as Regime 1 First because ADA scores are imperfect proxies for ideology we reran the regression shown in Column 1 and others presented in this paper using several other measures for ideology Using raw unadjusted ADA scores produced results that were essentially the same as or in some cases even better than mthose produced using real adjusted ADA scores Adjusted W NOMINATE scores also yielded strong support for our theory Because no measure of ideology is perfect scholars are better off using a variety of mea sures to test a theory The similarity across different measures in our analy sis demonstrates the robustness of our results Second the signi cance of our theoretical variables can be tested in two different ways Our theory predicts that the coef cients of these variables 19The NOMINATE scores were rescaled from 0 conservative to 1 liberal so they would be consistent with our other measures Using the average of a president s ideology score over the course of his term produced results that were nearly identical to those found in Column 1 Using the presi dential ideology scores of Segal Howard and Hutz 1996 which are based on expert assessments yielded similar results SUPREME COURT NOMINATIONS 1083 Table 3 Predicting the Ideology of Supreme Court Nominees Column 1 Column 2 Column 3 Column 4 Presidential ADA Scores 083 087 092 083 in Regime 1 only 023 025 053 025 Senate s Indifference Point 020 006 062 in Regime 2 only 032 030 039 Court Median 118 142 123 096 h1Reghne3n y 129 174 161 129 Presidential ADA Scores 004 in Regimes 2 and 3 042 Senate s Indifference Point 055 in Regimes 1 and 3 030 Court Median 018 in Regimes 1 and 2 047 Presidential ADA Scores 004 in all Regimes 042 Senate s Indifference Point 055 in all Regimes 030 Court Median 018 in all Regimes 047 Constant 013 011 011 012 010 025 025 010 Adjusted R2 041 049 049 031 SEE 026 024 024 028 DurbinWatson 1 49 1 99 199 1 21 Numbers in parentheses are standard errors N 28 should be positive and signi cantly different from zero In other words our model predicts that these theoretical variables should have a positive effect on the nominee s ideology However our theory also predicts that a oneunit change in the relevant independent variables should produce a correspond ing one unit change in the dependent variable To test this second prediction the proper null hypothesis is that the coef cient is equal to one and failure to reject this null constitutes support for our theory Both predictions are im portant We cannot simply test Whether the coef cient differs from one be cause a coef cient indistinguishable from one also might be indistinguish able from zero eg a coefficient of 100 with a standard error of 200 1084 Bryan J Moraski and Charles R Shipan Similarly we cannot test only whether the coef cient differs significantly from zero since such a coefficient might be statistically distinguishable from one e g a coef cient of 020 and a standard error of 004 In the re mainder of the paper we will report p values based on the null hypothesis that the coef cient is equal to zero but we will also note whether the coeffi cients differ signi cantly from one It is extremely important and support ive of our theory to note that in every instance where we nd that the coef ficients are signi cantly different from zero we also find that they are not significantly different from one20 Alternative Hypotheses While the most basic test of our theoretical model is the one presented in Column 1 our theory also produces additional hypotheses In particular the theory not only predicts that the president s ideology should be a signi cant in uence on the nominee s ideology in Regime 1 it also predicts that the president s ideology should not be a significant in uence on the nomi nee s ideology in Regimes 2 and 3 Similarly the Senate should play no role in Regimes l and 3 and the Court median should be insignificant in Re gimes l and 2 In other words instead of expecting that the variables for the president Senate and Court should be significant and positive and equal to one in all three regimes the theory predicts regimespecific effects The equation to be estimated is thus Nl30l31D1P32D2ISB3D3JB4D2D3P BSD1D3ISB6D1D2J 2 where we expect 31 32 and 33 to be signi cant and 34 35 and 36 to be in signi cant Column 2 in Table 3 presents the results for the alternative speci cation 2 Again the results are supportive of our theory The president s ideology is signi cant in Regime l but not in Regimes 2 and 3 and the Court median is signi cant in Regime 3 but not in Regimes l and 2 Both variables also are indistinguishable from one Once again the Senate s indifference point is not a signi cant in uence on ideology in Regime 2 although it curiously pro duces a somewhat signi cant but negative coef cient for Regimes l and 3 While this alternative speci cation follows directly from our theoretical model a number of other hypotheses also are plausible For example the nominee s position might be a function of the preferences of the