ANRV377-PL12-15 ARI 2 May 2009 12:49

Negative Campaigning

Richard R. Lau and Ivy Brown Rovner

Department of Political Science, Rutgers University, New Brunswick, New Jersey 08901; email: [email protected]

Annu. Rev. Polit. Sci. 2009. 12:285–306 Key Words The Annual Review of Political Science is online at attack advertising, demobilization, meta-analysis, negativity polisci.annualreviews.org

by Rutgers University Libraries on 11/01/10. For personal use only. This article’s doi: Abstract 10.1146/annurev.polisci.10.071905.101448 The past two decades have seen an explosion of social science research Annu. Rev. Polit. Sci. 2009.12:285-306. Downloaded from www.annualreviews.org Copyright c 2009 by Annual Reviews. on negative political advertising as the number of political observers All rights reserved complaining about its use—if not negative campaigning itself—has 1094-2939/09/0615-0285$20.00 also grown dramatically. This article reviews the literature on negative campaigning—what candidates are most likely to attack their opponent, under what circumstances, and most importantly, to what effect. We also discuss the many serious methodological issues that make studying media effects of any kind so difficult, and make suggestions for “best practices” in conducting media research. Contrary to popular belief, there is little scientific evidence that attacking one’s opponent is a par- ticularly effective campaign technique, or that it has deleterious effects on our system of government. We conclude with a discussion of whether negative political advertising is bad for democracy.

285 ANRV377-PL12-15 ARI 2 May 2009 12:49

INTRODUCTION also must decide what messages to provide in which media markets. Political candidates invest heavily in strategic One of the most important decisions can- campaign communications. Whereas press re- didates make is whether to run on their own leases, public appearances, stump speeches, and merits—that is, their own policy ideas, past ac- interviews are geared toward garnering free complishments, and personal strengths, which media coverage, carefully crafted commercial most observers would call a “positive” appeal; or or paid communications represent an indus- if instead their campaign will concentrate on the try unto themselves. Substantial percentages of perceived weaknesses of their opponent’s policy campaign war chests are dedicated to such var- proposals, prior policy failures, and/or personal ied forms of political communication as lawn peccadilloes, which in this essay we call a “neg- signs, bumper stickers, direct mail, candidate ative” appeal. Candidates are not restricted to web sites, and television and radio appeals. Al- one or the other, of course, and most campaigns though all of these elements are integral to a employ a combination of both techniques. But successful campaign media strategy, television in our review, at a very general level, if the ap- advertisements remain the most visible, expen- peals a candidate makes are primarily positive, sive, and presumably cost-effective form of paid we refer to their campaigns as positive, whereas political communication. if the appeals they make are primarily nega- As one might expect, television advertising is tive, we refer to their campaign as negative.1 by far the most pervasive communicative tech- We use the word “tone” to refer generically to nique studied in the political communications the balance of positive and negative appeals in literature. In just half a minute, a candidate can a campaign. deliver a precise message to a broad audience. A single ad can increase a candidate’s name recog- nition, inform voters of salient issues and can- Why a Focus on Negative didates’ policy concerns, or motivate previously Political Campaigns? reluctant individuals to vote through emotional appeals—if it is repeated often enough. Therein This essay focuses on negative campaign com- lies the rub: effective political advertisements munications, a widespread phenomenon that are expensive to produce, and they become in- has attracted increasing attention from the press creasingly costly with each ad buy. The more and social scientists alike. Although politics in often an ad is shown, the wider its audience, America was never a gentleman’s sport, an ac- and the more often people are exposed to an celerated proliferation of negative advertising over the past 30 years—or at least a rise in the by Rutgers University Libraries on 11/01/10. For personal use only. ad, the more likely they are to process and re- member its message. In theory, there should be number of people complaining about it—seems apparent. But that is the most we can say. Before

Annu. Rev. Polit. Sci. 2009.12:285-306. Downloaded from www.annualreviews.org a point of decreasing marginal utility at which everyone who is going to see an ad has already 1996 there are not, in fact, any good data on the seen and processed it, and additional exposures actual prevalence of negative campaigning in are ignored or, worse, become annoying. How often (if ever) this point is reached in a political 1Campaigns that employ approximately equal numbers of ad campaign is not clear. But even the wealthiest positive and negative appeals could be called “mixed,” but and best financed candidates are limited by the for simplicity we assume here that all campaigns are either size of their war chests, and consequently must primarily positive or primarily negative. How one could ac- tually get an overall negativism “score” for a campaign that decide which messages will be conveyed to the included television ads aired different numbers of times in dif- public, what media tactics should be employed, ferent media markets, another set of radio ads similarly aired and how often they can afford to present each different numbers of times in different markets, newspaper advertisements, billboards, lawn signs, bumper stickers, and message. If the electoral district is large enough personal appearances by the candidate, etc., is a question no to include multiple media markets, candidates one we know of has ever tried to address.

286 Lau · Rovner ANRV377-PL12-15 ARI 2 May 2009 12:49

the United States. Geer (2006) reports a steady appeared in any of these newspaper sources growth in the percentage of negative appeals in in 1980, but this figure grew to 210 stories the television advertisements produced by the in the millennium year before falling back a major party presidential candidates from 1960 bit in 2004. Figure 2b shows the number of through 2004 but has no evidence on how of- academic articles providing some evidence on ten each ad was shown. Buell & Sigelman (2008) the effects (broadly construed) of negative po- report no growth in the negativism of presiden- litical advertisements or negative campaigns. tial elections between 1960 and 2004, as judged The first article we could find in the social sci- by the tone of statements from all presiden- ence literature was published in1984, and the tial candidates reported in the New York Times first study by a political scientist appeared in between Labor Day and election day. Simi- 1990. At last count (near the end of 2006) there larly, Lau & Pomper (2004) report no growth were 110 books, chapters, dissertations, and ar- from 1988 to 2002 in the proportion of state- ticles addressing this broad topic empirically, ments reported in newspapers from major party and many more exploring other aspects of neg- senatorial candidates (or their spokespersons) ative campaigns. The upshot of the great ma- criticizing the opponent. But this is secondary jority of these articles and news stories is that evidence, filtered through the lens of the news- negative campaigning has become a pervasive paper reporter, and tracks only what the can- and often corrosive aspect of the American po- didates are saying. It is only with the advent litical scene—“the electronic equivalent of the of “ad detector” technology developed by the plague” (West 1993, p. 51), to quote one of the Campaign Media Analysis Group, which tracks more colorful characterizations. Negative cam- satellite-based feeds of political advertisements, paigning and the use of attack ads have been that we have been able to gather any solid evi- criticized for reducing politics to its least com- dence on how often different televised ads are mon denominator. Some scholars have gone so shown in different media markets. As seen in far as to attribute diminishing public trust and Figure 1, in this relatively short timeline, neg- ever-decreasing turnout rates to the growing ative political advertising in the United States use of negative ads. has not grown at all (at least at the federal Why all this attention to campaign nega- level).2 tivism if there is no real evidence it is increas- We can be more definitive about increas- ing? The obvious answer is that coordinated ing interest in negative campaigns, however. candidate messages appear to grow increasingly Figure 2a displays the number of stories negative with each campaign cycle even if they about negative advertising or negative cam- actually are not, making negativism increas- by Rutgers University Libraries on 11/01/10. For personal use only. paigns appearing in the New York Times, the ingly difficult for any voter to avoid or ignore. Washington Post, the Christian Science Moni- This is due primarily to biases in the news me- Annu. Rev. Polit. Sci. 2009.12:285-306. Downloaded from www.annualreviews.org tor, and the Associated Press newswire dur- dia that make stories about negative advertise- ing presidential election years since 1980. ments much more likely to be written and/or Only 17 stories about either of these topics aired than stories about positive ads. For exam- ple, West (2005) reports that almost two thirds of the stories about political advertisements

