Comm 88 Research Methods
Comm 88 Research Methods
Popular in Course
verified elite notetaker
Popular in Communication
This 29 page Reader was uploaded by Jessica Rodriguez on Friday April 4, 2014. The Reader belongs to a course at University of California Santa Barbara taught by a professor in Fall. Since its upload, it has received 174 views.
Reviews for Comm 88 Research Methods
Why didn't I know about this earlier? This notetaker is awesome, notes were really good and really detailed. Next time I really need help, I know where to turn!
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: 04/04/14
Communication Science Tuesolay April D8 ZD14 Eamp3 AM RecaHquot Communication science uses empirical observations to test theories about communication processes Unique Characteristics of Science Scientific research is public 0 Published in peer reviewed journals 0 Opportunity to replicate studies Science is empirical 0 There are conscious deliberate observations 0 Many observations to combat overgeneralization problem Science is quotobjectivequot as opposed to subjective like we are with most everyday ways of knowing 0 The process should be unbiased 0 Personal biases should be controlled and removed o Explicit rules standards and procedures help do this Science is systematic and cumulative o Builds on prior studies and theories 0 New knowledge modifies the old Goals of Scientific Research What can science tell us Description o Describing the way things are o Looking for social regularities of aggregates 0 Science can tell us quotwhat isquot Explanation o Develop understanding of WHY patterns exist What causes what Prediction 0 Predict outcomes given certain factors Science can tell us what will be Science CANNOT settle questions of value 0 Can39t tell us what quotshouldquot be rightwrong goodbad moralimmoral 0 It can inform a debate but can39t settle it The Research Process Theories hypotheses and research questions The Wheel ofScience 2 Theories I S Starting at the observation point is Empirical INDUCTION I I generalizations Hypotheses HumanIstIc Interpretive science Uses qualitative methods Observations Lecture Page 1 Starting at the thesis point is DEDUCTION Traditional science Uses quantitative methods So what39s the difference between 1 Quantitative and qualitative methods Quantitative o Employ numerical measures and data analysis 0 Adhere strongly to scientific goals and principles objectivity empirical data etc Examples surveys experiments content analysis Qualitative 0 Also called interpretive research or field research 0 A quothumanisticquot form of social science I Values some aspects of science especially empiricism I But also values researcher subjectivity Examples participant observation depth interviewing conversation analysis Note there39s also purely humanistic research in comm called quotcritical studiesquot example rhetorical criticism feminist analysis cultural analysis Science is an ongoing process one could argue that you started at a different point on the wheel 2 Basic and applied research quotBasicquot theoretical research 0 Testingbuilding theoriesconceptual ideas 0 Advancing what we know about a topic quotAppliedquot practical research 0 Applying research to solve practical problems eg testing effects of an ad campaign policy change new school program company technology etc Note even the most theoretical research has practical value and the most applied research uses theoretical reasoning and arguments to form hypotheses etc Using theories in research Theory an attempt to explain some aspect of social life 0 A schoar39s ideas about howwhy eventsattitudes occur 0 Includes set of concepts and their relationships Lecture Page 2 The Research Process Thursolay April 1 Z 14 E83 AM Theories hypotheses questions etc Using theories in research Theory an attempt to explain some aspect of social life 0 A schoar39s ideas about how andor why things occur 0 All theories have concepts What are concepts 0 Terms for thingsideasparts of the theory 0 Researchers must define them Ex Social Cognitive theory Bandura we learn by watching model behavior Requires attention retention motor reproduction motivation rewardspunishments What are some concepts involved here Retention model behavior motivation rewards punishments learning Concepts are studied as variables 0 They have variations that can be measured Ex motivation How does it vary Rewards vs punishment What is the amount of rewardpunishment Ex model Are children more likely to listen to a TV character or their parent How do they identify with a hero or a villain Does the degree of likeability or similarity to viewer have any effect Scientific theories should be falsifiable O Able to be tested empirically with data 0 There is some result that if