president the Senate s indifference point and the Court irrespective of the regime The expectation for this alternative hypothesis would be that as the ideologi 20We also nd that our constant term is always indistinguishable from zero which the theory also predicts SUPREME COURT NOMINATIONS 1085 cal scores of any of these actors increase so would the ideological score of the nominee To test this alternative hypothesis we added these three variables to Col umn l of Table 3 ie we removed the parenthetical terms from the last three interactions in Equation 2 The results of this test shown in Column 3 once again provide support for our model The president s ideology in Re gime l and the Court s median in Regime 3 again have coef cients that are in the predicted direction are signi cantly different from zero at p lt 05 and are also indistinguishable from one Interestingly the Senate s indifference point in this specification also approaches standard levels of signi cance p lt 10 and is indistinguishable from one lending an additional measure of support to our theory At the same time little support is given to the alternative hypothesis None of the noninteracted variables are signi cant in the predicted direction and a joint Ftest shows that they are jointly insigni cant at the 05 level21 These results like those presented in the other columns lend support to our theoretical argument and to the robustness of our results22 As the theory predicts the president s choice of a nominee appears to be a function of the president s preferences in Regime l the Senate s indifference point in Re gime 2 and the Court s median in Regime 323 21Furthermore we can reject the hypothesis that these other variables are equal to one As in the previous column the variable measuring the Senate s indifference point was weakly signi cant albeit with a negative coef cient This implies that for example if P lt IS lt J as IS moves to the right the president will choose a nominee farther to the left We offer no explanations for this result but simply note that it presents an opportunity for future research 22Similarly we allowed the constant to vary across regimes by omitting the constant from the estimation and instead including D1 D2 and D3 None of these noninteracted regime dummy vari ables were signi cant while the interaction of D1 with the president s preferences and the interac tion of D3 with the Court s position remained signi cant at the p lt 05 level 23The text reports two of the alternative hypotheses that we tested whether the president Sen ate and Court matter across all regimes and whether these three variables matter in regimes in which we expect them not to matter We also examined a variety of other null hypotheses including one in which the nominee s ideology is regressed on the president and Senate alone one where it is re gressed on the president and the distance between the president and the Senate one where we in clude the president alone and one where we include variables about the president s popularity and time remaining in of ce and the nominee s quali cations discussed below with respect to Table 4 We also included the regimespeci c theoretical variables in these models While there were some differences across these speci cations in general the results showed that the regimespeci c inter action terms were signi cant and the alternative variables were not The main exception to this oc curred when we included the president in all three regimes with the three regimespeci c theoretical variables In this case due to collinearity nothing was signi cant We discuss collinearity below We also note that even if these alternative models had not generally produced insigni cant results for variables we expect to be insigni cant and signi cant results for variables we expect to be signi cant it can be argued that none of these alternatives are welldeveloped enough to provide speci c alternative hypotheses or theoretical interpretations 1086 Bryon J Moraski and Charles R Shipan While these results support our theory we wish to introduce a bit of caution in interpreting them First standard hypothesis testing is more ap propriate for testing whether a particular variable is signi cant ie when the null hypothesis is that the variable is not different from zero than for situations where the model holds that a variable should be insignificant Thus conventional levels of signi cance should not be used to determine whether these coef cients are indeed equal to zero as this might result in a Type II error One approach that social scientists take to deal with this prob lem is to use a more stringent test of p lt 50 for a onetailed test e g Hall and Grofman 1990 Even according to this much more stringent test none of the variables we expect to be insigni cant turn out to be signi cant in a positive direction A second problem is more serious and has to do with collinearity be tween the variables In both Columns 2 and 3 collinearity is a problem24 Unfortunately two of the standard ways