2These data are available from the Wisconsin Advertising broadcast on the CBS evening news between Project at the University of Wisconsin. We thank John Sides 1972 and 2004 involved a negative advertise- for providing the numbers in Figure 1, and John Geer and ment. Unless predicated on false information, Shana Gadarian for data in Figure 2a. We may soon have new evidence on how much money presidential candidates from positive communications provide little if any before 1996 spent advertising in the different states, and even entertainment value. After all, how many stories possibly how often specific ads aired in different locales, from on a candidate’s love of God, country, family, data currently being gathered by Scott Althaus and Daron Shaw from presidential libraries around the country. Stay and all-American values can a single media out- tuned. ... let run? By contrast, negative communications

www.annualreviews.org • Negative Campaigning 287 ANRV377-PL12-15 ARI 2 May 2009 12:49

dangle like red meat in front of journalists hun- stands out against a generally more positive gry for a new angle. It is therefore not surprising background; the motivational “cost-orientation that negative ads get covered (and consequently hypothesis,” which is presumably part of our ge- replayed by the media) much more frequently netic makeup, states that most people are inher- than positive ads—leading to the common im- ently more strongly predisposed to avoid costs pression that negative ads are more pervasive than they are to seek gains. By this account, than they are. To the extent that political prac- negativity explains both the apparent growth of titioners recognize these media biases, they cer- campaign negativism and its alleged corrosive tainly take advantage of them, running negative effect on our political system. ads within the Washington Beltway media mar- In this article, we review the evidence on just ket in order to capture the attention of the po- how effective negative advertising is in achiev- litical news media. ing its goals, and just how corrosive those at- These same biases contribute to the impres- tacks are on our political system. We begin by sion that most negative ads are ad hominem at- trying to carefully define what a “negative polit- tacks and misleading exaggerations if not out- ical campaign” is. Is our very general definition and-out lies. Some, surely, are, and the media do of attacking one’s opponent (rather than talk- their best to bring the most outrageous exam- ing about oneself) sufficient, or do we need a ples to our attention, but it would be a mistake more precise and nuanced definition? Wemight to assume that all or most or even a large mi- believe (echoing Justice Stewart’s famous state- nority of negative political advertisements have ment about pornography) that we know it when these characteristics. we see it, but negative campaigning turns out to Even if it could be definitively shown that be a bit more difficult to define precisely than negative ads are increasing, such evidence one might think. Having established a work- would not explain why candidates attack their able definition of the phenomenon of inter- opponents. We can assume the answer is that est, we then look at the literature on when candidates and political practitioners believe at- candidates choose to “go negative.” There is tack ads are effective—and there are sound the- a fair amount of agreement between the for- oretical reasons to think they may be right. So- mal and the empirical literature, but we also cial psychologists have pointed to “negativity”: find some clear theoretical predictions that have the tendency for negative information to be been disconfirmed by empirical evidence. more influential than equally extreme or equally Then we turn to arguably the most impor- likely positive information (Kanouse & Hanson tant question: Do negative campaigns work? 1971). The “equally likely or equally extreme” Clearly, paid campaign consultants and the re- by Rutgers University Libraries on 11/01/10. For personal use only. part of the definition is crucial, and is what porters who cover campaigns believe they do— makes negativity an interesting psychological otherwise they would not continue to use them. Annu. Rev. Polit. Sci. 2009.12:285-306. Downloaded from www.annualreviews.org phenomenon rather than mere common sense. As Laura Mansnerus, writing in the New York If we describe a candidate as having a pleas- Times about the highly negative 2005 New ant smile but holding foreign policy positions Jersey gubernatorial campaign, explained: “The that will lead to nuclear war, of course no one people who produce these ads and the consul- predicts that the combination of these two bits tants who hire them know that negative cam- of information will result in neutral evalua- paigning works. These people are paid way too tions. Kernell (1977) and Lau (1982) have care- much to be mistaken about whether poison is fully documented negativity effects in candidate effective” (Mansnerus 2005). As social scien- evaluation and voting behavior, and Lau (1985) tists, we seek more convincing evidence than has provided evidence for two different expla- this. nations for negativity, one perceptual and one There is a long history in political science of motivational. The perceptual “figure-ground doubting that the media affect the political pro- hypothesis” states that negative information cess. But as the “minimal effects” view has been

288 Lau · Rovner ANRV377-PL12-15 ARI 2 May 2009 12:49

replaced by the “media matter” view (Iyengar & and effectiveness of negative political adver- Kinder 1987), some within the academy have tisements, trending toward a simple positive- come to believe that negative political cam- negative dichotomy in which ads that focus on paigns matter so much that they can affect the their sponsors are branded “positive” and those very nature or our political system, discourag- that address the opposition are “negative.” Each ing voters from participating in individual elec- advertisement is treated as a single unit of tions and weakening their faith in the political analysis and weighted equally. This dichotomy process itself over time. In fact, this is one of may lead many people intuitively to catego- the few instances that we can think of where rize positive ads as patently “good” and nega- an academic finding has been picked up by the tive ads as inherently “bad,” but Jamieson et al. press and become part of the conventional wis- (2000) warn that doing so is both incorrect and dom about political campaigns. The impact of misleading: negative campaigns on the American political system is the fourth major topic of our review. Many scholars mistakenly assume that “at- Although we mention some methodologi- tack” is both “negative” and “dirty.” Conflat- cal issues as we discuss our first four topics, ing these terms obscures the important dis- the penultimate section of the article focuses tinction between legitimate and illegitimate on a collection of methodological concerns that attack and minimizes the likelihood that the plague the study of political campaigns (and of deceptions found in supposedly “positive” dis- media effects more generally). In the conclud- course will be probed. (p. 44) ing section we ask a normative question: Are negative campaigns bad for democracy? Some Although the positive-negative dichotomy may political scientists argue that, far from demo- be conducive to experimental research—where bilizing the electorate, negative political cam- isolating and manipulating a single variable paigns may even stimulate participation. They (e.g., negative communication) in order to mea- believe that negative ads educate voters by pro- sure its effect is the end goal—Jamieson et al. viding critical information about a candidate’s suggest that content analysis that fails to treat policy positions, character, and personal his- “contrast” or “comparative” advertising as a tory, and in so doing, allow voters to make separate and unique category of political dis- more informed political decisions. A review of course not only excludes a key communica- the literature suggests that categorizing nega- tive ingredient, but also portrays a simpler tive campaigns as patently good or bad is at the landscape than what, in fact, exists. We agree, very least an oversimplification and at most a when it comes to creating typologies of indi- by Rutgers University Libraries on 11/01/10. For personal use only. gross oversight. The question of whether nega- vidual political advertisements. This distinction tive campaigns are good for democracy, like the loses most of its power when applied to en- Annu. Rev. Polit. Sci. 2009.12:285-306. Downloaded from www.annualreviews.org decision to launch a negative attack, is compli- tire campaigns, however. Campaign commu- cated and multilayered. Nonetheless, as nega- nications include television, radio, and print tive campaigns continue to have a sizeable pres- ads, as well as the content of official campaign ence in our political system, an assessment of web sites, speeches, debates, interviews, and di- their intended and unintended influence on po- rect mail (fund-raising) appeals. Unless a can- litical behavior is warranted. didate completely ignores his or her opponent (as may occur when a challenger fails to rep- resent a legitimate threat), all campaigns by WHAT IS A NEGATIVE their very nature are comparative. Not forget- CAMPAIGN? ting the tremendous difficulties of measuring In order to examine the role of negative cam- and combining the positive or negative tone of paigning, we need to carefully define it. The every different aspect of a major political cam- literature has emphasized analysis of the use paign, a unidimensional scale (100% positive to

www.annualreviews.org • Negative Campaigning 289 ANRV377-PL12-15 ARI 2 May 2009 12:49