you got it would show the theory is wrong Note you can never quotprove a theory truequot you can only gain supportevidence What about the Social Cognitive Theory Can it be tested All of it Orjust certain aspects What about the theory of man made global warming Evidence suggested in the press warmer temperatures colder temperatures less ice more ice droughts floods more hurricanes fewer hurricanes No matter what happens there is an explanation Is it really a theory if there39s an explanation for everything From prior findings andor theories we derive a testable hypothesis 0 A specific prediction about the relationship between variables in your study Ex using the social cognitive theory to make a prediction about the effects of TV violence Hypothesis TV violence viewing will produce more aggressive behavior than will non vioent TV viewing What are the variables involved here What if theory or previous research doesn39t lead to a specific prediction Or if the previous findings conflict or are inconclusive Pose a research question instead of a hypothesis Ex To what extent will children imitate the behavior of a TV character who they don39t like or relate to Research question Will there be gender differences in children39s imitation of violence EXAM PLE TESTING A HYPOTH ESIS Researcher A Researcher B Testing Social Cognitive theory Testing Catharsis Theory watching others Hyp watching TV violence increases children39s aggressive behave in a way you are feeling allows purging behavior of pent up emotions IJ 39 IL J IJ Il I39l I II I ll Lecture Page 1 0 rs an vvouu v I Hyp watching TV violence increases children39s aggressive behavior Has a lot of money Method Controls how much violent tv is viewed Measures aggression on playground Aggression TVViolence Conclusion TV violence increases aggression What researcher do you trust Researcher A Tests on a larger scale Allows generalization Kids observed for longer time period Natural behaviors observed Conclusion should instead say TV violence is related to aggression Types of hypotheses and research questions Hypotheses and research questions can be o v behave in a way you are feeling allows purging of pent up emotions Hyp watching TV violence will reduce kids aggressive behavior Has limited resources Method Kids watch one of four clips each with different amount of violent acts in them 0 5 10 20 Measures number of times kids hit toys aggression Q 0 5 10 20 TV violence Conclusion TV violence decreases aggression v I uuv 39 Researcher B Correlation seen with researcher A doesn39t lead to causation What outside factors make kids in Researcher A group act the way they do In this case the conclusion is correct for the given variables and observations 0 Causal state how one variable changes or influences others 0 Correlational state mere association between variables Lecture Page 2 The Research Process Tuesday April 15 Z 14 Eamp3 AM Types of Hypotheses and Research Questions Hypotheses and RQ39s can be 0 Causal state how one variable changes another 0 Or correlational state mere association between variables Example H1 TV violence will produce more aggressive behavior than will non vioent TV watching Vs H2 The more TV violence children watch the more aggressive they39ll be see page 106 in your textbook on quotcontinuousquot and quotdifferencequot statements Different Methods for Different Hypotheses Surveycorrelational Research Eg Researcher A 0 Tests correlational types of hypothesesRO39s mere relationshipassociation I Measure some variables and relate them compare existing groups etc 0 Great for external validity I Ability to generalize results to other people andor to quotnormal life3939 settings 0 Poor for causality I Measuring two things at the same time makes it difficulty to seeknow which came first Experimental Research Eg Researcher B 0 Tests causal hypothesesRO39s I Manipulate variablesgroups control everything else and measure effects 0 Great for internal validity I Ability to establish that X caused Y rules out other explanations 0 Poor for generalizability I Control calls for a more artificial setting not like real world Less representative sample Defining Concepts and Variables Variables in Experimental Research Independent variables 0 The quotcausequot in cause effect relationships 0 Variable manipulated by the researcher Dependent variable 0 The effect or outcome 0 Variable changedaffected by IV Example hyp Greater physical attractiveness will create impressions of greater friendliness IV physical attractiveness DV impressions of friendliness Variables in Survey Research IV is considered a quotpredictorquot variable DV is what is being predicted by IV Example hyp Stronger quotfanquot identity predicts greater participation in online forums IV fan