creating impressions: an experimental investigation of political advertising on television

Upload: elizabeth

Post on 07-Apr-2018

215 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/4/2019 Creating Impressions: An Experimental Investigation of Political Advertising on Television

    1/25

    Creating Impressions: An Experimental Investigation of Political Advertising on TelevisionAuthor(s): Kim Fridkin Kahn and John G. GeerSource: Political Behavior, Vol. 16, No. 1 (Mar., 1994), pp. 93-116Published by: SpringerStable URL: http://www.jstor.org/stable/586484 .

    Accessed: 08/09/2011 03:21

    Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

    JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of

    content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms

    of scholarship. For more information about JSTOR, please contact [email protected].

    Springeris collaborating with JSTOR to digitize, preserve and extend access to Political Behavior.

    http://www.jstor.org

    http://www.jstor.org/action/showPublisher?publisherCode=springerhttp://www.jstor.org/stable/586484?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/586484?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=springer
  • 8/4/2019 Creating Impressions: An Experimental Investigation of Political Advertising on Television

    2/25

    Political Behavior, Vol. 16, No. 1, 1994

    CREATINGMPRESSIONS:NEXPERIMENTALNVESTIGATIONFPOLITICALDVERTISINGN TELEVISIONKimFridkinKahnand John G.Geer

    Using an experimental design, this paper addresses a few basic, but important, questionsabout the influence of televised political advertising. How effective are different kinds ofpolitical spots in creating impressions of a candidate among viewers? Do negative adswork better than positive ads in creating favorable impressions? Do spots that focus onissues create more favorable impressions than spots that stress the traits of a candidate?Do two ads work better than one ad in creating impressions? Can the effects of a spot beundercut by a follow-up advertisement from the opposition? This paper offers some tenta-tive answers to these questions.

    In a few states during the 1988 presidential election, citizens watchedthe now-famous Willie Horton ad. The airing of this spot generated a tre-mendous amount of controversy. So much in fact that pundits, when spec-ulating about the recent 1992 elections, often wondered whether we wouldsee another "Willie Horton" campaign. Even though the Clinton-Bush-Perot contest did not yield any such (in)famousads, television advertisingcontinues to generate a lot of interest. For those who study political behav-ior the interest is somewhat different, centering around trying to deter-mine whether and how spots affect the political attitudes of citizens. Candi-dates, obviously, think ads influence the views of voters, as shown by thetremendous amounts of money and time they spend in crafting and airingthese messages for the public.' Journalists, too, think they matter, as sug-gested by the recent explosion in the coverage of ads (Adatto 1990).It is not clear, however, how much political ads actually do influence thepublic's views. Many observers, for instance, claimed that George Bush'sspots in 1988 were an important reason he defeated Michael Dukakis. Per-

    Kim Fridkin Kahn, Iepartment of Political Science, Arizona State University, Tempe, AZ85297-2001; John G. Geer, I)epartment of Political Science, Arizona State University, Tempe,AZ 85297-2(X)1.93

    019)-9320/94/(a3(X-(X)93$07.(X)/0 1994 Plenuml Puiblishing Cororation

  • 8/4/2019 Creating Impressions: An Experimental Investigation of Political Advertising on Television

    3/25

    KAHNANDGEERhaps. But more systematic examinations of the effects of political spots haveled scholars to adopt more tempered conclusions (e.g., Patterson and Mc-Clure 1976; Kaid and Sanders 1978; Garramone 1984; Just, Crigler, andWallach 1990; Newhagen and Reeves 1991; West 1991).In this paper we use an experimental design that offers one look atwhether political advertising influences the attitudes of citizens. But ratherthan assessing the import of ads in the middle or near the end of a cam-paign, as most previous studies have done, we focus our attention on thevery first stages of an electoral struggle. Specifically, we test to see if adscan create impressions of a single, unknown candidate among our subjects.In so doing, we compare the effects of negative and positive ads, issue andtrait ads, and, finally, we examine how different combinations of ads influ-ence voters' perceptions of our candidate. Even though the analysis is ad-mittedly exploratory, the findings should improve our understanding ofhow political advertising affects the electorate's attitudes toward candi-dates, especially as the first volleys of a campaign commence.THEPROBLEM

    Political advertising can influence the views of voters in a number ofways. To start, political commercials provide informationto voters that mayalter the considerations of citizens when they evaluate candidates. Even in30 seconds, office seekers can provide a wealth of information, rangingfrom their views about unemployment to their previous political experi-ence. While pundits often complain about the content of political advertis-ing, previous analyses of the actual content of ads show that these spotsoften stress relevant issues and personal traits of the candidates (oslyn1980, 1981; Kaid and Davidson 1986; Kern 1989; Shyles 1984a, 1984b).These bits of information, therefore, can help voters learn something aboutthe contenders, which, in turn, can alter citizens' overall evaluations ofthem (Markus 1982; Markus and Converse 1979; Kinder 1986; Brady andJohnston 1987).Of course, the relative importance of this informationfor voters will be aproduct of how much prior knowledge they have about the contenders. Buteven for well-known politicians, ads may influence voters' views. Spots al-ways have the potential of providing new information about candidates,even presidential contenders. In the fall of 1992, for example, many voterswere apparently unaware of Bill Clinton's personal background, making theads that stressed such informationpotentially significant to them.

    Political ads are also important because they may recast the relevance ofpre-existing information. That is, ads, by emphasizing certain traits or is-sues, can "prime" he attitudes of voters (Iyengar, Peters, and Kinder 1982;Iyengar and Kinder 1987). So, if a candidate runs a set of advertisements

    94

  • 8/4/2019 Creating Impressions: An Experimental Investigation of Political Advertising on Television

    4/25

    CREATINGMPRESSIONSstressing the environment, one might expect such concerns to figure moreprominently in voters' choices. In fact, one can think of the 1988 GOPcampaign against Michael Dukakis as an example of politicians trying to"prime" the electorate. Bush and his associates felt that if crime could be-come an important issue in the campaign, it would help the vice-presi-dent's cause (Blumenthal 1990). Consequently, the GOP ran the "revolvingdoor" ad in an effort to activate those concerns about crime among mem-bers of the electorate.While there are good reasons to expect ads to influence the attitudes ofvoters, there is, surprisingly, a limited amount of previous research thattests these effects. The first work that systematically tackled this topic wasPatterson and McClure's The Unseeing Eye. In this pioneering book, theseresearchers used survey research to examine the impact of ads. Otherstudies, too, have used surveys to explore the role of ads in campaigns (see,for example, Atkin, Bowen, Nayman, and Sheinkopf 1973; Atkin and Heald1976; West 1991).Although these studies are informative and important,2 some centralquestions about how ads affect public opinion cannot be addressed ade-quately with survey research. In particular, it is difficult to isolate causalvariables with data from surveys. This limitation has led a number of re-searchers to turn to experimental designs to test the effects of ads (e.g.,Garramone 1986; Just, Crigler and Wallach 1990; Kaid and Sanders 1978;Ansolabehere and Iyengar 1991). In this paper, we also adopt the experi-mental method. Our reasoning, like previous researchers, is simple: Withcontrolled experiments, one can hold all other variables constant and onlymanipulate the subjects' exposure to political ads. By so doing, one canestablish causality with more authority and hence detect more effectivelyhow commercials influence the political views of respondents.While the previous experimental work has advanced our understandingof political spots,:3much work remains. In particular, one basic question hasnot received adequate attention: Can ads create impressions of a candidate?To determine whether ads can create impressions, we focus on the ad-vertising for a candidate not known to our subjects. In this way, the adswill be writing on a clean slate, allowing us an opportunity to examine theability of spots to generate attitudes towards candidates. One might arguethat the central question concerning political advertising is how ads alterexisting impressions of candidates. That question, of course, is important,especially for presidential elections, but unknown candidates are a frequentand integral part of our electoral landscape. Even presidential hopefuls of-ten begin their quest for the White House as "unknowns" e.g., Gary Hartand Jimmy Carter). Such candidates are even more common in nonpresi-dential elections, where challengers are usually unfamiliar to voters.Focusing on a previously unknown candidate has an important meth-