to x collinearity are not plausible options we can neither add more observations nor delete independent vari ables LewisBeck 1995 Thus we cannot directly address the problem of collinearity The primary effect of multicollinearity of course is to increase the size of standard errors Thus it provides a bias against nding signi cant results We cannot know whether coef cients are indistinguishable from zero be cause the model is correct or because of multicollinearity and random varia tion producing spurious results Therefore it is even more surprising that our primary theoretical variables the president in Regime 1 and the Court me dian in Regime 3 remain signi cant under a wide range of speci cations In addition these variables remain signi cant for alternative measures for each variable as noted earlier Moreover the variables that we do not expect to be signi cant also turn out to be insigni cant under a wide range of speci cations and operationalizations25 Although we cannot ignore the problems 24For example the bivariate correlations range from 01 to 89 And when we regress Presiden tial Ideology in Regimes 2 and 3 on the other independent variables in Column 2 we nd a high adjusted R2 083 Other diagnostic tests were much more encouraging In Columns 2 and 3 the DurbinWatson statistic allowed us to reject the possibility of autocorrelation in Column 1 the sta tistic is inconclusive Furthermore plotting the standardized residuals of Column 1 against its stan dardized predicted values demonstrates that the analysis was not driven by outliers To complement this visual inspection of the data we relied on Cook s distance test which is a measure of how much the residuals of all cases would change if a particular case were excluded from the calculation of the regression coef cients A large Cook s D indicates that excluding a case from computation of the re gression statistics changes the coefficients substantially In the case of only one observation Stephen Breyer did this test indicate the existence of a potential outlier Removing Breyer from the dataset actually improved our results it increased the overall t of the model and made our theo retical variables more signi cant 25Comparing the degrees of collinearity our theoretical variables encounter to the degrees of collinearity encountered by the variables representing the alternate hypotheses yields further support SUPREME COURT NOMINATIONS 1087 potentially caused by multicollinearity the stability of our results across a wide range of tests lends additional support to our ndings In the end how should the ndings be interpreted We will not be sur prised if future theoretical work demonstrates a more important role for the president in other regimes Because of the collinearity between the presi dential variable and the other variables it is dif cult conclusively to reject this possibility And in the next section we explore a couple ways in which the president might be able to be more in uential in Regimes 2 and 3 At this point however the evidence suggests that the president is signi cant in Regime l and the Court median is significant in Regime 3 while these variables and the Senate are not signi cant in other regimes Reclassifying Borderline Observations One finding from the previous columns that has not been consistent with our theory is the lack of signi cance of the Senate s indifference point in Regime 2 No doubt this is due at least in part to the small number of cases in this regime However it also calls for a closer examination of the three observations that occur in this regime And it turns out that in all three cases the observations might justi ably be reclassi ed as belonging to one of the other regimes26 First let us look at the nomination of John Paul Stevens in 1975 At that time President Ford s adjusted ADA score was 18 the Court s median in the wake of Justice Douglas s retirement was 511 and the adjusted ADA score for the Senate median was 502 Based on the latter two of these values the Senate s indifference point was thus 493 Because Stevens s predicted vote was fairly conservative 380 Senate approval of his nomination would have led to a new median of 493 Since the Senate fell in between the exist ing and the potential new Court medians ie J and J we classi ed this observation as belonging to Regime 2 However given that the new median and the Senate s indifference point are equal and given that P is less than both S and J we could just as well have classi ed Stevens as belonging to for our theory For example in Column 3 of Table 3 the Regime 1 Presidential ADA Scores vari able suffers from a greater degree of collinearity an adjusted R2 of 87 when regressed on the other variables than any of the nontheoretical variables yet it is signi cant Furthermore when we use unadjusted ADA scores collinearity decreases a great deal yet the results remain the same 26We hasten to add that we examined all observations to determine which if any classi ca tions were borderline Only those classi ed