. afiaefi .- .1 I (3: :hiio: J.“ . 0.1131713 . .nve Lsih ha .. .3 I :5" :1 . t 4h. 13...... 4.3. I, .2 ‘ 1.1.4. £531)... .4 .r 3: 3.7.1 l. *1}!!! 11...} . 1...... . I, 1...). v... 1. .1114... 3;... :13? Z a! 1,... .. 1.. ssh» x pa. #:535an .5 .,.:~»q.. Wufifiwfix .x .‘Q If; .23., 'lo‘.’ .233»... % DI) ’4 fl! r. 3 . lit!- » 13...; l. 331! sillxs If)? .xt! hue...» filfly |~ ,4...1IVO?~$:2|S!. 3.53.... '5‘"?! 1r; 1.51:}... X 1 4h... r main . Id, (.0. I II . 5’3 , L . , , .10 . ‘ .. .Yvn .9 . .‘A. .t-Hctt I . ,d‘gmfi.“ 53.4.13; ‘ ta. , @545? 1?“? £3 23. .irr..un.b.1; .. I! ll- 1.....- 1?” LIBRARY W Michigan State University This is to certify that the dissertation entitled The Role of Attributions on Voter Response to Political Advertising presented by Karen M. Lancendorfer has been accepted towards fulfillment of the requirements for the Ph.D. degree in Mass Media 75cm m5 046 (at/CL Major Professor’ 5 Signature :Decarvtbar 5,, 3005 Date MSU is an Affirmative Action/Equal Opportunity Institution .i- _W_—--—......—--- —-.-.—1- - — n- PLACE IN RETURN BOX to remove this checkout from your record. To AVOID FINES return on or before date due. MAY BE RECALLED with earlier due date if requested. DATE DUE DATE DUE DATE DUE 2/05 p:/ClRC/DateDue.indd-pt1 THE ROLE OF ATTRIBUTIONS ON VOTER RESPONSE TO POLITICAL ADVERTISING By Karen M. Lancendorfer A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Advertising 2005 00L adV Win ABSTRACT THE ROLE OF ATTRIBUTIONS ON VOTER RESPONSE TO POLITICAL ADVERTISING By Karen M. Lancendorfer Political advertising has become an indispensable campaign medium as a way of reaching voters. Over the last twenty years, political advertising has surpassed news stories and other traditional political sources as the most important source of voting information, according to the majority of voters. Its rolehasbeenincraasinglycriticaltoebctionoutcomes, asparty-based campaigns have been transformed into media—based ones. Political advertisements, byfocusingon certain issuasand ignoring others, workto set the public agenda for the campaign; and advertisements help candidates promoteparticularimpressionsofthermelves and alterthedynamicsofelections. With the outcome ofvoting decisions having an enduring effect on public policies andebaedoffidalsmtembofadverfishginmaelectomlsystunisbecaning increasingly important Bearing this in mind, the research presented in this dissertation addressesoneofthamajorand longstanding issuesinpolitical communication research; namely, haw campaign advertising influences voter attitudes and behaviors. Thepurposeofthisdiaaartation istodeterminewhetherattributiontheory couldbeuaedtoaxplainthepmceubywhichvotersexposedtopolitical advertising rnessagesform attitudestoward candidatesforoflica. With this pmwymmmmdmmmmmmdmrmlmm attitudes, the current study proposed and tested a structural equation model with specific hypotheses in order to examine the role of both intrinsic and extrinsic attributions on voters' attitudes and voting intentions. Additionally, individual difference factors (gender and political party affiliation) were considered to see if they presented a moderating effect on voters’ attitudes and intentions. The resultsoftheresearch suggestthatattributiontheorycanbeumdto evaluate voter responses to positive and negative political advertising, and in particular that intrinsic attributions of candidate motive directly affect voters” evaluations of the sponsoring candidate, while mediating the effects of political advertisingonvobrattitudesandbehaviors. Further, with ragardstothe moderating variables, findings suggest a moderating effect of gender in that, only formles, posifiwadvertisenmtsweremorelikelyhan negativeadvertisements to generate extrinsic attributions. Considering political party affiliation, when votem’offlneopposingpaflygenemteinbinsicaflfibufiommwardmespmsonng candidate, voter attitudes become significantly unfavorable. However, attitude towardthecandidatsisnotsignificanfiyaffedadwheneiflnroemocmtsor Republicans generate extrinsic attributions. Givanthatanunderstandingoftheprocessingofpersuasivecorrtentin political advertising messages can provide important insights that will help resaarcharstoexplainwhypoliticaladvertisinghascertaineffects, this dissertation has significant implications for the further development of attribution theoryresearch. lmpIicationsforpoIiIicalcandidabsandcampaignmanagers, along with W for future research are presented. This dissertation is dedicated to my family for their loving support and encouragement. iv ACKNOWLEDGEMENTS This study would not have been possible without the guidance and assistance of numerous individuals. First and foremost I must thank my advisor, Dr. Bonnie B. Reece, for her unflagging support. She has been a source of insight and strength throughout my graduate studies at Michigan State University. I was able to enjoy and survive my graduate training through her encouragement and patience. A very special thanks also goes to Dr. Keith Adler, who was my adviser during my masters program. His expertise, questions, and kind critiques gave me direction. I would also like to express my appreciation to the members of my committee. The encouragement and insightful comments provided by Drs. Sandi Smith, Nora Rifon, and Teresa Mastin were crucial to the final dissertation. I must thank my friends, Byoungkwan Lee and Jungsun Ahn, who provided me with constant encouragement and assistance, without which this work could not have been completed. Thanks also to my wonderful family and children for their emotional support throughout the long, long process. I will also never find words enough to expressthegratitudethatlowetomyhusband, Jim. Eachandeverytimel struggled, he found the magic words that helped me to finish. Thank you again to everyone. TABLE OF CONTENTS LIST OF TABLES ......................................................................... viii LIST OF FIGURES ....................................................................... ix CHAPTER ONE: INTRODUCTION ................................................... 1 CHAPTER 1W0: LITERATURE REVIEW .......................................... 7 Political Advertising ............................................................... 8 The Effects of Political Advertising ........................................... 10 Negative Political Advertising .................................................. 14 The Demobilization Hypothesis ............................................... 20 A Review of Political Advertising .............................................. 24 CHAPTER THREE: CONCEPTUAL FRAMEWORK .............................. 28 The Foundations of Attribution Theory ...................................... 29 Review of Attribution Literature ................................................ 36 Application of Attribution Theory to Political Advertising ................ 43 Theoretical Concepts and Hypotheses ...................................... 44 Formation of Attributions of Candidate Motive from Candidate Advertising .................................................. 45 Negative versus Positive Candidate Advertising ................. 47 Political Cynicism ......................................................... 50 Relationship between Candidate Attributions and Outcome Variables ....................................................... 51 Individual Difference Variables as Moderators: Gender and Political Affiliation ................................................... 53 Proposed Model ................................................................... 54 CHAPTER FOUR: METHOD ............................................................ 56 Participants and Experimental Design ....................................... 56 Focus Group Promdure ......................................................... 58 Stimulus Materials ................................................................ 61 Administration ...................................................................... 63 Measures ............................................................................ 63 CHAPTER FIVE: RESULTS ............................................................ 67 General Sample Characteristics .............................................. 67 Scale Constuction and Reliability ............................................ 69 Hypotheses Testing ............................................................... 70 Hypothesis One ........................................................... 70 Hypothesis Two ........................................................... 74 Hypothesis Three ......................................................... 77 Hypothesis Four .......................................................... 78 Hypothesis Five ........................................................... 79 Hypothesis Six ............................................................ 79 Hypothesis Seven ........................................................ 86 Fit of the Structural Model ...................................................... 91 CHAPTER SIX: DISCUSSION, LIMITATIONS, FUTURE RESEARCH, AND CONCLUSIONS ................................................... 94 Overview ............................................................................. 94 Discussion and Implications .................................................... 95 Fit of the Model ..................................................................... 103 Limitations ........................................................................... 104 Suggestions for Future Research ............................................. 105 Conclusion .......................................................................... 106 APPENDICES Appendix A: Transcripts .......................................................... 109 Appendix B: Instrument .......................................................... 112 LIST OF REFERENCES .................................................................. 121 vii Table 1. Table 2. Table 3. Table 4. Table 5. LIST OF TABLES Intrinsic and Extrinsic Attributions of Motives .......................... 60 Demographic Characteristics of the Sample ........................... 68 Scale Descriptives for Independent and Dependent Variables ........................................................................ 70 Scale and Items of Intrinsic and Extrinsic Attributions ............... 73 One Sample T-tests for Intrinsic and Extrinsic Attributions ......... 74 viii Figure 1. Figure 2. Figure 2a. Figure 2b. Figure 3a. Figure 3b. Figure 4a. Figure 4b. Figure 5a. Figure 5b. LIST OF FIGURES Conceptual Model ........................................................ 55 Proposed Structural Model ............................................. 76 Bush-Sponsoring Structural Model ................................... 76 Kerry-Sponsoring Structural Model ................................... 77 Gender-Based Multiple Comparison Analysis for Bush- Sponsoring Structural Model ............................................ 83 Gender-Based Multiple Comparison Analysis for Kerry- Sponsoring Structural Model ............................................ 86 Political Party-Based Multiple Comparison Analysis for Bush-Sponsoring Structural Model .................................... 87 Political Party-Based Multiple Comparison Analysis for Keny-Sponson’ng Structural Model .................................... 90 Revised BushSponsoring Structural Model ......................... 93 Revised KerrySponsoring Structural Model ......................... 93 ix CHAPTER ONE INTRODUCTION Political advertising has become an indispensable campaign medium as a way of reaching voters. Over the last twenty years, political advertising has surpassed news and other traditional political sources as the most important source of voting information, according to the majority of voters (Media Studies Center 2000). Its role has been increasingly critical to election outcomes, as party-based campaigns have been transformed into media-based ones (Ansolabehere and Iyengar 1995). Political ads, by focusing on certain issues, and ignoring others, work to set the public agenda for the campaign (Atkin and Heald 1976; Bowers 1973), and ads help candidates promote particular impressions of themselves and alter the dynamics of elections (Kern 1989; Sabato 1981). With the outcome of voting decisions having an enduring effect on public policies and elected officials, the role of advertising in the electoral system is becoming increasingly important. Bearing this in mind, the research presented in this dissertation addresses one of the major and long-standing issues in political communication research; namely how campaign advertising influences voter attitudes and behaviors. “Political advertising is now the major means by which candidates for the presidency communicate their messages to voters,” states Dr. Kathleen Hall Jamieson, Dean of the Annenberg School for Communication at the University of Pennsylvania and Dimctor of the Annenberg Public Policy Center. “As a conduit dtm audit mos Camp; nmma EEWS (Crew of this advertising, television attracts both more candidate dollars and more audience attention than radio or print,” with the spot ad being the most used and most viewed of all forms of political advertising (Crawford 2004). This increased useoftelevision as atooIforconveying candidate ideas and perspectives to voters has led researchers to explore the impact campaign advertisements have onbofliindividualcampaignsandformepolificalsystemasawhole(Kingand McConnell 2003). The 2004 Presidential Race Theoostofcandidatas'commercialsnawconsumesagreatarportionof campaign budgets than ever. The Center for Mia and Public Affairs, a nonpartisaniesearchandeducationalorganwation, reponthatthecostof televised political advertisements has more than quadrupled since 1982 (Crawford 2004). At $1.2 billion, the 2004 Presidential Election was the most expensive in history (Harper 2004). In the first quarter of that year alone, President George W. Bush's campaign smnt $15.3 million, and Sen. John Kerry's spent $7 million according to Nielsen Monitor—Plus, a unit of Nielsen Media Research (Whitman 2004). By the time Election Day rolled around, spending reached $345 and $310 millionforBushand Kerry, respectively, acoordingtofiguresreleased bythe District of Columbia-based Center for Responsive Politics (2004). Furtherenalysis revealadthat, accordingtoCampaign MediaAnaIysis Group,anonpartisan organization, atleast70peroentofadsninbyGeorgeW. Bush (lite: ran a to pc on K histc prim; 6X30, and: exam Esues Bush were critical of Kerry, while only 25 percent of ads run by John Kerry directly criticized President Bush (www.pbs.orq). Although the Bush campaign ran a few positive advertisements early in March, the focus of his campaign was to portray the negative aspects of the Kerry campaign, with particular emphasis on Kerry's ‘flip flopping” with regards to taxes and defense spending. Political historians state that, while it is unusual for an incumbent president to run a primarily negative campaign, it is not unprecedented in certain situations. For example, Lyndon Johnson’s 1964 campaign used attack advertising against Barry Goldwater to divert attention from the president’s problems with Vietnam and civil rights legislation (“The Living Room Candidate' 2004). Starting with the primaries, John Kerry’s campaign messages were largely positive in tone, focused on the candidate’s biography, and emphasized domestic issuessuchasjobsand health care. “Thestrategywastodiawattentionto issuesconsideradfavorabtetothekmocraticcandidate, andto introduce Kerry to a voting public that has already formed strong opinions about President Bush” (“The Living Room Candidate’ 2004). However, as the campaign progressed, and ass resultofstiongattacksonthepartoftl'ie Bush campaign, Kerry's ads becainemuchmasaggiesshreintone,freqmnflyamaddngPresideMBushon the economy and Iraq (“The Living Room Candidate” 2004). ArecenttwistwiflmtheZOMelactionisduetothenechCain—Feingotd law-which set rules on raising and spending campaign funds—resulting in the Pmidential candidates running astetementorappeaiing intheirown adstosay “Iapprovedthismessage'. Theideabahindthisrulingmstl'iatcandidates would ta attacks c Ségcrfica C marketir (Norwitz candidat has cons Ddifical l Mikel a, ad‘iemsml would take responsibility for what they claim in their ads, and engage in fewer attacks or “mud-slinging" against their competitor (www.pbs.org). S'gnificance of Current Research Consumer behavior researchers have long been interested in both the marketing (Homer and Batra 1994; Newman and Sheth 1985) and the evaluation (Morwitz and Pluzinski 1996; Simmons, Bickart, and Lynch 1993) of political candidates. As political advertising has grown, a substantial body of research hasccnsideredtheeffectsofpositiveand negativepoliticaladvertising inthe political process. However, little research has been undertaken to examine voteis’ cognitive responses to political advertisements, and, in particular, the concept of voter attributions of candidate motives in the political arena. Moreover, although political advertising's impact on campaign dynamics has been a much discussed and even overly discussed topic, relatively few empirical studies have beancmdudedmanpaiedtostiasonoflterpolificalcempaignmediasuchas televisionnewsand newspapers. Whileafewstudieshavefocusedsolelyon politicaladveitising, nnsbumystudbsexaminedtheeffectofpolitical advertising in conjunction with other traditional campaign media, having aside theuniquecharacteristicsofadvertisingasacampaign medium. Theempiricalquedonthenisraised, howdoprospectivevoterslookat candidatesinpolitimladvertising? Specifically, howdoprospectivevoters Moorsuasiveadvertisingmessagesfliatrelatetopolifical candidates? Do Mmmmmmmmr In addition,doesthe mes atfitu purses explai polruca researc examine the lurth 5" the roi message processing by voters and the attributions endorsed influence the voters’ attitudes toward the candidate, and ultimately their voting intentions? These questions will provide the focus for the remainder of this dissertation. With these questions in mind and in order to address the previously noted research gap, the purpose ofthis study is to determine whether attribution theory can be used to explain the process by which voters exposed to political advertising messages form attitudes toward candidates for office. Given that an understanding ofthe processing of persuasive content in political advertising messages can provide important insights that will help researchers to explain why political advertising has certain affects, an examination of the role of attribution theory might have a significant implication in the further development of attribution theory research. With this primary interest intheroleofattributions inthefon'nationofpolitical advertising attitudesthe cunentshrdyproposesandtestsastructuralequationmodelwiflmsmcific hypotheses in order to examine the role of both intrinsic and extrinsic attributions onvoters'attitudesandvoting intentions. Throughthismethod,tl'iecunentstudy might substantially contribute not only to the theoretical accumulation of attribufiontheoryliterahrrebrfialsotomoreeffecfivedesignofpolifical advertising. Thus, the study has relevance for academic researchers, political candidams, and campaign managers. This manuscript begins with a review and discussion of political advertising,withemphasisontheaffectsofpolitimladvertising in ChmterTwo. Cmmmmmmmmmm providesthe general ooni reminder 0' conceptual r Chapter F on the hypothes presented in results, pres general conceptual framework for studying this specific type of advertising. The remainder of Chapter Three is devoted to the presentation of hypotheses and a conceptual model based on the literature from Chapters Two and Three. In Chapter Four, details of an experimnt designed and implementd to investigate the hypotheses and model are discussed. Results of the experiment are presented in Chapter Five. Finally, Chapter Six provides a discussion of the result, present limitations ofthe current study, and proposesfuture march. advertising. sections. Se advertising a Research in dietenl per: advertising (I CHAPTER TWO LITERATURE REVIEW The purpose of this chapter is to review the recent research on political advertising. For ease of understanding, this chapter has been divided into five sections. Section one present a short conceptual background regarding political advertising and servesasan introductiontothegeneral asmctofthistopic. Research in political advertising in recent years has been examined from many different perspectives, from research examining effect of female candidate advertising (Hitchon and Chang 1995; Hitchon, Chang, and Harris 1997), media coverageofpoliticaladvertising(Benwart, Bystrom, and Roberton2003; Lariscy et al. 2004), and political advertising in other countries (Chang 2003; O’Cass 2002), to content analyses exploring the differences and similarities among various kinds of political advertisemnt (Benoit, Pier, and Blaney 1997; Johnstonand Kaid2002). However, thecurrentfocus, in lightofthe Pfeponderanoe of the research, is on the effects of general candidate political advfiftisingasitoccursinpresidentialelectionirand nationalissuecampaigns. In addition, because Faber (1992) provided a comprehensive review of political advOffising,thecurrerrtexaminationspecificallyfocusesonresearchthathas been conducted since that time. Insecfiontwomeeffectofpolificaladverfisingingenemlaieconsideied, withanemphasisonnegativepoliticaladvertising researchaddiessedinsection "tree. Secfionfourconsidarsaspecificwbsetofnegativepolifimladverfising, Pei:- V‘vt Unte adve throU. ad‘veg VOiEII Ifnlng from t Voters d9081i namely, those studies based on the “demobilization” hypothesis. Chapter Two concludes with an assessment of the relevant research, a discussion of pertinent research in politics that has considered “attributions,” and a delineation of important gaps in the literature that may be addressed by the cunent research. PoliticalAdvertisng Sincetheappearanceoftelevisedpolificaladvertisinginflte19503,mass media advertising has emerged as the dominant form of communication in the UnitdStatesbetween political candidates and voters(Kaid 2004). Political advertisinghasbeendeemedtohavebecomesoirnportantthatsorne through political advertising (Pinkleton 1992). One mason may be that political adverfidnghasmebenefitofprovidingcandidateswimameflrodofreaching votersfliatisunmediatedbyflrepressmecausetheadverfisingmessageand fimingremainunderthecontrolofmacandidateorpany(Penoff2002). However, frOtnanonnativepointofview,politicaladvertisementaredesignedtoinform vofianabouttheissuesinthecampaign,inorderforvoteistomakereasoned decisions(Poiioir2002). Manyconcephializafionsanddefinifiutofpolibcaladverfisinghavearisen Sinoethefirstreviewofresearchonpoliticaladvertising in 1981. Atthattime, Wadverfisingwcsdefinedas,‘meconmunicafionpmcessesbywhicha “"709 (usually a political candidate or party) purchases the opportunity to exDOUGreceiverstiiroughmasschannelstopoliticalmessageswiththaintended effect of influencing their political attitudes, beliefs, and/or behaviors” (Kaid 1981, p. 250). In an effort to provide a much broader and more modern conwptualization of political advertising, Kaid (1999) suggested that ‘the defining characteristics of modern political advertising arc (1) control ofthe message and (2) use of mass communication channels for message distribution” (p. 423). This interpretation of political advertising requires dissemination of the advertising message through a large variety of mass media channels, while disassociating it from interpersonal communication between voters and candidate political Speeches (Kaid 2004). Benoit, Pbr, and Blaney(1997) reviewed priorrewarch in political commercial adwrtising that had focused on comparisons of issue versus image cds,andpositiveveisusnegativeads, inanefforttodefinethefunctionsof P0litical commercials. As a result of their analysis, the authors proposed a tYI’Ologyofactivitiesthatarlethefunctionsofpolitical advertising: ‘acclaiming (arguing that they have desirable aocornplishment and trait), attacking (pointing toobjectionableactionsandcharacteristicsoioppononts), and defending (mponding to attacks from opponent)“ (Benoit, Pier, and Blaney 1997; p. 16). Theaufliotapplbdflleirtypdogymananalystofpmsidenfialpolificalthvtion commercials from 1980-1998. Result from an analysis of 206 campaign Wtshowedfllatattadtsfocusedmoreonpolicyissuasfllandid Wins, suggesting perhaps that candidates wanted to minim-c the amealtimeofmud—slinging byavoiding attackingtheiropponent‘scharacter. Political advertising has arisen as the focus of modern contemporary campaigns and has been identified by political consultant as influencing the agendas for news, debates, and interpersonal discussions (Perloff 2002). This focus and importance of political advertising has contributd to the plethora of research conceming the particular effect of political advertising messages. The Effects of Political Advertising Overthe wars, politicalcandidats have increasingly relied upon advertising to reach and influence voters; and, while advertising enables candidatestopursuemulfipleobjecfivesmteowraugoalofmepolificianis alwaysto influence voterdecisions(Ansolabehereand Iyengar 1994). Wrththe 50mcnnwersaryi120040ffltefittuseofpdificaladverfisingontlevtionin Amenca,debatesaboutmeeffectofpolitical advertising stillabound,with mceMnntanalysesidenfifyingoverwmcamhstudbsinvoMngflleinpactof political advertising on voterparticipation (Allen and Burell 2002; Lauetal. 