    95

  • 8/4/2019 Creating Impressions: An Experimental Investigation of Political Advertising on Television

    5/25

    odological advantage. If we use the spots from a well-known politician, weface the confounding effects of previous information about the candidate.That is, a "trait"ad about a known politician, such as Ted Kennedy, maywork differently than a similar spot for someone like Robert Dole. But thereason for the difference could be the ad, preexisting views about the can-didate, or an interaction between the two. Our approach, in contrast, pro-vides a way to control for these differences.Note also that we limit our attention to a single candidate. We do notconsider the relative impact of ads among a set of candidates competing ina given election. Obviously, how ads influence the public's respectiveviews on such candidates is important, especially in shaping the final out-come of an election. But we are interested in the early phases of campaignswhen unknown candidates are trying to build favorable impressions amongthe electorate. At this stage of the contest, candidates are primarily inter-ested in their own standings than how they are faringagainst the opposition(see, for instance, Kessel 1992).RESEARCHOBJECTIVES

    The questions tackled in this study are fueled to a large degree by thestrategies open to candidates and their consultants as they enter a politicalcampaign. For most state and national elections, contenders must decidewhat type of commercials to air on television. Among their decisions arewhether to develop "positive"or "negative"spots and whether they want tostress the candidate's views on issues or the contender's personal charac-teristics. Obviously, these choices are not mutually exclusive. But thesecrude dichotomies offer a way to think about their decisions.With these distinctions in mind, this paper tackles three specific (ues-tions concerning whether ads can create impressions of an unknown candi-date. First, do ads about issues have different effects on respondents' eval-uations of the candidates than do ads about traits? The answer to this(uestion will have obvious import for contenders about to embark on acampaign. The importance of this answer to politicians becomes evengreater in light of previous scholarship that shows people's judgmentsabout a candidate's personal characteristics and positions on issues affecttheir choices in elections (Abramowitz and Segal 1992; Kinder 1986;Markus 1982; Miller, Wattenberg, and Malanchuk 1986).Previous scholars, examining the impact of political advertisements, haveaddressed this question, but the results are mixed. Garramone(1986), Gei-ger and Reeves (1991), and Kaid and Sanders (1978), for example, testedpeople's perceptions of unknown candidates in an experimental setting andfound that issue ads created more favorable impressions than trait ads.

    96 KAHNANDGEER

  • 8/4/2019 Creating Impressions: An Experimental Investigation of Political Advertising on Television

    6/25

    CREATINGMPRESSIONSMeadow and Sigelman (1982), on the other hand, failed to find such differ-ences. Given these conflicting results, we reexamine this basic question.

    In addition, when examining the impact of trait and issue ads on people'sperceptions of candidates, past scholars have restricted their analyses toevaluations of the candidate's personal characteristics (see Geiger andReeves 1991; Kaid and Sanders 1978; Meadow and Sigelman 1982). How-ever, since candidates' personal traits represent only one part of how citi-zens evaluate candidates, this approach is problematic. Studies examiningvote choice in national and statewide elections demonstrate that assess-ments of candidates' views on issues (and in some cases their electability)also affect overall impressions of the candidate (Bartels 1988; Brady andJohnston 1987; Markus 1982; Wright and Berkman 1986). Furthermore,these various dimensions of evaluation are not independent (Markus andConverse 1979; Page and Jones 1977). For instance, judgments about acandidate's position on issues may influence views about the contender'spersonal traits. Given the importance of these dimensions, as well as theirinterdependence, we explicitly consider assessment of traits, issues, andoverall evaluations of the candidate in this study.The second question of our study is: Do negative ads create differentkinds of impressions than positive ads? Given the public debate overwhether negative ads are "fair," his question is important not only to ob-servers of the political process but to those who create and air politicalspots. From a theoretical point of view, it is important to assess the impactof critical and favorable information since people do not treat all informa-tion equally when developing impressions of others (Cantor and Mischel1977; Kaplan 1971; Lau 1985; Schneider, Hastorf, and Ellsworth 1979).Past research suggests that while negative spots are remembered betterthan positive commercials (Basil, Schooler, and Reeves 1991; Lang 1991;Newhagen and Reeves 1991), it is not clear whether negative ads generatemore favorable impressions. In particular, the results of two experimentalstudies suggest that negative advertisements may be counterproductive,creating a backlashagainst the candidate (Basil, Schooler, and Reeves 1991;Garramone 1984). We compare the effectiveness of positive and negativespots, examining whether negative ads "boomerang"on the attackingcandi-date.Given the distinction between positive and negative ads-as well as thepossible differences between issue and trait ads-it is important, when de-veloping a better understanding of how the intial stages of a campaignworks, to examine the impact of each of the four types of commercials onpeople's impressions of a candidate. For example, certain types of negativeads, such as those focusing on issues, may be more effective than negativespots that address traits because voters may consider negative issue ads to

    97

  • 8/4/2019 Creating Impressions: An Experimental Investigation of Political Advertising on Television

    7/25

    be fairer than negative trait ads Johnson-Cartee and Copeland 1989). Infact, experimental work by Roddy and Garramone (1988) and Shapiro andRieger (1992) indicate that this may be the case. Shapiro and Rieger, forexample, using radio commercials, find that negative issue ads resulted inmore positive attitudes toward the sponsor than among negative image ads.In this study, we explicitly compare all four types of ads to see if we canprovide evidence regarding how a commercial's tone and substance influ-ence its ability to forge impressions of a candidate.Finally, we look at whether different "combinations"of ads influencepeople's perceptions of an unknown candidate. In campaigns, voters oftenwatch a number of ads even during the course of a single evening of televi-sion. Yet, we do not know whether two ads for the same candidate helpmore thanjust a single ad. Similarly, we do not know whether ads from theopposition will influence the effectiveness of the candidate's initial spots.Zaller's (1992) work, which demonstrates that the amount and sequence ofinformation influence people's attitudes, suggests that different combina-tions of ads will have varying effects on people's views of candidates. Andgiven that ads are seldom viewed in isolation during an electoral campaign,we present some initial tests of how different combinations of ads affectpeople's impressions of a candidate.EXPERIMENTALESIGN