Table 1 Perceived fairness of different types of attacks follow-up study examining the repercussions of Percent saying counterattacks reveals that voters ascribe sim- Criticism criticism is fair ilar standards of fairness to rebuttals. The less Talking one way and voting another 80.7 “fair” the response, the more willing voters are His/her voting record 75.8 to vote for the initial attacker, regardless of the His/her business practices 71.0 perceived fairness of his or her original charge. Taking money from special interest groups 70.7 But do voters’ theoretical opinions of cam- Taking money from individuals with ethical problems 63.0 paign negativism translate into the reality of Current personal troubles with drugs or alcohol 56.1 mass political behavior? Not necessarily, ac- Current extramarital affairs 45.1 cording to Sides et al. (2003). They report a Political actions of his/her party’s leaders 37.1 study of the 1998 California gubernatorial elec- Past extramarital affairs 27.8 tion, in which Democrat Lieutenant Gover- Past personal troubles with drugs or alcohol 25.9 nor Gray Davis opposed Republican Attorney Personal lives of his/her party’s leaders 19.1 General Dan Lungren. They link rolling cross- Behavior of his/her family members 7.7 sectional survey data from 2902 respondents

Source: 1998 Sorenson Institute for Political Leadership Survey of Virginia Voters, as interviewed during the last six weeks of the cam- reported by Freedman et al. (1999). paign with data on the actual tone of the adver- tising aired by the two candidates during each 100% negative) seems in theory the best way to day of that period. This design allows Sides think about the negativism of a campaign. et al. to determine how closely citizens’ per- But these are all subtle differences in aca- ceptions of a campaign’s negativism are tied to demic definitions that may be meaningless to the actual nature of that campaign. Not sur- the general public. Do laypeople know when prisingly, there was a very strong partisan com- they have seen a negative ad? Is there a limit ponent to these perceptions, with Democrats to what they will tolerate in a political cam- perceiving Davis’s campaign to be much less paign? Drawing on two surveys conducted in negative than Lungren’s campaign, and Repub- Virginia, Freedman et al. (1999) maintain that licans perceiving the opposite pattern. What is voters are indeed capable of distinguishing be- much more surprising is that the realities of the tween mudslinging and fair play, and that they campaign—how negative Davis’sand Lungren’s neither condemn nor condone campaign nega- campaigns actually were—had no effect on per- tivism outright. When asked to rate the fairness ceptions. Indeed, the only significant predictor of 12 charges that one candidate might make in the equations for both Davis and Lungren against another, respondents revealed a clear hi- was partisanship. Clearly, negativism remains in by Rutgers University Libraries on 11/01/10. For personal use only. erarchy of tolerance for attacks (see Table 1). the eye of the beholder. It seems that few vot- Generally speaking, voters believe that attack- ers during actual political campaigns come any- Annu. Rev. Polit. Sci. 2009.12:285-306. Downloaded from www.annualreviews.org ing a candidate for his actions in office is fair where close to the objectivity that a researcher game, whereas bringing up prior personal prob- would apply in judging the content or tone of a lems should be out of bounds. campaign. Freedman et al.’s findings also reveal educa- Sigelman & Kugler (2003) agree, suggest- tional and partisan biases. Respondents’ thresh- ing that scholars’ systematic examinations and olds for tolerating different types of attacks in- even our definitions of campaign negativism creased with education, political information, are out of sync with voter perceptions. Our and Republican party identification. That is, objective, systematic coding of ads implicitly more educated, better informed, and Repub- assumes that scientific realities guide percep- lican voters were all somewhat less likely to tions, but that assumption suggests that vot- say that any given charge was unfair. Freedman ers consume campaign information uniformly et al. did not, however, find significant varia- and as judiciously as scholars conduct their own tion with respect to either gender or race. A research. Sigelman & Kugler point out that

290 Lau · Rovner ANRV377-PL12-15 ARI 2 May 2009 12:49

whereas scholars seek out the breadth of po- causing a re-evaluation of both candidates’ track litical communications and afford each state- records and capabilities. ment equal weight, voters do not; instead they But do purely factual, unemotional ads even seek the information that is most concise and exist? Doesn’t every ad seek to arouse either readily available, often at the expense of accu- confidence in its sponsor or anxiety about the racy, and pay disproportionate attention to that competition? Even “objectively” pointing out with which they most agree. Given those dif- an opponent’s weak record on crime is an ef- ferences, it is not surprising that Sigelman & fort to create anxiety about his or her ability to Kugler uncover inconsistencies between voter address the issue in the future. Similarly, high- perceptions of negativity and scholarly classi- lighting one’s own proven track record is in- fications. Their comparison of ANES survey tended to instill confidence and security. Brader data against the newspaper coverage of three (2006) maintains that, though rare, “unimpas- 1998 gubernatorial races reveals asymmetrical sioned ads” do exist—if only so that their spon- impressions of campaign tone, as well as incon- sors may appear to function above the politi- sistent perceptions among voters within each cal fray. But isn’t that an enthusiastic appeal, state. However, although they attribute differ- painting oneself as a candidate of whom we can ences among voters to partisanship, political all feel proud? Brader points to Ross Perot’s efficacy, and above all, political sophistication, 1992 advertisements and 30-minute “infomer- Sigelman & Kugler stop short of prescribing a cials” as an example of such authentic commu- solution for this problem. nications, yet Perot’s ads tapped into a disgust The challenge lies in reconciling the rigors and frustration with the Republican status quo of systematic social science research with the and garnered enthusiasm for a more respon- reality that voters consume political informa- sible fiscal conservatism—so much enthusiasm tion erratically and with bias. Perhaps our def- that he received more votes than any third-party inition of tone should go beyond written tran- candidate in U.S. history. Even his dry format scripts of televised ads and include the visual triggered emotions among the electorate. Tobe images and accompanying music often aimed fair, Brader concedes that “ads are almost never at triggering voters’ emotions. For example, stripped completely of emotional content”; he Brader’s (2005, 2006) analysis of emotional ap- categorizes as “unimpassioned” those ads which peals reveals frequent, deliberate attempts to do not overtly appeal to voters’ emotions, ac- evoke enthusiasm or fear among the electorate knowledging that minimal residual sentiments through the strategic use of visual images, mu- are inevitable (Brader 2005). sic, and voiceover tone. These components do Every candidate’s goal is to win. Therefore, by Rutgers University Libraries on 11/01/10. For personal use only. not change the ads’ message but rather clar- just as candidates understand that it is unreal- ify that message so that it “strikes an emotional istic to expect all voters to make decisions on Annu. Rev. Polit. Sci. 2009.12:285-306. Downloaded from www.annualreviews.org chord,” as his title suggests. Unlike analyses that the merit of facts and the content of campaign measure tone in terms of ad content, Brader communications alone, scholars must appreci- measures the ads’ intended emotional outcome. ate the role of audiovisual elements in emotion- He describes enthusiastic ads—such as Ronald ally priming vote choice. This is not to sug- Reagan’s famous 1984 “Morning in America” gest that political decisions are fundamentally spot—as those which appeal to hope, enthusi- purely emotional, but rather that emotional asm, sentimentality, and national pride. Such cues can take the voter to a place that facts alone ads aim to capture the attention of a candidate’s may not always reach. As Brader explains, emo- base by motivating participation and ultimately tional elements may reinforce content, but they turnout. By contrast, fear ads do not automati- do not override it. Ultimately, Brader’s research cally elicit a transfer of allegiance from one can- reminds us that researchers who code the tone didate to another, but rather trigger anxieties of a political ad or campaign based solely on among undecided or weakly committed voters, the direction of the candidates’ appeals, without