1999). Miiaitiscamnonlyacceptdmatpolificaladverfisingrepresentadirect attemptbypolificianstopresentmeircampaignmessagesandtopackage candidatstvotersfiteresultareoonflflngcstotheintendedeflector unintended consequencesofpoliticaladvertising. Overthepasttwodecades abne,resecrdiethavecomideredfltembandinfluenceofpolificalparfiesin election campaigns, aswell asthe impactofcandidate ads during campaigns. Amwrtextensivereviewofpofificaladverfisingreceamhexaminedhree disfindstgesofdevcloprncrnsinoemehcepfionofpofificaladverbsing 10 research i wnsldere through ii in its mo have bee uncovere research lon'ng be Willem ilvolverr may Ont} lr ads tn Candida MCCILlre adverts and lea: and Sci (Faber, 1990; G Wilma: l 30130” research in the 19503, from the pre-advertising stage where advertising was considered as a form of mass communication with little effect on voting intentions, through the advent of televised political commercials and submquent analyses, to the most recent decade where specific individual level and situational variables have been considered (Faber 1992). Overall, Faber’s (1992) research uncoveredanumberofvariablesthathavebeenused inthepastdecadesof research including: exposure, awareness, knowledge, candidate preference, and voflngbehafiamiflwafiousstdiesindicafingtatcanmerdathadflweabilflyto influence awareness, knowledge, and voter choice. This is similar to the high involvementhierarchyofeffectsmodelfound ingeneralproductadvertising, and may only apply to an election where involvement was high. ln particular, prior research identified by Faber (1992) showed that political adsmnpmparemanyvoterstomakevofingdecisionsbyinfonningfliemabout candidate and their issue positions (Johnson-Canes and Copeland 1991; McClure and Patterson 1974; Patterson and McClure 1974). Negative advertis‘mg ismostinfluentialonthosevoterswhosupportthecourcecandidate, and law influential on independent and low involvementvoters (Faber, Tlms, and Schmitt1990; Merritt 1984). Variables such as demographics, involvement (Faber, Tlrns, and Schmitt 1990), and partisanship (Faber, Time, and Schmitt 1990; Ganamone 1985; Merritt 1984) are important considerations in explaining political advertising effect. Wrthwtevidencethatpolificaladverlisinghaseffectsonvoters, little additional researchwould havem conducted overtheyearssinoe Faber's 11 (1992) re candidaii the mark' spots tr postal 6 Va candidate Elevision 2032. We Wine! 1 Suggests 1 resulted fl ex.Dosure Pel "58d sum V0150 385 5905 can lute efiec and 'YEllg; lhao and I ”"908an (1992) review. Such evidence is not hard to find, and it support the decisions of candidates who spend millions on campaigns that they “are not completely off the mark” (Kaid 2004, p. 166). Perloff (1998, p. 374) states that “clearly, political spot can am voters' evaluations of candidates and their interpretations of political event.” Various studies haveproposedthatpaid advertising isa betterpredictorof candidate recognition and recall of candidate issue knowledge and salience than televisionnewsornawspapers(8riansandWattengberg 1996; Holbertetal. 2002; West 1994). However, other studies suggest that television news may sornelirnesbeabetterpredictorofvoterknowledgalevels(Chaffee, Shao, and Leshner 1994; Weaver and Drew 2001; Zhao and Chaffee 1995). Kaid (2004) WM likeallresearch inmediaeffect, contradictoryflndingsmayhave resulted from differences in rmasurement, particular concerning variables of exposure and attention. Perhapsmorerelevanttothecurrentstudy, a largebodyofresearch has used survey and experimental methods in a continuing examination of how voters assess candidates. Their findings confirmed that exposure to political spotscanaflectandidatei'nageevaluations. Politicaladvertisingappearstobe quite effective at promoting issue based evaluations ofcandidates (Ansolabehere and Iyengar 1995; Briana and Wmmerg1996;Pfau et al. 2002; West 1994; Zhaoand Chaffee 1995).Votersexposedtopoliticaladvertisingwillvoteasthe advertising message advocates (Bowen 1994; Goldstein and Freedman 2000), 12 pamculari and Mia: Of ! (2002). K wages the 2000 | pamwlar Dolmcal or 0% N Gor indicated militate adveriislr tongue“ negative I seqion. particularly among those voters who are the least aware (Valentino, Hutchings, and Vtrllliam 2004) or late deciders (Bowen 1994). Of particular note are two recent studies by Kaid (2002) and Tedesco (2002). Kaid (2002) experimentally tested channel effects for political advertising messages though a comparison of Internet versus traditional media channels in the 2000 presidential Mon. Channel effect were definitively noticed, in particular with undecided voters, in that undecided voters who were exposed to political commercials via the lntemet subsequently indicated an intention to vote forAlGore,whileundecidedvoterswhosawfliesamecommerciatontelevtion indicatedan intntiontovoteforGeorgeW. Bush. Tedesco (2002) examined political advertising effects on candidate image evaluations, emotions, and cynicism during the 2000 Robb-Albn senatorial election in Vlrginia. He used a perception analyzer in order to track participant’ second-by-second reactionstothecommercials. Ashypothesized, simplybeing exposed to an advertising message increased positive evaluations for both candidate, although cynicism did not influence evaluations of the candidates. In recentyearstherehasarisena particularfocuson negativepolitical advertisingeffects, resultinginasubstantialbodyofresearchfllatspeciflcally considers the effect on candidate images and voting behavior from exposure to nagaflvepolitimladvertisement. Thismsearchisoonsideredinthefollowing section. 13 advertiser frequently landldate Highilghel over idem Mng o Communl upswing i Campaigr Cartee ar EleCllOlt y 0039 m0 Negative Negative Pom l Advertising The most distinctive feature of contemporary political campaign advertisement is the negativity of their content and tone. Political advertisers frequently engage in so-called negative advertising in which the opposing candidate’s program and performance are criticized and even ridiculed. Highlighting the opponent’s liabilities and weaknesses usually takes precedence over identifying the sponsor's program and strengths. In the most comprehensive trackingofcampaignadvertisingtodate, scholarsattheAnnenberg Schoolof Communication havefoundthatsuch"negative" advertising hasbeen onthe upswinginrecentyears, and nowmakesupapproxirnatlyona—thirdofall campaign ads used in presidential campaigns (Jamieson et al. 1998; Johnson- Cartee and Copeland 1991; Kaid 1994). Millions ofdotlars are spenteach ebcfionyearinfliemarkefingofpoliflcalcandidates,wifliagmatpercentgeof thosemoniesbeingspentonnegativeadvertising, becauseofthebeliefthat negative information is more influential than positive information (Johnson-Cartee and Copeland 1991; Lau 1985; Pinkleton 1997). This has resulted in the majority ofrecentresearchdebatingthepersuasivenessofmgafiveads. Researchers haveammptdtopmvidetypobgiesofnegafiveadverfising(KaidandJotmston 1991; Koltz 1998), experimentally assessed the effects of negative advertising (Garramoneetal. 1990; Kahn and Geer1994; Thorson, Christ. andCaywood 1991), and surveyed voters concerning negative political advertising (Faber, Tlms, and Schmitt 1993; Weaver-Lariscy and kaham 1996). 14 Wit and lyengz voters (Fri lsoontinu 1902. Tlnl negat've a likelihood Negatve i mainly be Diocese: 1982, 192 Niel a it al. 199 1991; Ne 305 are 5 evaluate M While such advertising is consistently disliked by voters (Ansolabehere and Iyengar 1995; Hill 1989) and thought to alienate large numbers of potential voters (Freedland 1994; Rothenberg 1990), it effectiveness can be inferred by its continuing and increasing use at every level of political campaigns (Jamieson 1992; Tinkham and Weaver-Lariscy 1997). In it simplest form, the purpose of "egafiveadverfisingisto‘aeatealessfavombleimageofanddecreasethe likelihood of voting for the targeted candidate” (King and McConnell 2003, p. 844). NeQativeadvertisingcontantisthoughttobemorepersuasivethanpositiveads mainly because researchers have contended that negative content is noticed and Processed moreMplysothatitexertsmoreofan impact(l(emall1977; Lau 1982. 1935). Research within this stream notes that expmure to negative Political advertising resultin higher levels ofvoterrecallthan positiveads(Basil et al- 1 991; Johnson-Canes and Copeland 1989; Kahn and Kenney 2000; Lang 1991; NeMiagenandReeves1991),andfliatvoterswhomcallnegativepolitical “'8 are subsequently more likely to use knowledge acquired from these ads in "a'Uating candidates (Briana and Wattenberg 1996). Meta-analyses are split, however, concerning the relative influence and prooeasingorpositiveandnegativeadvertising,withtheconclusionthattrereis 3““ a needtorfurtherresearch. Lauetal. (1999)concludethat"thereissimply meVidencahflicresamehlitraMrematnegafiveadverfisancntareanymore effective than positive ads” (p. 857). Allen and Burrell (2002), on the other hand, Wuaematnegauvemromtaampmducasalargereriectonopinionromahon 15 As a recent meta—analysislreview examined much of the research from the 19908 and earlier, this section will review research concerning negative political advertising that has been published since the Lau et al. (1999) article. Although thefoousthenisontheextensiveresearchfromthepastflveyears, afew additional articles that were not included in the Lau et al. (1999) meta-analysis have been included here in anefforttoprovidethebroadedexaminationoftln tOpic- Abroadrangeofresearch hasarisan inreoentyearsindicatingthat negative political advertising can influence candidate attitudes and voting behavior. Negative political advertising is thought to lower voter evaluations of ttiraeteid candidates (Budesheim, Houston and DePaola 1996; Faber, Time and Schmitt 1993; Jasperson and Fan 2002; Pinkleton 1997, 1999), as well as to affect voting preferences (Ansolabehere and lyengar1995). Faber, Tms, and Schmitt (1993) continued their work exploring the ”We between involvement and voting, with results from the 1988 Minnreset-Senatoracaindicatingthathigherlamlsofinvolvemantresult in Water effectsfornegativeadsonvoterdecisions. Eligiblevoterswho resDer-dadtoatalephonesurvaypracadingthealectionwereaskedirthayhad ”an eachoffourtelevised negative ads,twosponsoredbytheincumbent, and twosponaoredbythachallenger,andwhetherseeingeachadmadetlemmore "less likelytovoteforthesponsoroftheadandthetargetofthead. ”Waspmposedfllatvoterinvowemninpofificsarldauenfimtnews W potiucswould moderatetheimpactofthenegativeads. Result indicated 16 that gene poi-tit c advertisi 8 negative different Ssue an then eva M poli Diocese r9500M 303% itven d .4 that general interest in politics, interest in the current campaign, and attention to polities on television news were all associated with a stronger impact of negative advertising. Budesheirn, Houston, and DePaola (1996) utilized different types of negative campaign advertising in order to assess the persuasiveness of the different messages, including issue based, character based, and combination issue and character attack advertising. Respondent were asked to read and than evaluatspeechesbycandidatestl'latbothopposedandsuppontl'leir 0W" Political ideology. Contrary to the hypotheses, subjects systematically Processed all types ofcandidate advertising, with candidates who shared the We political ideology being held to a higher standard, because their Mmenlypersuasive iftheywerewelljustifiad. In 2002, the Journal ofAdvertising presented a special issue on political ad"Brtisingwitharr numberofarticlesthatareparticularlyrelevanttothecurrent ”Search. To begin, Jasperson and Fan (2002), examined the dual effects (Mandunil'ltendedlbacklash)ofnegativepolitical advertising inareal "Md campaign scenario by examining actual candidate commercial buy data in "der totrackmediaplacementofpolitical commercialsfromJanuaryto "Ovemoer 2000, and subsequent shift in candidate favorability with voters. Results indicatasfliatmeetfectofnegativeinfonnationwasappmximatelyfour ““93 greater flian positive information when evaluating favorability of candidates With We, butsorneevidenceofbacklashonthesponsoringcandidatewas also found. 17 post We ad mm 932 ion em! W 3N ‘30! Big ads Pinkleton, Um, and Austin (2002) experimentally assessed the effects of positive, negative, and negative comparative political pn'nt advertisements in order to determine the effects of the advertising messages on key variables of negativism, cynicism, efficacy, and apathy. Following exposure to the advertisements, a sample of 246 undergraduates listed the droughts they had about each candidate in a thought listing procedure, as well as completing post— test scales. Contrary to the hypotheses, although participants found negative adverfisinglessusefulflianpocifiveadverfisingandwemmorenegafivetoward the campaigns, therewas noeffectofthe negative political advertising on Participanh’ cynicism. efficacy. or apathy— Meirick (2002) compared comparative and negative political advertisings from thezooo Minnesota congressional racebetween Kline and Lutherwiththe 909' ofidentifying differences in responsesbetweenthetwotypesofpolitical Mug. Sixty undergraduate students viewed the commercials, which were embedded within a track with other consumer product commercials. Following the Viewing, participantsprovidedtheirtl'loughts regardingthecommercials, as we" as answered questions related to the measurement or candidate favorability 1|"til Voting intention (while controlling for political affiliation). Overall, corI‘inarativeadsprovokedtewersouroederogations, promptedmoresupport “mummmmmmmMmmwmmnegm ads. MJheusoofnegafivepolificeladverfisingcenalsocleebbaddash “WW unsponsorottheadvertisingmage, resultinginthesponsorsbeing 18 subject to negative responses themselves (Pinkleton 1998). Much of the research in past years has found evidence of a backlash or boomerang effect with candidates who sponsor negative ads being subject to negative responses themselves (Faber, Tims, and Schmitt 1990; Garramone 1984; Roddy and Garramone 1988). A recent study found that repeated exposures of negative adverfisingmessagesbadtoincreasinglynegafivemsponsesamongwomenas emosure increased (King and McConnell 2003). As Garramone (1984) cautions: “Negative political advertising may achieve its intended effects, but it may also produce boomerang effects. A strong attack on a candidate, if perceived by the audience as untruthful, undocumented, or in any way unjustified, may create more negative feelings toward the sponsor, rather than the target. Similarly, an attack perceived as unjustified mysneratemorepositivefeelingstowardthetargetflp. However, whiletheraisabundantevidencethatnegativepotitical advertising can influence candidate attitudes, there are a few experimental Miss indicatingthatpositiveadsaremoreerrectivethannegativeor co'flmarative ads in shaping attitudes toward candidates (Houston, Doan, and Roskerwoldsen 1999; Kahn and ceer1994; Shen and Wu 2002). in particular, Houston, Doan, and Roskos-Ewoldsen (1999), in a similar Woeduretothetussdbyeudesheimetal. (1996), hadsubjectsreadsix adVertisementsthatweredescribedascomingfromtwowndidates(onealiberal, the't-‘rths-sraconservrrrtive)irrau.s. Senatecampaign,witheachcandidate WUCting eidler a positive campaign or a negative campaign. Consistent with the"‘Wlmthesas(whichwerebasedonapproach-approachandavoidance- amidancecontlicts),bothmndidatesreceivedrelativelyhighevaiuationswhen 19 they each r2 when they r candidates more realist Inthe 'demotiiza 399’9989 Wt ’93! candid they each ran a positive campaign, and they received relatively low evaluations when they each ran a negative mmpaign. Unfortunately, no tests where run with candidates running both a positive and negative campaign in order to examine a more realistic campaign scenario. The Demobilizationmlflgis Observerswhodecrytheriseofnegativeadvertising usuallyworryabout itse‘l‘fectonthepolitical process asawhole.Thequestionofwhether negative POIiticaladverfisingdoesinfacthannflledanoaaficpmcesshasbeendebated i"recentyears,withthemajorityotresearchrocusedonwhethernegative adverusng reduces votertumout. Unqusstionably, the possibilitythat negative advertising sets up a spiral ofcynicism that drives people awayfrom polities is Muttoresearcl'lersinthefield(Perloff2002). However,theresultsofrecent Mrchareequivocal. inureirdassicartds,Ansolabehereetal.(1994)hrstrocmedteirsights 0" theeffectsofnegativecampaignadvertisingonvotertumout(the “Weston-Wammasetotexpenmentalsmdiesandmenan aggregate analysis ofresults oithe 1992 v.3. Senateelections. Results of "’8' Gandidates and themes, indicated thatexposure to negative advertising MMWdWWWMSWTMMfism rep'icateoinananalysisofmeadverfisingmneofnewspaperarfidesmeadiof “348mm1992u88enateseatswerecontested.flledependent 20 varabies eiect indi vote in th Resuits a percent a decrease pmsible astudy t Hamel t Mont Kennel (. a meta-a MEI des variables were actual voter turnout and ballot roll-off, “... a campaign-specific effect indicating the degree to which people who were sufficiently motivated to vote in the presidential election chose to abstain in the Senate race” (p. 833). Results again revealed that negative mmpaigns reduced voter turnout by 4 percent and increased roll-off by 1.2 percent. The authors suggested that a decreaseinpoliticalefficacyassociatedwithviewingnegativeadsisone POSSible mechanism by which those ads may affect turnout Ansolabehere and lyengar (1996) followed up their initial experiment with a Sttidy that combined of experimental data from California campaigns and nationalelectiondata. Theyagain concludedthatnagativeadvertisingreduces election turnout by approximately 4.5 percent This result was most noticeable a"'Ong non-partisans. Ruearch rollowing Ansolabehere’s work by Kahn and Kenney (1999) also found that negative advertising suppresses turnout. Finally, a "mamas by Allen and Burrell (2002) revealed a slight diminishing ofthe Mar desire to vote as a result of negative political advertising. Kehnand Kenney(1999)analyzedarandornsampleofcitizensinstates ”"1 U-s. SenateeMonsin 1990 using theAmerican National Electionsmdy. P°°|ed malaction.Ashypottrosizedbytlreautrrors, controllingrorvarious “More nomrallyassociatedwithturnout, nogativepolitical ads stimulated interest “the canpaignsandsubsequenttumout. However, campaignsthatinvolved a 9°06 deal of mud-slinging (as judged by the campaign managers) decreased “but Thscsmpaignelrectswerestrongeston political independents, novices, 21 and those least interested in polities. The authors suggested that voters can discriminate between legitimate and unjust attacks, and they respond accordingly. Few other studies, however, have found support for the conclusion that reduced voter turnout results from negative advertising exposure (Freedman and Goldstein 1999; Wattenberg and Brians 1999). In particular, three recent longitudinal studies (Finkel and Geer 1998; Geer and Lau 1998; Vavreck 2000) further dispute the Ansolabehere hypothesis of a demobilization effect. Freedman and Goldstein (1999) introduced two methodological innovations to the study of political advertisements: (1) an ”ad detector" teel'lnologythat analyzes satellite transmissions to uniquely identity all lid"e“:iwsements broadcast in a particular media market and yields an accurate r”finding ofhowoften, when, andwtlerediflerentadswereaired, and (2) anew set 0f surveyitemsasking respondentsabouttheirtypicaltelevisionviewing habits todsvisea highly individual measureoflikelyexposuretodifferent campaign ads. Thiscombinationofmethodswasappliedtoarandom sarnpleof witheredvotersmute1997wginiagubematorialelecuon,toshowmat Wretonegafivepollficeladverfisinghadastrongmobilizingefiecton turnout.evenarnongpoliticalindependents.WattenbergandBrians(1999) “My contested the Ansolabehere at al. (1994) findings and disputed the Mlizabflflyofflls‘demobflizaflon'effectoutsideofenexperinontal setting. They used National Electionsmdydatatoshowthattherewasnodemobilization m from exposure to negative advertising. 22 The research by Finkle and Geer (1998) is important to the debate over whether negative political advertisements demobilize the electorate because it presents a set of theoretical arguments explaining why negative ads ought to stimulate, rather than demobilize, voters. In support of negative advertising, Finkel and Geer (1999) hypothesized that negative advertising can lead to greaterknowledgeorthecandidates and higherlevelsofcaringaboutthe outcome of the election. A detailed coding of every political advertisement aired during the U.S. presidential ebctions of 1960 through 1992, combined with an “U358 of both aggregate and survey data from these election years, revealed "0 relationshbbetweenwgreeofnegativityinthecanpaign advertising and tur“curt. in particular, and in direct contrast to the results of Kahn and Kenney (1999). therawerenodirrerentialsrrectsotnegativeadson independents, nor among those voters exposed to high amounts of mass media. In the same year, G°°r andLau(1999)againusedacombinationoraggregatedatarrom presidenfialelectionsand National ElectionStudysurveysanddeterminedthat “‘9 amountornsgativsadvertisingwas, intact, associatedwithgreater, notless, voter turnout Similarly, Vavreck (2000) used NES data from 1976 to 1996, with ”suit: indicafingfliatnegaflvepolificeladvenisingdidnotappearmbwlovels °f interest in the campaign, attention to the campaign, nor participation in voting. Overall, the demobilization hypothesis reasons that voters who are exPOSGdtonegativepoliticaladwrlisingmaybecomecynicaltoward politicians “M “‘0 polficelprooeuJeelthatmeydonothaveinputintothemnningofdle ”unity. and ultimately decreasing their voting. 23 are pror l Dottie “negat'l concep too bra multidlr implied We 01 risearc the" at exam ofiler: meSam Mid A Review of Political Advertisi Given the clear ambiguity in the discussed research of political advertising, it seems sensible to look at the research findings in more detail, in order to discern possible explanations for conflicting conclusions. Three such reasons are proposed: 1. One explanation for the conflicting findings concerning negative political advertising effects may stem from the researchers’ conceptualizations of “negative” political advertising. Richardson (2001) argues that the conceptualizations of negative advertising in the academic literature are entirely too broad. Somersmrchconsiderednegativeadvertisingtobea multidimensional construct, consisting of direct attack, direct comparison, and implied comparison appeals (Johnson-Cartee and Copeland 1991; Perloff 2002), while others defined it simply as “attack advertising“ or “mudslinging'. 2. Asecorldrationalefortl'leconflictingresultsmaybethat rescarchershave notoontrolledforthe responses based on differentsectors of tl'leiraudience, mostinportantly, politicalpartyaffiliation. Withsomenoted exceptions (Faber, Time and Schmitt 1990; Meirick 2002; Tedesco 2002), much Oflhe researchintlnpafldecedehasnotcontrdled resultsforpartisanship. 3. Lastly, there are often problems in measurement of variables mmmmmsmhmmchmaypurponmbeemining the samething, actualmeasurementofthevariablesmaybediflerent, resulting 5" mdifierentrosults. Thisismostnotableinrnessurementofcsndidate 24 evaluation it variable. Ace platelet theoretic; relied upl From all lol the 9' Cognitive atlbutlc Millie evaluation with a number of different scales being used to assess this critical variable. Accordingly, the current study was motivated to revisit some of the previous findings about political advertising for both methodological and theoretical reasons. On the methodological side, much of the prior research has relied upon student samples, which may limit the generalizability of the results. F rorn a theoretical standpoint, research has considered a number of rationales for fl'cemctsofpoliticaladvertising, andwhiletheyhavejustbeguntotouchon cognitive responses and “attributions” in political advertising, the application of attribution theory (as discussed in Chapter Three) presents a relatively untapped avenue of research as applied tothe field ofpolitical advertising. As a media effects outcome, election ‘interpretations' of “attributions” about politicalmessagec representa shift inthestandard modelofmediael'fects Presented by Lanrsfcld, Bewison and Gaudet (1948), which traditionally focuses on the direct impact of messages on attitudes, beliefs, opinions, and behaviors (Hall and Cappella 2002). Although the» studies do not explicitly consider political “advertisingftheirfocuson media messages is relevant in that “Militias” are considered ill each context. For example, Iyengar (1990) exl>e|'il'l‘lentallyexaminedthe't'npactofthenewsmedia by assessing W’ attributions, orcausalexplcnations, ofthecausasofpoverty. 'yega'(1991)followed his 1990mmmmasetotexperimentsthatdiscovered t"attl‘swavintrlhichnriwswasrrarnedintiuencedviewers'peroeptionsor Warm, social, and economicconditions. Cappellaand 25 researcr 3min“ tie'ssu hm v01 Weiude Willa! Llnbau: mam their un Umbeu oi”Com Me In 5% ll WM 5 Jamieson (1997) investigated coverage of political candidates and policy issues in the media. Similar relationships were explored in Iyengar and Kinder’s (1987) work on news coverage of the president, and in Kinder and Sander's (1990) research conceming the public’s perceptions of affinnative action. In each case, attn'butions of responsibility or causality influenwd respondents’ understanding of the issues. Only a few studies have begun to consider “attributions” as a rationale for how voters evaluate candidates and political outcomes, and they provide a prelude to the following section. Hall and Cappella (2002) investigated audience attributions of the 1996 presidentialebction asa resultofexposuretotheflamesthey received in political tak radio. In particular, over a nine-month period, listeners ofthe Rush Limbaugh radio program, listeners to other political talk radio, consumers of mainstreamnewsmedia, andnon-consumersofnewsmediawereqmliedasto theirunderstandingsofthe causesofthe election results. Results indicated that Limbaugh mdiofistermweremorelikelytodiscountsubstantiveelection outcorneathibutionstinndidollerrespondents, inthatattributionsof resporlsitiflityforfllemofdleelecfionbyoobwueatuibutedtofactors WIN addressed during Lirnbaugh’s broadcasts. Schenck-l-lanlin, Procter, and Rurnsey (2000) explored how framing of issuesinnegativepoliticaladwrtising influencedtheextenttowhichtl'le messageifiumcedmepiblic'satnibufionsofmsponsibflflyforpmblunsilflle politicelsystem. Three hundred sbrtyundergraduatestudentswerepresented 26 ”with USJ'igE aim vase tame resm V0305 alibi 970634 with a variety of negative political advertisements and were asked to respond, using a thought listing procedure, to the following question, ‘When you hear or read about America’s probbms, what do you think are the most important muses of those problems?“ As hypothesized, results indicated that political advertising was able to influence attributions made by the respondents, in that different framesemphasized intheadvertising messagesledtoattributionsof responsibility for problems in the political system. Rudolphand Grant(2002)testedamathematicalmodelofeconomic voting in which attributions of responsibility for the economy and vote choice in the2000 presidentialelectionwereanalyzed. Specifically,theauthorswantedto know, 'Towhom did theArnerican electorate attribute responsibilityforthe nation’s economy and what impact. if any, did these respondbility judgments have on presidential vote choice?’ (p. 806). Results indicated that, as hypothesized, attributionsofresponsibilityforthestabilityofthecountrydid inflmnce subsequent voting behavior. Asdiscussed above, llnitadacademic research cunentlyexists inwhich attribution theory is applied conceptually to political advertising. In addition, the pmblansnotedwiulpnormseamhinpofificaladverfisingpmvideanincenfiveto continuethework. Accordingly, the purposeofthisstudyistoextend research in poiflmladwfishgbyapplyingaflibufiondnowhorderbfllusfleteflieefiects ofattributional processingonvoterattitudesand voting intentions, while addressingnotadgapaintheliterature. Thefollowingchapteroutlinashaw attribution theory can be mpliad to examine political advertising. 27 Swot theoryai GWEN r andNeu 31?