    Our concern about traits versus issues and negative versus positive infor-mation yields four basic types of commercials that comprise the core of ourexperimental design.' The first type, "PositiveTraitAds," stresses favorablythe candidate's personal characteristics; the second type, "Negative TraitAds," attack the opponent's personality. The next two types of ads focus onthe candidate's or the opponent's views on issues: "Positive Issue Ads"and"Negative Issue Ads."To sort out the effects of these four types of ads and combinations ofthem, we developed an experiment with nine conditions. These experi-ments represent a partial factorial design with three independent factors:valence of the ads (positive/negative), substance of the ads (issues/traits),and combinations of the ads (e.g., positive trait ad and positive issue ad).As Table 1 indicates, the first four conditions provide evidence aboutwhether various types of ads can create different impressions of the candi-date. The next two conditions examine how two ads from the same candi-date influence subjects' perceptions. And the last three conditions provideevidence about whether ads by an opponent undercut the effectiveness of acandidate's original commercial.5

    98 KAHNAND GEER

  • 8/4/2019 Creating Impressions: An Experimental Investigation of Political Advertising on Television

    8/25

    CREATINGMPRESSIONSTABLE1. Descriptionof ExperimentalConditionsCondition Typeof PoliticalAdvertisementConditionI Candidate'sPositiveTraitAdCondition2 Candidate'sNegativeTraitAdCondition3 Candidate'sPositiveIssueAdCondition4 Candidate'sNegativeIssueAdCondition5 Candidate'sPositiveTraitAd + Candidate'sPositiveIssueAdCondition6 Candidate'sNegativeTraitAd + Candidate'sNegativeIssueAdCondition7 Candidate'sPositiveTraitAd + Opponent'sNegativeAdCondition8 Candidate'sPositiveIssue Ad + Opponent'sPositiveAdCondition9 Candidate'sNegativeTraitAd + Opponent'sNegativeAd

    Experimental ManipulationIn designing our experiment, we had to make an important decision be-tween what Iyengar (1991) refers to as "realism versus precision." Specifi-cally, should we use actual spots from a campaign to have a realistic experi-mental manipulation or should we create our own spots to control the exactcontent of the ad? There is, of course, no ideal solution. But like Iyengar

    (1991), we opted for using a realistic experimental manipulation(i.e., actualpolitical advertisements used in a campaign). To minimize any loss in preci-sion, we made efforts to ensure that our ads were as comparable as possi-ble."Given our preference on this matter, the second decision involvedchoosing a specific set of ads. Here, we faced three constraints. First, be-cause of the questions posed above, the candidate had to be unknown toour subjects. Second, the ads could not make explicit partisan or ideologi-cal appeals that might activate the underlying predispositions of our re-spondents. Third, we had to find a set of advertisements that representedthe four "types" of ads described above. Fortunately, the University ofOklahoma has an impressive archive of political commercials that yieldedone set of ads that met our constraints:7 he 1986 Casey-Scranton campaignfor governor of Pennsylvania."With the ads in hand, the next issue became how best to show them toour subjects. In an effort to enhance the external validity of the study, weembedded the advertisement(s) in a tape of the 1990 season opener ofNBC's comedy "Cheers." We chose "Cheers"because it is a popular half-hour show that was likely to interest our subjects." The insertion of thead(s) was done in a professional manner, making it impossible to tell thatthe original tape had been altered in any way."'We tried to simulate watch-

    99

  • 8/4/2019 Creating Impressions: An Experimental Investigation of Political Advertising on Television

    9/25

    KAHNANDGEERing television "at home' by providing popcorn and soda during the after-noon sessions and doughnuts during the morning sessions. We also encour-aged subjects to take time during the program to get some refreshments."

    ParticipantsWe chose as subjects 303 undergraduates enrolled in an Introduction toAmerican Politics class at Arizona State University. We randomly assignedthese subjects to the nine experimental conditions. The randomizationwaseffective: the groups were not statisticallydifferent from each other on suchmatters as age, sex, political interest, and partisanship.Although reliance on a sample of students may reduce the gener-alizability of our results, the use of undergraduates has one possible advan-tage. Students may actually represent a "tough test" for whether politicalads can create impressions. A number of scholars have shown that as peo-ple become more informed and educated, they become less influenced bythe mass media (Iyengar and Kinder 1987; Miller, Goldenberg, andErbring 1979; Zaller 1992). As a result, college students may be more re-sistant to the effects of political commercials than a more representativesample from the general population.ProcedureWe conducted our experiment during the fall of 1990. We chose thistime frame because students were in the midst of state and local campaignsfrom their respective districts. As a result, our subjects would not find apolitical ad embedded in the episode of Cheers unusual, further lesseningthe artificialityof the experimental setting. Of course, our respondents hadno intrinsic interest in the Casey-Scranton campaign, which might lead one

    to question the generalizability of these findings. But this lack of naturalinterest should be more than offset by that generated by the experimentalcondition itself.Participants were scheduled for one of several daily sessions, averagingabout 18 students per meeting.12 When students arrived for the class, theywere given the following cover story by the experimenter:I amconducting studywitha colleagueof minefromthe University f Pennsyl-vania. We are studyingviewers'responsesto prime-timetelevisionprogram-ming.Mycolleaguehasalready onducted hisstudywith University f Pennsyl-vania undergraduatesand we are replicating that study here with ourundergraduates.Whatwe need you to do is view the followingbroadcast s youwouldview anyTV show. Sincewe wantto representan actualviewingexperi-

    100

  • 8/4/2019 Creating Impressions: An Experimental Investigation of Political Advertising on Television

    10/25

    CREATINGMPRESSIONScnce, we have sodaandpopcorn doughnuts)hatyou can get duringthe tapedbroadcast.This is the sametapethatwasshown o the University f Pennsylvaniaparticipants.To strengthen this cover story, the instructor of the class told his stu-dents one week earlier that the study was being conducted by faculty inSociology. To legitimize further this story, we used a classroom for oursessions from that department and our experimenter, a 45-year-old gradu-ate student, looked like a possible member of that faculty.After the introduction to the study, students watched the 30-minute epi-sode of"Cheers." Following the broadcast, the experimenter handed out a

    questionnaire to each respondent. The questionnaire contained standardmeasures of political attitudes, demographic questions, and a series ofquestions about Bob Casey. We also included a large number of filler (lues-tions about the "Cheers"episode and other commercials in the tape. Thesequestions were included to reinforce the cover story, thereby lessening"demand characteristics"(Orne 1962).We asked participants at the end of the study to report what theythought was the purpose of the study. Although some students believed wewere interested in political advertising, the vast majorityof subjects simplyrepeated our cover story."'To test further the possibility of "demand char-acteristics," we examined whether students paid more attention to the po-litical commercials than to other advertisements on the tape. Fortunately,no differences appeared. For instance, 90 percent of our subjects re-sponded to an open-ended question that asked about their impression ofthe political ad(s). When asked about a spot that dealt with an upcomingmovie, "PacificHeights," 91 percent answered the open-ended item. Theseproportions suggest that students did not focus more attention on the polit-ical commercials than on other spots.To tap subjects' perceptions of Robert Casey, we asked three questionsabout his competence to handle issues related to education, health care,and economics. " We chose education as one of the issues, since one of theads focused on this topic. The other two issues were chosen because theyare often central concerns in gubernatorial elections (Kahn 1991). We alsoasked four questions about Casey's traits, borrowing the format from theNational Election Studies (NES). Specifically, we had questions aboutwhether Casey was a strong leader, trustworthy, hardworking, and knowl-edgeable. " Again, we chose these items because they tap themes presentin the ads and are potentially important criteria for voters. We relied on asingle item to tap the subject's assessments of Casey's viability."' The finalquestion borrowed the NES's classic feeling thermometer, which allowedus to measure students' overall judgment of Casey on a scale from 0 to 100