as belonging to Regime 2 were even remotely border line no observations classi ed as belonging to other regimes were close to the regime borders Thus we stress that we are not reclassifying Regime 2 observations in order to provide post hoc ex planations for the lack of signi cance of the Regime 2 variable Rather because the theory predicts such sharp cutpoints and because there are only three cases in this regime we believe it makes sense to examine these cases individually to see how securely they t into Regime 2 1088 Bryan J Moraski and Charles R Shipan Regime 127 More generally any time a nomination is this close to the bor der of two regimes classi cation of regime type is uncertain since the ide ology and voting scores should be considered to measure true preferences with some error and since any uncertainty on the part of the political actors could lead to a change in regime classi cation Examination of the Bork and Kennedy nominations similarly shows that these also could be classi ed differently In both cases the Senate median is less than but very close to the Court median with S 482 and J 486 respectively Since the president s ideal point is much less than either of these other points we classi ed these nominations as belonging to Regime 2 However because the values for S and J are extremely close both Bork and Kennedy could very plausibly be reclassi ed as belonging to Regime 3 which calls for the president and Senate to be on opposite sides of the Court median28 Based on these reclassi cations we re estimated Equation 1 In Column 4 of Table 3 Stevens is moved into Regime 1 while Bork and Kennedy are placed in Regime 3 The results again show support for our theory Both the Court median in Regime 3 and the president s ideology in Regime l are sig ni cantly different from zero and indistinguishable from one29 Controlling for the President s Bargaining Advantage Several other factors outside of our model might affect the choice of a nominee We would not however expect these factors which we can think of as the president s bargaining advantage to be important under all condi tions Rather like the theoretically derived independent variables discussed earlier we expect that if these variables matter they do so only in certain re gimes We offer these as additional alternatives to our model alternatives that derive from the descriptive and historical work on nominations More speci cally in Regime l we would not expect these other factors to matter whereas we might expect them to matter in Regimes 2 and 3 In Regime l the president already can nominate someone who shares his ide ology He cannot use any political capital he might have to move the nomi nee closer to his own position In Regimes 2 and 3 however the president must choose a nominee located some distance from his own most preferred 27Furthermore since the Court median and the Senate median are separated by only one per centage point it could also be the case that J lt S or at least that actors perceive that J lt S in which case this could be classi ed as belonging to Regime 3 28In addition aggregate civil liberties voting scores probably understate the conservative na ture of the Rehnquist Court Baum 1995 Lee Sandstrum and Weisert 1996 29We also performed a switching regimes regression using LIMDEP on this equation This procedure similarly yielded signi cant results for these variables SUPREME COURT NOMINATIONS l 089 point30 Several factors might however enable him to choose a nominee such that P lt N lt IS in Regime 2 or P lt N lt J in Regime 331 We propose and test three such variables presidential approval the president s year in office and the qualifications of the nominee Each of these variables was recoded for placement on the interval from zero to one The approval ratings which register the percentage of people approving of the president were acquired for the month of each nomination from the Gallup Opinion Poll32 Gallup asked the same question in each month for each year in our sample Do you approve or disapprove of the way the cur rent president is handling his job as president The descriptive literature on nominations suggests that the greater a president s popularity the more likely he is to be able to choose a nominee closer to his own ideal point Similarly the more recently a president has been elected the more suc cessful he should be in bargaining with the Senate Consequently we include a variable measuring the percentage of the president s term that remains to be served This variable takes on the value of 1 in the president s rst year in of ce 075 in the second year and so on Finally the more quali ed a nominee is the harder it will be for the Sen ate to reject her or him Thus a president will be able to nominate someone who is closer to P if that nominee is highly quali ed33 These quali cation 30Thus we have another example of regimespeci c variables By including the variables dis cussed below without interacting them with speci c regimes we would be biasing the results