}wa CHAPTER THREE CONCEPTUAL FRAMEWORK This chapter details the conceptual framework of attribution theory for the study of political advertising. In the first section the foundations of attribution theory are reviewed in order to provide a strong theoretical background for the current research. Research that has utilized an attributional framework in recent years has been examined from many different perspectives, both psychological and marketing based, and has included research examining the effects of merging ofthird person perceptions and attribution theory (Hoffneret al. 2001; RucinskiandSalmon 1990), aswellastl'leapplicationofattributionsin management, team, and sales scenarios (LePine and Van Dyne 2001; Taggar andNeubert2001, 2004). However,thefocushereisontheeffedsof attributions on respondents” attitudes towards communication messages and, more specifically, their responses to advertising messages. Because Folkes (1988) provided a comprehensive review on attribution remrchinthereaknofmarketing andconsumerbahavior,thecunent examination specifically considers research that has been conducted since that time. Therefore, sediontwoprovidesa leviewofthe relevantliterature, which includesdiscussionsofcurrentattributional research, the lntersectionof comparafivecdverfisklgandathibufioneandfimllyflienndefingofamibufion theoryasapplledtocause—related marketing. Sectionthreediscussesthe applicwiliworattributiontneorytopolitical advertising. Theohapterconcludes 28 mm pnord how th severe 90005 mate: Togethi Ruskin. Wane: mean Man thtier with the proposal of hypotheses and a structural model that is consistent with the prior discussion of political advertising and the cunent review of attribution theory. The Foundations of Attribution Theory Attribution theory is concerned with how individuals interpret events and how these interpretations relate to their thinking and behavior. It is, in actuality, several theories that share core assumptions. Heider (1958) was the first to prom a psychological theory of attribution, by suggesting that people are like amateur scientists, trying to understand other people's behavior by piecing together information until they arrive at a masonable explanation or cause. Building on the work of Heider, Jones and Davis (1965), Kelley (1973) and Weiner (1992) developed a theoretical framework that has become a major research paradigm of social psychology. Consumer behavior research suggests that attribution theory provides a valuable framework for predicting behavior. A synopsisofthemaincomponentsofthistteoryloolcssomething likethis(eachof these will be discussed in further detail below): . Heider(1958) arguedthatpeopletrytoidentifythecausalproperties that underlie observed behavior and do so by attributing behavior either to external or internal causes. . Jones and Davis (1965) built on Holder’s work andfocused on the conditionsunderwhich peopleobserveanagent'sbehaviorandeither doordonotattributeacausalexplanationtotheagent 29 'nai've . Kelley (1967) theorized in detail about the information processing people engage in when explaining social events. His model describes the rational analysis of patterns of covariation among three elements— a person acting toward a stimulus in particular circumstances—and derives the conditions under which people make attributions to the person or the stimulus. . In studying attributions for achievement outcomes, Weiner and his colleagues (1992) found that people rely notonly on the person— situation dimension of causality but also on the dimensions of stability and controllability, and these three-dimensional causal judgments mediate some of people's emotions and motivations in response to social outcomes. Heider’s “naive M. Heider's perspective is commonly known as "naive psychology“ (Folkes 1988; Kelley and Michaela 1980). Focusing on inwrpersonal relationships, Heider believed maple were "naive psychologists" Mlosoughtcarvnon-senseanswerstounderstandflleworldamundmemand the behaviorofothers (Mizerski et al. 1979; Weiner 1990). Attribution theory is "bamdontheconvictionthatifwecancapmrethenaive understandingsofthe persononthestreet,wecanacculetelyinfer. . . hisotherexpectationsand actions'uones 1985, p. 89). Attrbutiontmorydealswith howthe‘social perceiveruses informationtoaniveatcausalexplanationsforevents'(Fiske and Taylor 1991, p. 23) through an examination ofwhct infonnm isgathsred and how it is combined to form a causal judgment In other words, how a person explains other people’s behavior is based on that individual’s own perceptions. Attribution theory assumes that people try to determine why people do what they do; i.e., attribute causes to behavior. Therefore, assuming that receivers are constantly scanning incoming information for the underlying motivesofthebehaviorofothers, itshould bepossibleforreceiverstoinferthe reason forthatbehaviorbased on characteristics related tothe motives ofthe sender. Laddngdiredknowledgeofthesemotives, observersfeel impelled to inferthesemotivessothattheymaybetterorder, organize, andthus understand their environment (Smith and Hunt 1978). Message recipients are assumed to continually generate expectancies abouttheposifionthata communicatorwill advocateon certain issues, and to believe that a certain aspect ofthe communicators situation or personality is likely to influence the communicators position and message. In simplest mime, Eagly, Wood, and Chaiken (1978) suggest that an individual’s explanations mgardingwhycornmunicatorsadvocateparficularposifionsafiectmessage persuasiveness. An importantconsequenceofthistheoretical proposition isthat inferencesleadtobehavior, i.e.,youwillorwillnotbehaveincertainwaystoward meactorbasedonyourinferencesandyouwiflfonnexpectafionsastohowflie actor will behave. ltisbelievedthatflleaveregepersoniscontinuouslyandspontaneously generating inferencesthatlinkeventsthrough causal relationshipsnhae inferencesarebefiefsMaflowformWrstandingandpmdicfionofflre 31 re. Ac mc hit at: the em. det. Dar 9X16 3cm: drcu 0539: in Na (Hind, observable world. One key element of attribution theory that is particularly relevant here is the distinction beMen internal and external attributions. According to Heider (1958), these two types of factors can shape attributions of motive. lntemal attributions occur when people attribute the causes of actions to internal, controllable, characteristics of the actor (intrinsic motives), while external attributions involve attributing causes of actions to situational factors external to the actor (extrinsic motives). Heider argued that both these personal forces and environmentalfactorsoperateonthe"actor," andthebalanceofthese determines the attribution of responsibility or motive for the ach'ons (Lewis and Daltroy 1990). This process holdstruewhethertheattributoris observing his or herown actionsorthoseofothers inthatindividualstendtoattributetheirown actionsto external factors and the actions of others to internal characteristics. In hot, peopleintheU.S. aresopronetoplacing responsibilityonthecharacterof actors rather than on an interaction between character and environmental circumsmnces, socialpsychologists have namedthetendencythe FAE, or “fundamental attribution error” (Ross and Fletcher 1985). In situations where an obsewerobsuvesanundesirabbbehaviorjrndannntalathbufimarormsults hiblarnhgflieaaor(ordievicfim)forflienegafiveconsequencesofmeacfion (Hindman 2003). W. Correspondent lnferenceTheorywasdevelopedbyJonesand Davis(1965)asafurther Wdeiderscausalinferencesinanefionmdesaibeparfiwlar 32 types of am theory. Jon intentions a he behavlc another’s pl may infer 3 correspond Libets (89, Katie 50me0flhe various facfl types of attributions rather than causal inferences in general. According to this theory, Jones and Davis described how an ”alert perceiver" might infer another's intentions and personal dispositions (personality traits, attitudes, etc.) from his or her behavior. Perceivers make correspondent inferences when they infer another‘s personal dispositions directly from behavior. For example, perceivers may infer a disposition of mean-spiritedness from a mean act lnferences are correspondent when the behavior and the disposition can be assigned similar labels (e.g., mean). Kellexs Model of Attribution Dam. Kelley (1967, 1973) has discussed someofthewaysinwhicheffectsproduced byanactionareattributedtothe various factors present in the situation. His addition to attribution theory concerns the subjective experience of attributional validity and asks the question, “how do individuals establish the validity of their own or of another person’s impression of an object?” Under many circumstances, an individual will have access to multiple instances of the same or similar events. With information aboutmulfipleevents,wecanernployacovariation principletoinferthe causes ofevents. Covariationistheobservedco—occunenceoftwomnts, orin other words,weobsarveaneventscovanationwithvanouspotentialcausesand attributetl'leet'fecttothecausewithwhich itmostcloselycovaries. According to Kelley, pcopleusesscovariation inforrnationacmssthreedimensions relevant totheentitywhose behaviorthey aretrying to explain. Consensus, consistency overtimeandmodality, anddistinctivenessinflwncevvhetlerpeopbattributean effecttotheperson, thestimulus, orthesituation. 33 1.C 2. C re 3. Di St; 1. Consensus: Do all or only a few people respond to the stimulus in the same way as the target person? 2. Consistency over time and modality. Does the target person always respond in the same way to this stimulus? 3. Distinctiveness: Does the target person respond in the same way to other stimuli as well? Kelley (1973) argued that the ways in which people make causal attributions depend on the information available to them. When you have much relevant information from several sources, you can detect the covariation of observed behavior and its possible causes. However, in everyday life, we often only have information from a single observation to guide us in making a causal attribution, asconsumersoften lackthenecessarytimeand motivationto make multiple observations. In these single inference situations, configuration principles, asopposedtocovariation pnnciplesflareevoked Thecovariation principle is most applicable for understanding how people learn to make attributionsinflndedinfonnationprocessingsituations, butitisgenerallytoo unmanageable for understanding specific consumer responses (Mizerski, Golden, and Kernan 1979). However, ‘consumersrapidlyleamtoassociatecauseswithevents, andto generalize across similar attribution situations. These generalized causal expadancies and the attributional niles governing the inference procedure are captured in the derivatives of the covariance model referred to as causal schemata or ‘configuration’ concepts” (Mizerski, Golden, and Kernan 1979, p. 128). lnthiscase, iftheonlyinforrnationisasingleoccunenceoftheeventthe obsavermuflfaflbadronoflmshategiesormlesofcausalinferencefiske and Tayi principie Ti how cau MMer intrinsic r Thereforr men the (Caider a “m fat and Taylor 1991). One such method of “configuration” is the discounting principle. The discounting principle (Kelley 1973) represents one type of view about how causes are related. This principle explains that consumers discount or minimize an explanation if an alternative explanation exists, and further that intrinsic motivation is discounted when extrinsic motivation explains an event Therefore, a person is more likely to attribute an intemal motivation to an actor when there are no plausible external, altemative explanations for an action (Calder and Bumkrant 1977). For example, ‘whena productendorserhasextemal reasonstoaccountfor favorable comments about a product, recipients of the communication often believe the product less worthy than when endorsement involves minimal or no external incentives. Thus internal reasons for liking the product are discounted when an alternative reason for endorsement is presented“ (Folkes 1988, p. 553). my s Motivational Research. Weiner (1992) further advanced atuibufionmeorybypmposhgacabgonzafionsdiemefliatclassifiescauseson the basis ofthree dimensions: locus of causality, controllability, and temporal stability. Locusisa referencetowhstherthecauseofactionis‘intemal'or “external" to the actor and closely mirrors Heider’s distinch'on between intrinsic andextrinsicmotivations. Controflability,ontheotherhand, isbasedonwhether an actor'sactionwasincontroloftheactorornot. Anactionisconsidered volifionalorconholhbbifitwaspemehndtobeundertakenasawiflfirldloice, whemas ifanactionwasunavoidableorwasconstrained, then itis likelytobe perwivedasuncontrollable. Temporalstabilityisthethirdcausalinference 35 proposec time oris proposed by Weiner (1992), and it refers to whether a cause remains stable over time or is a temporary phenomenon. In recent years, researchers have applied each of these components of attribution theory in different ways from Weiner’s continued work in satisfaction, to other researchers applying small parts of the overall theoretical base (e.g., utilizing only the discounting principle). Additionally, much of the past research has focused on interpersonal situations. However, recent work in attributions has included organizational communications, and specifically advertising, as an antecedentofattlibutions. Despitethefactthatsomeofthe research reviewed below does not deal smcifically with advertising, nevertheless, an investigation of it reveals implications for admrtising. ReviewofAttributionLiterature AttributiontleorywasfirstextendedtoapromotionalsituafionbySettle andGolden(1974)whohypothesizedthatreadersofadvertisingmessages wouldevokeaflibufionstointemretflwevalidityoffienmagedaims. More specifically consumers were expected to attribute the promotional claims to either meadverfisarswishmsellmepmdudwmommfliemaldiamctensfics of the product (internal). lftherespondentsmade internal attributions i.e., massagedainswereaflrbutedtomeacmalcharactensficsofthepmduct, consumerswouldbemoreconfideMinmeadvertisingclairnsanddevelopmore favorableattitudestowardthebrand. Ontheotherhand,ifthemessageclairns mattrbutedtoflleadverfisersdesiretoseflthapmduct,i.e.,extemal the We attribution: favorable i Usi: (1974) pre Produds tr Promoted ‘ WCh wen relatively L Promoted 1 summer or DFOducts w The b00kie Claims (is. i” the book these treat; to the that “Van “imports HoweVer, proceSSes Tn; i mumtale p attributions, consumers would be less confident in the claims and form less favorable attitudes toward the brand. Using an experimental procedure with business majors, Settle and Golden (1974) presenbd two different versions of print advertisements for five different products to the respondents. One advertisement (non-varied product claims) promoted the product as superior on five preselected characteristics (three of which were thought to be important and two of which were thought to be relatively unimportant). The other advertisement (varied product claims) promoted the product as superior on the three important characteristics but not superior on the two unimportant characteristics. The advertisements forthe five products were then combined to form a booklet and given to subjects to read. The booklet contained different combinations of varied and nonvaried product claims (i.e., there could be from zero to five varied product claim advertisements inthebooklet, and allpossiblecombinationswere used). Following exposureto these treatments, measures were taken of the importance of each product claim totherespondents, andtheirconlidenceineachclaim. Theauthorsooncluded that advertisers should be willing to disclaim ['discount’] superiority on an unimportantcltaraWtoincrmethepemived credibilityofthe source. However, since Settleand Goldendid notverifytheexistenceofattributional processes, this conclusion cannot be confirmed. Tripp, Jensen, and Carlson (1994) used attribution theory to suggest that multipleproductendorsementsresultindiflerencesinconswners’ perceptionsof theendorser. Theauthors utilized in-depthinterviewsinordertodevelop 37 descriptior advertising Results inc celebrity rr attributions attributed t PTOduct en enSierider r i”moment t parTimpani: descriptions of the attributional processes that operate when consumers view advertising messages with a celebrity endorser endorsing multiple products. Results indicated that consumers’ perceptions of liking for and expertise of the celebrity may be tied to the number of products endorsed by the celebrity through attributions of trust. In addition, in the advertising messages, consumers attributed the celebrity's motive for the endorsement to money, and for multiple product endorsements, more money. The money motive did not appear to engender negative attributes toward the endorser. While interesting in itself, it is important to note that the authors conducted the in-depth interviews with only ten participants, so the results may not be applicable to consumers at large. Stern (1994) contrasted classiml TV advertising from vignette (sequential “stories? advertising and proposed that the two have different effects on consumer attribufions. In particular, Stern proposed that vignette advertising relies upon the thme components (disfinctiveness, consistency, and consensus) of Kelley's (1973) covariation principle to induce consumer attributions about a product. Ontheotherhand, classical advertising utilizestheconfiguration principle to influence consumers to make attributions of products in that individuals assign causality on the basis of a single observation. This is importanttothecunent research becausetheconsumers inthe cunent research are notemosedtovignetteadvertising, butrathertoclassical commercial messages. ltisexpected thattheconfiguration (ratherthan the covariation) principles will be in effect. 38 03‘ between c response advertiser ethical atti indicated 1 Significant advertising DUl’Chase i Davis (1994) utilized attribution theory to explore the relationships between consumers’ attributions of advertiser motivations and consumer response using two types of attributions that relate to the perceived ethics of the advertiser (advertiser ethical attributions) and the advertising message (message ethical attributions) in environmental product advertising. A national mail survey indicated the existence of both advertiser and message ethiml attributions. A significant impact of both types of attributions was found on measures of advertising response (attitude toward the ad, attitude toward the brand, and purchase intent). Lee (2004) has recently added to the attribution research with her analysis of attributions of responsibility in crisis communication messages. Individuals from Hong Kong responded to hypothetical crisis communication messages describing a plane crash. Utilizing an internal (the crisis is perceived to be within the boundaries of the company) versus external (the crisis is perceived to be outside the realm of the company) locus of responsibility in the communication message, respondents were asked to attribute responsibility for the crisis event. As hypothesized, attribution of internal responsibility for the crisis resulted in a negative impression toward the organization, while external attributions of responsibility resulted in a degree of sympathy and trust in the company. As with some of the additional research discussed here, this study shows the applicability of. attribution theory, although original formatted for an interpersonal context, to an organizational context 39 An oomparat did not to discountir discount 5 the adver guilt Carr Ming tat adlimiter Persuasio attn'bution. 9etleral, d noncompE La An additional stream of research has considered the intersection of comparative advertising and attributions. Although each study discussed here did not formally utilize attribution theory, they all employed the concept of the discounting principle. Coulter and Pinto (1995) reported that message recipients discount a guilt-provoking ad tactic when they associate manipulative intent with the advertiser, particularly when the adverb'ser attempts to instill high levels of guilt. Campbell (1995) similariy showed that, when an advertiser uses attention- getting tactics, such asbrand name delay, consumers doubtthemotivesofthe advertiserand areledtoperceivetheadvertiserasmanipulative,whichaffects persuasion negatively. Jain, Buchanan, and Maheswaran (2000) used a sinilar attributional framework specific to comparative advertising and reported that, in garteral, directcomparativeadvertisementsarecounterargued rnorethan are noncomparative advertisements. Lastly, Jainand Posavac(2004) reportedtheresults offourstudies (2 lab studies and 2 field studies) with each study examining the mediational role of consunerattributionsotthaadvemmonestand objectivaversus unfairand biased)onconsumerattitudeatowardpositiveand negativecomparative advertising. Results indicatedthattheet'iectivenmofcomparativeadvertising is indeed mediated byconsumerattributionsdiouttheadvertiserandthatnegative attributions result from negative comparisons. Amoentstreamofresearchassassingattributionsofmotivesincause— relatedmarkefingcampaigmisapplicabletomecunernmaeemhasmodelsof conswneratbibufionsofmotiveshavebeanpmposedandhavelaidfliegmund 40 work for the current research. In the initial study of this type, Webb and Mohr (1998) conducted in-depth interviews in order to explore how consumers think and bel about muse-related marketing. Using basic distinctions between extrinsic and intrinsic motivations, the authors coded respondents' thoughts about the firms motives into one of four categories: “rewards sought for the firm itself, rewards sought mostlyforthefirrn butpartlyforothers (such asthe community or the environment), rewards sought mostly for others but partly for the firm, and rewards sought solely for others.” (p. 231). Almost half of the respondents indicated that they believed that companies engaged in cause- related marketing for serfish reasons, while the other half believed that companies had mixed motives (both self-interest and altruism). From this analysis, Webb and Mohrdeveloped atypologyofconsumers (identified as skeptics, balancers, attribution-oriented, and socially concerned) based on attributions consumers make in mlation to cause-related marketing. Dean (2002) built upon Webb and Mohr’s work by experimentally assesshgconsumerattributionsofthemotivationsforsponsorships and how these attributions affect corporate community relations. On the basis of attribution theory, Dean (2002) proposed that consumers would employ negative attributions of company self-interest to explain the sponsorship activity, and that ashucturalequafionmodelwouldindicataapafltfiomflrenagafiveatbibufions to the outcome variable of pemived corporate community relations. Results indicatedhatconsumemformboflrposifiveandnegaflveatflbufionsabmn 41 corporate motivations for sponsorships, and that the respondents’ views of these motives do mediate their perceived views of the company's community relations. In order to increase external validity for his 2002 study, Dean (2003) assessed consumer perceptions of consumer donations in a cause-related marketing scenario. He used a 3 (types of company: scrupulous, average, or irresponsible in social matters) x 2 (type of donation: conditional or not conditional upon corporate revenue) factorial design to examine corporate donations on the dependent variables of consumer regard for the company, perceived mercenary intent of the company, and whether the social performance ofthecompanyisperceived as‘good” management. Resultssuggestedthat "(1)finns with a starting reputation for social responsibility have little to gain by engaging in a single episode of charitable donation; (2) firms with a reputation for social irresponsibility may significantly increase their favor with consumers by engaging in a single episode of charitable donation; (3) firms with an average reputation for social responsibility are perceived differently by consumers depending upon which type of donation the company pursues; and (4) a single charitable donation will not raise the imageofan irresponsiblecompanytothatofa scrupulous company.” (p. 101). Rifonetal. (2004) utilized attributiontheorytodevelopandtaMa structural equation model of sponsorship effects that builds upon the prior research by Dean (2002). The authors “evaluated the effects ofthe congruence betweena sponsorand camandtheuseofcommnyversus brand narneson consunleratuibufionsofcorporatemofiveformesupponofahealdrcauseand resulfing consumer perceptions of the sponsor” (p. 39). Results of the Wauggestedfliatconsumeraflribufionsofammsficsponsormofives 42 can result from a good fit between a company and the cause it sponsors, and ultimately enhances sponsor credibility and attitude toward the sponsor. Application of Attribution Theog to Political Advertising Folkes (1988) suggests that “because many commercials require viewers to make inferences about characters’ intentions and goals, this sort of analysis [attributional] should provide guidelines for understanding advertising emcts” (p. 559). Therefore, consumers’ attributions as to why a communicator takes a particular position in a message is important in determining whether a consumer accepts or rejects the message (Gotlieb and Sarel 1991). In this regard, attributiontheorywould suggestthatthe recipientsofa political advertising message would seek to explain the undertying motives of the candidate in the advertisement Here,headvertisingrrmagerepresentsanobserved behavior, and consumers may attribute certain motivations to the actor (Smith and Hunt 1978). Attributionsofmotivemaybeafunctionofpastexperiencesand individual characteristics, but they may also be a function of the characteristics of the advertising strategy and message (Rifon et al. 2004). While it is unlikely that consumers will have specific knowledge of a candidate’s motives, voters are Iikelyto understandthatcandidates produceadvertising in orderto influence voting behavior. HallandCappeIIa(2002)shessfl1eimportanceofsmdyingatUibutional interpretations because they may disclose relationships and consequences that would notbeobvioustothereseareinrifattihrdesmtheorflymeasures. 