    101

  • 8/4/2019 Creating Impressions: An Experimental Investigation of Political Advertising on Television

    11/25

    KAHNANDGEERdegrees. We used all nine items when examining how the various politicalads altered respondents' assessments of this gubernatorial hopeful.RESULTS

    Do political spots create favorable impressions of an unknown candidate?For our respondents, the answer is "no." After viewing the four types ofpolitical ads, subjects' overall impressions of Casey were decidedly nega-tive (see Table 2). The average rating on the feeling thermometer rangedbetween 32 and 42 degrees-hardly an overwhelming endorsement. Infact, only 12 percent of the total sample gave Casey a rating of over 50degrees on the thermometer. Subjects' judgment of Casey's personal traitswere also less than stellar. When assessing whether the term trustworthydescribed Casey, only 7 percent of the total sample thought that term fithim very well. Similarly, respondents' evaluations of Casey's ability to dealwith issues were tempered, seeing him as only "somewhat"competent todeal with these concerns. In short, our subjects were not enthusiastic aboutCasey's candidacy. Only a handful of students raved about this guber-natorial hopeful, suggesting that unknown candidates may have to marshallan extensive campaign in the media before the public is likely to respondpositively to their campaign.While none of these ads generated positive impressions of Casey, someadvertisements were more successful than others. First, Casey's "positive"ads generated more favorable impressions than his negative spots. As Table2 reports, each positive ad created more positive views of Casey than dideither of his negative commercials.'7 While all the differences in the tablewere not statistically significant, the results showed that the positive adsconsistently led subjects to develop more favorable views of Casey's traitsand his ability to handle particular issues than the negative spots. For in-stance, respondents who viewed the positive ads considered Casey a stron-ger leader than those who saw the negative commercials (p < .02). A simi-lar pattern arose when comparing how subjects' rated Casey's competenceto handle economic issues (p < .03).Besides the tone of the advertisement, the substance of these spots alsoappeared to play an influential role. For example, respondents who saw thecommercial promoting Casey's concern for education viewed the candidateas more competent at dealing with that issue when compared to respon-dents in the other three conditions (p < .01).When we restrict comparisons to those respondents exposed to the posi-tive ads, we fail to find any consistent differences in the impact of the traitand issue ads. For each of the nine indicators, no clear pattern arises.Sometimes the positive trait ad led to a higher rating and at other times the

    102

  • 8/4/2019 Creating Impressions: An Experimental Investigation of Political Advertising on Television

    12/25

    CREATINGMPRESSIONSTABLE2. The Effect of the FourBasic Ads on Evaluationsof BobCaseyA. Ratingson Casey'sCompetenceon IssuesEducation Health EconomicPositiveIssueAd (n = 39) 2.49 3.21 3.10PositiveTraitAd (n = 37) 2.73 3.11 2.86NegativeIssue Ad(n = 38) 3.05 3.14 3.14NegativeTraitAd (n = 37) 3.61 3.45 3.50p-Value* p = .00 p = .33 p = .02B. Ratingson Casey'sPersonalityTraitsTrust Hardworking Leader KnowledgePositiveIssue Ad(n = 39) 2.41 2.21 2.28 2.21PositiveTraitAd (n = 37) 2.46 2.30 2.22 2.16NegativeIssue Ad (n = 38) 2.49 2.32 2.43 2.24NegativeTraitAd (n = 37) 2.87 2.55 2.74 2.42p-Value p= .06 p= .27 p= .03 p= .47C. Thermometer ndViabilityRatings orCaseyThermometer ViabilityPositive Issue Ad 42.4 (n = 39) 2.46 (n = 39)Positive Trait Ad 39.2 (n = 37) 2.31 (n = 36)Negative Issue Ad 38.7 (n = 38) 1.97 (n = 37)Negative Trait Ad 32.4 (n = 37) 2.62 (n = 37)p-Value p = .22 p = .00

    *Thep-value s basedon the F-statistic.

    ad concerning issues led to more favorable impressions. And in none ofthese cases do the differences reach statistical significance. Thus, at leastfor this experiment, there is no reason for unknown candidates to choosebetween a positive trait or issue ad-both worked about the same.For negative ads, in contrast, the subject matter did lead to importantand consistent differences in evaluations of Casey. The negative spot at-tacking the traits of Casey's opponent, when compared to the other threeads, was clearly the least effective commercial. This attack on Scranton ledparticipants to rate Casey the most negatively on each of the nine items.Furthermore, when we restrict comparisons to only those subjects who sawthe negative ads, we find that the personal attack encouraged respondentsto develop less favorable views of Casey. Subjects who saw the negativetrait ad viewed Casey as less trustworthy and less able to handle such is-sues as education or the economy (p < .10). 'People seem, in sum, to react negatively to Casey when he raised a

    103

  • 8/4/2019 Creating Impressions: An Experimental Investigation of Political Advertising on Television

    13/25

    series of personal attacks against his opponent. But when Casey directedhis attack to a specific issue, the audience responded more favorably. Nev-ertheless, the differences between negative and positive spots suggest thatthe latter are "safer bets" for unknown candidates.

    Are TwoAds BetterThanOne?Our findings suggest that when respondents see two ads from the samecandidate, it can build more favorable impressions-but only if the two adsare "positive." As the data in Table 3 show, the two positive ads led tomore favorable evaluations of Casey than just a single positive spot. People

    who saw, for instance, both positive ads rated Casey more highly on allthree issues and each of the four personal characteristics. Not all the differ-ences were significant, but the pattern was consistent. Moreover, of allnine conditions in our experimental design, the dual positive ads yieldedthe highest overall thermometer ratings of Casey (51 degrees).More may not be better when it comes to negative ads, however. Sub-TABLE 3. The Cumulative Impact of Two Positive Ads on Evaluations ofBob CaseyA. Ratings on Casey's Competence on IssuesEducation Health EconomicPositive Issue Ad (n = 39) 2.49 3.21 3.10Positive Trait Ad (n = 37) 2.73 3.11 2.86BOTH (n = 33) 2.18 2.88 2.79p-Value* p = .02 p = .20 p = .26B. Ratings on Casey's Personality TraitsTrust Hardworking Leader KnowledgePositive Issue Ad (n = 39) 2.41 2.21 2.28 2.21Positive Trait Ad (n = 37) 2.46 2.30 2.22 2.16BOTH (n = 33) 2.12 1.94 2.12 1.91p-Value p = .15 p = .10 p = .70 p = .01C. Thermometer and Viability Ratings for CaseyThermometer ViabilityPositive Issue Ad 42.4 (n = 39) 2.46 (n = 39)Positive Trait Ad 39.2 (n = 37) 2.31 (n = 36)BOTH 51.3 (n = 33) 1.91 (n = 33)p-Value p = .04 p = .01

    *Thep-value s basedon the F-statistic.