www.annualreviews.org • Negative Campaigning 291 ANRV377-PL12-15 ARI 2 May 2009 12:49

also considering the role of emotional overlays, herself, against the greatest threat. This was ev- are accepting a broad and less precise defini- ident in the 1992 presidential election: George tion of campaign negativism, one that can be H.W. Bush and Bill Clinton focused their at- potentially misleading. On the other hand—a tacks on each other, rather than against third- very important hand—simple directional cod- party candidate Ross Perot. However, as Perot ing of campaign appeals can be accomplished began to chip away at Bush’s base toward the very easily and reliably. The more subjective end of the campaign, he found himself on the and complicated a coding scheme, the more receiving end of Bush-sponsored attacks, which difficult (and thus expensive) it is to use, and is counter to the predictions of the model. typically the less reliable are its results. The rebuttal calculus also matters. How does a candidate’s decision to go negative with- out provocation compare to his willingness to THE DECISION TO ATTACK: retaliate against a negative attack, in light of WHEN, HOW, AND WHO? poll positioning, closeness of the race, or the When does a candidate decide to inform vot- number of opponents? Why, for example, did ers of his or her opponent’s alleged negative at- John Kerry fail to rebut a series of negative ads tributes, despite the risk of alienating potential throughout August 2004? Where is the line be- and existing supporters? The decision to attack tween staying above the fray and being an oppo- is a political calculation, based on the presump- nent’spunching bag? There are costs either way. tion that its execution will damage the intended On the one hand, a candidate’s failure to refute target more than it will jeopardize the status of attacks can leave the information environment the candidate sponsoring the attack. dominated by the charges from the opponent. Skaperdas & Grofman (1995) use formal On the other hand, responding to the allega- theory to identify factors that motivate a candi- tions gives them credibility, and it can also put date to go negative. Assuming that candidates the candidate on the defensive and take him or are aware of their relative status in the polls her off message, which can prove equally dam- and that future vote shares remain indetermi- aging (Kern 1989, Jamieson & Campbell 1983). nate until the day of the election, and most im- After the candidate is attacked, the campaign portantly that campaign attacks provide a net faces not only a question of strategy (whether benefit to the attacking candidate, Skaperdas & to go negative) but also a question of tactics Grofman examine the use of negative advertis- (how). ing in both two-candidate and three-candidate Harrington & Hess (1996) disagree with the contests. In a two-candidate race, they find that assertion that a candidate’s willingness to attack by Rutgers University Libraries on 11/01/10. For personal use only. the frontrunner typically uses a greater propor- reflects his or her status in the polls. Instead, tion of positive ads and his opponent more fre- their theoretical model suggests that a candi- Annu. Rev. Polit. Sci. 2009.12:285-306. Downloaded from www.annualreviews.org quently goes negative. However, as the gap be- date’spersonality and ideology are stronger pre- tween the two candidates narrows, the leading dictors. Any connection between the use of neg- candidate’s use of negative advertising will in- ative ads and poll position, they maintain, is crease in order to maintain his or her front- purely coincidental; both are driven by a candi- runner status. Multi-candidate races are less date’s weakness in terms of personal attributes. straightforward. A candidate who is trailing by Advertising tactics, they argue, are driven by the a definitive margin and thus only serving as a ideological position of undecided voters rather “spoiler” is predicted to run positive ads exclu- than the relative strength of an opponent. Can- sively. Further, in any three-way race, no can- didates will use positive messages to reposition didate should launch negative attacks against themselves as ideologically compatible with un- the weaker of the two opponents. Instead, all decided voters and will likewise use negative ads attacks should be directed toward the frontrun- to create an ideologically based wedge between ner or, in the case of the frontrunner him- or their opponent and potential swing voters.

292 Lau · Rovner ANRV377-PL12-15 ARI 2 May 2009 12:49

Inherent to this theory is an assumption that Rather than examining the campaign in the candidate with the more desirable personal its entirety, Haynes & Rhine (1998) focus attributes or larger “valence index score” will on those variables—namely “competitive po- present the more positive campaign, regardless sitioning, reward factors, and media related of his position in the polls. conditions”—that predict negative communi- In order to reconcile these opposing ar- cations in the early primary season, a time guments and identify whether actual cam- when the intraparty candidate field is crowded paign activity follows theoretical prediction, with ideologically similar candidates, and lack Theilmann & Wilhite (1998) ran a series of of funding may prevent some candidates from experiments using members of the American purchasing any airtime. Using content analy- Association of Political Consultants, exposing sis, they discover that underfunded candidates this unique sample to a series of hypotheti- seek to increase name recognition through cal campaign scenarios and assessing their rec- personality-driven, self-promotional (positive) ommendations regarding the willingness to ads, and utilize a variety of “intermediated” attack. Theilmann & Wilhite report confir- vehicles—including press releases, television matory evidence for Skaperdas & Grofman’s interviews, stump speeches, and debates—to theory of relative strength, but they find no launch attacks on their opponents. These al- connection between personal attributes and ternative media provide a low-cost opportunity negative ads, as suggested by Harrington & for lower-tier candidates to garner free (some Hess. They also examine the role of partisan- would call this earned ) media coverage by com- ship, testing Ansolabehere & Iyengar’s (1995) paring themselves to the frontrunners while hypothesis that positive ads are more effec- presenting a positive and more dignified image tive with Democratic voters whereas negative through advertising. ads more often resonate with Republicans. The All of the empirical data on the nature of Republican consultants in their sample were in- campaigns reported thus far have come from deed more open to implementing attack strate- studies of relatively few campaigns, typically gies than were their Democratic counterparts at the presidential level. In contrast, Lau & overall, more inclined to employ them early in Pomper (2001, 2004) examine the tone of the a campaign, and more likely to use them when campaigns of virtually every contested U.S. a client is comparatively underfunded. Senate election between 1988 and 2002. They Intrigued by this decision-execution calcu- propose seven hypotheses about when candi- lus, Buell & Sigelman (2008) examined all ma- dates will go negative. In particular, they pro- jor party presidential tickets between 1960 and pose that candidates who are behind, candidates by Rutgers University Libraries on 11/01/10. For personal use only. 2004 to identify how competing tickets de- in close elections, challengers, candidates with cide not only when, but also how, to attack. relatively few campaign resources, Republicans, Annu. Rev. Polit. Sci. 2009.12:285-306. Downloaded from www.annualreviews.org Buell & Sigelman find support for some of males, and candidates whose opponents are at- Skaperdas & Grofman’s predictions, but this tacking them have higher levels of negativism in support is primarily limited to races they con- their campaigns. Lau & Pomper report signif- sider blowouts—the elections of 1964, 1972, icant support for all seven of these hypotheses and 1984. Although the Skaperdas-Grofman at the bivariate level. Only three of these hy- model can play a useful organizing function, pothesized effects maintain their strength in a when put to empirical test it “served pri- multivariate analysis, however. Controlling on marily as a foil for one contrary finding af- the other explanations, Republicans, candidates ter another.” The model “abstracts away too with less money than their opponents, and can- much and thereby ignores a multitude of fac- didates whose opponents are attacking them tors that figure importantly in candidates’ deci- are all significantly more likely to attack their sions to go negative” (Buell & Sigelman 2008, opponents. Particularly noteworthy in Lau pp. 441–42). & Pomper’s findings is the almost one-to-one