43 “Failing to consider interpretations can bad one to overlook an element of audience members’ mental representations of an event that shapes their behavior or their evaluation of related targets” (p. 335). Certainly, analysis of the general principles of how people interpret motivations and actions of actors that are suggested by attribution theory offers important basic insights on consumers' interpretations of persuasion-related material, such as advertising (Friestad and Wright 1994). Further, Mizerski (1978) notes that causal attributions in much of thepastresearich haveseldom been measured directly, andthatattribution measures have not allowed “for an examination of the number of perceived causes, northe allocation ofattribution among causes” (p. 221). It isthe goal of thecrmentresearchtoaddressthesegapsinthefield. Theoretical Conm and mm Thepurposeofthecurrentstudyistoextend research in political advertising by applying attribution theory in order to illustrate the effects of attributionalprocessingonvoterattitudesand voting intentionsthrough a controlled experiment Although most analyses of political attitudes and behavior involvesurveyresearch,contmltedexperimentsaresometineschosenin political advertising as a more precise way to measure political advertising muse (Ansolabehere and lwngar 1995; Pfau et al. 2001; Pfau at al. 2002; Valentino, Hutchings, and Williams 2004). An experimental design was chosen infliecuneMsmdybecauseitenabbdmereseardiertomanbmmesfimulus materialmfinrflnanrelyingonparficipants’memoryaboutthepolifical commercials they may have seen. Despite the great deal of research in political advertising, however, there is still the question of exactly how, how much, and under what conditions political advertising matters. It is believed that attribution theory will shed light on the effects of political advertising as a whole and begin to answer those unanswered questions. Noting Jamieson’s (2000) call for “scholarship that sts light on the nuances of both media messages and voter reactions” (p. 17), this study addresses both discourse in the form of advertising and voter decision making in the 2004 presidential election. Formation of Attributions of Candidate Motive from Candidate Advertising Attribution theory proposes a cognitive process through which individuals might assign an underlying cause or explanation to an observed event (Kelley 1973), such as a political advertisement A prominent assumption in attribution theory is that individuals regularly engage in attributional activities, based on the definition espoused by Heider(1976) that links attribution with spontaneous cognizing of the environment (Harvey and Weary 1984). Davis (1994) utilizes flieeXplanafionomefl(1981)todesuiJehowindividualsmovefiemobse~ing an event to attribution formation: 'An individual: (1) is exposed to, comprehends, and encodes a set of stimuli (such as overt behaviors, language, etc). These stimuli are labeled the anmdent event. 45 (2) constructs or infers a tentative set of attributions which are felt to be the most probable explanation for the reasons or motivations underlying or wusing the observed stimuli. (3) evaluates the tentative attributions in light of additional information, observations, or past knowledge. (4) modifies or adopts the attributions. (5) stores the attribution in memory.” (p. 874) The above process results in the formation of attributions that may be applied to political advertising and, in the cunent research, to the development of attributions regarding political candidates. Therefore, it is proposed that by applying this general process to the formation of attributions, attributions of candidatemotivemaybefonned in responsetotheadvertising message. Attributionsofcandidatemotivecan bedescribed asavoter’s atternptto detennineflieundenyingmotivafionsofflrecandidateintheadverfisement, inan effort to understand and predict the observable world, and may be either extrinsic orintn'nsic. lnaccordancewithattribufiontheoryitisbelievedthatathibutionsof candidatemotiveswillbemadeasthevotersortsthrough and interprets incomingfifiomlafionandfirferscausalityinordertomakesenseofh’sorher environment. Priorreseareh hasshowntl'latthefcnnationofattributions requires an Observed antecedent event (Kelley 1973). In addition, higher involvement with thSeventorcommunicsrtion, alongwithgreabrlevelsofpereeived importance a"d mlavance to an individual’s life, increases the level of message processing (Celsi and Olson 1988; Chaiken 1980; Petty and Cacioppo 1986) and the formation of attributions (Weiner 1986). As a result, it is believed that attributions of candidate motive are more likely to be formed when the voter regards the advertising as highly involving, important, and/or personally relevant. Furthermore, political advertising is highly prevalent in the days leading up to an election, with exposure on the part of the vobrs likely. Thus, it is hypothesized that: H1: Voters who are exposed to political advertisements will attribute motives (both intrinsic and extrinsic) to the candidate who sponsors the ad. N_egative versus Positive Candid§t_e Advertisigg Political advertising has often been categorized into positive advertising and negative advertising based on whether the candidates advanced their own strengtl'isorcriticizedtheiropponent'sweaknesses(Chang 2003). Areviewof prior research indicates little agreement as to a specific definition for positive and negative ads (Ansolabehere et al.1994;.lohnson—Cam and Copeland 1991; Shapiro and Rieger 1992; Tinkham and Weaver-Lariscy 1993). In this study, Positiveandnegativeadvertisingwasconcepmalized accordingtoarecentstudy by Chang (2003) because this appears to be the most concrete considered advertising that promoted ‘candidates’ issue policies, themby highlighting the candidates’ capability, while negative advertising ms viewed as ’anaanngopponerm'iasuapoiciaammhyramrmgmeopponone' 47 incompetence” (Chang 2003, p. 57). Such advertising is often termed “mudslinging' or “attack advertising” (Pinkleton 1998), but is distinct from that form of advertising which involves character assassination of one candidate against the other candidate’s image. When considering the effects of negative versus positive candidate advertising, we must take into account that prior research has noted that people tend to dislike negative ads, with 75% of respondents in Garramone’s (1984) research and 65% of Johnson-Cartee and Copeland’s (1989) participants expressing disapproval of negative ads. Further, a body of research has indicatedthatnegativeadvertising may resultin a“backlasheffect’againstthe sponsoring candidate (Garramone 1984; Merritt 1984) with voters viewing the sponsoring candidate as mean spirited (Pinkleton 1998). Prior research involving comparisons beMen positive and negative politicaladvenisinghasrevealedthatparticipantsexposedtonegative advertising generated more source candidate derogations and were more negative toward the political candidate in general than were participants exposed to positive political advertising (Hill 1989; Pinkleton, Um, and Austin, 2002). Furthermore, abodyofrssearch inconsumerproductsadvertisinghas consistently indicated that consumers gererate significantly more negative foistedstabmentswlnnexposedtonegativeversusposifiveadsbecsuseof differing perceptions ofthe advertisers’ motivations (Belch 1981; Gom and Waiters; Swinyard 1981; Wibon and Mudderisogle1980). Specifically, consumersexposedtothe negativeadvertising mayfocustheirattentiononthe negative information and may be more likely to see the advertiser (or candidate) as biased or “self-serving.” Lastly, Kahn and Geer’s (1994) experiment assessing the effectiveness of positive and negative advertisements showed that positive advertisements yielded warmer feelings toward the sponsoring candidate than the negative advertisements. Thefindings ofthepriorstudies suggestthat negativeadvertisements should yield attributions of intrinsic candidate motive more than would positive ads (Meirick 2002), because they would be attributing the cause ofthe advertisements to the “mean spiritedness” of the candidate, rather than outside forces contributingtothecornmercialmessages. Therationalebehindthisisthat intrinsic motivations are seen as internal to the candidate and controlled by the candidate, whereas extrinsic motivations are seen as external or situational (i.e., beyond the control of the candidate). Therefore, H2: Voters who are exposed to negative political advertisements will attribute different motives to the candidate who sponsors the ad than voters who are exposed to positive political advertisements. More specifically, H2a: Voters who are exposed to negative political advertisements will be more likely to attribute intrinsic motives to the candidate who sponsors the ad than voters who are exposed to positive political advertisements. H2b: Voters who are exposed to negative political advertisements will be less likely to attribute extrinsic motivestothecandidatewhosponsorstheadthanvoters whoareexposedtopositivepoliticaladvertisements. 49 Polrtical Cynicism Political cynicism is perhaps one of the most frequently explored aspects of political disaffection and is often defined as “a person’s perception that his or her opinion is not important to political leaders, that governmental institutions and political leaders are not trustworthy, or that leaders do not always act in the interests of their constituents” (T edesco 2002, p. 39). Prior research suggests that cynical citizens have given up on the political prowss based on their lack of confidence in and a feeling of distrust toward the political system (Crotty and Jacobson 1980; Dionne 1991; Perloff and Kinsey 1992; Schenck-Hamlin, Procter, and Rumsey 2000). Further, there is increased concern that cynicism contributes to low voter tumout as shown through conelational studies by Answehere, Behr, and Iyengar (1993), and Crotty and Jacobson (1980). According to Scheneck-Hamlin, Procter, and Rumsey (2000), once cynicism was activated, voterswhowereexposedtopolitical advertising regarded politicians as being responsible for the country‘s problems and treated politicians as a whole with greater contempt Whilepriorresearchhasproposedthataspiralofcynicism mayexist in political advertising tends to decrease cynicism (Ansolabehere and Iyengar 1995), lirnitedresearchhasconsideredpoliticalcynicismasanantecedent variable that may impact voter attitudes and attributions. In one exception, Tedesco(2002)reportedthatpre—testcynicism levelsservedasapredictorof post-tadevaluationsofcandidates. ltisconceivablethatasvotersbecome 50 more informed about the “dark side of politicians” they may become more cynical (Bowen, Stamm, and Clark 2000). Political consultants agree and worry that negative advertising increases distrust of politicians (Perloff and Kinsey 1992). Cynicism as a variable is often used to refer to a lack of confidence and a feeling of mistrust toward politicians (Austin and Pinkleton 1995; Dionne 1991; Perloff and Kinsey 1990; Pinkleton, Um, and Austin 2002; Tedesco 2002), which may result in a differentiation between intrinsic and extrinsic attributions. Furthermore, research has shown effects of other individual difference variables, such as existing preferences, partisanship, and general attitudes toward negative ads, on voter responses to political advertising. Therefore, in order to further explore how political cynicism as an individual difference variable may impact voter attitudes, the following hypothesis is proposed: H3: Regardlessofadtype, politicalcynicism isdirectlyrelamd to the type of attributions of candidate motive voters make for the sponsoring candidate. Further, H3a: Political cynrcrsmwrll haveapositiveimpactonthe attributions of intrinsic candidate motives for the sponsoring candidate. H3b: Political cynicism will have a negative impact on the attributions of extrinsic candidate motives for the sponsoring candidate. RM‘ i2 between Qndidate Attributions and Outcome Variables Acantralcomponentofattributiontheoryisthat, onceattributionsare fomied,thereisarelafionshipbetweenmoseattribudonsandsubsequent 51 attitudes and behaviors. This belief that attributions have consequences for attitudes and behaviors is explicated in the three-stage model of attribution theory presented by Kelley and Michaela (1980), in which consequences of the attributions (which include changes in affect and behavior) proceed directly from the attributions themselves. Harvey and Weary (1984) indicate in a review of attribution research that “attribution theorists typically assume, either explicitly or implicitly, that attributions directly influence behavior or mediate the relationship batman other factors and behavior“ (p. 445, emphasis added). Kelley (1973) further articulated the general position that attributions affect related attitudes and behavior, by stating that “(clausal attributions play an important role in providing the impetus to actions and decisions...” (p. 127). Thus, if candidate attributionsarefonnad, wewould axpecttham toaxert an influencaon relevant attitudes and behavior. Therefore, it is hypothesized that, H4a: lntrinsicattributionsofthe sponsoring candidate advertisements will have a negative impact on voter attitude toward the sponsoring candidate. H4b: Extrinsic attributions of the sponsoring candidate advertisements will have a positive impact on voter attitude toward the sponsoring candidate. H4c Attributions of candidate motive (both intrinsic and extrinsic) for the sponsoring candidate will mediate the relationship between exposure to political advertising and voter attitudes toward the sponsoring candidate. Thereisasubstantialbodyofresearchthatspecificallyaddressastl'ra affectsofattitudatoward candidatesonvoting behaviorwith receivarvariables (suchasatfihrdetowammecandidata)phyingamleinvofingbahavbr(aowen 52 1994; Goldstein and Freedman 2000; Kaid 2004; West 1994). In his study of 1992 California Senate races West (1994) identified affects of exposure to ads on voter attitudes and subsequent voter preference. Therefore, it is expected that voter attitudes will affect voter intentions. H5: Voter attitudes toward the sponsoring candidate will be positively related to intentions to vote for that candidate. Individual Difference Variables as Moderators: Gender and Political Affiliation This study also proposes two individual difference variables that may affect the interpretation and evaluation of the political advertisements. These factors are gender and political party affiliation. Each variable is described below and additional hypotheses offered. Priorresearch hasnotedevidenceofa“gendargap’batwean parties, with women more often voting for Democratic candidates (Chaney, Alvarez, and Nagler 1998; Mattel and Mattel 1998) and reacting more strongly to campaign advertising than men (King and McConnell 2003). In a national study by Kaid and l-loltz-Bacha (2000), women were more strongly affected by political broadcasts. Kern and Just (1997) concluded that, Women were more responsive than men to negative attackmessages; in particularthaywere more Iikelyto blamatheauthor ratherthan the objectofthe attack" (p. 111). Considering the effects ofgender, the following hypotheses are offered: H6a: Women will be more likely to attribute intrinsic candidate motivesforthesponsoring candidatethanwill men. Hob: Gendarwill moderate the effect of political advertising on generation of attributions of intrinsic and extrinsic candidate motive for the sponsoring candidate. 53 Partisanship and political party affiliation have often been mentioned as variables impacting voter decisions. In previous studies, the persuasiveness of political ads varied depending on viewars’ political party affiliation (Pfau at al. 2001), with the direction of change in candidate vote choice as a result of exposure to advertising massages highly related to partisanship (Faber, Tims, and Schmitt 1990; Merritt 1984; Robidaaux 2002). Voters’ evaluations of ads were also impacted by party affiliation (Robidaaux 1998, 2002). Merritt (1984) observed that those voters who identified with a targeted candidate’s political party were more likely to evaluate the sponsoring candidate lower. Considering the possible effect of political party affiliation, the following hypothesis is proposed: H7: Political affiliation will moderate the effects of attributions of candidate motive for the sponsoring candidate on voter attitudes and voter intentions. Hm Model In light of the literature on political advertising that has been reviewed here and the hypotheses that have been developed, attribution theory will serve asafoundationforproposing andtestingacausalmodelofthaeffectsof attributions of candidate motive on voter attitudes and voting intentions (the dependent variables). Figure 1 depicts the relationships among the variables. In sum, this model posits direct relationships between political advertising, political cynicism, and intrinsic and extrinsic attributions of candidate motive, and indimct relationsh'ps between political advertising and theirsubsequenteffectson voter reactions to the advertisements. In addition, this effect is dependent on gender and political affiliation (moderators). Fg‘ ure 1. Comtual Model On a conceptual level, a study of attributions can extend our understanding of the types of inferences consumers make about political advertising messages and the relationships between these inferences and attitude and behavior change. From a practical standpoint, research indicating that voters do evaluate candidates based upon their advertising messages, and that these evaluations affect voting attitudes and intentions, provides candidates with a strong motive for carefully considering the content and tone of their advertising messages. Although previous research has confirmed the effect of political advertising on voters” evaluations and intentions, it is believed that attributiontheorycan helpustounderstandtheprocess better. 55 CHAPTER FOUR METHOD The purpose of this chapter is to outline the method used to investigate the study's hypotheses and structural model. Specifically, this chapter will cover the experimental participants and procedures, a pre-experiment focus group, stimulus materials, and independent and dependent variables of interest. Participants and E_xpgn_'mental Desg’ n External validity of an experiment is realized if its findings are readily generalized to the population at large. In general, external validity can be improved when naturally occurring rather than artificial stimuli are used, when the situation is representative ratherthan atypical ofthe range ofsituations ofthat type, and byensuringthatthe sample ofparticipants is representative ofthe population of interest. Typically, group research employing randomization will initially possess higher external validity than will studies (e.g., case studies and single-subject experimental research) that do not use random selection/assignment (Campbell and Stanley 1963). In orderto increase the external validity of this experiment, registered voters were recruited to serve as participants. Additionally, full-time undergraduate students were specifically excluded in anefforttotapthepotentialofafinalsamplethathad avested Winflrspresidenfialebcfionand,flierefom,maybemomfikelymacfively promos the advertising messages. Registered voters were recruited from community groups throughout Michigan (church groups, work groups, parents involved in scout groups, school PTAs, etc.) using a technique in which contact was made with each group leader or administrator in order to recruit during group meetings or through the group listserv. A target sample of 300 participants was desirable for the experiment, so a pool of volunteers in excess of this number was recruited. Through a combination of personal contact at group meetings and introductory e-mails, 510 registered votersprovided e-rnail addressestothe researcherwith the understanding that they would be contacted in the later months of the presidential election and asked for their evaluation of political advertising messages. An e-mail experiment and data collection procedure have the benefit of combining “the most advantageous features of postal communications, such as eliminating synchronous interaction and interviewer effects, with the most advantageous features of telephone communications, such as the ability to experimentally manipulate questions and secure rapid response time' (Best and Krueger 2002, p. 73). Because these features can often be obtained with little personnelandexpense, itisanattractivealmmativetopostalortelephone communications. In the past, e-mail surveys have been used successfully to investigate topics such as electronic democracy (Fisher, Margolis, and Resnick 1998), government elections (Taylor et al. 2001), and news sources for political invomm (Althaus and Tewksbury 2000). In the current experiment, e-mail 57 transmission facilitated the transmission of text, graphics, and video, so that e- mail users were able to assess campaign commercials. In the week immediately preceding the presidential election in November, 2004, the potential participant pool of 510 registered voters was transmitted an e- mail message encouraging recipients to participate in the study and providing a hyperlink to the actual survey written in Hypertext Markup Language (htrnl) accessible by most cunent browsers. The e-mail also included a name and e- mail address for recipients to contact if they were unable to access the experiment and questionnaire, or if they had questions about completing the questionnaire. As an incentive to encourage participation, each participant was entered into a lottery for a chance to win one of five $20 gift certificates from Amazon.com. Once participants accessed the webpage, they were asked to provide an on-line informed consent, after which they were able to proceed with the experiment. Focus Group Procedure In orderto construct scale items ofattributions ofcandidate motives as perceived by voters (a dependent measure discussed below), an informal focus group was conducted. A convenience sample of prospective voters (4 female, 4 male) from the same general population as that used in the final experiment viewed twelve candidate television commercials and discussed their feelings about the advertisements. A focus group method was utilized in order to provide rich, qualitative data that facilitate opinion-sharing and in-depth responses. The focus group discussion lasted approximately 1 and V2 hours, and was held in an informal environment in order to promote social group interaction. Each advertisement was played multiple times upon the request of the participants in order to completely explore all possible attributions that might be generated from the advertising messages. An initial list of attributional statements was developed from this focus group. A panel of consumer behavior researchers reviewed this list in order to be certain that it included the widest possible range of reasonable attributional statements. Subsequently, sevenmen, five-point, Likert-type scale items were created (see Table 1) and divided into intrinsic attributions (motivations internal to the candidate) and extrinsic attributions (motivations external to the candidate or situational). 59 Table 1. Intrinsic and Extrinsic Attributions of Motives Variable Items II' 'lll' «A The candidate ran this commercial because he wants the voters to know all the facts about the issues. The candidate ran this commercial to discredit the opposing candidate. The candidate ran this commercial because he cares about the country. The candidate ran this commercial because he doesn’t want the opposing candidate to win the election. The candidate ran this commercial because he is trying to mislead the voters about the issues. The candidate ran this commercial to persuade me to vote for him. The candidate ran this commercial because he believes he is the best person for the office of President. The candidate ran this commercial because he wants the voters to question or doubt the opposing candidate. The candidate ran this commercial because he wants the power ofthe presidency and will say anything to get there. PPNF’QPP’N Extrinsic Motives 10. The candidate ran this commercial because a PAC pressured him do it. 11. The candidate ran this commercial to respond to allegations made by the opposing candidate. 12. The candidate ran this commercial because the opposing candidate made misleading statements that had to be corrected. 13. The candidate ran this commercial to discuss an issue that voters think is important. 14. The candidate ran this commercial because a PAC was attacking him. 15. The candidate ran this commercial because he was behind in the polls. 16. The candidate ran this commercial because his political party wanted him to do it. 17. The candidate ran this commercial to tell the voters what they wanted to hear. N etc: The above items were measured on a five-point Likert-type scale for agreement with the following statements (1=strongly disagree, 5=strongly agree). 