    KAHNANDGEER04

  • 8/4/2019 Creating Impressions: An Experimental Investigation of Political Advertising on Television

    14/25

    CREATINGMPRESSIONSjects who watched both the negative trait and issue ads had less favorableimpressions of Casey than those who watched a single negative ad (seeTable 4). For all four ratings of Casey's personal traits, the dual negativespots led to lower ratings of the candidate. These differences are not alwayslarge, but the pattern suggests that multiple negative ads created morenegative impressions of Casey.The differences between dual negative and positive ads become evenclearer when one compares the two conditions directly. For instance, therewas a 14 percentage point difference in the feeling thermometer ratings ofCasey in the two conditions (51 degrees to 37 degrees, p < .05). We seethe same pattern when looking at subjects' assessment of Casey on the fourpersonality traits and the three issues.These results offer some straightforwardconclusions. To begin, whenrunning positive ads, more is better. It appears that when respondentshave no previous knowledge of a candidate, they react favorably to addi-tional "positive"information. Citizens at this beginning stage may act muchlike sponges, absorbing information that casts a positive light on the candi-TABLE4. The Cumulative mpactof Two NegativeAds on Evaluations fBobCaseyA. Ratingson Casey'sCompetenceon IssuesEducation Health EconomicNegativeIssueAd(n = 38) 3.05 3.14 3.14NegativeTraitAd (n = 37) 3.61 3.45 3.50BOTH(n = 25) 3.40 3.40 3.40p-Value* p = .06 p = .33 p = .24B. Ratingson Casey'sPersonalityTraitsTrust Hardworking Leader KnowledgeNegativeIssue Ad(n = 38) 2.49 2.32 2.43 2.24NegativeTraitAd(n = 37) 2.87 2.55 2.74 2.42BOTH(n = 26) 3.04 2.73 2.69 2.65p-Value p = .03 p = .18 p = .25 p = .10C. Thermometer ndViabilityRatings orCaseyThermometer ViabilityNegativeIssue Ad 38.7 (n = 37) 1.97(n = 37)PositiveTraitAd 32.4 (n = 38) 2.62 (n = 37)BOTH 37.1 (n = 26) 2.07 (n = 26)p-Value p = .43 p = .00

    *Thep-value s based on the F-statistic.

    105

  • 8/4/2019 Creating Impressions: An Experimental Investigation of Political Advertising on Television

    15/25

    date. But interestingly, this absorption of information works differently for"negative"ads. People, when confronted with critical statements about anunknown opponent, did not react positively to the author of those attacks.Thus, negative ads appear to exact a price when candidates air them, mak-ing them a risky investment for an unknown candidate.

    Do Ads from Casey's Opponent Hurt?All our previous analyses have focused on ads from the Casey campaign.But, of course, Casey did not run alone. He had an opponent, WilliamScranton, who also aired a series of spots in an effort to persuade the public

    to support him. For our experiment, we examined two of those ads. Onead from Scranton stressed his ability to help improve the education systemin Pennsylvania (i.e., "Opponent's Positive" Ad). The second spot attackedCasey's dedication to his job by noting, for instance, that he earned$100,000 from a law firm while drawing a salaryfrom the government (i.e.,"Opponent's Negative" Ad). These two ads allow us to explore some simplecombinations that will provide hints about whether counterpunches fromone's opponent can negate a candidate's own advertising.Given that one of Scranton'sads concerned education, one might expecta change in the effectiveness of Casey's ad about education. But, inter-estingly, no significant differences arose for any of the nine items amongthose who watched Casey's education ad and those who watched both spotsthat dealt with education (see Table 5). Even when judging Casey's compe-tence to handle education, the Scranton spot did not depress students'views of Casey. Of course, it is possible that the Scranton ad boosted hisown ratings on the subject, but our attention here focuses just on assess-ments of Casey.Scranton's other ad, which attacked Casey's character, may underminethe effectiveness of Casey's positive ad that stressed his own personal char-acteristics (see Table 6). Although there is a small decline in evaluations ofCasey on items assessing traits for those who saw the two ads, none of thedifferences came close to statistical significance. Thus, as before, the Scran-ton ad did little to alter assessments of Casey.The final comparison concerns how the audience responded to negativeads from both camps. It appears that Casey's and Scranton's negative traitads created a more negative impression of Casey than just the single nega-tive ad. For each of the four questions about Casey's traits, people who sawboth negative trait spots had a lower rating of him than those who saw justCasey's attack on Scranton (see Table 7). These differences were statis-tically significant at p < .06. In fact, when we look at all nine experimental

    106 KAHNANDGEER

  • 8/4/2019 Creating Impressions: An Experimental Investigation of Political Advertising on Television

    16/25

    CREATINGMPRESSIONSTABLE 5. The Interaction of Casey's Positive Trait Ad and Scranton's NegativeTrait Ad on Evaluations of Bob CaseyA. Ratingson Casey'sCompetenceon IssuesEducation Health EconomicPositiveTraitAd(n = 37) 2.73 3.11 2.86PositiveTraitAd +Opponent'sNegativeAd(n = 24) 2.89 3.17 3.19p-Value* p = .56 p= .81 p = .17B. Ratingson Casey'sPersonalityTraits

    Trust Hardworking Leader KnowledgePositiveTraitAd(n = 37) 2.46 2.30 2.22 2.16PositiveTraitAd +Opponent'sNegativeAd(n = 24) 2.58 2.42 2.47 2.08p-Value p = .55 p = .59 p = .24 p = .67C. Thermometer ndViabilityRatings orCaseyThermometer ViabilityPositiveTraitAd 39.2 (n = 37) 2.31 (n = 36)PositiveTraitAd +

    Opponent's Negative Ad 36.7 (n = 25) 2.28 (n = 25)p-Value p = .63 p = .88*Thep-value s based on the T-statistic.

    conditions, students who saw both negative trait ads gave Casey the lowestratings on the feeling thermometer (37 degrees).These results suggest that ads can, at times, have some interactive ef-fects. Specifically, it seems that the attack ads from the respective campsworked together, producing the most negative assessments of Casey. Itcould be argued that the impression formed by the initial negative ad wasreinforced by the story line Scranton developed against Casey. But, inter-estingly, the same pattern did not emerge for the ads on education andtraits. In particular, the Scranton ads-even the negative spot-did notseem to damage the image of Casey generated by his own positive ads.Even though the ads by Scranton in these two conditions tackled roughlythe same topic as Casey's commercial, they did not dampen people's im-pression of the Pennsylvania Democrat. These results suggest that a gooddefense against an opponent's attack may be a positive spot stressing the

    107

  • 8/4/2019 Creating Impressions: An Experimental Investigation of Political Advertising on Television

    17/25

    KAHNANDGEERTABLE 6. The Interaction of Casey's Positive Issue Ad and Scranton's PositiveIssue Ad on Evaluations of Bob CaseyA. Ratings on Casey's Competence on IssuesEducation Health EconomicPositive Issue Ad (n = 39) 2.49 3.21 3.10Positive Issue Ad +

    Opponent's Positive Ad(n = 30) 2.57 3.20 3.27p-Value* p = .72 p = .98 p = .40B. Ratings on Casey's Personality Traits

    Trust Hardworking Leader KnowledgePositive Issue Ad (n = 39) 2.41 2.21 2.28 2.21Positive Issue Ad +Opponent's Positive Ad(n = 30) 2.27 2.17 2.20 2.30p-Value p = .40 p = .79 p= .64 p = .62

    C. Thermometer and Viability Ratings for CaseyThermometer ViabilityPositive Issue Ad 42.4 (n = 39) 2.46 (n = 39)Positive Issue Ad +

    Opponent's Positive Ad 39.4 (n = 31) 2.33 (n = 30)p-Value p = .52 p = .50*Thep-value s basedon the T-statistic.

    candidate's strengths. Positive ads, even when coupled with the opponent'snegative ad, produced favorable impressions of unknown candidates.CONCLUSION

    These experimental results indicate that paid political advertising variesin the kinds of impressions it generates among our subjects. Positive ads,for instance, consistently led our subjects to develop more favorableimagesof the unknown candidate than negative spots. Furthermore, among nega-tive commercials, ads focusing on traits and issues led respondents to offerdifferent evaluations of our candidate. In particular, we found that subjectswere more tolerant of attack ads that focus on a specific issue and provideevidence to support the attack. Spots that criticized the opponent on per-sonal grounds, in contrast, appeared to be counterproductive, leading re-spondents to form more negative impressions of the attacker.