toward zero since we would be mixing a regime where we expect to nd no effect Regime 1 with regimes in which an effect might exist Regimes 2 and 3 Not surprisingly when we did include the bargain ing variables without regimespeci c effects the results were insigni cant 31In effect while we do not formally model it these factors if important should improve a president s position in a bargaining game with the Senate Since the president already can nominate someone located at his ideal point when faced with a Regime 1 con guration he need not bargain and these factors should not be important in that regime However we should note that allowing these variables to matter across all three regimes did not alter the results shown below 32Gallup lacked data on Reagan at the time of his nomination of Kennedy in November 1987 As a result we used his approval rating for the last month they had it available which was August 1987 Ratings were obtained from The Gallup Opinion Index October November 1980 Report Number 182 and The Gallup Poll Public Opinion 1981 1995 33All values were rescaled to account for the location of the president If for example P lt J in Regime 3 then we would expect an increase in popularity to decrease the value of N whereas if P gt J then we would expect an increase in popularity to increase the value of N Because the ex pected signs differ according to whether P is to the left or to the right of J D2Approval is inappro priate The most straightforward way to correct for the directional issue is to multiply this interaction term by the difference between P and the Court for Regime 3 or P and the Senate s indifference point for Regime 2 For example for popularity the measures are D2ApprovalP IS and D3ApprovalP J This controls for not only direction but also for the distance between the president s ideal point and the other points Equivalent changes were made for the other variables We control for distance as well as direction because we hypothesized that for example a very popu lar president in Regime 3 located far from J could lead to a greater change in the nominee s position 1090 Bryon J Moraski and Charles R Shipan scores were obtained from The Supreme Court Compendium Epstein et al 1996 and were computed in the same way as the SegalCover ideology scores The results presented in Table 4 demonstrate little systematic support for these factors34 It appears that in Regime 3 there is at least some reason to believe that these variables affect the choice of a nominee in the expected manner High approval ratings proximity to an election and a more highly qualified nominee all seem to allow a president in Regime 3 to nominate someone whose predicted vote is on the president s side of the Court s me dian However even in this regime the results are not strong with onetailed tests the p values range from 07 for approval and quali cations to 15 for years remaining in of ce Two nal points are important to note here First we do not doubt that these factors sometimes matter What we show here is that they are not sys tematic regular and significant in uences on the president s choice Sec ond the Years Remaining variable taps into what might be called the elec toral context whereas most of this paper has been devoted to looking at the institutional context These results as well as other unreported results pro vide little evidence of a signi cant in uence from the electoral context35 Additional Interpretations What additional substantive interpretations can be drawn from our re sults particularly those found in Columns 1 2 and 3 of Table 3 First the model makes it clear that presidents are faced with different constraints at different times Two of the rst three justices that Ronald Reagan nominated were extreme conservatives William Rehnquist and Antonin Scalia Yet his final nominee to the bench was the relatively moderate Anthony Kennedy One difference between these contexts was the regime For Reagan s rst three appointments he was unconstrained ie Regime 1 His party con trolled the Senate and he was able to choose any nominee he wished know than a very popular president located a very short distance from J We also tried using a signed dummy variable eg instead of P J we used 1 if J gt P and 1 if P gt J This produced no sub stantive differences in the results 34Because of collinearity and limited degrees of freedom we have chosen to report these vari ables separately rather than together When we include all the variables shown in Table 4 in one equation the only variables that are signi cant are presidential ideology in Regime l and the Court median in Regime 3 35We also examined whether the choice of a nominee was affected by a host of other related variables such as whether it was a president s rst term whether he was running for reelection whether he was in his last year in of ce whether he was in his last two years in of ce and whether he was an unelected president ie Johnson and Ford None of these variables were signi cant Nor were variables controlling