60 Stimulus Materials 099' inal Evaluation of Advertising Stimuli. Initially, forty campaign television commercials (20 for each candidate) were identified and subsequently reviewed by the researcher in order to determine a set of possibilities for the final experiment. The spots were downloaded from a number of sites, including Bush and Kerry campaign websites, from www.mlitichScom, which features links to television commercials run by the presidential candidates, and from w. During the course of the review, it was determined that each candidate was running a very different campaign, with Bush having few straight issue advertisements, and Kerry having few commercials that spoke negatively of Bush. Ultimately, however, it was possible to narrow this pool of commercials down to a final group of 12 (6 for each candidate), which were comprehensively issue-based and providedagood subsetofboth positive and negative advertising messages. Although image advertising plays a strong part in the political process, the cunent research focused primarily on issue-based advertising messages with the belief that they might generate the greatest h umber of attributional responses. According to West (1997) approximately half of the broadcast advertisements from 1952 to 1996 provided statements about tree candidalss'posilionsonissuss; and, contrarytothesmculationthatads havebecomelesspolicyorientedandmorepersonalitybased in recentyears, the prominence of issues in political advertising campaigns has increased since 1 980. Lastly, commercialswereeliminatedfiomconsflerationiftheywerenotin 61 English (Kerry ran a few Spanish speaking commercials) and if file messages were limited to a particular geographical area rather than running in national media (Bush run a subset of commercials in Ohio only). Pre—testigg of Advertisigg Stimuli. An additional function of the focus group discussed above was to conduct a pre-test prior to the main stage of the consumer research in order to clarify and identify the respondents’ perceptions of the commercials and to explore attributions made based on the messages they were exposed to. The pre-test informed the choice of television advertisements clips to use in the main study. Final Advertisigg Stimuli. The final stimulus materials consisted of two spots each for the major 2004 presidential candidates, George W. Bush and John Kerry (one negative and one positive advertisement for each) (see Appendix A for the transcripts of the commercials). Spot ads provide information about candidate positions on various issues, while also attempting to influence voters’ impressions of the candidates and their voting intentions. When using real advertisements, the researcher runs the risk that participants may have already been exposed to the advertisement and that they may have pre-existing knowledge, attitudes, and beliefs about the advertisement or the candidate. Although the use of existing advertisements introduces these potentially °°nf°unding factors, the use of a naturalistic context enhances realism and generalizability, because a specifically designed advertisement may be deemed ‘°° artificial. In addition, the research was conducted during an ongoing political 62 campaign and featured real candidates who were relying heavily on television advertising. Administration Based upon a randomization algorithm written in JavaScript and embedded within the online survey page, the participants were assigned to one of the four conditions (positive Kerry ad, negative Kerry ad, positive Bush ad, or negative Bush ad) after they completed the first section of the questionnaire, which contained pre—exposure questions related to the measurement of political cynicism and voter attitudes toward the candidate. While it may seem as though a fifth conditionwithnocommercialsmighthavebeenusedasacontrol, sucha condition would not test attributions as a result of exposure to political advertising. After exposure to the stimulus, participants answered questions relative to the measurement of the dependent variables. To insure the confidentiality of the participants, all identifying information “'38 replaced with a serial number after eliminating multiple submissions. W The independentvariableswereadtype(positive adlnegativeadforeach candidate) and political cynicism. Because there were both positive and negative ““8 for each andidate, candidate may be considered another independent ”mama and as such some of the subsequent analyses reflected this distinction. V°ter attitudes toward the candidate mre measured both pre- and post- 63 exposure to the stimulus. The remaining dependent variables, attributions of candidate motive and voter intentions, were measured after exposure to the stimulus. The moderating variables (gender and political party affiliation), along with additional general demographic information, were measured after exposure to the stimulus. Indggndent Variables In order to capture voter’s political cynicism, this study employed a well- used Political Cynicism Measure, which had been adapted from the National Election Studies and which included measures of political efficacy and distrust (Kaid 2002; Kaid, McKinney, and Tedesco 2000; Rosenstone, Kinder, and Miller 1 997). This scale provides the most representative and highly reliable measure currently being used in political cynicism. Participants were asked to strongly disagree or strongly agree on a five-point Likert-type scale, which was summed to form a unidimensional measure of political cynicism, with higher scores indicating higher levels of political cynicism. This scale has achieved acceptable reliability in prior research (.75 at a minimum). It consists of eight Mus: (1) Whether I vote or not has no influence on what politicians do, (2) One never really knows what politicians th'mk, (3) People like me don’t have any say about What the government does, (4) Sometimes politics and government seem so °°mi>licated that a person like me can’t really understand what’s going on, (5) One can be confident that politicians will always do the right thing (reverse W). (6) Politicians often quickly forget their ebction promises alter a political campaign is over, (7) Politicians are more interested in power than in what the people think, and (8) One cannot always trust what politicians say. Dependent Variables. There are several dependent measures incorporated in the experimental design. As discussed above, seventeen, five-point, Likert-type scale items of attributions of candidate motive as perceived by voters were created. In order to measure voter attitudes toward the candidates a semantic differential scale of candidate image was utilized. This scale included 12 bipolar adjective pairs (Kaid, Leland, and Whitney 1992; Tedesco 2002): qualified- unqualified, sophisticated-unsophisticated, honest—dishonest, sincere-insincere, successful-unsuccessful, attractive-unattractive, calm-excitable, aggressive- unaggressive, strong-weak, passive-active, friendly-unfriendly, and believable- unbelievable, and was applied to both candidates for president Each pair had five intermediate points, and higher scores indicated more favorable traits, with One item (passive-active) being reverse coded to minimize response bias. This scale has achieved acceptable reliability in prior research (.82 at a minimum) and was summed to create a total score to use in the statistical equations. A traditional feeling thermometer scale (0-100), which has been used in past mar Ch (T edesco 2002), was not utilized, as it did not fit with the nature of the Current research. As a measure of voter intention, participants were asked “how likely' it Was that they would vote for each candidate (Chang 2003). 65 Other Variables. The posttest questionnaire also contained several demographic and individual difference variables including age (an open ended question); marital status ("What is your cunent marital status? Single, Married, Divorced, Wldowed); ethnicity ("What racial or ethnic group best describes you?” Caucasian, African-American, Asian, Hispanic, Native American, or other); gender (female, male); occupation (Professional, White Collar, Blue Collar, Student (part-time), Retired, Other); and political party affiliation (“Generally speaking, do you usually think of yourself as a Republican, a Democrat, an I ndependent, or something else?"). A copy of the measurement instrument is attached as Appendix B. The following chapter details the results of the experiment. CHAPTER FIVE RESULTS The results of the experiment are presented here in several sections. The first section presents general sample characteristics. The second section contains reliability measures for the scales used in the experiment. Hypotheses are tested in order in the next section. Lastly, the fit of the proposed structural model is discussed. General Sample Characteristics Registered voters were recruited from community groups throughout Michigan (church groups, work groups, parents involved in scout groups, school PTAs, etc.) and represented a cross section of Michigan voters who, on Election Day. would have to choose between the candidates whose political commercials they viewed. A total of 326 respondents participated in the study. The original Sample was 336; however, 10 respondents experienced technical difficulties and did not complete the study. They were dropped from further analysis. Of the Overall recruiting list of 510 potential participants, 326 actual participants represents a 64% response rate. Table 2 presents the demographic Characteristics of the sample. 67 Table 2 Demographic Characteristics of the Sample‘I Characteristic N Percent Gender Male 165 50.6 Female 158 48.5 A96 20-30 83 25.4 31-40 93 28.5 41 ~50 76 23.3 51-60 52 15.9 61-70 10 .03 71+ 8 .02 Marital Status Single 79 24.2 Married 221 67.8 Divorced 26 8.0 Wldowed 0 0.0 Ethnicity Caucasian 285 87.4 African-American 15 4.6 Asian 17 5.2 Hispanic 3 .9 Native American 1 .3 Other 5 1.5 Occupation Professional 158 48.5 White Collar 62 19.0 Blue Collar 14 4.3 Student (part-time) 19 5.8 Retired 21 6.4 _Other 52 16.0 Polltical Party Affiliation Republican 141 43.3 Democrat 89 27.3 I"’deDendent 69 21.2 Other 27 8.3 .The NS Within characteristic groupings do not add up consistently to the sample use of missing data for some respondents. Nbeca 68 One hundred sixty-five males (50.6%) and 158 females (48.5%) participated in the study (3 participants did not indicate gender). Their ages ranged from 20 to 77 years (2 participants did not indicate age), with an average of 39 years. Further, the participants were predominately married (67.8%) and Caucasian (87.4%). Republicans represented the largest group in the current study at 43.3%, with Democrats and Independents represented at 27.3% and 21.2%, respectively. Although the sample was not randomly selected, the participants were generally representative of Michigan demographics (49% male, 51% female, with a median age of 35.5 years old according to Michigan wnsus data). Scale Construction and Reliabil'gy Before addressing the specific hypotheses and research questions, the scales used to measure the independent and dependent variables were checked for internal consistency and unidimensionality. Cronbach’s reliability analysis was used for internal consistency and confirmatory factor analysis (CFA) was used to test for scale unidimensionality (Hunter and Gerbing 1982). Table 3 summarizes the mean, standard deviation, and Cronbach’s coefficient alpha for each independent and dependent composite variable. 69 Table 3 Scale Descriptives for Independent and Dependent Variables Independent Variable Scale Mean Std. Dev. ct Political Cynicism 25.63 4.503 .714 Dependent Variables Attitude Toward the Candidate: George W. Bush 33.03 9.924 .919 John Kerry 32.87 7.732 .867 It is important to note that all of the measures used here are summated scales, not averages across items. Since cynicism was a 5-point, 8—item scale, scores could range from 8 to 40, with the sample mean of 25.63 being slightly above the midpoint of the range. Additionally, attitude toward the candidates was a 5-point, 12-item scale and scores could range from 12 to 60. Scores of 33.03 and 32.87 for George W. Bush and John Kerry, respectively, represent scores above the mean. Additionally, it is interesting to note that the mean for George W. Bush is higher than that for John Keny, representing a slight advantage. 61mm Hypothesis One This study tested H1 predicting that exposure to political advertisements would lead voters to make both intrinsic and extrinsic attributions of candidate motive. This prediction was constructed based on the a priori theoretical assumption of the previous conceptualization of the nature of the intrinsic and extrinsic attributions, rather than the exploratory statistical classification. Therefore, the present study performed a confirmatory factor analysis to obtain 70 more stringent discriminant validity between the two constructs with AMOS 5.0 (Arbuckle 2003) using maximum likelihood estimation. As expected, the two theoretically distinct factors were confirmed, supporting H1. However, eleven items either had low factor loadings or indicated a high modification index (showing a possibility of being cross-loaded on two factors), so they were removed in this confirmatory factor analysis process. Thus, only three items were found to be valid for each construct. Multiple fit indices were used for evaluating this two-factor model. The fit of this model (and all subsequent analyses) was evaluated using the chi-square statistic, which is sensitive to sample size, as well as additional multiple fit indices. The GFI is considemd an absolute fit index as it measures the proportion of model fit improvement in the hypothesized model compared to no model at all. The GFI is roughly similar to the square multiple conelation in SEM (or the R2 in multiple regression) as it attempts to explain the proportion of observed correlations in the model. The goodness-of-fit index [GFI] was .98 and the adjusted goodness-of-fit index [AGFI] was .95. The data in the present study yielded a comparative fit index [CFI] of .97, a normal fit index [NFI] of .95, and a root mean square error of approximation [RMSEA] of .07. In general, values of .8 or above for GFI and AGFI, higher than .9 for NFI, closer to 1.0 for CPI, and less than or equal to .08 for RMSEA are considemd to be indications of a good fit (Kelloway 1998; Kline 1998). All factor loadings on the two factors were also statistically significant and ranged from .40 to .86 (p< .0001), indicating that the two-factor model was found 71 to be consistent with the data very well. The internal reliability of each factor also indicates that both intrinsic attributions (a = .79) and extrinsic attributions (a = .58) were internally reliable. Although the value for extrinsic attributions falls below the usual threshold of acceptable reliability, values below .70 have previously been deemed acceptable when used in exploratory research (Hair et al. 1995). Thus, the items of each factor were summed into a single score for the subsequent analyses. Table 4 summarizes the means and standard deviations of the scales and their component items, along with their alpha coefficients. Also included in Table 4 are the loadings for each of the items that were dropped from further analysis. 72 Table 4 Scale and Items of Intrinsic and Extrinsic Attributions candidate made misleading statements that had to be corrected Scalelltem a, Mean SD Factor locdlltg_ Voter Attributions (f: 21.5 df = 8, N==326) Intrinsic Attributlgg .79 9.67 2.88 The candidate ran this commercial because he is trying to 3.28 1.17 .82 mislead the voters about the issues. The candidate ran this commercial to discredit the opposing 2.79 1.07 .70 candidate. The candidate ran this commercial because he wants the 3.61 1.17 .64 voters to question or doubt the opposing candidate. ‘The candidate ran this commercial because he wants the .79 power of the presidency and will say anything to get there. ‘The candidate ran this commercial because he doesn’t .43 want the opposingiandidate to win the election. The candidate ran this commercial to persuade me to vote .24 for him. ‘The candidate ran this commercial because he cares -. 80 about the county. ' The candidate ran this commercial because he believes -. 26 he is the best mrson for the allies of the President. ‘The candidate ran this commercial because he wants the -.21 voters to know all the facts about the issues. lc Attributlo .58 8.97 1.94 The candidate ran this commercial because his political 3.43 .92 .52 party wanted him to do it. The candidate ran this commercial because a PAC 2.62 .81 .45 red him to do it The candidate ran this commercial to respond to allegations 2.93 .91 .40 made by the opposing candidate. 'The candidate ran this commercial because he was behind .71 in the polls. 'The candidate ran this commercial to tell the voters what .59 they wanted to hear. ‘The candidate ran this commercial because a PAC was .25 attacking him. ‘The candidate ran this commercial to discuss an issue that -. 26 voters think is important. ‘Tha candidate ran this commercial because the opposing -. 19 *lndicates items that were removed after CFA because modification indices of the CFA indicamd that either their loading was low or indicated a high modification index (showing a possibility of being cross-loaded on two factors). In addition to the CFA procedure illustrated above, one sample t-tests were also performed to test if the averaged mean scores on each of the intrinsic 73 and extrinsic attribution scales were significantly different from the default position of 1, indicated by “Strongly Disagree.” The results of the t-tests showed that the final three-item scale used as the intrinsic attribution measure (M=3.23, SD=.96) and the final three-item scale used as the extrinsic attribution measure (M=2.99, SD=.65) were significantly different from the default position of 1 (t=41.88, #325, p<.001 and t=55.64, df=325, p<.001, respectively). Table 5 summarizes the results of the one sample t-tests. Table 5 One Sample T-testc for Intrinsic and Extrinsic Attributions Test Value = 1 Mean 95% Confidence Interval of t df Sig. (2-tailed) Difference the Difference Lower Upper Intrinsic Attributions 41.879 325 000 2.22495 2.1204 2.3295 Extrinsic Attribution 55.640 325 .000 1.99080 1.9204 2.0612 Hypothesis Two H2a and H2b predicted that those participants exposed to negative candidate advertisements would make more intrinsic attributions and fewer extrinsic attributions than those exposed to positive candidate advertisements, respectively. As hypothesized in H2a. an independent sample t-test showed that intrinsic attributions (M=11.16, SD=1.92, N=170) of the sponsor‘s negative candidate advertisements were significantly greater than the intrinsic attributions (M=8.05, SD=2.88, N=156) of the sponsor’s positive advertisements (t(324)=11.59 , P<.001). Consistent with the prediction of H2b, another t-test also demonstrated that extrinsic attributions (M=8.61, SD=1.56, N=170) of the 74 sponsor’s negative candidate advertisements were significantly lower than of the extrinsic attribution (M=9.37, SD=2.23, N=156) of the sponsors positive advertisements) (t(324)=-3.57, P<.001). Consequently, these findings supported H23 and H2b. In order to examine the relationships hypothesized by H23 and H2b, this study also performed a path analysis using AMOS 5.0 and examined the hypothesized relationships. Structural equation modeling has many advantages over traditional analytical techniques for assessing measurement issues. F irst, SEM allows researchers to use observed variables to construct unobserved (latent) constructs, which have the strength of correcting for measurement and thus creating a ‘true score” of a construct. Second, using maximum likelihood (ML) estimation, SEM generates parameter estimates for multiple variables in a model, which are calculated simultaneously, allowing testing of a causal model and of predictive validity between measures. Finally, multiple group analysis in SEM allows the researcher to test the equality of models across multiple population groups (e.g. men versus women, Democrats versus Republicans) by testing for group invariance. In the proposed structural model (See Figure 2), political advertising type (negative vs. positive ads being dummy-coded 0 and 1, respectively) and political cynicism are proposed to directly affect both intrinsic and extrinsic attributions, which have an influence on voter intention indirectly through voter's attitudes toward each candidate (Bush vs. Kerry). In order to assess this model, the pmpoaedmodelisalsodividedintotwosub—models,specifyingfl1eefiectsof 75 positive vs. negative ads on voter attributions, attitudes, and intention toward each candidate: one model is for Bush and the other is for Kerry. Figure 2 Proposed Structural Model Political .. ’ A'WW We] . + of Candidate ' Positive Motive Voter + Attitudes , Voter . + Toward the Intentions P m. i _ AW Candidate Cynicism of Candidate Motive In the first structural model, Bush-negative advertisements (coded 0) were found to lead to more voter attributions of intrinsic candidate motives than Bush- positive advertisements (coded 1) (B = -.66, p< .01), while Bush-negative advertisements (coded 0) were found to produce fewer voter attributions of extrinsic candidate motives than Bush-positive advertisements (coded 1) (B = .20, p< .01) (See Figure 2a). Figure 2a Bush-Sponsoring Structural Model (N=166) . . .. Intrinsic PM , '66 e Attributions , Ming. OfBUSh Negative! pm Candidate Voter “ Voter 20“ 1 Bush '79 ' I _ sic oward Toward Bush Political .11 Attributions Cynicism * otBush Candidate Note: ‘CoatliclarrtscrestatlcalyeignihccntctP<.05 level. "CoafiiclantccresteticehysignificcntetP<.01 level. ChiSqucre - 39.0. df - 6. P < .01. GFI - .93. AGFI -.76. CFI = .90, NFI 3.89. and RMSEA - .18 76 Consistent with the first model, the second model also showed that Keny- negative advertisements were more related to voter’s attributions of intrinsic candidate motives than were Kerry-positive advertisements (B = -.35, p< .01), while Keny-negative advertisements were found to produce fewer voter attributions of extrinsic candidate motives than Kerry positive advertisements (B = .19, p< .01). Thus, along with the t-tests above, the path analyses also confirmed H2a and H2b (See Figure 2b). Figure 2b Kerry-Sponsoring Structural Model (N=160) Intrinsic Attributions , of Kerry Candidate 63“ Extrinsic Attributions of Kerry ‘ Candidate Note: ‘CoetficlentsarestaticclysignificentctP<.05 level. “CoetllclantscrecteticcllycignificcntetP<.01 level. Chi-Square = 51.2, dt I 6. P < .01, GFI = .92, AGFI 2.70, CFI 8 .80. NFI =79, and RMSEA = .22. Hypothesis Three H3a and H3b hypothesized that, regardless of the advertisement type, political cynicism would be positively related to intrinsic attributions of the sponsoring candidate advertisements and negatively related to extrinsic attributions of the sponsoring advertisements, respectively. A bivariate correlation analysis demonstrated that political cynicism had a significant positive relationship with intrinsic attributions of the sponsoring candidate advertisements 77 (r =.19, p<.01), supporting H3a. However, another bivariate correlation analysis demonstrated that political cynicism was not significantly associated with extrinsic attributions of the sponsoring candidate advertisements(r=.02, p>.10), disconfinning H3b. Hypothesis Four Hale and H4b hypothesized that voter attitudes toward the sponsoring candidate would be negatively related to intrinsic attributions of the sponsoring candidate advertisements and would be positively related to extrinsic attributions of the sponsoring candidate advertisements, respectively. As expected, the path analyses showed that, for those who were exposed to either Bush-sponsoring advertisements (B =-.42, p< .01) or Kerry-sponsoring advertisements (B =-.49, p< .01), their intrinsic attributions of the sponsoring candidate was negatively related to their attitude toward the sponsoring candidate. However, the path analyses showed atthesametimethat, forthosewhowere exposedtoeither Bush or Kerry sponsored advertisements, their extrinsic attributions did not have a significant relationship with their attitude toward the sponsoring candidate, disconfirrning H4b (See Figures 2a and 2b above). H4¢statesthatbofl1intrinsicandexhinsicatuibufionsmediatethemctof advertisement type (negative vs. positive) on voter attitude toward the sponsoring candidate oftheadvertisement. Based on thefindingsofHZa, H2b, H4a, and H4b, the path analyses suggest that voter attitude toward the sponsoring candidate becomes unfavorable (lower) only when voters develop intrinsic 78 attributions about the sponsoring candidate of the advertisements. However, the non-significant effect of extrinsic attributions on voter attitude toward the sponsoring candidate indicates that extrinsic attributions do not mediate the effect of advertisement type on the voter attitude. Thus, these findings partially support H4c. Hypothesis Five H5 hypothesized that voter attitudes toward the sponsoring candidate would be positively related to voter intention. As expected, the path analyses showed that, for those who were exposed to either Bush-sponsored advertisements (B = .79 p< .01) or Kerry-sponsored advertisements (8 =63, p< .01), their attitude toward the sponsoring candidate had a positive impact on their intention to vote for the sponsoring candidate (See Figures 2a and 2b above). Hypothesis Six H6a posited that, regardless of advertisement type, women would make more attributions of intrinsic candidate motives for the sponsoring candidate than will men. An independent samples t-test was employed to test this hypothesis for positive and negative ads. As a result, the Host demonstrated that women (M=10.27, SD=2.63, N=158) endorsed more intrinsic athibutions than men (M=9.15, SD=2.98, N=165) when exposed to both advertisements, confirming H6a (t(321)=3.59, P<.001). I 79 Further, H6b predicted that gender differences would moderate the effects of political advertisement type (negative versus positive) on attributions of candidate motive for the sponsoring candidate. To test the moderating effect of gender on intrinsic and extrinsic attributions, multiple-group structural equation analyses (Bollen 1989; Joreskog and Sobrom 1993; Scott-Lennox and Lennox 1995) were applied for each sponsoring candidate group (Bush and Kerry) to examine whether the parameter estimate between advertisement type (a dummy variable being coded 0 for negative and 1 for positive advertisement) and attributions (intrinsic and extrinsic) differs across gender type (female vs. male). Multiple-group structural equation modeling deals with moderators indirectly. In other words, the empirical criterion is whether there are different values for structural parameters at different values of a moderator. Thus, in order to test H6b and H7, the subjects were divided into groups according to their gender (female vs. male) and political party affiliation (Republican vs. Democrat). The procedure for H6 is as follows: First, the sample was divided into the two gender groups. For each subsample, AMOS calculated a covariance matrix and then estimated the parameters for each subsample using maximum likelihood estimation. Then, the pairwise comparison of the path coefficients between political advertisement type (negative/positive) and attributions (intrinsic and extrinsic) was conducted, particularly based on the chi-square difference between the two models. Specifically, the two path coefficients between political advertising type and intrinsic attributions and between political advertising type and extrinsic attributions were constrained to be equal across females and males 80 in one model, whereas the two path coefficients were allowed to be freely different across the gender type in the other model. The difference of the two model’s statistical significance was used as a test for the equal path coefficients by gender type, that is, whether the unconstrained model produced a better fit than the equality constrained model. If this was the case, then, the hypothesized moderating effect of gender type is supported. Following this procedure, the pairwise comparison between females and males was conducted for each sponsoring candidate group (Bush and Kerry). For the group that saw Bush commercials (N for female = 80 and N for male =83), the pairwise comparison between females and males demonstrated that the chi- square for the unconstrained path coefficients between advertising type and attributions was 46.5 (df=12) and the chi-square for the constrained path coefficients was 59.3 (df=14). Given that the critical value of chi-square statistical difference with two degrees of freedom at the .05 level is 7.68 (e.g., critical value for one dogma of freedom at the .05 level is 3.84), the chi-square difference between thetwo models (12.8fortwo degrees offreedom) indicates thatthe chi- square estimate was better when the two parameters were allowed to be different rather than constrained to be equal. Thus, gender was found to moderate the effect of political advertisement type (positive/negative) on attributions in the Bush sponsoring candidate model. In ordertotestwhich path coefficient (between advertisementtype and intrinsic attributions vs. between advertisement type and extrinsic attributions) plays a critical role in the moderating effect of gender, two pairwise comparisons 81 were further conducted separately. The findings of the analyses indicated that gender type significantly moderates the relationship between advertisement type and intrinsic attributions toward Bush (chi-square = 55.0, df=13) as well as the relationship between advertisement type and extrinsic attributions toward Bush (chi-square = 50.3, df--13). Specifically, in the unconstrained model (chi-square = 46.5, df=12), the path coefficients between political advertisement type and intrinsic attributions was -.52 (p< .001) for females and -.76 (p<.001) for males and the path coefficients between political advertisement type and extrinsic attributions was .04 (p> .10) for females and .33 (p< .001) for males. The two path coefficients between political advertisement type and intrinsic attributions across gender were statically significant, while only one coefficient between advertisement type and extrinsic attributions across gender was statistically significant (See Figure 3a). As indicated by these findings, males are more likely than females to endorse intrinsic attributions when both males and females were exposed to the negative advertisements, even though both males and females tend to endorse intrinsic attributions when both were exposed to the negative advertisement. At the same time, interestingly, the findings also suggest that, only for males, positive advertisements were more likely to generate extrinsic attributions than negative advertisements, when both males and females were exposed to positive political advertisements. However, for bmales, the two ad types did not make any difference in the level of extrinsic attributions. 