    108

  • 8/4/2019 Creating Impressions: An Experimental Investigation of Political Advertising on Television

    18/25

    CREATINGMPRESSIONSTABLE 7. The Interaction of Casey's Negative Trait Ad and Scranton's NegativeTrait Ad on Evaluations of Bob CaseyA. Ratings on Casey's Competence on IssuesEducation Health EconomicNegative Trait Ad (n = 38) 3.61 3.45 3.50Negative Trait Ad +Opponent's Negative Ad(n = 24) 3.92 3.96 3.62p-Value* p = .29 p = .08 p = .62B. Ratings on Casey's Personality Traits

    Trust Hardworking Leader KnowledgeNegative Trait Ad (n = 38) 2.87 2.55 2.74 2.42Negative Trait Ad +Opponent's Negative Ad(n = 24) 3.37 3.00 3.29 2.96p-Value p = .02 p = .06 p =.01 p = .01C. Thermometer and Viability Ratings for CaseyThermometer ViabilityNegative Trait Ad 32.4 (n = 38) 2.62 (n = 37)Negative Trait Ad +Opponent's Negative Ad 29.0 (n = 25) 2.68 (n = 25)p-Value p = .53 p = .77

    *Thep-value s basedon the T-statistic.

    Running a series of ads also improved our candidate's image, but onlywhen those spots were positive. In the experiment, respondents exposedto two positive commercials rated the candidate most favorably. Multiplenegative advertisements, on the other hand, did not improve people's im-pressions of Casey. Respondents who watched two negative spots had lessfavorable views of the candidate than those who witnessed the single nega-tive ad.Finally, the effectiveness of a candidate's political advertising can beblunted by the spots aired by the opposition. But, again, there were impor-tant differences for positive and negative commercials. Specifically, the im-pact of a candidate's positive commercial was not depressed by the pres-ence of an ad from an opponent. Our candidate's negative commercials,however, became less effective when coupled with the opposition's nega-tive ads. We found, for instance, that attack ads from each campaignworked together to produce the most negative evaluations of Casey.

    109

  • 8/4/2019 Creating Impressions: An Experimental Investigation of Political Advertising on Television

    19/25

    KAHNAND GEERThese results suggest that unknown candidates may want to adopt aninitial campaign strategy that emphasizes positive ads over negative ads.

    Positive commercials created more favorable impressions-both individu-ally and when aired with other positive spots. Positive advertisements alsoappear to be an effective defense against attacks by one's opponent. Theimpact of an opponent's attack can be muted when coupled with the candi-date's own positive spot. This advice, however, applies only at the start of acampaign when a candidate is still unknown to the public. Once votersbegin to develop some impressions about a contender, the informationfrom ads may interact with these impressions, producing different effects.These initial findings, of course, are just one step in what we hope willbe a continuing effort to sort out the impact of political ads. Additionalstudies need to examine other ads and other candidates to replicate ourresults. Further research, for example, could test whether different kindsof appeals lead to alternative impressions of candidates. That is, are someads about issues more effective than others? Does the style of presentationmatter? So, for instance, do ads about education work better with a talkinghead format or with the candidate shown talking with students?Given that the 30-second spot is a centerpiece of the modern campaign,we need to increase our understanding of the impact of political advertis-ing. With candidates spending well over half of their campaign treasurieson political advertisements, our failure to know more fully how political adswork limits our understanding of elections. And since elections are a cen-tral mechanism connecting the elected with the electorate, it is importantthat we close this gap as quickly as possible.

    Acknowledgtments.he authors hankAllanMeyer, JenniferCrookes,and KateLehman or their help with our experiment.We alsoacknowledge he kindassis-tance of JulianKanterand the PoliticalCommercialArchivesat the UniversityofOklahoma.Finally,we expressappreciationo Peter Esaiasson,JamesGeer, RickHerrera,Pat Kenney,WarrenMiller,VincentPrice, and WendyRahnfor usefulcommentson this manuscript.Thispaperwas in every way a collaborativenter-prise. We gave an earlierversionof this paperat the 1991MidwestPoliticalSci-ence AssociationMeetings.APPENDIX:Content of the Political Ads

    Below is a brief summary of the six different political advertisements we usedfrom the 1986 Casey-Scranton gubernatorialelection. The purpose is simply to pro-vide readers with a sense of these ads. For those people who want additional infor-mation abut these 30-second spots, please contact us.

    110

  • 8/4/2019 Creating Impressions: An Experimental Investigation of Political Advertising on Television

    20/25

    CREATINGMPRESSIONS(1) "Positive Issue Ad"Bob Casey stressed his concern over Pennsylvania's system of education. Hewanted to improve the quality of education (i.e., recruit better teachers) in the stateso as to ensure the "economic future" and the "hopes of our children." He wouldfight for better teachers and make sure they educate "our kids." As he made theseclaims, the ad showed kids working on science projects and studying in class. At theend of the ad, Casey appeared in a classroom to show his commitment and dedica-tion to education.(2) "Positive Trait Ad"This ad talked of Casey's character, noting that he wants to help the state, ishardworking, possesses "integrity,"wants to move the state "ahead,"and help chil-dren. During the ad, the announcer described two endorsements Casey receivedfrom majornewspapers in the state.As these statements were made, Casey walked with his wife in a pleasant setting.The next scene showed him making a speech. At the end, Casey was walking in afield with a number of children.(3) "Negative Issue Ad"In this commercial, Casey attacked Scranton for not meeting the responsibilitiesof his job as Lt. Governor. Casey provides a number of statistics, noting his oppo-nent's failure to attend important meetings. The ad then goes on to say that Scran-ton only had that job because of his father-a former governor of the state. Thebasic point is that Scranton cannot be counted on, which is driven home at the endwhen the closing slogan reads: "Casey:A Governor we can count on."(4) "Negative Trait Ad"An announcer begins this ad by stating that this campaign has come down to one(luestion: "Character."That word appears on the screen by itself. At this point, thead launches into an attack on Scranton. The majortheme is that Scranton is runninga "dirty" campaign, spending "millions of dollars" for "negative ads" and "dirtymail." The letters from Scranton have smeared Casey and have been "lies."The adthen goes on to question whether Scranton thought he could get away with suchtactics. Finally, the announcer notes that Scranton must think the rules are differ-ent for him.This ad was based on a series of still shots that sought to convey the messagesnoted above.(5) "Opponent's Positive Ad"Scranton walks near a school, noting that kids frequently cannot even read anewspaper, which is a "scandal for our schools." He proposes new standards toincrease the quality of education, like ensuring discipline and tougher rules to grad-uate. In addition, he plans to keep parents informed about how the schools areworking. Schools, he then notes, are the key to our future and we must make surethey work.