identity DV fan forum participation Lecture Page 1 Could the V39sDV39s be the other way around YES Defining Concepts and Variables Conceptual definition 0 A working definition of what the concepts means for the purposes of investigation usually based on theoryprior research Example fear What is it Operationaldefinition 0 How exactly the concept will be measured in a study MeasurementOperationalizing Variables Types of measurements Physiologicalmeasurements o Ex blood pressure brain imaging heart rate etc Sef report measures o Ex items on a questionnaire Lecture Page 2 The Research Process Thursday April 17 ZQJ14 Eamp3 AM Measuring Operationalizing Variables Types of measures 0 Physiological o Behavioral 0 Sef report Levels of Measurement Nominal categoricaldiscrete 0 Variable is measured merely with different categories 0 Not a good strategy when using variation of degree Examples gender malefemale ethnicity yesno questions TV violence TV use highlow 0 Categories must be mutually exclusive I Subjects shouldn39t be able to choose both 0 Categories must be exhaustiveall encompassing I Every option must be available for the subjects to choose from include an quototherquot category for them to pick 0 Nominal measurements are for comparing differences Ordinal 0 Variable is measured with rank ordered categories Example rank top 5 favorite TV shows most to least important political issues Interval 0 Variable is measured with successive points on a scale with equal intervals Example measure on immigration policy opinion quotThe US should build a fence along the borderquot Strongly oppose 1 2 3 4 5 6 7 Strongly agree The numbers are only relative to themselves they have no real value They39re not anchored to anything Ratio 0 Interval measurement with a true absolute zero point Example time in hours weight in pounds age in years test scores if 0 is possible 0 You can compare people and results with each other because the scale is real Interval and Ratio measures are quotcontinuousquot variables they allow a continuous hypothesis more x more y etc Measures should capture variation Use continuous variable for dependent variables where possible have good quotconceptualquot fit with variables in the hypotheses and research questions minimize potential quotsocial desirabilityquot effects Using Questionnaire Items as Measures common for V39s and DV39s on surveys common for DV39s in experiments where IV is a manipulation Types of questionnaire items Openended 0 Respondents give their own answers to questions Closedended 0 Respondents select from list of choices Lecture Page 1 choices should be exhaustive and mutually exclusive Some ClosedEnded formats Likerttype items respondents indicate their agreement with a particular statement Example Parents should talk openly about sex with their children Strongly agree 1 2 3 4 5 6 7 Strongly disagree 0 Other response options possible opposefavor not at allvery much almost neveralmost always Semantic differential 0 Respondents make ratings between two opposite bipolar adjectives Example quotMy best friend isquot Warm Cod Intelligent Unintelligent putting numbers in assigns a value to the quality and can make people nervous Lecture Page 2 Measurement in the Research Process Tuesday April 22 Z 14 Eamp3 AM Measurement Operationalizing Variables Types of measures Physiological Behavioral Sef report Levels of Measurement Nominal different categories Ordinal rank ordered Interval successive points on a scale with equal intervals Ratio interval scaling but with a true absolute 0 Composite measures Use multiple items for one variable combine those items into an quotindexquot aka scale Example variable perceived credibility of a speaker As a single item measure quotThe speaker I just heard isquot Credible 1 2 3 4 5 6 7 Not credible As a Composite measure quotThe speaker I just heard isquot Knowledgeable 1 2 3 4 5 6 7 Not knowledgeable Experienced 1 2 3 4 5 6 7 Inexperienced Trustworthy 1 2 3 4 5 6 7 Untrustworthy Honest 1 2 3 4 5 6 7 Dishonest Unbiased 1 2 3 4 5 6 7 Biased Competent 1 2 3 4 5 6 7 Incompetent 0 To get the final score take the average of all scores I Uni dimensiona index all items are added or averaged into one overall score I Muti dimensiona index group different items into different quotsubscaesquot Ie breaking down into different aspects on credibility III It39s separating the different dimensions of a variable III Benefits the possibility of scoring high on one thing and low on another Example knowledge experience and competence make the quotexpertisequot dimension of credibility trustworthiness honesty and bias make the quottrustworthinessquot dimension of credibility How GOOD is your measurement Reliability and Validity Reliability of Measurement are you measuring