82 Figure 3a Gender-Based Multiple Comparison Analysis For BushSponsorlng Structural Model (For Female: N = 80) Political -.52“ Intrinsic Advertising: + Agbutiom . Positive . Candidate Voter Intentions .04 _ Toward Bush .. Extrinsic Political .17 Attrbutions Cynicism of Bush ‘ Candidate (For Male: N = 83) Political --73“ Advertising: Posiive .74.. ' VO‘Ol’ Toward Bush Political Cynicism '06 Note: 'Coafficiantscreshticclyalgnificenth<.05 level. "CoefficientsarestatlccllysignlficcntatP<.01 level. Modal Flt Chi-Square = 48.5, of s 12, P < .01, GFI s .92, AGFI =.72. CFl :- .90, NFI 2.87, and RMSEA = .13. The similar pattern of gender moderating effect was found in the Keny sponsoring candidate model (N for female = 78 and N for male =82). Specifically, for the group exposed to Kerry commercials, the pairwise comparison between females and males demonstrated that the chi-square for the unconstrained path coefficients between advertising type and attributions was 59.0 (dfi12) and the chi-square for the constrained path coefficients was 69.2 (df=14). Given that the critical value of chi-square statistical difference with two degrees of freedom at 83 the .05 level is 7.68 (e.g., critical value for one degree of freedom at the .05 level is 3.84), the chi-square difference between the two models (10.2 for two degrees of freedom) indicates that the chi-square estimate was better when the two parameters were allowed to be different rather than constrained to be equal. Like the Bush-sponsored ad model, gender was found to moderate the effect of political advertisement type (positive/negative) on attributions in the Keny ad model. In order to test which path coefficient (between advertisement type and intrinsic attributions vs. between advertisement type and extrinsic attributions) makes a contribution to the moderating effect of gender, the same pairwise comparison procedures as above were also employed. The findings of the analyses indicated that gender type significantly moderates the relationship between advertisement type and extrinsic attributions toward Kerry (chi-square = 68.6, dfi13) while gender type did not moderate the relationship between advertisement type and intrinsic attributions toward Kerry (chi-square = 59.0, df=1 3). Specifically, in the unconstraimd model (chi-square = 59.0, df=12), the path coefficients between political advertisement type and intrinsic attributions was -.35 (p< .001) for females and -.33 (p<.001) for males and the path coefficients between political advertisement type and extrinsic attributions was - .07 (p> .10) for females and .39 (p< .001) for males. The two path coefficients beMen political advertisement type and intrinsic attributions across gender were statically significant, while only one coefficient between advertisement type and extrinsic attributions across gender was statistically significant (See Figure 3b). As a result, like the Bush-sponsored advertisement model, the findings also suggest that only males tend to endorse extrinsic attributions when both males and females were exposed to positive political advertisements. However, unlike the Bush-sponsored advertisement model, these findings suggest that both males and females would not endorse different levels of intrinsic attributions when both were exposed to Kerry-sponsomd negative advertisements. Thus, based on these multiple group analyses, this study partly confirmed H6b predicting gender differences would moderate the effects of political advertisement type (negative versus positive) on the attributions of candidate motive for each sponsoring candidate. 85 Figure 3b Gender-Based Multiple Comparison Analysis For Kerry-Sponsoring Structural Model (For Female: N = 78) Political Advertising: Mgativel Poshive VOW' Intentions Toward Keny Political Cynicism Political '33“ mm“ Advertising: Negative! Positive Voter Intentions Toward Kerry Political _ 04 Cynicism ' Nob: 'CoefficiantsarestaticalyslgnficentatP<.05 level. "CoefficientsarestaticellyslgnificentatP<.o1lavel. Model Fit Chi-Square = 59.0, (if = 12, P < .01, GFI = .91, AGFI =.68, CFI = .81, NFI =.78, and RMSEA = .16. Hypothesis Seven H7 hypothesized that differences in political party affiliation would moderate the effects of attributions of candidate motive for the sponsoring candidate on voter attitude. The same multiple-group comparison analyses procedure as conducted for gender was applied for testing this hypothesis. Specifically, for the Bush sponsoring candidate model (N for Republican = 72 and N for Democrat = 47; see Figure 4a), the pairwise comparison between those affiliated with the Republican party and those with the Democratic Party demonstrated that the chi-square for the unconstrained path coefficients between attributions and voter attitude was 56.7 (dfi12) and the chi-square for the constrained path coefficients was 64.6 (df=14). The chi-square difference between the two models (7.9) is greater than the critical value of Chi-square statistical difierence with two degrees of freedom at the .05 level (7.68), which indicates that the Chi-square estimate was better when the two parameters were freely calculated rather than being constrained to be the same. Figure 4a Political Party-Based Multiple Comparison Analysis For Bush-Sponsoring Structural Model (For Republican: N = 72) Political «70“ Wins“ Advertising: P m . pm" . 3 Candida” Voter “ Voter 19 1 rd Bush .36 ' . i . 0W8 . , Toward Bush Political .201 Attributions WW ’ or Bush . Candidate (For Democrat: N = 47) Political ~53" , ”am," "I?” Negativel' °' 3“" ‘ 15 Intentions - b Toward Bush Political .10 Attributions Cynicism of Bush Candidate Notcz'Coelllcientsarestaticcly atP<.05Ieval. “Coefficientsarestaticallysignlficanta P<.01Ievcl. ModelFlt: Chi-Square-56.7,df= 12, P< .01,GFI-= .88.AGFI=.58. CFI=.66. fit-18.65. andRMSEA= .18. 87 In order to test which path coefficient (between intrinsic attribution and voter attitude vs. between extrinsic attribution and voter attitude) plays a critical role in the moderating effect of the political party affiliation, two pairwise comparisons were further conducted separately. The findings of the analyses indicated that political party significantly moderates the relationship beMen intrinsic attributions and voter attitude toward Bush (Chi-square = 64.4, dfi13) while political party does not significantly moderate the relationship between extrinsic attributions and voter attitude toward Bush (Chi-square = 56.8, df=13). Specifically, in the unconstrained model (chi-square = 56.7, df=12), the path coefficients between intrinsic attribution and voter attitude was .03 (p> .10) for those of the Republican party and -.40 (p<.01) for those of the Democratic party and the path coefficients between extrinsic attribution and voter attitude was -.05 (p> .10) for those of the Republican party and .03 (p> .10) for those of Democratic party. In addition, the findings also indicate that the path coefficient from intrinsic attributions to voter attitude was only statistically significant when the political party was Democratic (B =-.40, p< .01), while the rest of the path coefficients between intrinsic attributions and voter attitude and between extrinsic attributions were not statistically significant (See Figure 4a above). These findings suggest that voter attitude toward the sponsoring candidate, Bush, becomes significantly unfavorable when a Democratic rather than a Republican voter endorsed intrinsic attributions. However, the voter attitude toward Bush would not be significantly affected when either Democrats or Republicans endomd extrinsic attributions. The same multiple-group comparison analyses procedure was also applied to the other sub—model of the Kerry sponsoring candidate model (N for Republican = 69 and N for Democrat = 42; see Figure 4b) in order to test whether political party affiliation would moderate the effect of attributions on voter attitude toward the sponsoring candidate, Keny. Specifically, for the group exposed to Keny commercials, the pairwise comparison between those affiliated with file Republican party and those with the Democratic party demonstrated that the chi-square for the different path coefficients between attributions and voter attitude was 16.1 (df=12) and the chi-square for the equal path coefficients was 26.6 (df=14). The chi-square difference between the two models (10.5) is greater than the critical value of 7.68 for two degrees of freedom at the .05 level. This indicates that the chi-square estimate was better when the two parameters were unconstrained rather than being constrained to be same. 89 Figure 4b Political Party-Based Multiple Comparison Analysis For KerrySponsoring Structural Model (For Republican: N = 69) Polltl' “ca .30“ Intrinsic Negative! 0' K9"? Positive .40” Candidate 30“ Political Extrinsic . _ -_14 Attributions 070W" ’ of Kerry ‘ Candidate WWI .071 “ Voter Intentions Toward Keny Note: 'Coafficiantsarestatlcalysignificanth<.05 level. ”Coeffimnts arestatlcellysignillccnt at P <.01lavel. Modal fit: Chi-Square =16.1.df= 12, P > .10, GFI 8 .96, AGFI =.84, CFI = .97, NFI 8.92, and RMSEA = .06. Subsequently, two pairwise comparisons were separately conducted to test which path coefficient (between intrinsic attribution and voter attitude vs. between extrinsic attribution and voter attitude) significantly moderated the effect of attribution on voter attitude toward the sponsoring candidate. The findings of the analyses indicated that political party significantly moderates the relationship between intrinsic attributions and voter attitude toward Kerry (chi-square = 26.4, df=13) while political party does not significantly moderate the relationship 90 between extrinsic attributions and voter attitude toward Kerry (chi-square = 18.9, df=13). Thus, in the unconstrained model (chi-square = 16.1, dfi12), the path coefficients between intrinsic attribution and voter attitude were -.57 (p< .01) for those of the Republican party and -.01 (p> .10) for those of the Democratic party, and the path coefficients beMen extrinsic attribution and voter attitude were .15 (p> .10) for those of the Republican party and -.09 (p>.10) for those of the Democratic party (See Figure 4b above). These findings suggest that voter attitudes toward the sponsoring candidate, Keny, became significantly unfavorable particularly when Republicans endorsed intrinsic attributions but not when Democrats did. However, voter attitude toward Keny was not significantly affected when either Democrats or Republicans endorsed extrinsic attributions. FE of the Structural Model The multiple fit indices suggest that the proposed structural model did not fit the data very well. Specifically, first, for the Bush-sponsored ad model, Chi- square was 39 (df=6), P < .01, GFI was .93, AGFI was .76, CFI was .90, NFI was .89, and RMSEA was .18 (See Figure 28 above). Secondly, the multiple fit indicesforthe Kerry-sponsomd ad modelalsosuggestthattheproposed Structural model was not consistent with the data: chlasquare was 51.2 (df=6), P < .01, GFI was .92, AGFI was .70, CFI was .80, NFI was .79, and RMSEA Was .22 (See Figure2babove). Thus,theproposed modelwas re-specifiedtofit the data. 91 To respecify a recursive or nonrecursive model, there are two ways to test the model: model trimming and model building (Klein 1998). Model trimming refers to a strategy where the researcher begins a path analysis with a just- identified model and by eliminating paths, simplifies it. The Chi-square increases as paths are trimmed, indicating that the fit of the model becomes worse. In the second method, model building, paths are added to a null model. The chi-square decreases as paths are added to the model, indicating an improved model fit. With either strategy, the eliminating (or adding) of paths is determined based on theoretical considerations or face validity, although it may be guided by empirical information obtained from a statistical program (i.e., a modification index or the chi square difference test). In the re-spacification process, two new significant paths emerged in the Bush-sponsoring candidate model; from political advertising type to voter attitude toward the sponsoring candidate and from intrinsic attributions to voter intention to vote for the candidate (See Figure 5a). Now, the multiple fit indices suggest the revised model is consistent with the data very well: chi-square was 1.8 (df=4), P>.10, GFI was .99, AGFI was .98, CFI was 1.0, NFI was .99, and RMSEA was less than .001. These findings suggest that political advertisements (negative vs. positive) affect voter attitude toward the sponsoring candidate directly as well as indirectly through the attributions that voters endorse. Also, these findings suggest that the attributions influence voter intention directly as well as indirectly through the voter attitude toward the sponsoring candidate. 92 Figure 5a Revised Bush-Sponsoring Structural Model (N=166) Poltical Advertising: 1 Postiva VONI’ Intentions Toward Bush Political Cynicism Note: " Coefficients are steticcly significant at P <.05 level. " Coefficients are staticaly significant at P <.01 level. -> paths emerged Modal Ft Chi-Square = 1.8, dis 4, P >.10. GFI s .99, AGFI 8.98. CFI = 1.00, NFI =.99, and RMSEA <.001 However, as opposed to the Bush-sponsoring candidate model, adding the same two paths to the Kerry-sponsoring candidate model did not improve the model fit (See Figure 5b). The multiple fit indices suggest the revised model is still not consistent wit the data: chi-square was 47.9 (df=4), P<.01, CF I was .92, AGFI was .58, CFI was .81, NFI was .80, and RMSEA was .263. Figure 5b Revised Kerry-Sponsoring Structural Model (NI-160) Political Advertising: ~ Negative! Poalive V0” Intentions Toward Kcrry Political Cynicism Nobz‘Cocfficlantsarestatlcclysignlficanth<.05Iavel. "CoefficientsarestatlccllysignlficantatP<.01level. ModalFltChl-Squares47.9,df=4, P<.01,GFl-.92,AGFI-.58,CFI=.81, NFI=.80,andRMSEA- .263 93 CHAPTER SIX DISCUSSION, LIMITATIONS, FUTURE RESEARCH, AND CONCLUSIONS Overview The purpose of this study was to determine whether attribution theory could be used to explain the process by which voters exposed to political advertising messages formed attitudes toward candidates for office. Although most analyses of political attitudes and behavior involve survey research, controlled experiments are sometimes chosen in political advertising as a more precise way to measure political advertising exposure (Ansolabehere and Iyengar 1995; Pfau et al. 2001; Pfau at al. 2002; Valentino, Hutchings, and Williams 2004). An experimental design was chosen in the current study because it enabled the researcher to manipulate the stimulus material and to control exposure rather than relying on participants’ memory about the political commercials they may have seen. Despite the considerable amount of research in political advertising, however, there are still the questions of exactly how, how much, and under what conditions political advertising matters. It was bellemd thatattributiontheorywould shed lightontheeffects ofpolitical advertising asa Whole and begin to answer those unanswered questions. With this primary interest in the role of attributions in the formation of POlitical advertising attitudes, the current study proposed and tested a structural Nuafionmodelwith specifichypotheses in ordertoexaminethe roleofboth intrinsic and extrinsic attributions on voters’ attitudes and voting intentions. Additionally, individual difference factors (gender and political party affiliation) were considered to see if they presented a moderating effect on voters’ attitudes and intentions toward the candidate. Through this attempt, the current study might substantially contribute not only to the theoretical accumulation of attribution theory literature but also to the practical issues of designing more effective political advertising. Before discussing the results of the present study, it should be noted that it is difficult to determine the effectiveness of political campaigns (Thorson, Christ, and Caywood 1991). Ultimately, the final measure and bottom line for political campaigns is whether the candidate won or lost. Further, because many things happen simultamously in a real election, “it is difficult to isolate the impact of political advertising” (Thorson, Christ, and Caywood 1991, p. 483). The present study, however, revealed many interesting effects of political advertising. Discussion and Imam ' ns Overall, the results of Hypothesis One are consistent with the application of attribution theory in that individuals are believed to regularly engage in c°9|1itive processes in which they assign underlying causes or explanations to observed events. Most relevant to the current research is that individuals also aPI‘W these attributional activities to advertising messages (Davis 1994; Folkes 1988; Gotlieb and Sarel 1991; Settle and Golden 1974; Smith and Hunt 1986) in orClel'to explainthe underlying motives ofthe sourceoftheadvertisemant. Inthe 95 cunent study, it was believed that attributions of candidate motive would be formed in response to the political advertising message when voters’ attempted to determine the underlying motivations of the candidate in the advertisement. After being exposed to political advertising commercials, respondents did indeed endorse attributional statements of candidate motive, supporting H1. Furthermore, it was believed that in response to the candidate messages that a two-factor model of attributions (intrinsic and extrinsic) would emerge. The results of a confirmatory factor analysis supported this premise with factor loadings on the two factors being statistically significant, and multiple fit indices indicating that the two-factor model fit the data very well. However, it should be nobd that all of the intrinsic attributions that were retained through CFA were negatively phrased statements. They representadvelse viewsofthe sponsoras a person. This unfavorable orientation should be kept in mind as the other results of the study are considered. Furthermore, the results of the factor analysis indicate that the attributions endorsed by voters may not simply fall into theoretical categories of intrinsic and extrinsic, but instead there are subtleties within each of these categories. As mentioned above, because the intrinsic attributions were negative, it had an overalleffectonthemodel. Future research should considerand more fully explore the factors that may emerge. Hypothesistwoproposedthat, asa resultofexposureto negative candidate advertisements, voters would endorse more intrinsic than extrinsic attributions. This hypothesiswasploposadonthebasisofpriorresearch involving comparisons between positive and negative political advertising which revealed that participants exposed to negative advertising generated more source candidate derogations and were more negative toward the political candidate in general than were participants exposed to positive political advertising (Hill 1989; Pinkleton, Um, and Austin 2002). The findings of the prior studies suggested that negative advertisements would yield attributions of intrinsic candidate motive more than would positive ads (Meirick 2002), because they would be attributing the cause of the advertisements to the “mean spiritedness” of the candidate, rather than outside forces contributing to the commercial messages. The rationale behind this was that intrinsic motivations would be seen as internal to the candidate and controlled by the candidate, whereas extrinsic motivations are seen as external or situational (i.e., beyond the control of the candidate). The cunent research supported Hypotheses 2a and 2b through two different analyses: Meets and structural equation modeling, indicating that the results were consistent for both candidates. Hypotheses 3a and 3b proposed that, regardless of the advertisement type, political cynicism would be positively related to intrinsic attributions of the sponsoring candidate advertisements and would be negatively related to extrinsic attributions of the sponsoring advertisements, respectively. Although results indicated that political cynicism was significantly related to intrinsic attributions of sponsoring candidate motive in support of H3a, no support was found for a relationship between political cynicism and extrinsic attributions. Cynicismasavanablehasoltenbeenusedtorefertoaladtofconfidenceanda feeling of mistrust toward politicians specifimlly (rather than the political system 97 as a whole) (Austin and Pinkleton 1995; Dionne 1991; Perloff and Kinsey 1990; Pinkleton, Um, and Austin 2002; Tedesco 2002). This may have resulted in the differentiation between intrinsic and extrinsic attributions, such as was found in the present case. Although cynicism directly related to increased intrinsic attributions (in effect “blaming” the candidate), cynicism did not relate to extrinsic attributions in which the blame was placed on external factors, such as the political system as a whole. Garramone et al. (1990) has argued that political advertising and cynicism would not be significantly related for two main reasons: first because political advertising may be deemed informative, and second because political advertising may increase involvement. Other researchers have argued that ”blame-placing” messages provide a basis for evaluation of candidate performance and useful criteria in making voting choices so that political information could be viewed as a useful source of criteria for making political choices (Bowen, Stamm, and Clark 2000; Jones and Davis 1965; Lau 1982), rather than contributing or being related to cynicism. This prior research provides one explanation for why cynicism was related to intrinsic attributions, which place the blame on the candidates themselves, rather than an increase in extrinsic attributions, which attribute motives for actions to the overall political process. This finding provides an important note for political consultants and candidates in that as voters become increasingly cynical, there is the tendency to “blame“ the political candidates personally for actions during the campaign, and particularly for the political advertising messages being broadcast. Candidates should 98 remember this propensity for “blame” on behalf of the voting public when preparing political messages. Hypotheses 4a, 4b, and 4c proposed that both intrinsic and extrinsic attributions would affect related attitudes toward the candidates and voting behavior through a mediation process such has been found in prior attributional research in which consequences of the attributions (which include Changes in affect and behavior) proceeded directly from the attributions themselves (Harvey and Weary 1984; Kelley 1973). As hypothesized, when exposed to either Bush or Kerry advertisements, intrinsic attributions of candidate motive were related to voters’ attitudes toward the sponsoring candidate. This supports the belief that voter attitudes toward the sponsoring candidate in the advertising message become increasingly unfavorable when voters endorse intrinsic attributions about the sponsoring candidate, in effect blaming the candidate personally for the advertising message. However, H4b was disconfirmed in that extrinsic attributions of candidate motive did not affect voters’ attitudes toward the candidates. Further, in connection with H4c, while path analyses support a mediation effect of intrinsic attributions on voter attitudes, a similar effect of extrinsic attributions on voter attributes was not found. As discussed in connection with Hypothesis three, it appears that, although voters’ do agree with extrinsic attributions of candidate motive in which they acknowledge that external forces (such as the political process, a particular political party, or PACs) have a role in the content of political advertising messages (see discussion of hypothesis one above), these extrinsic attributions do not affect their attitude toward the 99 sponsoring candidate. The voters in the current research were more likely to attribute responsibility for the advertising content to the politicians themselves, ultimately leading to negative attitudes toward the sponsoring candidate. This finding may suggest that it would be more appropriate for negative political ads to be visibly sponsored by political parties or PACs rather than the candidate themselves due to a lack of correlation between extrinsic attributions and attitudes. In prior research, ads sponsored by groups other than the candidate themselves (is. ‘soft-money' or “issue advocacy” ads) enhanced overall attitudes toward those candidates supported in the advertisements, and they elicited more positive perceptions of those candidates’ competence and character (Pfau et al. 20002). Further research has shown that these ads are seen as more credible and persuasive than candidate-sponsored versions of the same appeal (Groanendyk and Valentino 2002). Hypothesis five supported an oft-found result in political advertising research, that attitudes toward candidates will affect voting behavior. This finding is important to note because, although prior research has indicated a strong correlation between attitudes and intention, it is important that researchers continue to investigate under which conditions such conelations exist. Hypotheses six and seven considered whether individual difference variables, gender and political party affiliation, would moderate the effect of the attributions on voter attitudes and intentions. HypothesesBaandeproposedthat, becausewornenaremorestrongly affected by political advertising and are considerably more likely to blame the 100 sponsoring candidate for the political message, women would be more likely to endorse intrinsic attributions than men, and that a moderating effect of gender would emerge. As expected women were more likely to endorse intrinsic attributions than men when exposed to either positive or negative advertisements, confirming