for the number of previous nominees or appointments SUPREME COURT NOMINATIONS 1091 Table 4 Controlling for Additional Factors in the Nomination of Supreme Court Justices Column 1 Column 2 Column 3 Theoretical Variables Presidential ADA Scores in Regime 1 080 081 078 023 024 023 Senate s Indifference Point in Regime 2 030 074 006 110 4 53 161 Court Median in Regime 3 104 106 095 030 032 033 Additional Factors Presidential Approval in Regime 2 027 223 Presidential Approval in Regime 3 091 059 Years Remaining in Of ce in Regime 2 010 850 Years Remaining in Of ce in Regime 3 053 05 1 Nominee s Quali cations in Regime 2 025 186 N ominee s Quali cations in Regime 3 100 065 Constant 014 013 015 010 010 010 Adjusted R2 041 038 041 SEE 026 026 026 DurbinWatson 154 151 159 Numbers in parentheses are standard errors N 28 ing that the Senate would be likely to approve of him or her However after the Senate changed hands and was controlled by Democrats after the 1986 election Reagan was faced with a different situation Once in Regime 3 he was no longer able to nominate an extreme conservative but rather was con strained because the Senate s preferred point was on the other side of the Court median Bill Clinton on the other hand disappointed some Democrats by not choosing justices who were more liberal And indeed Stephen Breyer s 1092 Bryan J Moraski and Charles R Shipan prenomination record was perceived to be fairly moderate On the other hand Ruth Bader Ginsburg s SegalCover score is liberal and even Breyer s score was far more liberal than all but one of the nominees from the previous twenty ve years Interestingly shortly after Breyer was approved Clinton s party lost control of the Senate This changed the political context to Regime 3 which would further constrain Clinton s ability to nominate a more liberal jurist The model also allows us to speculate on the substantive in uence of regimes and of elections If George Bush had defeated Bill Clinton in the 1992 election Bush certainly would have chosen nominees who were more conservative than Ginsburg and Breyer But at the same time the nomination context instead would have been Regime 3 compelling Bush to choose a moderate nominee rather than a strongly conservative nominee Speculating even further if Bush had been in office when the Republicans gained control of the Senate in 1994 we would have switched back to Regime 1 where Bush would have been unconstrained In this situation a justice like William Rehnquist might have realized that conditions were ripe for a successor who shared his ideology and that it might be a strategically opportune time to retire 6 DISCUSSION Presidents view Supreme Court nominations as important policy choices To help ensure Senate approval of their nominees and to move the policy orientation of the Court closer to their own preferences they must take into account the preferences of the Senate and the ideology and behav ior of continuing members of the Court Our theoretical model which is supported by our empirical results spells out the way in which presidents engage in such strategic behavior One potential extension to our analysis would be to incorporate the Sen ate Judiciary Committee which holds hearings on each nominee into the theoretical model Including this committee would change the president s calculus No longer could he worry about only the Senate median he would also have to worry about the committee median36 However while the Judi ciary Committee undoubtedly plays an important role in the confirmation process and while the president undoubtedly takes into account the prefer ences of its members the committee does not act as a gatekeeper Even if the committee disapproves of a nominee it still can send that nominee s name to the oor for a vote by the entire Senate37 For example neither Clarence 36See Nokken and Sala Forthcoming for such a model 37If the committee approves a nominee it automatically sends the nomination to the floor If the committee does not approve the nominee it can choose whether or not to send the nomination to the floor SUPREME COURT NOMINATIONS 1093 Thomas nor Robert Bork won a majority of the vote within the committee yet the nominations of both of these men were forwarded to the oor Instead of acting as a gatekeeper the committee s primary role is to make a recommendation that sends a signal to the rest of the Senate Whether the committee acts as a binding constraint and absolutely needs to be in cluded in this analysis is therefore an open question On the one hand the Senate has never approved a nominee who was voted down by the commit tee with Thomas s committee vote of 7 7 being the only near exception On the other hand the committee does not gatekeep thus its function is prob ably better handled by a model of information transmission Second our paper was motivated by the observation that in this century presidents have been enormously successful in getting their nominees past the Senate However it is worth noting that failed nominations