    111

  • 8/4/2019 Creating Impressions: An Experimental Investigation of Political Advertising on Television

    21/25

    KAHN AND GEER(6) "Opponent's Negative Ad"Scranton suggests that Casey has not met his responsibilities as auditor general.Casey failed to attend some important meetings. Moreover, Casey, while on thegovernment payroll, has been earning $100,()00 in private legal fees. Casey repre-sents the "old"politics. And Pennsylvania can do better: Scranton.

    NOTES1. In 1990, for instance, candidates for the U.S. Senate spent $60 million on television andradio advertising (Alston 1991). This amount of money becomes even more impressive,given that candidates in senatorial elections are spending about 60 percent of their totalcampaign treasuries on television advertising (Luntz 1988). The same proportion of spend-

    ing on advertising applies for gubernatorial elections (Luntz 1988).2. For instance, these studies show that exposure to ads increases voters' informationalxout,and interest in, the campaign.3. A number of prior experimental studies have isolated variables that mediate the impact ofpolitical advertising. l)onohue (1973) and Garramone (1984, 1986), for instance, find thatthe motivation of voters sometimes influences what they learn from ads. Rothchild andRay (1974), looking at the political context of the races, finds that spots are only effectivein contests where voters are not very interested and where they possess little informiationabout the candidates.4. Previous work on political advertising shows that these four "types"of spots appear fre-quently on television (Garramone 1986; Joslyn 1980; Kaid and l)avidson 1986; Kaid andSanders 1978; Kahn 1993).5. We do not have conditions that represent all possible combinations of ads because of thelimited number of respondents available. Instead, we limit our attention to the mosttheoretically interesting conditions. A full factorialdesign-exploring all combinations ofads-would have required an additional 12 conditions.6. So, for instance, our ads from the opposing candidates dealt with the similar issues andtraits. See the Appendix for a discussion of the ads.7. Ideally, we would have preferred to have found additional ads that met our requirementsto extend the generalizability of our findings. But none existed. As a result, we will takecare to be sure not to overgeneralize from these findings.8. Robert Casey, former state auditor general, won the governorship in this election. It washis fourth attempt to capture this office. Casey's opponent, William Scranton III, was amoderate Republican who was makinghis first bid for the governorship. Scranton, the sonof a popular former governor and the lieutenant governor for Governor Richard Thorn-burgh, stressed during the campaign that an economic recoverv was under wav in Penn-sylvania. Casey, on the other hand, claimed that eight years of Thornburgh and Scrantonhad left the state economically depressed and a change of leadership was needed. Caseywon the election with 51 percent of the vote.9. Recent work by Kaid, Chanslor, and Hovind (1992) indicate that the context of the admatters. By placing political spots in a show like "Cheers," we may produce differenteffects than by inserting them in a New York Met's baseball game. These findings areobviously relevant. But given the proliferation of half-hour comedies on television, wehave, at least, chosen a common "context."Of course, additional tests in other settings aredesirable, but are beyond the scope of this paper.10. For those conditions with just a single political advertisement, the spot was inserted in

    112

  • 8/4/2019 Creating Impressions: An Experimental Investigation of Political Advertising on Television

    22/25

    CREATINGMPRESSIONSthe second commercial break. For those conditions with two ads, we inserted the spot inthe first and second commercial break.11. Subjects did take us up on our offer. They routinely left their seats during the program,often during commercials, to get something to eat or drink. To simulate the environmentat home, the refreshments were placed in the corner of the room so that subjects had toleave their seats to get the food.12. For each of the nine conditions, we scheduled two sessions. By using two sessions percondition, we were able to increase the realism of the setting by limiting the number ofparticipants watching the broadcast at one time. We staggered the sessions in the morn-ings and afternoons, making sure that no condition had two morning sessions or twoafternoon sessions. Before collapsing the two sessions of each condition, we made surethat the timing of the session did not affect how subjects reacted to the ads.13. We did test to see whether correctly guessing the purpose of the experiment altered theresponses of subjects. No statistically significant differences arose.14. The exact wording of these questions is as follows: What is your best guess about BobCasey's competence in dealing with issues related to education? What is your best guessabout Bol) Casey's competence in dealing with health-related issues? What is your bestguess about Bol) Casey's competence in dealing with economic issues? For each question,students could circle (1) "very competent," (2) "competent," (3) "somewhat competent,"(4) "somewhat incompetent," (5) "incompetent," and (6) "very incompetent.15. The question about Casey's traits read as follows: "How much would you sav 'trustworthyfits your impression of Bol) Casey: (1) a great deal, (2) somewhat, (3) a little, or (4) not atall?" In place of "trustworthy we inserted "provides strong leadership," "hardworking,""knowledgeable."16. The specific question was: "Whatis the likelihood that Bol) Casev will win the election forgovernor this November? (1) very likely, (2) somewhat likely, (3) not verv likely and (4)not at all likely."17. We ran analyses of covariance (ANCOVA) to see whether the impact of our ads varieddepending on the political interest, sex, and party identification of the subjects. Ourresults did not change when we controlled for these three variables.18. Of course, these attacks by Casev may have led to very negative assessments of Scranton.The purpose of our experiment, however, was to examine changes in judgments of Casey,not comparative evaluations of both candidates-a topic for future experiments.

    REFERENCESAbramowitz, Alan I. and Jeffrey A. Segal (1992). Senate Elections. Ann Arbor:University of Michigan Press.Adatto, Kiku (1990). Sound bite democracy: Network evening news presidentialcampaign coverage, 1968 and 1988. Research paper R-2. The Joan ShorensteinBarone Center at Harvard University.Alston, Chuck (1991). Forcing down cost of TV ads appeals to both parties. Con-gressional Quarterly Weekly Report 49: 647-652.Ansolabehere, Stephen, and Iyengar, Shanto (1991). Why candidates attack:Effects

    of television advertising in the 1990 California gubernatorial campaign. Paperpresented at the Annual Meeting of the Western Political Science Association.Atkin, Charles, Lawrence Bowen, Oguz B. Nayman and Kenneth G. Sheinkopf

    113

  • 8/4/2019 Creating Impressions: An Experimental Investigation of Political Advertising on Television