H6a. Of interesting note are the results of H6b, which explored the moderating effect of gender flirough structural equation analysis, and specifically considered whether ad type (positive vs. negative) would interact with gender. Findings indicated that men were more likely than women to endorse intrinsic attributions when both males and females were exposed to the negative advertisements, even though bofll males and females tend to endorse intrinsic attributions when both were exposed to the negative advertisement. At the same time, interestingly, the findings also suggested that, only for males, positive advertisements were more likely to lead to extrinsic attributions than negative advertisements when both males and females were exposed to positive political advertisements. One explanation for this gender gap might be as a result of gender-related differences in socialization (Kern and Just 1997; King and McConnell 2003) or the different political attitudes of women and men as a consequence of different life experiences (Kaid and Hollz-Bacha 2000). Quite often the issues that are focused on in political campaigns are among those that demonstrate gender differences, such as social security in the current campaign. “The keys to any successful advertising campaign—in the political arena or elsewhere—are reaching the right target audience and “pushing the right buttons" (King and 101 McConnell 2003, p. 854). As men and women react differently to the personalitiesfimages of the candidates and to campaign advertising, candidates and campaign managers would be well advised to plan their campaign strategies with these differences in mind. Hypothesis seven further explored an individual difference variable, that of political party affiliation. Partisanship and political party affiliation have often been mentioned as variables impacting voter decisions. In previous studies, the persuasiveness of political ads varied depending on viewers’ political party affiliation (Pfau at al. 2001), with the direction of change in candidate vote choice as a result of exposure to advertising messages highly related to partisanship (Faber, Tims, and Schmitt 1990; Merritt 1984; Robidaaux 2002). Voters’ evaluations of ads were also affected by party affiliation (Robidaaux 1998, 2002). As expected, political party affiliation did moderate the effect of attributions on voter attitudes in both the Bush and Keny models. Further analysis was conducted in order to identify whether intrinsic or extrinsic attributions played a more critical role in moderating the effect of the political party affiliation. On the basis of additional pairwise comparisons it was found that voter attitudes toward the sponsoring candidate, Bush, became significantly unfavorable particularly when Democratic rather than Republican Voters endorsed intrinsic attributions. However, the voter attitude toward Bush was not significantly affected when either Democrats or Republicans endorsed extrinsic attributions. This result was the same for the Kerry model in that voter attitudes toward the sponsoring candidate, Keny, became significantly 102 unfavorable particularly when intrinsic attributions were endorsed by Republican rather than Democratic voters, and not affected when either Democrats or Republicans endorsed extrinsic attributions. Fit of the Mgel Because are results of initial path analyses showed that the hypothesized model did not fit the data, the model was respecified. The current research proposed an alternative model based on the respecification of the original model. In the le-specification process, two new significant paths emerged in the Bush ad model; from political advertising type to voter attitude toward the sponsoring candidate and from intrinsic attributions to voter intention to vote for the candidate (See Figure 5a). These findings suggest that political advertisements (negative vs. positive) affect voter attitude toward the sponsoring candidate directly as well as indirectly through the attributions that voters endorse. Also, these findings suggest that the attributions influence voter intention directly as well as indirectly through the voter attitude toward the sponsoring candidate. However, as opposed to the Bush ad model, adding the same two paths to the Keny-sponsoring candidate model did not improve the model fit (See Figure 5b). This result is interesting in that it points out possible differences between either ( 1) the advertisements run by the candidates, or (2) something fundamentally different beMen Bush/Kerry supporters. Further research should continue to eXplore these differences in an effort to identify possible rationales. 103 Limitations The current research is subject to the usual limitations of experimental research. In the “real world” of political campaigns, voters have many different sources of information about the candidates—television news, newspapers, peer groups, etc., with competing messages transmitted via different channels amid a variety of social and cultural influences. It is improbable that potential voters would see only negative or positive advertising sponsored by one candidate. A more likely occurrence would feature a mix of positive, negative, and comparative political advertising, along with commentary in the news media regarding the accuracy of campaign messages aired by the candidates. In an experimental setting, each of these influences and messages are necessarily limited in order to examine the relationships between independent and dependent variables. This lack of context hinders the external validity of experimental research. By controlling for extraneous variables, important sources of influence and their interaction with other campaign elements are eliminated. The result is an increase in intemal validity at the expense of external validity. A second limitation involves measurement issues. One important dependent variable (Voter Intentions) was measured utilizing a single-item measure. When considering that multi-item measures are more reliable and Valid than single-item measures, the weaknesses of this measure’s reliability and validity arises as one of the limitations of this study. 104 In addition, subject responses to test items cannot be accepted as the equivalent of their political behavior or attitudes. While scaled items may be taken as indications of behavior and attitudes, there is a substantial difference beMen projecting responses onto five-point scales and the actual behavioral and attitudinal outcomes that are likely to result from exposure to political advertising. In addition, these measures are collected at a single point in time, immediately after exposure to experimental stimuli. Such measurement, while a necessary part of the design of the experiment, may not be accurate in its attitudinal and behavioral representation. Suggestions for Future Research Future research examining the application of attribution theory to political advertising should attempt to provide a greater depth of understanding concerning the findings disclosed in this study. Further information needs to be gathered in order to confirm or disconfirrn these results and provide additional information concerning this important field of research. At the most basic level, additional research should be conducted in an attempt to replicate these findings, correcting for the weaknesses existing in this study. An important aspect of replication involves the use of alternative media, such as radio and print, to study political advertising effects. In this instance, an examination of different types of political advertising messages would provide additional information concerning the role of communication modality in political advertising effects. 105 Finally, research efforts should be made that attempt to examine the influences of comparative political advertising, in addition to the standard comparisons of positive and negative political advertising. While some social scientists have suggested that comparative advertising is necessarily negative advertising (Johnson-Caries and Copeland 1989, 1991), it is reasonable to expect that are differences in voter responses among the three. In this instance, a three-way comparison among positive, negative, and comparative advertising would be useful in separating the effects of comparative information from the influences of positive and negative information in political advertising. Conclusion How voting decisions are made and whether political campaigns matter are long-standing questions in the field of politics. Perloff (2002) laments that researchers have not sufficiently probed the mechanisms that mediate political ad effects on candidate attitudes. Although the “minimal effects“ model has had a strong influence on the field, an emerging consensus holds that, given the right conditions, political campaigns and political advertising can have an influence on individual voting behavior and electoral outcomes. One of the aims of the current study was to further explore a theoretical basis for the effects of political advertising on voter attitudes and behavior, namely that of attribution theory. In addition, important mediating and moderating variables were proposed and explored. The primary conclusion that followsfiomtheresearch presented hereisthatattributiontheorycanbeuwdto 106 evaluate voter responses to positive and negative political advertising, and in particular that intrinsic attributions of candidate motive directly affect voters’ evaluations of the sponsoring candidate, while mediating the effects of political advertising on voter attitudes and behaviors. As mentioned above, although decisive conclusions are not drawn to determine the impact of negative and positive political advertising, the findings of the present study have implications for politicians, political consultants, and advertising agencies. The findings raise doubts about the wisdom and value of using negative political advertising in a political campaign. As Merritt (1984) noted, respondents’ negative attitudes toward both candidates and their overall disapproval of negative political advertising resulted from the increasing use of negative political advertising during the 80's, producing cynicism toward politics and declining political participation. Because negative political advertising that identifies the sponsor and the target hurts both candidates, when a candidate uses such advertising, it would be better not to identify the sponsor. However, the law now requires that the sponsor be identified. Garramone (1984) suggested that ”independent political action committees sponsoring negative advertising offer the candidates they help this anonymity advantage. Independent sponsors may contribute the additional benefit of greater credibility" (p. 259). 107 APPENDICES 108 Appendix A Transcripts Bush Positive Ad Full screen shot of George Bush Supertext: WWW.GEORGEWBUSH.COM Full screen shot of George Bush Full screen sot of house. Full screen shot of woman turning on open sign in coffee shop. Full screen shot of older woman and man talking with children. Full screen shot of George Bush. Two shots of workers smiling into camera. Full screen shot of George Bush. Full screen shot of factory workers Full screen shot of George Bush I’m George W. Bush and I approved this message. GEORGE W. BUSH: One of the most important parts of a reform agenda is to encourage people to own something. To own their own home. Own their own business. Own their own health care plan. Own a piece of their retirement. Reforms that trust the people. Reforms that say government must stand on the side of people. Cause I understand that if you own something You have a vital stake in the future of America. Small print: Approved by President Bush and Paid for by Bush-Cheney '04, Inc. Bush Negative Ad Full screen shot of George Bush Supertext: WWW.GEORGEWBUSH.COM Small shot of John Kerry. Supertext: KERRY ECONOMIC RECORD. Medium shot of two elderly man I’m George W. Bush and l approved this message. VOICE OVER NARRATOR: John Kerry’s economic record. Troubling. Kerry voted to increase taxes on social Supertext: TAXES ON SOCIAL SECURITY security benefits. BENEFITS 109 Full amen shot of woman turning on open sign in coffee shop. Supertext: OPPOSED TAX CREDITS FOR HEALTH CARE. Full screen shot of person pumping gas Supertext: SUPPORTED 50¢IGAL GAS TAX. Medium shot of young couple. Supertext: RAISE TAXES. Title: $900 BILLION Small screen shot of John Kerry. Supertext: THE FIRST 100 DAYS. $900 BILLION. And he voted against giving small businesses tax credits to buy health care for employees. Kerry even supported raising taxes on gasoline. 50 cents a gallon. Now John Keny’s plan will raise taxes by at least $900 billion dollars his first 100 days in office. And that’s just his first 100 days. Small print: Approved by President Bush and Paid for by Bush-Cheney ’04, Inc. Kerry Positive Ad Medium shot of John Keny Supertext: JOHN KERRY Supertext: KEEP AMERICA SECURE Supertext: DEFEND AMERICAN JOBS Title: LEARN MORE ABOUT JOHN KERRY’S PLAN FOR AMERICA JohnKerry.com JOHN KERRY [to camera]: As President, I'll set a few clear national priorities for America. First, we will keep this country safe and secure. Second, I'll put an end to tax incentives that encourage American companies to ship jobs overseas. VOICE OVER: And third, we’ll invest in education and healthcare. KERRY: My priorities are jobs and healthcare. My commitment is to defend this country. I'm John Kerry and I approved this message because together we can build a stronger America. Small print: Approved by John Kerry and paid for by John Kerry for President Kerry Negative Ad Closeup of small girl swinging on swing. Small shot of George Bush superimposed over picture of factory VOICEOVER NARRATOR: Under George Bush and Right Wing Republicans we’ve 110 Supertext: JOBS LOST Supertext: GEORGE BUSH AND RIGHT WING REPUBLICANS: WORST RECORD SINCE HOOVER Shot of woman looking into camera Supertext: NEW JOBS PAY $9000 LESS. Closeup of small girl swinging on swing. Shot of factory worker looking into camera. Supertext: HEALTH CARE COSTS SKY ROCKET. Screen shot of older couple Supertext: HIGHER DEDUCTIBLES AND CO-PAYS Screen shot of person pumping gas Supertext: GAS PRICES SOAR Small shot of George Bush superimposed over picture of office Supertext: GEORGE BUSH AND RIGHT WING REPUBLICANS Screen shot of factory worker on the line Supertext: EXPORT JOBS Closeup of small girl swinging on swing Screen shot of Enron logo Supertext: CORPORATE I-IANDOUTS Swen shot of couple with baby Supertext: SQUEEZE THE MIDDLE CLASS Closeup of small girl swinging on swing Small shot of John Kerry Supertext: IT'S TIME FOR A NEW DIRECTION lost 2.7 million manufacturing jobs. The worst jobs record since Herbert Hoover. New jobs pay $9000 less. Health care costs skyrocket. Higher deductibles and co-pays. Gas prices soar. George Bush and Right Wing Republicans Give tax breaks for companies that export jobs. Handouts to Hallaberton and Enron But they put the squeeze on the middle class. It’s time for a new direction. l’m John Kerry and I approved this message. Small print Approved by John Kerry and paid for by John Kerry for President —_* 111 Appendix B Instrument WELCOME. SELECT YOUR ANSWERS BY CLICKING ON THE APPROPRIATE CIRCLE NEXT TO EACH QUESTION. Pollflcal Cynicism PhasekubabmyouagreemdbagmemeadldmefdangWSebamW byclickingontheappropriatecirclenexttoeachquestion. M . W M” Disagree Disagree AgreeIDisagiee Disagree Disagree 1 2 3 4 5 Whetherlvoteornothasnoinfluenceon O O O O O whatpoliticiansdo Oneneverreallyknowswhatpoliticiansthink O O O O O Peoplelikemedon'thaveanysayaboutwhat O O O O O thegovemmentdoes Sometimespoliticeandgovemmentseemso O O O O O complicatedthetapersonlikemecan'treally understandwhat’sgoingcn Onecanbeconfidentthatpoliticianswillalways O O O O O dotherightthing Politiciansoftenqdcklyfomettlniralection O O O O O promisesalterapolitlcalcampaignisover Politiciansaremoreinterestadinpowerthanin O O O O O whatthepeoplethink Onecannotalwaystrustwhatpoliticlanssay O O O O O AttitudcsaboutAdvcrIisIng Belowisasetofwordpairs. PIaasadidrflrecirdedoseubuleadjectivewhidryoubefievebastreflacts wurbdingsabmnadmtsiminm.memapme8ntadjewwm,memmm adjectiveyoucickthecircle. Advertising In general Good 0 O O O 0 Bad Unpleasant O O O O 0 Pleasant Favorable O O O O O Unfavorable Unconvinclng O O O O O Convincing Believable O O O O O Unbelievable Biased O O O O O Unbiased 112 CandldateEvaluatIons Belowisasetofword pars. Whenthinkingaboutthechubllcanprccldantialcandldab,GcorgeW. Bush.clickthecirdeclosedtotheadjectivewhichyoubelievedescribesthecandidatebetter. 'Ihemore appropriatethatadjediveseems.theclosertotheadjectiveyou clickthecircle. GEORGE W. BUSH Qualified O O 0 O O Unqualified Sophisticated O O O O O Unsophisticated Honest O O O O O Dishonest Sincere O O O O O lnsincera Successful O O O O O Unsuccessful Attractive O O O O O Unattractive Calm O O O O O Excitable Aggressive O O O O O Unaggressive Strong 0 o o o 0 Weak Passive O O O O 0 Active Friendy o o o o o Unfriendly Believable O O O O O Unbelievable Unconvinclng O O O 0 O Convinc'mg Biased O O O O O Unbiased Belowisanothersetofword pairs. Now.whenthinldngabouttheDcmocratlcprccldcntialccndldatc, mKuwCidtmedrdedmmunadjedimwhidlwubelbvedesaibamcmdmm. The mmappropnatefliatadjecfiveseummiedosertomeadjecfiveyoudickmedrde. JOHN KERRY Qualified O O O O O Unqualified Sophisticated o o o o o Unsophisticated Honest O O O O O Dishonest Sincere O O O O O lns'moere Successful O O O O O Unsuccessful Attractive O O O O O Unattractive Calm O O O O O Exoitabla Aggressive o o o o o Unaggressive Strong 0 o o 0 0 Weak Passive O O O 0 0 Active 113 Friendy O O O O O Unfriendly Believable O O O O 0 Unbelievable Unconvincing O O O O O Convincing Biased O O O O O Unbiased Wehavesomefurtherquestionsaboutspecificcendidates.Wewould liketoknowhowyouperceivelhem. lnthefollcwingquestions.youwillseehenameofacandidateandflenmadasuiesofstatennntsaborfl thatcandidate. Please mdicatemryouagmeudisagmewmieadidflnfolowhgstamfiebd yoursnswersbydickingontheappropfidecirclenexttceachquesfion. GEORGE W. BUSH Disagree Disagree AgreeIDisagree Disagree Disagree 1 2 3 4 5 'lhecandidaiehasagreatamountof O O O O O lbustthecendidate O 0 O O O Thecenddateisskilledlnwhathedoes O O O O 0 Thecandidaiehasgreetexpertise 0 O O O O Thecandidaieishonest O O O O O Thecendidatedoesnothavemuch O O O O 0 experience. ‘l'hecandidaiemakestmthfulclaims O O O O O ldonotbelievewhetthecendidatebllsme O O O O O JOHNKERRY Strongly Neither . Strongly Disagree Disagree AgreelDisagreeDisagreeDlsagree 1 2 3 4 5 Thecandidatehasagrealemountol O O O O 0 experience. ltrustthecandldate O O O O O 'l'hecandidateissk‘lledinwhalhedoes O O O O O Thecendidatehasgreatexperiise O O O 0 O Thecendidefeishonest O O 0 0 O Thecendidatedoesnothavemuch O O O O 0 experience. ‘l'hecendidaiemakestruthfulclairm O O O O O Idonotbelievewhetthecandidatetellsme O O O O O 114 Welcomebeclt. flunexteetofquesdonshsstodowlththecommerclalyoujustviewed. VoterAttrlbutions Pleasereadeachstatementandselectyouranswersbyclickingontheappropriatecirdenexttoeach question. Strongly Neither Strongly Disagree Disagree Agree/Disagree Disagme Disagree 1 2 3 4 5 Thecendidateranthiscommercialbecause O O O O O hewantsthevoterstoknowallthefacts abouttheissues. Thecandidateranthiscommereialto O O O O O discredittheopposingcandidate. Thecandidateranthiscommercialbeceuse O O O O O hecaresaboutthecountry. The candidate ran this commercial because 0 0 O O O hedoesn’twanttheopposingcendidateto wintheelection. fliecandidateranthiscommercialbeceuse O O O O O heistryingtomislewmevotersabouttheissues. Niecendidateranthiscormiercialto O O O O O pereuademetovoteforhlm. Thecandidateranthiscommercialbeceuse O O O O O hebeiievesheisthebestpersonfortheoflice ofPresident. fliecendidateranthiscommercialbecause O O O O O hewantsthevoterstoquestionordwbt theopposingcandidate. Thecendidate ran this commercial because 0 O O O O aPACpressuredhimdolt. ‘lhecendidateranthiscommercialto O O O O O respondtoalegetionsmadebythe opposingcandidate. Thecendidateranthiscommercialbeceuse 0 O O O O the opposing candidate made misleading statements that had to be corrected. 'l'hecandidateranthiscommercialto O O O O O discussanissuethatvotersthinkis Wm Thecandideteranthiscommercial O O O O O becauseaPACwasattackinghim. Themdidateranthiscommercial O O O O O becausehewasbehindinthepolls. Thecanddaterenthlscommerdal O O O O O becarisehispoliticalpenywanted himtodoit. 115 Themdidateranthiscommercial O O O O O totellthevoterswhattheywanmd toheer. Thecendidateranthiscommercial O O O O O becausehewantsthepowerofthe presidencyandwillsayanything togetthere. AttitudetlovvarddleCommerclal Belowisasetofword pairs. Pleasedidflredrdedosedtoflmeadjecfimwhidwoubelievebestreflects yourfeelingsaboutthecommercialyoujustsaw.1hemoreappropriatethatadjectiveseems.lhecloserto theadjectiveyouclickthecircle. The Commercial Informative 0 O O O O Uninfonnative Believable O O O 0 O Unbelievable Persuasive O O O O O Unpersuasive Like 0 O O O O Dislike Pleasant o o o o o Unpleasant Truthful O O O O O Deceptive Accurate 0 O O O 0 Inaccurate Ethical O O O O O Unethical Good 0 O O O 0 Bad Candidate Evaluations Belowisasetofwordpairs. WMWMMWHWQMWW. Buh.didrmedrdedoeeflmmeadjecfiwwhidlwubdievedesaibesmmbedm Themore appropriatematariecfiveseansjledosermmeadjeaiveyoudickmedrde. GEORGE W. BUSH Qualified O O 0 O 0 Unqualified Sophisticated O O O O O Unsophisticated Honest 0 O O O O Dishonem Sincere O O O O O lnsincele Successful O O O 0 0 Unsuccessful Attractive 0 O 0 O O Unattracfive Cairn O O O O O Exdtable Aggressive O O O O O Unaggresslve Strong 0 o o o 0 Weak Passive 0 O O 0 0 Active 116 Friendly 0 O O O O Unfriendly Believable O O O O O Unbelievable Unconvincing O O O O O Convincing Biased O O O O O Unbiased Below is another set of word pairs. Now. when thinking about the Democratic presidential candidate, John Keny.didtdndrdedosestmmeadjecfivewhidiywbdievedescnbesmmidate better. The more appropriate that adjective seems, the closer to the adjective you click the circle. JOHN KERRY Qualified O O O O O Unqualified Sophisticated O O O O O Unsophisticeted Honest O O O O O Dishonest Sincere O O O O O lnsincera Successful O O O 0 O Unsuccessful Attractive O O O O O Unattractive Calm O O O 0 O Excitable Aggressive 0 O O O O Unaggressive Strong 0 o o o 0 Weak Passive O O O O 0 Active Friendly 0 o o o o Unfriendly Believable O O O O O Unbelievable Unconvincing O O O O O . Convincing Biased O O O O O Unbiased Wehavesomefurtherquestionsaboutspeclficcandidates.Wewouldliketolmowhowyoupereeivethem. In thefollowing questions, you will seethe name ofa candidate and then read a series ofstatements about that candidate. Please indicate whetheryou agree ordisagree with each ofthe following statements. Select youranswersbydicldngontheappropriatecirclenexttoeachquestion. GEORGE W. BUSH Strongly Neither . Strongly Thecandidatehasagreatamountof O O O O 0 experience. ltrustthecendidate O O O O O Thecandidateisskilledinwhathedoes O O 0 O O The candidate has great expertise 0 O O O O 117 Thecendidateishonest O O O O O Thecandidatedoesnothavemuch O O O O 0 experience. Thecandidatemakestruthfulclaims O O O O O ldonotbelievewhatthecandidatetellsme 0 O O O O JOHNKERRY Strongly Neither Strongly Disagree Disagree Agree/Disagree Disagree Disagree 1 4 ‘l'hecandidatehasagreatamountof O O O O 0 experience. ltrustthecendidfle O O O O O Thecendidateisskilledinwhathedoes O O O O O Themdidatehasgreatexpertise O O O O O Thecandidateishonest O O O O O 'lhecendldatedoesnothavemuch O O O O 0 experience. The candidate makes truthful claims 0 O O O O ldonotbelievewhatlhecandidatetellsme O O O O 0 Political Involvement Belowisasetofstatementpa’rs. Pleasedickhecircleclosedtothestatennntwhflyoubelievebest reflectsyourfeelings. lsnewsaboutpoliticssomethingyoutrytopeyattentionto.orisitsornethingyoujusthappentoleemabout becauseitisinthemedia? ltrytopayattentiontopolitics O O O 0 O Politicsisjustsomethmgl leamaboutbecauseitisin thernedia. lspdificssonnflfingyouikebtakabmfiadoyouaiydbansflfisaneomdsebfingsiup? lliketomlraboutpolitics O O O O O Ionlydiscusspoliticsif someoneelsebringsitup. HowdoselyhaveyoufollowedthecurrentUSOpreddentialrace? Veryclosely O O O O Notatallclosely HowconcemedaleyouwithwhowinstheraceforthePresident‘? Veryconcerned O 0 0 O O Notatallconcemed Voterlntentions Howlikelyisitthatyouwillvoteforrepublicenpreudendalcandidate, GeorgeW. Bush? 118 Howlikelyisitthatyouwillvoteiordemocraticpresidentialcandidate. John Kerry? 0 Very likely O Likely O Undecided O Unlikely 0 Very unlikely lftheZOO-t Presidential Electionwere heldtoday.whowould you votefor? 0 George W. Bush 0 John Keny 0 Ralph Nader O Undecided Howsatisfied areyouwithyomchoiceforPresident? 0 Very satisfied 0 Satisfied 0 Neither satisfied nor dissatisfied 0 Dissatisfied 0 Very dissatisfied AboutYou. Nowjustafewpersonalquationstohelpusclassilyyourresponses. Areyou? OFemale OMale Whatisyourage? Whatisyourcurrentmaritalstatus? OSingle OManied ODivorced OWidowed What is your occupation? 0 Professional 0 White Collar 0 Blue Collar 0 Student 0 Retired 0 Other Whatracialorethnicgroupbeddescribesyou? O Caucasian O African-American 0 Asian 0 Hispanic 0 Native American 0 Other Gemfiyspeaking.dowuuwalwmmkdwundfaaRepublben.aDunwatmIMependm.m somethingelse? ORepublican ODemocrat Olndependent OOtller Howstrongisyourattachmenttoyourpoliticelparty? Verysb'ong O O O O O Notverystrong Thank you for participating in this study. 119 REFERENCES 120 References Allen, Mike and Nancy Burell (2002), “The Negativity Effect in Political Advertising: A Meta-Analysis,” in The Persuasion Handbook: Theory and Practice, James Price Dillard and Michael Pfau, eds., Thousand Oaks, CA: Sage, p. 83-98. Althaus, S.L. and D. Tewksbury (2000), “Patterns of Internet and Traditional News Media Use in a Networked Community,’ Political Communication, 17, 21-45. Ansolabehere, Stephen, Roy Behr, and Shanto Iyengar (1993), The Media Game, New York: Macmillan. Ansolabehere, Stephen and Shanto Iyengar (1994), “Riding the Wave and Claiming Ownership over Issues,” Public Opinion Quarterly, 58, 335-357. —-——-—-— and —-—-—-—- (1995), Going Negative: How Political Advertisements Shrink & Polarize the Electorate, New York: The Free Press. ,Adam Simon, and Nicholas Valentino (1994), “Does Attack Advertising Demobilize the Electorate?“ American Political Science Review, 88 (December), 829-838. Atkin, Charles K. and G. Heald (1976), “Effects of Political Advertising,“ Public Opinion Quarterly, 40, 218-228. Arbuckle, J. L. (2003), Amos 5.01 [Computer software]. Chicago: Smallwaters. Banwart, Mary Christine, Dianne G. Bystrom, and Teny Robertson (2003), “From the Primary to the General Election: A Comparative Analysis of Candidate Media Coverage in Mixed-Gender 2000 Races for Governor and US. Senate,” The American Behavioral Scientist, 46 (5), 658-878. Belch, George E. (1981), “An Examination of Comparative and Noncomparative Television Commercials: The Effects of Claim Variation and Repetition on Cognitive Response and MW Acceptance,” Journal of Marketing Research, 18(3), 333-349. Benoit, Wllliam L., P.M. Pier, and Joseph R. Blaney (1997), “A Functional Approach to Televised Political Spots: Acclaiming, Attacking, Defending,” Communication Quarterly, 45(1), 120 Best, Samuel J. and Brian Krueger (2002), “New Approaches to Assessing Opinion: The Prospects for Electronic Mail Surveys,” lnlemational Journal of Public Opinion Research, 14 (1 ), 73-92. 