the concept consistently Asserting Reliability For measures using questionnaire items interitem reliability 0 Administer the same items more than once test retest split half o Look at internal consistency of similar items in a scale Example Cronbrach39s alpha numerical formula that measures reliability Uni dimensiona indexes are likely to get a low CA poor reliability Muti dimensiona indexes are likely to get a higher CA because variables are computed separately for each subscale Lecture Page 1 Similar items should get similar scores For measures using coders eg behavioral observations 0 Intercoding reliability I Compare multiple coders 0 Intracoder reliability I Compare multiple observations made by the same coder Validity of Measurement Does your measure really capture the concept you intend to be measuring Assessing Validity Subjective types of validation 0 Face validity the measure lookssounds good quoton the face of itquot 0 Content validity the measure captures the full range of meaningdimensions of the concept Criterionrelated validation aka predictive validity 0 The measure is shown to predict scores on an appropriate criterionfuture measure Example SAT scores measure your quotpotentialquot to achieve College GPA measures your achievements Construct validation 0 The measure is shown to be related to measures of other concepts that should be related and not ones that shouldn39t Example relationship between verbal aggression and hostility Don39t worry about convergent and discriminant types of validation for the midterm Relationship between the two can you have a measurement that39s reliable but not valid Yes can you have a measurement that39s valid but not reliable No reliability is part of validity Lecture Page 2 Sampling Thursday April 24 2014 800 AM Sampling how we select participants or other units for a study Sample a subset of the target population whowhat you want to report about Example target populations voters facebook users married couples juries business owners etc OR TV shows magazine ads blog posts etc Representative Sampling probability Intended to be a quotminiature versionquot of the target population Key is random selection o Everyone in the population has an equal chance of being included in the sample How representative is it There will always be a quotsampling errorquot 0 Sample data will be slightly different from the population because of chance alone aka quotrandom errorquot 0 Statistically known as the margin of error Ex National poll of 1000 people is accurate give or take 3 o Larger sample size lower margin of error Representative Sampling Techniques Simple random sampling 0 Select elements randomly from the population I Listed populations random numbers on a table I Phone random digit dialing Systematic sampling 0 From a list of the population choose every n th element 0 Must have a random start and cycle through the entire list 0 Similar results as simple random sampling I But watch out for potential periodicity Stratified sampling 0 For getting population proportions even more accurate o Divide population into subjects quotstrataquot of a particular variable I Usually stratify for demographic variables Ex sex race political party etc Select randomly from each strata to get right proportions of the population Need prior knowledge of population proportions Increases representativeness because it reduces the sampling error BUT it39s more costly and time consuming Multistage cluster sampling Useful for populations not listed as individuals 0 First randomly sample groups quotclustersquot then randomly sample individual elements within each cluster Example sampling high school athletes Stage 1 randomly sample high schools Stage 2 randomly sample athletes from those schools 0 Reduces costs o But there39s a sampling error at each stage To reduce the error use both stratified and multistage sampling 0000 For all representative sampling techniques Will always have a sampling error Lecture Page 1 Can generalize to a larger population Caution avoid systematic error sampling bias 0 Systematic over orunder representation of certain segments of population 0 Caused by I Improper weighting I Very low response rate I Wrong sampling frame I Using nonrepresentative sampling methods Nonrepresentative Sampling can39t be generalized Convenience sample 0 Select individuals that are availablehandy Example standing outside of a mall and surveying people on shopping Purposive sample 0 Select certain individuals for special reason their characteristics Example Only asking CEO39s on the Fortune 500 list questions about business What about the smaller businesses Volunteer sample 0 People select themselves to be included Volunteers can have biased beliefs there39s a reason they want to