www.annualreviews.org • Negative Campaigning 293 ANRV377-PL12-15 ARI 2 May 2009 12:49

correspondence between the tone of a candi- gender and that of the opposition. Kahn and date’s campaign and that of his or her oppo- Kenney (Kahn 1993, Kahn & Kenny 2000) find nent. These days, attacking your opponent in a that gender does indeed factor into a candidate’s senate election is almost guaranteed to provoke decision to attack. However, their research has an approximately equivalent counterattack. yielded somewhat inconsistent results. Kahn’s In an unpublished manuscript, “Going Neg- 1993 study examining the campaign messages ative in a New Media Age: Congressional Cam- of 38 candidates for U.S. Senate in the 1984 and paign Websites, 2002–2006,” Druckman, Kifer, 1986 races revealed that although both male and and Parkin find similar patterns in political at- female candidates prefer policy-based messages tacks seen on the internet. Having analyzed a over candidate-oriented appeals, men make use dataset that includes 730 candidate web sites of negative ads less frequently and are partic- spanning three campaign cycles, Druckman ularly reluctant to go negative against female et al. concur that competition drives nega- opponents, for fear of being perceived as “beat- tivism, but they provide a somewhat differ- ing up on a woman” (Kahn 1993, p. 491). She ent explanation for this now familiar result. argued that regardless of candidate status (in- First, Druckman et al. argue that the audi- cumbent or challenger), the messages delivered ences for campaign web sites are highly self- by female candidates tend to reflect those of selective, falling into two distinct groups: jour- underdogs, exhibiting a greater proportion of nalists and politically active supporters of a negativism than their opponents demonstrate. candidate. Drawing on the emotions literature Less than a decade later, Kahn & Kenney (Brader 2006, Marcus et al. 2000), Druckman (2000) again compared the communicative et al. go on to argue that negative informa- styles of male and female senatorial candi- tion raises anxiety levels and encourages vot- dates, now spanning the 1988, 1990, and ers to seek out additional information about 1992 campaigns. They found female candidates all candidates and engage in compensatory de- consistently less likely than male candidates— cision making, wherein candidates are com- regardless of candidate status—to go nega- pared against one another across a variety of tive. Identifying five categories of negative ads dimensions in order to “choose the best alter- (attack, comparative, negative-trait, negative- native, regardless of cognitive demands.” Be- issue, and criticism), Kahn & Kenney found cause calm voters remain satisfied with the sta- that male challengers consistently launch more tus quo, incumbents prefer to avoid campaign negative appeals than their female counterparts, attacks except in highly competitive races. The and in particular are 10% more likely to run same is true of both frontrunners and members negative-issue ads. This finding is consistent by Rutgers University Libraries on 11/01/10. For personal use only. of the majority party. By contrast, challengers, with the bivariate results reported by Lau & nonfrontrunners, and members of the minority Pomper (2001, 2004). When closeness of race Annu. Rev. Polit. Sci. 2009.12:285-306. Downloaded from www.annualreviews.org party not only have an incentive to go negative, is taken into account, the differences are even but ultimately must go negative if they are to more stark; the frequency of male-sponsored convince the electorate that a change of course attacks greatly increases with competitiveness, is necessary. Nonetheless, as Druckman et al. whereas female candidates are more consistent. and others have pointed out, the greater the The two studies are not entirely contradic- competitiveness of the race, the more likely it is tory, however. Kahn and Kenney confirm that that both candidates will go negative. Ultimately male candidates remain far more aggressive to- competitiveness trumps incumbency, frontrun- ward other males than toward female oppo- ner, and majority-party status. nents, attacking 56% of the time as opposed A review of the circumstances under which to 39%. candidates go negative would be incomplete Kahn and Kenney’s inconsistent findings, without assessing whether candidates adopt and the null effect of gender in a multi- different message strategies depending on their variate analysis reported by Lau & Pomper

294 Lau · Rovner ANRV377-PL12-15 ARI 2 May 2009 12:49

(2001, 2004), cast doubt on any speculation that negative campaigning: It works. This is prac- female candidates adhere to a softer style and tically a mantra among political practitioners more stereotypically feminine cultural norms (Kamber 1997, Swint 1998) and has reverber- while male candidates are more aggressive and ated throughout the scholarly literature as well. cut-throat. This skepticism is supported by The only reason a rational candidate should Procter et al. (1994), who examine the use of choose any campaign strategy is because that negative advertising in eight 1990 U.S. Senate candidate believes the strategy will win votes. campaigns and eight gubernatorial campaigns Thus, if attacking one’s opponent is an effective from the same year, all of which featured a fe- campaign strategy, it must result in the attacker male candidate against a male opponent. Proc- receiving more votes than he or she would have ter et al. find that male and female candidates if some other (less negative) campaign strategy do not differ in their use of negative advertising. had been adopted. Assuming the vote choice is Female candidates are no less likely to strike little more than selecting the more highly eval- first, do not manipulate ad formats to soften uated candidate, attacks that lower mean evalu- their own images, and do not shy away from at- ations of the target of those attacks would con- tacking their opponent’spersonal ethics or issue stitute additional evidence for the effectiveness positions. Procter et al.’s findings illustrate that of negative campaigns. However, there is always decisions to go negative and decisions concern- a danger that evaluations of the attacking candi- ing how to strike are not reflections of a candi- date will also decrease (a so-called backlash ef- date’s gender but rather of a formulaic calculus fect: see, for example, Garramone 1984, Roese that is unaffected by gender. We should men- & Sande 1993). Candidates who attack their op- tion one important caveat here: Research on the ponent might be perceived as mean-spirited or contrasting campaign styles of male and female nasty, particularly by voters who were taught “if candidates when they oppose each other suf- you can’t say anything nice, don’t say anything fers from the absence of an important compari- at all.” But even if backlash effects against the son group, female-versus-female races. There sponsors of campaign attacks are frequent oc- are still too few examples of statewide races currences, attacking one’s opponent would still where major party female candidates oppose be an effective strategy if evaluations of the tar- each other. get of the attacks went down more than evalua- tions of the sponsor, such that the net differen- tial evaluation of the two competing candidates DO NEGATIVE CAMPAIGNS worked to the advantage of the attacker. WORK? Lau et al. (2007) have just published a meta- by Rutgers University Libraries on 11/01/10. For personal use only. If you ask observers of the American political analytic review of the social science literature on scene to think of a negative political advertise- the effects of negative political campaigns (see Annu. Rev. Polit. Sci. 2009.12:285-306. Downloaded from www.annualreviews.org ment, they might mention the infamous Willie Lau et al. 1999 for an earlier review). A meta- Horton ads from the 1988 presidential elec- analysis is a quantitative synthesis of research tion campaign, or remember seeing Michael findings from independent studies on a partic- Dukakis riding around in a tank during that ular topic, where the reported findings them- same campaign. Or they might come up with selves, rather than the raw data on which they the Swift Boat Veteransads from the 2004 presi- are based, are the data analyzed. A meta-analysis dential campaign. Somewhat older voters might is as comprehensive as possible, including both remember the daisy ad from the 1964 presi- published and unpublished studies so that well- dential campaign. In each case, the target of known publication biases (e.g., only publishing those attacks—Michael Dukakis in 1988, John papers with statistically significant findings), or Kerry in 2004, Barry Goldwater in 1964—lost the prejudices of the reviewer (e.g., deciding a the election. These impressions are perfectly particular study is “junk” and therefore should consistent with the conventional wisdom about not be included in the review), do not influence