were more common in the nineteenth century as were closely contested votes Were presidential appointments in general more contentious in the nineteenth cen tury than in this century Did presidents treat Supreme Court appointments differently While such questions are beyond the scope of this study they are certainly worthy of attention Our analysis represents a building block toward a fuller understanding of presidential appointments but we acknowl edge that future research should attempt to deal with these other sorts of questions Third instead of using the median voter of the Senate future work might consider using the member of the Senate located at the libuster pivot The reason for this of course is that the Senate can libuster on a nomina tion a tactic that the president will need to take into account While the use of a libuster on a Supreme Court nomination is a rare event its infrequency does not denote its unimportance38 Fourth we believe scholars should proceed cautiously when moving from our model to any predictions about Senate voting Even leaving aside the possibility that the president has made an error when choosing a nominee which presidents surely have done we would not expect all Senate con r mation votes to consist of bare majorities which this analysis might seem to indicate Events that happen between the president s nomination and the con firmation vote can in uence the Senate s action In the cases of Fortas and Thomas for example information made public subsequent to the nomination but prior to the confirmation vote undoubtedly shifted the votes of some members In addition individual senators might be in uenced by party lead ers norms or even something akin to a snowball effect causing a larger than expected majority to vote for a nominee Thus we recognize that the Senate 38The 1968 nomination of Abe Fortas to the position of chief justice was withdrawn after it was libustered in the Senate In addition it is entirely conceivable that Bill Clinton had the libuster pivot Krehbiel 1998 in mind when he nominated moderate Stephen Breyer to the bench 1094 Bryon J Moraski and Charles R Shipan might pay attention to factors other than policy positions when voting on Su preme Court nominees and recognize that incorporating such factors into fu ture models might make for a richer analysis While these and other changes might be made to our theoretical model and empirical tests we believe that our analysis sheds new light on the nominations process There is a reason most Supreme Court nominees eas ily pass the Senate presidents act strategically to choose the best nominee they can given the constraints that they face Manuscript submitted March I 6 I 998 Final manuscript received March 8 1999 REFERENCES Abraham Henry J 1991 Justices and Presidents A Political History of Appointments to the Su preme Court Oxford Oxford University Press Baum Lawrence 1995 Measuring Policy Change in the Rehnquist Court American Politics Quarterly 23373 382 Caldeira Gregory A and John R Wright 1998 Lobbying for Justice Organized Interests Su preme Court Nominations and the United States Senate American Journal of Political Sci ence 42499 523 Cameron Charles M Albert D Cover and Jeffrey A Segal 1990a Supreme Court Nominations and the Rational Presidency Presented at the American Political Science Association Annual Meeting San Francisco Cameron Charles M Albert D Cover and Jeffrey A Segal 1990b Senate Voting on Supreme Court Nominees A Neoinstitutional Model American Political Science Review 84525 534 Epstein Lee Jeffrey A Segal Harold J Spaeth and Thomas G Walker 1996 The Supreme Court Compendium Data Decisions and Development 2Ild ed Washington DC Congressional Quarterly Inc Groseclose Tim Steven D Levitt and James M Snyder 1999 Comparing Interest Group Scores across Time and Chambers Adjusted ADA Scores for the US Congress American Political Science Review 9333 50 Hagle Timothy M 1993 Strategic Retirement A Political Model of Turnover on the United States Supreme Court Political Behavior 1525 48 Hall Richard and Grofman Bernard 1990 The Committee Assignment Process and the Condi tional Nature of Committee Bias American Political Science Review 841149 1166 Hammond Thomas H and Jeffrey S Hill 1993 Deference or Preference Explaining Senate Con rmation of Presidential Appointments to Administrative Agencies Journal of Theoreti cal Politics 523 59 Krehbiel Keith 1996 Institutional and Partisan Sources of Gridlock A Theory of Divided and Uni ed Government Journal of Theoretical Politics 87 40 Krehbiel Keith 1998 Pivotal Politics Chicago University of Chicago Press Lee Emery G 111 Frances U Sandstrum and Thomas C Weisert 1996 Context and the Court Sources of Support for Civil Liberties on the Rehnquist Court American Politics Quarterly 24377 395 Lee LungPei and Robert Porter 1984 Switching Regression Models With Imperfect Sample Separation Information With an Application on Cartel Stability Econometrica 52391 418
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'