participate and can skew the results of the study Nonrepresentative sampling Techniques Quota sample 0 Select individuals to match the demographic proportion in the population nonrandom sampling It39s like stratified sampling but it can39t be generalized Networksnowballsample 0 Select individuals who contact other individuals who then contact other individuals etc The problem you only get a certain set of individuals Continuous vs Difference Statements Con nuous 0 Predicted based on a continuous variable Ex interval or ratio Difference 0 Predicted based on discrete separate and exhaustive groups Ex nominal Causal vs Correlational Variable Causal 0 Shows one variable directly affects another variable 3939If X then Yquot RelationalCorrelational 0 Shows there is a relationship between the groups Lecture Page 2 Survey Research Thursday May 1 2 14 8 AM Primary Goals Identifydescribe attitudes or behaviors in a given population Examine relationships between the attitudebehavior methods measures 0 Does X predict relate to Y Ex Does exposure to alcohol ads X predict teen drinking Y Do many factors together predict Y Ex Do alcohol ads X1 parent drinking X2 peer drinking X3 and risk taking X4 together predict teen drinking Y Administering Surveys Sef administered questionnaires Mail surveys online or emailed questionnaires handouts diaries O Relatively easy and inexpensive O No interviewer influence 0 Increased privacyanonymity BUT 0 Must be selfexplanatory 0 Very low response rate Ways to increase response rate 0 Have inducements 0 Make it easy to complete and return 0 Include persuasive cover letter andor do advance mailing sets up an expectation gives legitimacy 0 Send foow up mailings Interview Surveys Personalfacetoface 0 More flexible can probe for depth 0 Higher response rate BUT 0 More potential for interviewer influence 0 Higher costs Telephone 0 Quickest results 0 Compared to facetoface I Reduced costs I More privacy I More efficiency 0 Compared to selfadministered I More detail possible I Better response rate BUT what about call screening and cell phones Role of Time in surveys Crosssectionalstudies 0 One sample at one point in time Longitudinal studies 0 More than one point in time measured Lecture Page 1 Survey Research Tueselay May 6 Z 14 Eamp3 AM Recall from last time Role of Time in surveys Cross sectionastudies 0 One sample one point in time ie each variable is measured once Longitudinal studies 0 Variables measured more than one time to track changes over time Panel same people each time Trend different random samples from same population Eg Survey Americans every 5 years regarding their church going poll likely voters over the course of an election campaign tracking student use of alcohol and drug use Cohort different samples everywhere but of same quotcohortquot Cohort group of people anchored together by something in time Eg class of 2012 people living in New York at time of 911 Survey class of 2012 every 5 years regarding their employment since graduation How do we relate variables in research Recall the goals for surveys Identifydescribe attitudesbehaviors in a given population Looking if X and Y are related Relating Variables Depends on your hypRO39s and how your variables are measured When both IV and DV are nominalcategorical discrete All you can do is break down the percentages by category Ex Gallup survey on support of marijuana legalization Q Should it be legal 44 yes 54 no where39s the relation to another variable EX RQ Does gender predict support for legalization 2005 men 41 women 32 2009 men 45 women 44 Relation disappears What about political ideology Democrats 54 yes 45 no Republicans 32 yes 77 no If IV is categorical but DV is internalratio data Compare mean average DV scores for the different IV categories Comparing means Ex RQ Does political ideology IV predict support for legalization DV IV political ideology 0 Measured as categoricalnominal variable I consider myself check one Liberal Moderate Conservative 0 Measured as continuous variable but then collapsed to categorical Lecture Page 1 I consider myself to be Very liberal 1 2 3 4 5 6 7 Very conservative Divide participants into categories based on their scores Above median conservative Below median Iberal DV support for legalization 0 As continuous variable The recreational use of marijuana should be legal Strongly disagree 1 2 3 4 5 6 7 Strongly agree To relate variables in each IV category compute their mean score on the DV then compare means Ex conservatives M 23 Moderates M 42 Liberals M 61 RC1 does political ideology predict support for legislationlegalization YES not only do more liberals support it but liberals even feel more strongly about legalization than other groups Relating variables If both IV and DV are