    23/25

    KAHNANDGEER(1973). Quality v. Quantity in Televised Political Advertisements. Public OpinionQuarterly, 37: 209-224.Atkin, Charles and Gary Heald (1976). Effects of Political Advertising. Public Opin-ion Quarterly, 40: 212-228.Bartels, Larry M. (1988). Presidental Primaries and the Dynamics of Public Choice.Princeton, NJ: Princeton University Press.Basil, Michael, Schooler, Caroline, and Reeves, Byron (1991). Positive and negativepolitical advertising: Effectiveness of ads and perceptions of candidates. In FrankBiocca (ed.), Television and Political Advertising, Volume 1: Psychological Pro-cesses. Hillsdale, NJ: Lawrence Erlbaum.Blumenthal, Sidney (1990). Pledging Allegiance: The Last Campaign of the ColdWar. New York:Harper Collins, Publisher.Brady, Henry E., and Johnston, Richard (1987). What's the primary message:Horserace or issue journalism. In Gary R. Orren and Nelson W. Polsby (eds.),Media and Momentum. Chatham, NJ: Chatham House Publishers, Inc.Cantor, N. and W. Mischel (1977). Traits as Prototypes: Effects on RecognitionMemory. Journal of Personality and Social Psychology, 35: 38-48.Donohue, T.W. (1973). Impact of Viewer Predispositions of Political TV Commer-cials. Journal of Broadcasting. 18: 3-16.Garramone, Gina (1983). Issue v. Image Orientation and Effects of Political Adver-tising. Communication Research, 10: 59-76.Garramone, Gina (1984). Voter response to negative political ads: Clarifying spon-sor effects. Journalism Quarterly 61: 250-259.Garramonc, Gina (1986). Candidate image formation: The role of information pro-cessing. In Lynda Lee Kaid, Dan Nimmo, and Keith R. Sander (eds.), New Per-spectives on Political Advertising. Carbondale: Southern Illinois University Press.Geiger, Seth F., and Reeves, Byron (1991). The effects of visual structure andcontent emphasis on the evaluation and memory for political candidates. InFrank Biocca (ed.), Television and Political Advertising, Volume 1: PsychologicalProcesses. Hillsdale, NJ: Lawrence Erlbaum.Iyengar, Shanto (1991). Is Anyone Responsible? How Television Frames PoliticalIssues. Chicago: University of Chicago Press.lyengar, Shanto and Donald R. Kinder (1987). News That Matters. Chicago: Uni-versity of Chicago Press.Iyengar, Shanto, Mark D. Peters and Donald R. Kinder (1982). ExperimentalDemonstrations of the "Not So Minimal"Consequences of Television News Pro-grams. American Political Science Review, 76: 848-858.Johnson-Cartee, Karen S., and Copeland, Gary (1989). Southern voters' reaction tonegative political ads in the 1986 election. Journalism Quarterly 66: 188-193,986.

    Joslyn, RichardA. (1980). The Content of Political Spot Ads. Journalism Quarterly,57: 92-98.Joslyn, Richard A. (1981). The Impact of Campaign Spot Advertising on VotingDefections. Human Communication Research. 7: 347-360.Just, Marion, Ann Crigler, and Lori Wallach (1990). Thirty Seconds or Thirty Min-utes: What Viewers Learn from Spot Advertisements and Candidate Debates.

    Journal of Communication 40: 120-133.Kahn, Kim F. (1991). Campaign Messages in Statewide Elections. Working Paper,Arizona State University.Kaid, Lynda Lee and Dorothy K. Davidson (1986). Elements of Videostyle: Candi-

    114

  • 8/4/2019 Creating Impressions: An Experimental Investigation of Political Advertising on Television

    24/25

    CREATINGMPRESSIONSdate Presentation Through Television Advertising. In Lynda Lee Kaid, DanNimmo and Keith R. Sander (eds.) New Perspectives on Political Advertising,Carbondale: Southern Illinois University Press.Kaid, Lynda Lee, Chanslor, Mike, and Hovind, Mark(1992). The influence of pro-gram and commercial type on political advertising effectiveness. Journal ofBroadcasting and Electronic Media 36: 303-313.Kaid, Lynda Lee, and Sanders, Keith R. (1978). Political television commercials:An experimental study of type and length. Communication Research 5: 57-70.Kaplan. M. F. (1971). Dispositional effects and weight of information in impressionformation. Journal of Personality and Social Psychology 18: 279-284.Kern, Montague (1989). 30-Second Politics: Political Advertising in the 1980's. NewYork:Praeger.Kessel, John H. (1992). Presidential Campaign Politics. Pacific Grove, CA: Brooks/Cole Publishing Company.Kinder, Donald (1986). Presidential character revisited. In Richard Lau and David0. Sears (eds.), Political Cognition. Hillsdale, NJ: Erlbaum.Lang, Annie (1991). Emotion, formal features, and memory for televised politicaladvertisements. In Frank Biocca (ed.), Television and Political Advertising, Vol-ume 1: Psychological Processes. Hillsdale, NJ: Lawrence Erlbaum.Lau, Richard R. (1985). Two Explanations for Negativity Effects in Political Behav-ior. AmericanJournal of Political Science. 29: 119-138.Luntz, Frank I. (1988). Candidates, Consultants, and Campaigns: The Style andSubstance of American Electioneering. Oxford: Basic Blackwell, Ltd.Markus, Gregory B. (1982). Political attitudes during an election year: A report ofthe 1980 NES panel study. American Political Science Review 75: 538-560.Markus, Gregory B., and Converse, Philip E. (1979). A dynamic simultaneousmodel of electoral choice. American Political Science Review 73: 1055-1070.Meadow, Robert G., and Sigelman, Lee (1982). Some effects and noneffects ofcampaign commercials: An experimental study. Political Behavior 4: 163-175.Miller, Arthur H., Goldenberg, Edie, and Erbring, Lutz (1979). Type set politics:The impact of newspapers on public confidence. American Political Science Re-view 73: 67-84.Miller, Arthur, Wattenberg, Martin, and Malanchuk, Oksana(1986). Schematic as-sessments of presidential candidates. American Political Science Review 79: 359-372.

    Newhagen, John E., and Reeves, Byron (1991). Emotion and memory responses fornegative political advertising: A study of television commercials used in the 1988presidential election. In Frank Biocca (ed.), Television and Political Advertising,Volume 1: Psychological Processes. Hillsdale, NJ: Lawrence Erlbaum.Orne, M. T. (1962). On the social psychology of the psychological experiment: Withparticular reference to demand characteristics and their implications. AmericanPsychologist 84: 231-259.Page, Benjamin I., and Jones, Calvin (1977). Reciprocal effects of policy prefer-ences, party loyalties and the vote. American Political Science Review 73: 1071-1089.Patterson, Thomas E., and McClure, Robert D. (1976). The Unseeing Eye. NewYork:Putnam.Roddy, Brian L., and Garramone, Gina M. (1988). Appeals and strategies of nega-tive political advertising. Journal of Broadcasting and Electronic Media 32: 415-427.

    115

  • 8/4/2019 Creating Impressions: An Experimental Investigation of Political Advertising on Television

    25/25

    116 KAHNANDGEERRothchild, Michael, and Ray, Michael (1974). Involvement and political advertisingeffect: An exploratory experiment. Communication Research 1: 264-285.Rothenberg, Randall (1990). Newspapers watch what people watch in the TV cam-paign. New York Times, November 4.Schneider, David J., Hastorf, Albert H., and Ellsworth, Phoebe C. (1979). PersonPerception. Philippines: Addison-Wesley.Shapiro, Michael A., and Rieger, Robert H. (1992). Comparing positive and nega-tive political advertising on radio. Journalism Quarterly 69: 135-145.Shyles, Leonard (1984a). Defining the issues of a presidential election from tele-vised political spot advertisements. Journal of Broadcasting 27: 333-343.Shyles, Leonard (1984b). Defining "images"of presidential candidates from tele-vised political spot advertisements. Political Behavior 6: 171-181.West, Darrell (1991). Ads and Agenda-Setting in Elections Campaigns. Paper pre-sented at the Annual Meeting of the Western Political Science Association.Wright, Gerald C., Jr., and Berkman, M. B. (1986). Candidates and policy in U.S.Senate elections. American Political Science Review 80: 569-588.Zaller, John R. (1992). The Nature and Origins of Mass Opinion. Cambridge: Cam-bridge University Press.