www.annualreviews.org • Negative Campaigning 295 ANRV377-PL12-15 ARI 2 May 2009 12:49

the outcome of the meta-analysis. Any factor Controlling for a large number of factors in the analyst believes could influence the magni- a multivariate analysis, Kahn & Kenney re- tude of the reported results (e.g., experimental port that both incumbents and challengers were versus correlation designs; the perceived qual- liked significantly less when they were attacked ity of the reported study) can be coded and by their opponent than when their opponent tested empirically—if there are enough relevant stayed primarily positive. Twenty-three of the studies. 31 studies reported similar effects, 12 of which Lau et al. (2007) located 43 studies examin- were statistically significant. Only seven stud- ing the effects of negative campaigns (or expo- ies (two of which were significant) reported sure to particular negative advertisements) on the opposite effect, with evaluations of the tar- the actual or intended vote choice of those ex- get actually increasing after the attack—which posed to the campaigns/advertisements. For ex- might plausibly occur if many people perceived ample, King et al. (1998) exposed subjects to ei- an attack to be exaggerated, false, or otherwise ther a positive or negative ad from Bill Clinton out-of-bounds, and responded in part by sym- or Bob Dole near the end of the 1996 presi- pathetically increasing their evaluations of the dential campaign. Controlling for initial liking target of the unjustified attacks. In any case, we of the candidates, King et al. found that expo- are reasonably confident that attacking a po- sure to a negative ad from Clinton strongly de- litical opponent is generally likely to result in creased intention to vote for him, but exposure lower evaluations of that candidate. to a negative ad from Dole had no effect on like- But attacking one’s opponent is not a risk- lihood of voting for him. In neither case is there free enterprise, as evaluations of the attacker any evidence that negative ads are particularly might also decrease if voters are convinced the effective in winning votes. Of the 43 relevant attacks are unjustified or mean-spirited. This studies, 12 reported positive effects (that is, the same study by Kahn & Kenney (2004) reported more negative candidate received more votes), that evaluations of both incumbents and chal- four of which were statistically significant; 28 lengers decreased (the former significantly) af- studies reported unfavorable effects (that is, at- ter they sponsored attacks against their oppo- tacking the opponent was counterproductive), nent. All told, 31 of the 40 relevant studies (19 six of which were statistically significant. Three of which were statistically significant) reported studies reported that there were “no significant similar findings, whereas only six studies (two differences,” without providing any numbers or significant) found evaluations of the attacker in- hints as to which direction the nonsignificant creasing after the attacks. If anything, this back- results leaned; in such cases, results can only be lash effect appears to be a bit stronger and more by Rutgers University Libraries on 11/01/10. For personal use only. coded as having an effect size of 0. Looking at consistent than lowering evaluations of the tar- all of these studies together, no one could con- get of the attacks. On balance, then, there is Annu. Rev. Polit. Sci. 2009.12:285-306. Downloaded from www.annualreviews.org clude that negative campaigning is a particularly simply no support in the scientific literature for effective campaign strategy. the hypothesis that negative campaigns are any Many factors go into a vote decision, of more effective than any other type of campaign course, including habit, and it may be ask- strategy. ing a lot of a typical political campaign, never To this point we have not commented on mind exposure to a few ads in an experiment, the different designs that researchers employ to change that decision. But surely, one would to study the effects of negative campaigns. We think, evaluations of the targets of political at- assume the reader is generally familiar with the tacks must go down. Lau et al. (2007) located 31 relative advantages and disadvantages of experi- studies with relevant data. For example, Kahn & ments and correlational (or observational) stud- Kenney (2004) coded the tone of up to five ads ies. Despite what we often tell our undergrad- aired by major party candidates in 97 competi- uates, it is often possible to make reasonably tive senate elections between 1988 and 1992. confident causal statements from correlational

296 Lau · Rovner ANRV377-PL12-15 ARI 2 May 2009 12:49

data. But the very fact that we can reliably iden- strategies to different candidates) were no more tify several factors that explain when candidates likely to produce positive results than were decide to go negative raises a difficult but very studies using various nonexperimental designs important methodological problem when we where the endogeneity problem must be con- try to examine the success of different strate- fronted. Twelve of the 25 experimental stud- gies with observational data from real election ies showed positive effects (that is, the more campaigns. In statistical terms, this problem negative candidate received more votes), which arises because the choice of campaign strategy were statistically significant in four of the stud- is endogenous to the subsequent outcome of ies; three of the 13 negative findings were sig- the election itself, which makes it very difficult nificant, and the overall mean effect was –0.02. to determine the effect of the campaign on the Again, there is no way one could conclude that outcome of the election. Now, how can a cam- negative ads or negative campaigns are any paign strategy, which logically must take place more or less effective than more positive ads prior to election day, be endogenous to (influ- or campaigns. Still, this methodological issue enced by) the outcome of the election, which must be kept in mind when reviewing the avail- we only know after the campaign is over and able literature or when thinking about the suc- the polls close on election day? This is possi- cess of any particular negative campaign. ble because many of the factors that help deter- At this point one may reasonably ask why, mine the outcome of the election are widely if attacks are not particularly effective, political known long before election day, and there- consultants continue to urge candidates to at- fore are likely to influence candidates’ choice tack their opponents. There are several quick of campaign strategies in the first place. Most answers. First, as discussed above, few political incumbents running for re-election are better campaigns are in fact overwhelmingly negative. known, better liked, and better financed than We are more likely to hear about negative cam- their challengers, and thus can reasonably ex- paigns in the media, but relatively few ads are pect to win ( Jacobson 2004). As we have seen, entirely negative, and most campaigns, on bal- challengers and incumbents unexpectedly find- ance, are probably more positive than negative. ing themselves in close elections are more likely Second, to say that all or even most political to decide to go negative ...but they are also attacks are not effective is not to say that at- more likely to lose. This makes it extremely tacks are never effective. Even if all elections difficult to determine if such a candidate lost involved only two candidates, half of all polit- because of her choice of campaign strategy, or ical campaigns would have to fail. Rarely are in spite of it. This statistical problem does have political consultants trained as scientists, and by Rutgers University Libraries on 11/01/10. For personal use only. a statistical solution: instrumental variables or they have few incentives—and even less time two-stage least squares. As Bartels (1991) warns, and money—to study the efficacy of their ac- Annu. Rev. Polit. Sci. 2009.12:285-306. Downloaded from www.annualreviews.org however, this solution is no better than the tions. Like most of us, they are likely to repeat quality of the instruments. See Lau & Pomper what they have done in the past. Furthermore, (2002) for further discussion of this issue and the time horizon of a consultant for judging the details on one reasonably successful attempt to success of a particular ad campaign is probably provide such instruments. shorter than that of a researcher, who inevitably Does this methodological problem account is not looking at the campaign until the election for the lack of empirical evidence for the ef- is over. All of these factors combine to make ficacy of negative campaigning? Probably not; negative campaigns a steady part of the politi- in their meta-analytic review of the litera- cal scene, even if they sometimes, or even most ture, Lau et al. (2007) report that experiments times, do not ultimately produce the results (which avoid the methodological problem dis- that are desired by the candidates who employ cussed above by randomly assigning campaign them.