intervalratio data Compute correlation statistical value that relates two or more continuous variables Correlation Compute an quotrquot value quotPearson39s rquot R ranges from 1 O 1 0 R tells you I Type of correlation positive or negative I Magnitude strength of relationship Type of Relationship Positive r as x increases y increases looking at overall pattern Strength of relationship by how closely dots fit pattern Also called a quotdirectquot relationship Negative r as x increases y decreases quotmore conservative less support for marijuana Alco called an quotinversequot relationship Magnitude of relationship R ranges from zero to one both positive and negative The further from zero the stronger the relationship R 0 means no correlation no relationship What can you conclude from surveycorrelational data CAN conclude variables are relatedassociated CANNOT conclude causality Lecture Page 2 Relating Variables in Survey Research Thursday May 8 ZQJ14 Eamp3 AM What can you conclude from surveycorrelational data CAN conclude that variables are relatedassociated CANNOT conclude that one variable causes the other 0 Why not I What if the correlation works the other way around I What are the other factors Example Does more M39ing increase the quality of friendship or does a higher quality of friendship increase M39ing Do people who study more have higher grades or do people with higher grades study more What are the other factors 9 Liking school 0 Doing well makes you wanna keep doing better Remember to establish causality Variables must be related X correlated with Y Okay so far surveys can show that Must establish time order IV happened before DV Must rule out other explanationscauses So surveycorrelational research has two causality problems Causal direction time order o Does X cause Y or does Y cause X Does the chicken or egg come first Third variable problem o Does some 3rd variable explain the XY relationship Getting closer to causality To help solve the 3rd variable problem quotpartial correlationquot I Take out third variable that could be causing correlation I Measure potential third variables I Statistically quotpartial outquotcontro for effects of those third variables Can we get everyone to have the same level of interest in school I Then see if XY relationship still holds To help solve causal direction problem Need a longitudinal study quotcross agged panel designquot I Time 1 measure X amp Y variables I Time 2 measure X amp Y variables again later for the same people I Compute R39s for XampY but across the times measured measure both variables for the same people at different times then see which quotcrossquot relationship holds Lecture Page 1 Experimental Research Tuesday May 13 ZQJ14 Eamp3 AM Purpose Test hypothesis of cause and effect 0 Goal is to establish internal validity internal validity directly affected dependent variable 0 Willing to sacrifice external validity Remember to establish causality 0 Variables must be related 0 Must establish time order 0 Must rule out other explanations or causes Key elements to a true experiment Manipulation of causal variables independent variable o Divide IV into quotconditionsquot Ex IV New painkiller drug Half of the participants get the drug other half get a placebo while controlling all other variables subjects in each condition should be treated the same o Examine effects on dependent variable compare measures mean scores for subjects in each condition and see if differences exist Ex DV amount of perceived pain eg likert scale Random assignment of subjects to conditions o Everyone must have an equal chance of ending up in either condition 0 Why is it important It makes groups equal before manipulation Types of Experiments Design notation X IV manipulationtreatment O observation measures for DV R random assignment True Experiments 0 Posttest only control group design R X 01 group 1 R 01 group 2 gt no manipulation Example R X antismoking ad 01 beliefs about smoking R no ad 01 beliefs about smoking Compare average scores for each group If you get a difference between group means on 01 then the IV caused it Variations more groups several different treatments Example X1 personal cancer story X2 cancer statistics X3 tobacco industry 0 Pretest posttest control group design R 01 X 02 group 1 R 01 02 group 2 Example R 01 beliefs about smoking X antismoking ad 02 beliefs about smoking Lecture Page 1 R 01 beliefs about smoking 02 beliefs about smoking Difference between groups means IV caused it I Possible problem differences on 02 could be the result of manipulation with pretest 0 Solomon four group design R X 01 group 1 R 01 group 2 R 01 X 02 group 3 R 01 02 group 4 Compare the ones with no manipulation and the ones with manipulation I Twice the cost trouble