121 Bollen, Kenneth A. (1989), Structural Equations for Latent Variables, New York: John Wlley & Sons. Bowen, Lawrence (1994), “Time of Voting Decision and Use of Political Advertising,” Journalism Quarterly, 71 (3), 665-675. ---——--, Keith Stamm, and Fiona Clark (2000), “Television Reliance and Political Malaise: A Contingency Analysis,“ Journal of Broadcasting & Electronic Media, 44 (1), 1-15. Bowers, T. A. (1973), “Newspaper political advertising and the agenda-setting function,” Journalism Quarteriy, 50, 552-556. Brians, Craig Leonard and Martin P. Wattenberg (1996), “Campaign Issue Knowledge and Salience: Comparing Reception from TV Commercials, TV News, and Newspapers," American Joumal of Political Science, 40, 172-193. Budesheirn, Thomas L., David A. Houston, and Stephen J. DePaola (1996), “Persuasiveness of in-group and out-group political messages: the case of negative political campaigning,“ Journal of Personality and Social Psychology, 70 (3), 523-534. Calder, Bobby J. and Robert E. Bumkrant (1977), “Interpersonal Influence on Consumer Behavior. An Attribution Theory Approach,’ Journal of Consumer Research, 4 (1 ), 29-38. Campbell, Donald T. and Julian C. Stanley (1963), Experimental and Quasi- Experimental Designs for Research, Boston: Houghton Mifflin Company. Campbell, Margaret C. (1995), "When Attention-Getting Advertising Tactics Elicit Consumer lnferences of Manipulative Intent: The Importance of Balancing Benefits and Investments," Journal of Consumer Psychology, 4 (3), 225- 54. Cappella, J.N. and K.H. Jamieson (1997), Spiral of Cynicism: The Press and the Public Good, New York: Oxford University Press. Celsi, Richard and Jeny C. Olson (1988), “The Role of Involvement in Attention and Comprehension Processes,” Journal of Consumer Research, 15 (September), 210-224. Center for Responsive Politics (2004), “2004 Presidential Election,” www.0pensecretsorg. 122 Chaffee, S.H., X. Zhao, and G. Leshner (1994), “Political Knowledge and the Campaign Media of 1992,“ Communication Research, 21 (June), 305-324. Chaiken, Shelly (1980), “Heuristic versus Systematic lnfonnation Processing and the Use of Source versus Message Cues in Persuasion,“ Journal of Personality and Social Psychology, 39 (5), 752-766. Chaney, Carol Kennedy, R. Michael Alvarez, and Jonathon Nagler (1998), “Explaining the Gender Gap in US. Presidential Elections, 1980—1992,“ Political Research Quarterly, 51, 311-339. Chang, Chingching (2003), “Party Bias in Political-Advertising Processing,“ Journal of Advertising, 32(2), 55-67. Coulter, Robin H. and Mary B. Pinto (1995), ”Guilt Appeals in Advertising: What Are Their Effects?" Journal of Applied Psychology, 80 (December), 697- 705. Crawford, Darlisa (2004, May 21), “Television Primary Information Source for Most 2004 Voters,” available at http://usinfostategov. Crotty, Wllliam J. and Gary C. Jacobson (1980), American Parties in Decline, Boston: Little, Brown, and Company. Davis, Joel J. (1994), “Good Ethics is Good for Business: Ethical Attributions and Response to Environmental Advertising,“ Journal of Business Ethics, 13 (1 1), 873-885. Dean, Dwane Hal (2002), “Associating the Corporation with a Charitable Event Through Sponsorship: Measuring the Effects on Corporate Community Relations,“ Joumal of Advertising, 31 (4), 77-87. -——--—- (2003), “Consumer Pemption of Corporate Donations,“ Journal of Advertising, 32(4), 91-102. Dionne, E.J., Jr. (1991), Why Americans Hate Politics, New York: Simon and Schuster. Eagly, Alice H., Wendy Wood, and Shelly Chaiken (1978), “Causal Inferences about Communicators and their Effect on Opinion Change,“ Journal of Personality and Social Psychology, 36 (4), 424-435. Faber, Ronald J. (1992), “Advances in Political Advertising Research: A Progression from If to When,“ Journal of Current Issues and Research in AdVertising, 14 (2), 1-18. 123 , Albert R. Tims, and Kay G. Schmitt (1990), “Accentuate the Negative?: The Impact of Negative Political Appeals on Voting Intent,“ in Proceedings of the American Academy of Advertising, Patricia Stout, ed., Austin, TX: American Academy of Advertising, 10-16. —----—-, and -———-—-- (1993), “Negative Political Advertising and Voting Intent: The Role of Involvement and Alternative Information Sources, Journal of Advertising, 22(4), 67-77. Finkel, Steven E. and John Geer (1998), “A Spot Check: Casting Doubt on the Demobilizing Effect of Attack Advertising,“ American Journal of Political Science, 42 (2), 573-595. Fisher, 8., M. Margolis, and D. Resnick (1996), “Surveying the Internet: Democratic Theory and Civic Life in Cyberspace,“ Southem Political Review, 24, 399-429. Fiske, Susan and Shelley Taylor (1991), “Attribution Theory,“ In Social Cognition, McGraw-Hill, 22-56. Folkes, Valerie S. (1988), “Recent Attribution Research in Consumer Behavior: A Review and New Directions,“ Journal of Consumer Research, 14 (March), 548-565. Freedman, P. and K. Goldstein (1999), “Measuring media exposure and the effects of negative campaign ads,“ American Journal of Political Science, 42 (4), 1189-1208. Friestad, Marian and Peter Wright (1994), “The Persuasion Knowledge Model: How People Cope with Persuasion Attempts,“ Journal of Consumer Research, 21 (1), 1-31. Garramone, Gina M. (1984), “Voter Response to Negative Political Ads,“ Joumalism Quarterly, 61 (2), 250-259. , Gina M. (1985), “Effects of Negative Political Advertising: The Roles of Sponsor and Rebuttal,“ Journal of Broadcasting and Electronic Media, 29, 147-159. , Charles K. Atkin, Bruce E. Pinkleton, and Richard T. Cole (1990), “Effects of Negative Political Advertising on the Political Process,“ Journal of Broadcasting and Electronic Media, 34 (3), 299-311. Geer, J. and Richard R. Lau (1998), “A New Way to Model Campaign Effects,” Paper presented at the annual meeting of the American Political Science Association, Boston, August. 124 Gom, Gerald J. and Charles B. Weinberg (1984), “The Impact of Comparative Advertising on Perception and Attitude: Some Positive Findings,“ Journal of Consumer Research, 11 (2), 719-727. Gotlieb, Jerry B. and Dan SareI (1991), “Comparative Advertising Effectiveness: The Role of Involvement and Source Credibility,“ Journal of Advertising, 20 (1), 38-45. Groenendyk, Eric W. and Nicholas A. Valentino (2002), “Of Dark Clouds and Silver Linings: Effects of Exposure to Issue Versus Candidate Advertising on Persuasion, lnforrnation Retention, and Issue Salience,“ Communication Research, 29(3), 295-319. Hair, Joseph F. Jr, Ralph E. Anderson, Ronald L. Tatharn, and Wllliam C. Black (1995), Multivariate Data Analysis with Readings, 4‘" edition, New York: Prentice Hall. Hall, Alice and Joseph N. Cappella (2002), “The Impact of Political Talk Radio Exposure on Attributions about the Outcome of the 1996 US. Presidential Election,“ Journal of Communication, 52 (2), 332-350. Harper, Jennifer (2004, October 21), “Campaign Spending Nears $4 Billion, A Record Level,“ The Washington Times, www.washinqtontimes.com. Harvey, John H. and Gifford Weary (1984), “Current Issues in Attribution Theory and Research,“ Annual Review of Psychology, 35 (February), 427-459. Heider, Fritz (1946), “Attitudes and Cognitive Organizations,“ Joumal of Personality, 21, 107-112. --——--- (1958), The Psychology of Interpersonal Relations, New York: Wlley. HiII, Ronald Paul (1989), “An Exploration of Voter Responses to Political Advertisements,“ Journal of Advertising, 18 (4), 14-22. Hindman, Elizabeth Blanks (2003), “The Princess and the Paparazzi: Blame, Responsibility, and the Media’s Role in the Death of Diana,“ Journalism and Mass Communication Quarterly, 80 (3), 666-688. Hitchon, Jacqueline C. and Chingching Chang (1995), “Effects of Gender Schematic Processing on the Reception of Political Commercials for Men and Women Candidates,“ Communication Research, 22 (4), 430-58. ,and Rhonda Harris (1997), “Should Women Emote? Perceme Bias and Opinion Change In Response to Political Ads for Candidates of Different Genders, Political Communication, 14, 49-69. 125 Hoffner, Cynthia, Richard S. Plotkin, Martha Bushanan, Joel David Anderson, Stacy K. Kamigaki, Lisa A. Hubbs, Laura Kowalczyk, Kelsey Silberg, and Angela Pastorek (2001), “The Third-Person Effect in Perceptions of the Influence of Television Vlolence,“ Journal of Communication, 51 (2), 283- 299. Holbert, R.L., William L. Benoit, G.J. Hansen, and W-C. Wen (2002), “The Role of Communication in the Formation of an Issue-Based Citizenry,“ Communication Monographs, 69, 296-310. Homer, Pamela M. and Rajeev Batra (1994), “Attitudinal Effects of Character- Based Versus Competence-Based Negative Political Communications,“ Journal of Consumer Psychology, 3 (2), 163-185. Houston, D.A., Doan, K. and RoskocEwoldsen, D. (1999), “Negative political advertising and choice conflict,“ Journal of Personality and Social Psychology, 5(1), 3-16. Hunter, J.E. and D. W Gerbing (1982). “Machiavellian beliefs and personality: Construct invalidity of the machavellianism dimension,“ Journal of Personality and Social Psychology, 43, 1293-1305. Iyengar, Shanto (1987), “Television News and Citizens’ Explanations of National Affairs,“ Political Science Review, 81, 815-831. ---—--— (1990), “Framing Responsibility for Political Issues: The Case of Poverty,“ Political Behavior, 12, 19-40. —-----— (1991), Is Anyone Responsible? How Television Frames Political Issues, Chicago: University of Chicago Press. —-——--- and DR. Kinder (1987), News that Matters: Television and American Opinion, Chicago: University of Chicago Press. Jain, Shailendra P., Bruce Buchanan, and Durairaj Maheswaran (2000), “Comparative Versus Noncomparative Advertising: The Moderating Impact of Pre-Purchase Attribute Verifiability,“ Journal of Consumer Psychology, 9(4), 201-212. Jain, Shailendra P. and Steven S. Posavac (2004), “Valenced Comparisons,“ Journal of Marketing Research, 41 (February). 46-58. Jamieson, Kathleen H. (2000), Everything You Think You Know about Politics: Any Why You’re Wrong, New York: Basic Books. 126 —----—-, Paul Waldman, and Susan Sherr (1998), “Eliminate the Negative? Defining and Refining Categories of Analysis for Political Advertisements,“ Paper delivered at the Conference on Political Advertising in Election Campaigns, Washington, DC. Jasperson, Amy E. and David P. Fan (2002), “An Aggregate Examination of the Backlash Effect in Political Advertising: The Case of the 1996 US. Senate Race in Minnesota,“ Journal ofAdvertising, 31 (1), 1-12. Johnson-Carlee, Karen S. and Gary A. Copeland (1989), “Southern Voters” Reactions to Negative Political Ads in the 1986 Election,“ Joumalism Quarterly, 66(4), 888-893. ------ and ——-——- (1991), Negative Political Advertising: Coming of Age, Hillsdale, NJ: Lawrence Erlbaum. Jones, Edward E. and Keith Davis (1965), “From Acts to Dispositions: The Attribution Process in Person Perception,“ in Advances in Experimental Social Psychology, Vol. 2, Leonard Berkowitz, ed., New York: Academic Press, 219-266. Joreskog, Kari G. and Dag Sorbom (1993), Lisrel 8: Structural Equation Modeling with SIMPLIS Command Language, Chicago: Scientific Soflware lntemational. Kahn, Kim Fridkin and John G. Greer (1994), “Creating Impressions: An Experimental Investigation of Political Advertising on Television,“ Political Behavior, 16 (1), 93-116. Kahn, Kim Fridkin and P.J. Kenney (1999), “Do Negative Campaigns Mobilize or Suppress Turnout? Clarifying the Relationship between Negativity and Participation,“ American Political Science Review, 93 (4), 877-889. Kaid, Lynda Lee (1981), “Political Advertising,“ in Handbook of Political Communication, 0.0. Nimmo and KR. Sanders, eds., Beverly Hills, CA: Sage, 249-271. ---—-— (1994), “Political Advertising,“ in Handbook of Political Communication Research, Lynda Lee Kaid, ed., New Jersey: Lawrence Erlbaum Associates, 155-202. ————-—— (1999), “Political Advertising: A Summary of Research Findings,“ The Handbook of Political Marketing, Thousand Oaks, CA: Sage, 423-438. 127 -—----— (2002), “Political Advertising and Information Seeking: Comparing Exposure via Traditional and Internet Channels,“ Journal of Advertising, 31 (1), 27-35. (2004), “Political Advertising,“ in Handbook of Political Communication Research, Lynda Lee Kaid, ed., New Jersey: Lawrence Erlbaum Associates, 155-202. ----—-— and Christina HoItz-Bacha (2000), “Gender Reactions to TV Political Broadcasts,“ Harvard Intemational Journal of Press/Politics, 5 (2), 17-29. Kaid, Lynda Lee and A Johnston (1991), “Negative versus Positive Television Advertising in US Presidential Campaigns, 1960-1988,“ Joumal of Communication, 41 (3), 53-64. Kaid, Lynda Lee, Chris M. Leland, and Susan Whitney (1992), “The Impact of Televised Political Ads: Evoking Viewer Responses in the 1988 Presidential Campaign,“ Southern Communication Journal, 57 (4), 285- 292. Kaid, Lynda Lee, Mitchell McKinney, and John C. Tedesco (2000), Civic Dialogue in the 1996 Presidential Campaign: Candidate, Media, and Public Voices, Cresskill, NJ: Hampton Press. Kelley, Harold H. (1972), “Causal Schemata and the Attribution Process,“ in Attribution: Perceiving the Causes of Behavior, Edward E. Jones, ed., Morristown, NJ: General Learning Press, 151-174. -—--—-- (1973), “The Process of Causal Attribution,“ American Psychologist, 28 (February). 107-128. -----— and John L Michaela (1980), “Attribution Theory and Research,“ Annual Review of Psychology, 31 (February). 457-501. Kelloway, E. K. (1998), Using LISREL for structural equation modeling: A researchers guide, Thousand Oaks, CA: SAGE Publications, Inc. Kern, Montague (1989), 30-Second Politics: Political Advertising in the Eighties, New York: Praeger. —---—-- and Marion Just (1997), “A Gender Gap Among Vlewers?“, In Women, Media, and Politics, Pippa Nonis, ed., New York: Oxford University Press. Kemall, S. (1977), “Presidential Popularity and Negative Voting: An Altemative Explanation of the Mid-Term Congressional Decline of the President’s Party,“ American Polih’cal Science Review, 71, 44-66. 128 Kinder, DR. and L.M. Sanders (1990), “Mimicking Political Debate with Survey Questions,“ Social Cognition, 8, 73-103. Kline, R. B. (1998), Principles and practice of structural equation modeling. New York: NY, The Guilford Press. King, James D. and Jason B. McConnell (2003), “The Effect of Negative Campaign Advertising on Vote Choice: The Mediating Influence of Gender,“ Social Science Quarterly, 84 (4), 843-857. KoItz, Robert (1998), “Vlrtual Criticism: Negative Advertising on the Internet in the 1996 Senate Races,“ Political Communication, 15 (3), 347-365. Lariscy, Ruthann Weaver, Spencer F. Tlnkham, Heidi Hatfield Edwards, and Karyn Ogata Jones (2004), “The “Ground War“ of Political Campaigns: Nonpaid Activities in US. State Legislative Races,“ Journalism & Mass Communication Quarterly, 81 (3), 477-497. Lau, Richard R. (1982), “Negativity in Political Perception,“ Political Behavior, 4, 357-377. -——-——— (1985), “Two Explanations for Negativity Effects in Political Behavior,“ American Journal of Political Science, 29, 1 19-138. -———---, Lee Singelman, Caroline Heldman, and Paul Babbitt (1999), “The Effects of Negative Political Advertisements: A Meta-Analytic Assessment,“ The American Political Science Review, 93 (4), 851-875. Lazarsfeld, P.F., B. Berelson, and H. Gaudet (1948), The People’s Choice: How the Voter Makes up his mind in a Presidential Election, New York: Columbia University Press. Lee, Betty Kaman (2004), “Audience-Orienmd Approach to Crisis Communication: A Study of Hong Kong Consumers“ Evaluations of an Organizational Crisis,“ Communication Research, 31 (5), 600-618. LePine JA and L. Van Dyne (2001), “Peer responses to low performers: An attributional model of helping in the context of groups,“ Academy of Management Review, 26, 67-84. The Living Room Candidate (2004), “2004 Bush vs. Keny,“ http://|ivingroomcandidate.movingqimage.US. MacKenzie, Scott B. and Richard J. Lutz (1989), “An Empirical Examination of the Structural Antecedents of Attitude Toward the Ad in an Advertising Protesting Context,“ Journal of Marketing, 53 (April), 48-65. 129 Mattei, Laura R. and Franco Mattei (1998), “If Men Stayed Home: The Gender Gap in Recent Congrcssional Elections,“ Political Research Quarterly, 51, 41 1-436. ’ McClure, Robert and Thomas Patterson (1974), “Television News and Political Advertising,“ Communication Research, 1 (January). 3-31. Media Studies Center (2000), The Media and Campaign 2000, New York: The Media Studies Center. Meirick, Patrick (2002), “Cognitive Responses to Negative and Comparative Political Advertising,“ Journal of Advertising, 31 (1 ), 49-59. Merritt, Sharyne (1984), “Negative Political Advertising: Some Empirical Findings,“ Joumal of Adverh'sing, 13 (3), 27-38. Miniard, Paul W., Sunil Bhatla, Kenneth R. Lord, Peter R. Dickson, and H. Rao Unnava (1991), “Picture-Based Persuasion Processes and the Moderating Role of Involvement,“ Journal of Consumer Research, 18 (1 ), 92-107. Mizerski, Richard W. (1978), “Causal Complexity: A Measure of Consumer Causal Attribution,“ Journal of Marketing Research, 15 (2), 220-228. ----——-, Linda L. Golden, and Jerome B. Keman (1979), “The Attribution Process in Consumer Decision Making,“ Journal of Consumer Research, 6(2), 123-140. Momitz, Vicki G. and Carol Pluzinski (1996), “Do Polls Reflect Opinions or Do Opinions Reflect Polls? The Impact of Political Polling on Voters“ Expectations, Preferences, and Behavior,“ Journal of Consumer Research, 23 (June), 53-67. Newman, Bruce I. and Jagdish N. Sheth (1985), “A Model of Primary Voter Behavior,“ Joumal of Consumer Research, 12 (September), 178-187. Newell, Stephen J. and Ronald E. Goldsmith (2001), “The Development of 3 Scale to Measure Perceived Corporate Credibility,“ Journal of Business Research, 52 (3), 235-247. O’Cass, Aron (2002), “Political Advertising Believability and lnfonnation Source Value During Elections,“ Journal of Advertising, 31 (1), 63-74. Patterson, Thomas E. and Robert D. McClure (1974), Political Advertising: Voter Reactions to Televised Political Commercials, Princeton, NJ: Citizen’s Research Foundation. 130 Perloff, Richard M. (1998), Political Communication: Politics, Press, and Public in America, Mahwah, NJ: Lawrence Erlbaum. (2002), “Political Campaigns: Persuasion and Its Discontents,“ in The Persuasion Handbook: Theory and Practice, James Price Dillard and Michael Pfau, eds., Thousand Oaks, CA: Sage, p. 605-632. ———--—- and Dennis Kinsey (1992), “Political Advertising as Seen by Consultants and Journalists,“ Joumal of Advertising Research, 36 (May/June), 53-60. Petty, Richard E. and John T. Cacioppo (1986), “The Elaboration Likelihood Model of Persuasion,“ Advances in Experimental Social Psychology, 19, 123-205. Pfau, Michael, R. Lance Holbert, Erin Alison Szabo, and Kelly Kaminski (2002), “Issue-Advocacy versus Candidate Advertising: Effects on Candidate Preferences and Democratic Process,“ Journal of Communication, 52, 301-315. Pfau, Michael, David Park, R. Lance Holbert, and Jaeho Cho (2001), “The Effects of Party- and PAC-Sponsored Issue Advertising and the Potential of Inoculation to Combat Its Impact on the Democratic Process,“ The American Behavioral Scientist, 44 (12), 2379-2397. Pinkleton, Bruce E. (1992), “The Role of Comparative lnfonnation in Advertising Evaluations, Candidate Evaluations, Voting Preferences, and Election Involvement,“ Ph.D. Dissertation, Michigan State University. ------— (1997), “The Effects of Negative Comparative Political Advertising on Candidate Evaluations and Advertising Evaluations: An Exploration,“ Journal of Advertising, 26 (Spring), 19-29. ----—-- (1998), “Effects of Print Comparative Political Advertising on Political Decision-Making and Participation,“ Journal of Communication, 48 (4), 24- 36. -—--—-, Nam-Hyun Um, and Erica Weintraub Austin (2002), “An Exploration of the Effects of Negative Political Advertising on Political Decision Making,“ Journal ofAdvertising, 31 (1), 13-26. Richardson, Glenn W. Jr. (2001), “Looking for Meaning in All the Wrong Places: Why Negative Advertising is a Suspect Category,“ Journal of Commuicatr'on, 51 (4), 775-800. 131 Rifon, Nora J., Sejung Marina Choi, Carrie S. Trimble, and Hairong Li (2004), “Congnrence Effects in Sponsorship: The Mediating Role of Sponsor Credibility and Consumer Attributions of Sponsor Motive,“ Journal of Advertising, 33(1), 29-42. Robideaux, Douglas R. (1998), “Ad Attitude and Negative Political Advertising: Toward a Theoretical Understanding,“ Academy of Marketing Studies Journal, 2(1), 1-10. ------- (2002), “Party Affiliation and Ad Attitude Toward Political Ads,“ Journal of Marketing Theory and Practice, 10 (1 ), 36-44. Roddy, BL. and Garramone, Gina M. (1988), “Appeals and Strategies of Negative Political Advertising,“ Journal of Broadcasting & Electronic Media, 32, 415-427. Rosenstone, Steven J., Donald Kinder, and Warren E. Miller (1997), American National Election Study, 1996: Pre- and Post-Election Survey, University of Michigan, Center for Political Studies, Ann Arbor, MI. Ross, Michael and Garth J. O. Fletcher (1985), “Attribution and Social Perception,“ in The Handbook of Social Psychology, 3"I edition, volume ll, Lindzey and Aronson (eds), p. 73-98. Rudolph, Thomas J. and J. Tobin Grant (2002), “An Attributional Model of Economic Voting: Evidence from the 2000 Presidential Election,“ Political Research Quarterly, 55(4), 805-823. Rucinski, Dianne and Charles T. Salmon (1990), “The 'other' as the vulnerable voter: A study of the third-person effect in the 1988 US. presidential campaign,“ lntemational Journal of Public Opinion Research, 2, 345-368. Sabato, L. (1981), The Rise of Political Consultants, New York: Basic Books. Schenck-Hamlin, Wllliam J., David E. Procter, Deborah J. Rumsey (2000), “The Influence of Negative Advertising Flames on Political Cynicism and Political Accountability,“ Human Communication Research, 26 (1), 53-74. Scott-Lennox, Jane and Richard D. Lennox (1995), “Sex-Race Differences in Social Support and Depression in Older Low-Income Adults,“ in Structural Equation Modeling: Concepts, Issues, and Applications, Rick H. Hoyle, ed., Thousand Oaks, CA: Sage Publications, 199-216. Settle, Robert B. and Linda L. Golden (1974), “Attribution Theory and Advertiser Credibility,“ Joumal of Marketing Research, 1 1, 181-185. 132 Shapiro, Michael A. and Robert H. Rbger (1992), “Comparing Positive and Negative Advertising on Radio,“ Journalism Quarterly, 69 (Spring), 135- 145. Simmons, Carolyn J., Barbara A. Bickart, and John G. Lynch, Jr. (1993), “Capturing and Creating Public Opinion in Survey Research,“ Journal of Consumer Research, 20 (September), 316-329. Smith, Robert E. and Shelby D. Hunt (1978), “Attributional Processes and Effects in Promotional Situations,“ Joumal of Consumer Research, 5 (3), 149-157. SrUII, T. K. (1981), “Person Memory: Some Tests of Associative Storage and Retrieval Models,“ Journal of Experimental Social Psychology, 7, 440-462. Stern, Barbara B. (1994), “Classical and Vlgnette Television Advertising Dramas: Structural Models, Formal Analysis, and Consumer Effects,“ The Journal of Consumer Research, 20 (4), 601 -61 5. Swinyard, William R. (1981), “The Interaction between Comparative Advertising and Copy Claim Variations,“ Journal of Marketing Research, 18 (2), 175- 186. Taggar Simon and Mitchell Neubert (2001), “Teams: How one bad apple can affect the whole barrel,“ HR Professional, 70, 33-53. -——-—- and ---——— (2004), “The Impact of Poor Performers on Team Outcomes: An Empirical Examination of Attribution Theory,“ Personnel Psychology. 57(4), 935-969. Taylor, H., J. Bremer, C. Overrneyer, J.W. Siegel, and G. Terhanian (2001), “Touchdown! Online Polling Scores Big in November 2000,“ Public Perspecfive, 12, 33-35. Tedesco, John C. (2002), “Televised Political Advertising Effects: Evaluating Responses during the 2000 Robb—Allen Senatorial Election,“ Journal of Advertising, 31 (1), 37-48. Thorson, Esther, William G. Christ, and Clarke Caywood (1991), “Effects of Issue-Image Strategies, Attack, and Support Appeals, Music, and Vlsual Content in Political Commercials,“ Journal of Broadcasting 8. Electronic Media, 35 (4), 465-486. Tinkham, Spencer F. and Ruth Ann Weaver-Lariscy (1993), “A Diagnostic Approach to Assessing the Impact of Negative Political Television Commercials,“ Joumal of Broadcasting and Electronic Media, 37 (Fall), 377-399. 133 Tripp. Carolyn, Thomas D. Jensen, and Les Carlson (1994), “The Effects of Multiple Product Endorsements by Celebrities on Consumers’ Attitudes and Intentions,“ Journal of Consumer Research, 20 (4), 535-547. Vavreck, L. (2000), “How does it all ‘turnout’? Exposure to Attack Advertising, Campaign Interest, and Participation in American Presidential Elections,“ in Campaign Reform: Insights and Evidence, L.M. Bartels and L. Vavreck, eds., Ann Arbor. University of Michigan Press, 79-105. Valentino, Nicholas A., Vlncent L. Hutchings, and Dmitri \NlIIiams (2004), “The Impact of Political Advertising on Knowledge, Internet Information Seeking, and Candidate Preference,“ Joumal of Communication, June, 337-354. Wattenberg, Martin P. and Craig Leonard Brians (1999), “Negative Campaign Advertising: Demobilizer or Mobilizer?“ The American Political Science Review, 93(4), 891-899 Weaver, D. and D. Drew (2001), “Voter Laaming and Interest in the 2000 Presidential Election: Did the Media Matter?“ Journalism 8. Mass Communication Quarterly, 78(4), 787-798. Weaver-Lariscy, Ruth Ann and Spencer F. Tlnkham (1996), “Advertising Message Strategies in US Congressional Campaigns: 1982, 1990,“ Journal of Current Issues and Research in Advertising, 18 (1 ), 53-66. Webb, Deborah J. and Lois A. Mohn (1998), “A Typology of Consumer Responses to Cause-Related Marketing: From Skeptics to Socially Concerned,“ Journal of Public Policy 8. Marketing, 17 (2), 226-238. Weiner, Bernard (1992), Human Motivation: Metaphors, Theories and Research, Sage Publications, Newbury Park, CA -—---— (2000), “Attributional Thoughts about Consumer Behavior,“ Journal of Consumer Research, 27 (December), 382-387. West, Darrell M. (1994), “Political Advertising and News Coverage in the 1992 California US. Senate Campaigns,“ The Journal of Politics, 56, 1053- 1075. --—---- (1997). Air Wars: Television Advertising in Election Campaigns, 1952- 1996. Washington D. C: Congressional Quarterly Inc. Whitman, Janet (2004, May 27), “US Advertising Spending Rises; TV posts Biggest Gain,“ Dow Jones. 134 Wllson, R. Dale and Aydin Muderrisoglu (1980), “An Analysis of Cognitive Responses to Cornparativa Advertising,“ Advances in Consumer Research, 7, 566-571. Zhao, X. and SM. Chaffee (1995), “Campaign Advertisements versus Television News as Sources of Political Issue Information,“ Public Opinion Quarterly, 59, 41-65. 135