www.annualreviews.org • Negative Campaigning 297 ANRV377-PL12-15 ARI 2 May 2009 12:49

THE DEMOBILIZATION negative ad were 5% less likely to intend to vote HYPOTHESIS in the upcoming election compared to those exposed to a positive ad. A subsequent study, The scholarly debate over the impact of nega- combining analysis of raw aggregate turnout tive advertising has centered on the demobiliza- and ballot roll-off data from all thirty-four 1992 tion hypothesis, developed by Ansolabehere, senate races with a tone-specific content anal- Iyengar, and colleagues (Ansolabehere et al. ysis of the news coverage of those campaigns, 1994, Ansolabehere & Iyengar 1995). Seeking yielded similar results. (Roll-off occurs when an to establish a causal link between the presumed individual only votes in the race at the top of the rise of negative advertising and the steadily ballot—president—and abstains from voting in declining turnout in national elections, An- the senate election that was on the same ballot.) solabehere et al. suggested three reasons why Ansolabehere et al. go on to report that political negative campaigns might lower turnout. First, independents seem particularly susceptible to negative campaigns could be successful in low- demobilization from exposure to negative ads, ering the probability of voting for the target and further report a general increase in polit- of the attacks without simultaneously increas- ical cynicism and lower sense of political effi- ing the probability of voting for the sponsor of cacy among those exposed to negative ads. In the attacks, resulting in a decision to stay home concluding that exposure to negative advertis- on election day. Second, negative campaigns ing does in fact reduce turnout and erode con- could lower evaluations of both the target and fidence in the political process, Ansolabehere the sponsor of the attacks, thus decreasing the et al. warn that because “candidates with suffi- probability of voting for either of them. Third, cient resources can, through the use of negative negative campaigns could influence turnout in- messages, keep voters away from the polls,” the directly through an effect on civic attitudes such phenomenon poses a unique threat to democ- as trust in government and political efficacy, racy (Ansolabehere et al. 1994, p. 835). which in turn are often associated with turnout. This is wonderful research, with clear theo- Ansolabehere et al. (1994, 1995) tested these retical predictions and amazingly consistent re- hypotheses by conducting a series of controlled sults across two very different research designs, experiments on the effects of exposure to a sin- and it did what all wonderful research should gle positive or negative political ad. Partici- do: generate a lot more research. Finkel & Geer pants, all of whom were drawn from the Los (1998) were among the first to respond. While Angeles area, were shown a 15-minute local acknowledging the logic of Ansolabehere et al.’s newscast and systematically exposed to one of three reasons why negative campaigns might by Rutgers University Libraries on 11/01/10. For personal use only. two experimental political ads (either positive decrease turnout, Finkel & Geer counter with or negative) embedded midway through the three reasons why negative campaigns might

Annu. Rev. Polit. Sci. 2009.12:285-306. Downloaded from www.annualreviews.org broadcast. While manipulating tone by chang- actually turnout. First, negative ad- ing key words within the script, both ads used stimulate vertising often conveys large amounts of pol- the same announcer and were visually and the- icy and retrospective performance information matically identical. In order to accurately re- to voters, and knowledgeable citizens are more flect real-world ad exposure, Ansolabehere et al. likely to participate. Second, negativity effects used authentic political ads spanning a variety of could make attack ads more useful than com- California campaigns, including the 1990 gov- parable positive ads in helping voters discrim- ernor’s race, the 1992 senate races (primaries inate between the two candidates, and people and general election), and the 1993 Los Angeles with distinct alternatives are more likely to vote. mayor’s race. Upon comparing ad exposure to Third, negative ads may be more likely to pro- each individual’s intention to vote and general duce strong emotional responses that can get level of political participation, Ansolabehere people to care more about the outcome of the et al. found that participants exposed to the election, which again should increase turnout.

298 Lau · Rovner ANRV377-PL12-15 ARI 2 May 2009 12:49

Finkel & Geer (1998) then turn to an empir- units. These raw effect sizes can then be ad- ical investigation of whether the negativism of justed for sampling error (so that studies with presidential election campaigns affects turnout. larger sample sizes count more than studies with They measured the tone of presidential elec- smaller sample sizes) and unreliability of mea- tion campaigns by the percentage of negative surement of the dependent variable. By meta- themes in the campaign ads aired by the major analytic standards, there is one very large pos- party candidates between 1960 and 1992. Finkel itive effect in this literature, one moderately & Geer use this variable to predict turnout in large positive effect, one moderately large neg- a thoroughly specified multivariate equation. ative effect, and 53 findings that hover a little Controlling for the standard demographic and above or a little below zero but do not amount political variables, they find no systematic rela- to much. In sum, the research literature pro- tionship between turnout and tone. Thus, here vides no general support for the hypothesis that is one study that adopted a different method- negative political campaigning depresses voter ology and studied a different type of elections, turnout. In fact, if you had to bet on one or but found no support for demobilization due to the other, mobilization is a bit more likely than exposure to negative campaigns. demobilization. The major weakness of this study, as Finkel Some of the individual findings in this liter- & Geer (1998) readily admit, is that campaign ature are based on very large samples. As is al- tone only varies across election years, provid- ways the case with statistics, particular findings ing very gross measurement of the crucial in- can be statistically significant but not represent dependent variable. When this study was pub- very large effects. For example, the experimen- lished, there were no better data available. Geer tal work reported by Ansolabehere & Iyengar & Lau (2006) try to overcome this limitation (1995) is easily statistically significant with a t with a Bayesian analysis that treats these yearly value of −2.56, but because it is based on a sam- averages as a baseline around which one can ple of 2216 subjects, its effect size is only −0.10. estimate plausible variations of tone across dif- At the same time, the accompanying analy- ferent electoral contexts (states)—and find very sis of 34 senate election campaigns from 1992 similar results. (Ansolabehere et al. 1999) has a larger t value But there have been many more tests of the (–3.11), but this represents a much larger effect demobilization hypothesis. The meta-analytic (–1.27) because it is based on only 34 cases. Of review of the literature by Lau et al. (2007) course we might trust a data point based on mentioned above identified 55 additional tests >2000 cases a lot more than one based on 34 of demobilization. Of the results of those tests, cases, and it is conventional in the meta-analytic by Rutgers University Libraries on 11/01/10. For personal use only. 25 were in the demobilization direction (nine literature to adjust all effect sizes for sampling statistically significant results) and 29 were in error, which has the effect of weighting findings Annu. Rev. Polit. Sci. 2009.12:285-306. Downloaded from www.annualreviews.org the mobilization direction (eight of which were based on larger samples relatively more than significant). This hypothesis has been such a fo- those based on a smaller number of cases. Lau cus of research in the field that we want to very et al. (2007) report the raw, unadjusted averages carefully illustrate the lack of any clear find- along with adjusted means; never do the various ings. Figure 3 displays the adjusted effect size measures differ by very much. for all 56 findings reported in the social sci- It could still be the case that although voters ence literature. An effect size is a way of ex- are not turned off by all campaign negativism, pressing the results of many different studies when legitimate criticism crosses the line into in the same metric. For the prototypic study, it mudslinging, demobilization occurs. But when represents the difference in turnout between a exactly does a critical ad cross the line? group exposed to a positive ad (or campaign) Goodman (1995), writing in Campaigns and and another group exposed to a negative ad Elections, a magazine for political consultants, (or campaign), expressed in standard deviation attempts to clarify the point: “Does it feel