work participants I Rarely ever done Lecture Page 2 Experimental Research Tuesday May 2 ZD14 Eamp3 AM Key elements to having a true experiment Random assignment Manipulationcontrol o Both of these make internal validity Threats to Internal Validity If not a true experiment or if experiment was done improperly 0 Alternative explanations become possible explanations Selection bias History effect Reactivity effects 0 Placebo 0 Hawthorne o Demand characteristics How do we removecontrol these threats 0 Conduct a TRUE experiment I RA to proper conditions o Be sure to treat groups equally I All groups get equal time attention etc Threats related to pretesting to measure over time 0 Testing effect Maturation Statistical regression Instrumentation Mortality attrition Experimenter effectsbias o Experimenter39s behaviorsattributes rather than treatment of IV influences DV How to Control experimental effects 0 Same thing as with true experiments but also I Automatescript experiments I Have blind experiments I Have ignorant experimenter I 3939double blindquot is best but also more time consuming 0000 Example study Music and Learning RQ Does listening to music while studying hinder or enhance learning Possible experiment R Xmusic O1test score group 1 R no music O1test score group 2 Type of design True posttest only Data M165 M278 For this data music hinders learning Factorial Designs Purpose to examine the effects of two or more I V39s simultaneously quotFactorsquot are IV39s Each factorIV has at least two levels conditions Lecture Page 1 Example o Music factor musicno music AND Caffeine factor caffeineno caffeine M39 39l5 39 N n E E Caffeine 39 39 Each box IS called a cell Caffeme Caffeme You need participants No caffeine Music no No music no and data for each cell caffeine caffeine DV learning test score Data eventually entered into a cell A 2x2 design 2 levels of 2 different variables Add more boxes if there are more conditions Factorial designs test for O 0 Main effects Interaction effects Lecture Page 2 Factorial Designs continued Factorial designs test for Main Effects Interaction effects Main Effect The effect of IV individually on the DV Ex a 2x2 experiment on the effects of listening to music and caffeine on learning To test for main effects compare the quotmarginal meansquot of DV for each factorDV Music No music Caffeine M5O M6O gtM55 No Caffeine M 70 M 60 gt M 65 M 60 M 60 lt A Marginal means Main effect for caffeine greater leaning without caffeine than with no main effect for music no difference in marginal means Interaction effect The unique effect of the combination of IV39s The effect of one IV depends on the levels of the other IV39s Ex for a music x caffeine interaction Caffeine reduces earning only when combined with listening to music without music there is no effect To test for an interaction effect graph the cell means 90 so 70 393 No caffeine 60 gta 50 Caffeine I music no music There is an interaction effect if the lines aren39t parallel So although caffeine lowered scores overall the effect was worse when combined with music Music actively improved scores when without caffeine A perfectly crossed interaction each variable has equal and opposite effects on each other Sample mean Music No Music Caffeine M8O M6O No Caffeine M 50 M 30 A word about factors IV39s In one design can have as IV39sfactors o Manipulated variables Ex music exposure caffeine 0 Subject variables Ex gender personality traits TV usehilo Can only make causal conclusions about manipulated variables If no manipulated variables then it39s not an experiment it39s a survey with factorial type setup Lecture Page 1 For this example factorial design is a true experiment because a Manipulation of IV b Random assignment If one of these conditions doesn39t exist it39s Survey Quasi Experiments Not true experiments because there39s no random assignment but have decent quotcomparison groups Nonequivalent control group design pretest posttest with quasi equivaent groups 01 X 02 group 1 O1 O2 group2 Use pretest scores to quotmatchquot groups before manipulation Time Series designs 0 Track many observations over time before and after a manipulation quotSingle group interrupted time series designquot 01 O2 O3 04 X 05 O6 O7 08 group 1 lmproves upon the one group pretest posttest Example crime prevention program If there39s a change between 04 and 05 what else could be going on Was the trend already decreasingincreasing It39s only valid if there39s been a big change Solves some threats to internal validity 0 Variation take treatment away and measure again Lecture Page 2 Experimental Research cont Thursday May 29 Z 14 Eamp3 AM Laboratory vs Field Experiments Laboratory Experiments Bring subjects into a highly controlled environment High control gt high internal validity Artificial setting gt low external validity Must watch more carefully for experimenter amp reactivity effects o Experimenter problem with researcher39s qualities 0 Reactivity subjects answer to being studied Field experiments Don39t confuse with field research Manipulate V39s in the 3939real worldquot o Example quotShe Said Noquot TV Study I How do people react to media manipulation I Boomerang effect Media agenda backfires More natural setting gt higher external validity Less reactivity Still has validity even without representative sampling because of reality factor But harder to maintain experimental control Content Analysis Quantitative Systematically quantitatively examining the content of communication Used to o Describe how muchwhat kind of certain messages there are eg sex on TV types of tweets 0 Assess quotimagequot of particular groups in media eg stereotypes of race gender political parties age religion etc 0 Compare media content to 3939real worldquot o Examine message changes over time 0 Provide background for research on media effects also a method for codinganalyzing open ended data in surveysexperiments Important Issues Sampling o Define population of interest Ex Primetime shows FB discussions 0 Identify quotunit of analysisquot for coding Ex for TV shows code each episode Each scene Character For FB pages coding each entry Code each thread 0 Select representative sample ideally Coding transforming content into numerical categories 0 Conceptualize categories I Manifest content visible surface content I Latent content underlying meaning 0 Operationalize categories o Establish reliability Limitations Purely descriptive Lecture Page 1 0 Cannot explain why the content is that way 0 Cannot conclude anything about effects of the messages Very reductionistic 0 Reduces complex subject into key points 0 Reduces content to codeable concepts only Lecture Page 2 Qualitative Research Tuesday June 3 ZQJ14 Eamp3 AM Qualitative types of Message Analysis contrast with content analysis Subjectively analyze comm messages Researchers able to give scholarly educated opinion Ex media content conversations Rhetorical criticism 0 Critique form content imagery delivery of speechespop culture Ex use of metaphors in presidential speech themes in students drinking stories Goal greater understanding andor appreciation Critical theory aka cultural studies 0 Goal socialpolitical awareness and change Important issues Typically purposive types of sampling case studies common Construction of detailed field notes and records Finished when achieve quotsaturationquot more data will not add new data Qualitativefield interviewing Unstructured of semi structured 0 Open ended questions free to change Getting depth is key Focus groups Groups discuss an issue in presence of moderator Again open ended questions o Leader should facilitate not control Popular technique in marketing and political research The quottrustworthinessquot of data Qualitative research is NOT concerned with 0 Reliability amp validity of measurement 0 Internal and external validity Instead focus on quality of researcher39s interpretations 0 Should be credible trackable we reasoned 0 Good to quottrianguatequot qual methods ie participant observations with depth interview Lecture Page 1 Research Ethics Thursday June D5 Z 14 Eamp3 AM Treatment of Subjects federal government has ethics guidelines Must submit research proposal for approval from university39s Institutional Review Board 0 At UCSB Human Subjects Committee 0 Conducting research without approval gets you fired Must complete Human Subjects Training Guidelines for Using Human Subjects Participation must be completely voluntary Must obtain informed consent o Explain to participants I Purpose and procedures I Possible risks and discomforts I Ability to withdraw from study Should protect subjects from harm 0 Should not diminish self worth or cause stress anxiety or embarassment Should preserve right to privacy Should avoid deception both types 0 Outright deception deliberately providing false information 0 Concealment withholding key information Any deception must be justified by compelling scientific concerns Subjects must be adequately debriefed Treatment of Data and Research Reporting Ethical handling of data o Do not manipulate data 0 Keep all data Reporting findings 0 Support proper quotpeer reviewquot 0 Avoid conflicts of interest I Funding sources that explain certain findings What happens to research after its conducted Academic publication Practicalapplications 0 Marketing consulting policy Media coverage of hot topics 0 Journalists and consumers often lack understanding The quotFile Drawerquot Lecture Page 1
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'