www.annualreviews.org • Negative Campaigning 299 ANRV377-PL12-15 ARI 2 May 2009 12:49

negative? Is it mean-spirited? Is it crudely pro- negative information mobilizes participation, duced? Does it go beyond the pale of good taste particularly among independents and those and appropriate manners? Cross any of these who are generally more interested in or more lines, and you’ve crossed the viewer” (p. 23). knowledgeable about the political process. Besides the fact that production sophistication is hardly an operational measure, the author’s description of inappropriate negative advertis- METHODOLOGICAL CONCERNS ing reflects another problem: Like the percep- The major methodological issues plaguing the tion of campaign negativism, the “point of no study of political campaigns can be placed into return” is very much in the eye of the beholder. three broad categories: the lack of good data Although we can agree that presentations of on the very nature of political campaigns, the documented facts are fair game, we are also difficulties of measuring exposure to political aware that talented strategists are capable of campaigns, and the statistical assumptions that manipulating facts in such a way that they dis- are sometimes unwittingly made in our analysis tort the truth. Even topics that may be consid- of campaign effects. We are making progress on ered “off limits” (for example, those listed in the all of these fronts, however. The designs that bottom panel of Table 1) can be framed in such researchers use to study political campaigns are a way that they are perceived by some voters to one important vehicle for addressing some of be relevant. these problems. Yet Kahn & Kenney (1999) maintain that voters know the difference. They examine three distinct measures of the nature of the 1990 sen- Data ate election campaigns: the tone of up to five ads As discussed above, the most fundamental prob- aired by each candidate, the negativism of news- lem in the study of political campaigns involves paper coverage of the campaign, and, of particu- data: the lack, until very recently, of any good lar concern here, a judgment by the media con- evidence on exactly what candidates actually sultants involved in these campaigns of whether do when they are running for office. With re- the campaign degenerated into mudslinging. cent technological advances this information is Kahn & Kenney use all three of these measures becoming easier to gather, but there are still in a multivariate equation predicting turnout. major parts of campaigns that remain largely Controlling on a variety of demographic, psy- hidden from the researcher’s view. Before we chological, and political orientations, Kahn & had more objective measures of the actual con- Kenney find that both the negativism of the tent of political campaigns, some researchers by Rutgers University Libraries on 11/01/10. For personal use only. candidates’ political ads and the amount of asked survey respondents to recall an adver- criticism they received in newspaper cover- tisement from some campaign, code the na- Annu. Rev. Polit. Sci. 2009.12:285-306. Downloaded from www.annualreviews.org age were associated with higher turnout. Cam- ture of the remembered ad, and then try to paigns involving mudslinging, however, had judge the impact of that ad on the voter. Alter- significantly lower turnout. More fine-grained natively, researchers asked respondents to rate analyses suggest that political independents and how positive or negative a particular campaign those with little interest in politics were partic- has been. But memory is notoriously unreliable, ularly affected by mudslinging. Although other and we have already seen that subjective per- researchers have had difficulty replicating this ceptions of how negative a campaign is suffer finding (see Brooks & Geer 2007, Jackson & from serious partisan biases. Noisy data about a Sides 2006), Kahn & Kenney’s research pro- phenomenon are always better than no data at vides some evidence that voters are capable of all, but more objective, systematically gathered discerning malice from legitimate critiques, and measures should almost always be preferred to that it is the former that turns voters off from less reliable subjective measures. Today there politics, whereas legitimate, campaign-relevant should be no reason to have to rely on memory

300 Lau · Rovner ANRV377-PL12-15 ARI 2 May 2009 12:49

of political advertisements from any campaign supporters (Cornfield 2004; Druckman et al., as the primary measure of that campaign’s tone. unpublished manuscript). Limiting attention to We have already mentioned the Campaign only one aspect of a political campaign provides Media Analysis Group (CMAG), which since at best an incomplete image, and at worst a mis- 1998 has tracked every political advertisement leading picture, of the entire campaign. Yet few aired on television in the 75 largest media mar- researchers have the resources to consider more kets (which include 80% of the population) in than one or (rarely) two of these different as- the United States.3 Among campaign profes- pects of a campaign. It is vitally important that sionals, the television and radio advertisements any researcher who gathers data on the nature are often called the “air war.” But there is also a of any of the multiple components of a politi- “ground war,” which consists of the speeches cal campaign make that data available to other candidates deliver and campaign visits they researchers studying campaigns. make to different constituencies, the bulk mail- ings they send out, the billboards, lawn signs, bumper stickers, and get-out-the-vote phone Exposure calls. Any candidate blessed with sufficient re- In gauging the effects of any aspect of a cam- sources to advertise on television also typically paign, one also must be careful to distinguish invests heavily in different aspects of the ground the content or tone of the campaign from its war. In fact, because ground war activities are volume. In using the CMAG data to mea- not included in the definition of electioneering sure negative advertising campaigns, for ex- communication by the Bipartisan Campaign ample, some researchers have mistakenly em- Reform Act of 2002, there are now incentives ployed the number of negative ads shown in for parties to shift their campaign activities to- a media market as their primary measure of ward the ground (Dwyre & Kolodny 2003). campaign negativism. Without a comparable It is usually possible to obtain a fairly com- measure of the number of positive ads shown plete record of the campaign stops a candidate in that same market, this approach confounds makes from media accounts (see Shaw 1999), the nature of the advertising campaign with its but we have not seen anyone systematically col- magnitude, as Dan Stevens details in his unpub- lect this information except at the presidential lished manuscript, “More than Volume: Pro- level. Magelby et al. (2006) describe a new data portion, Volume, and the Effects of Exposure collection effort in which a random sample of to Negative Advertising.” The resulting regres- registered voters were asked to collect every bit sion coefficient will be estimating the effect of of campaign mail and log every political phone one more negative ad, all else (in the equation) by Rutgers University Libraries on 11/01/10. For personal use only. call and personal contact they received during equal, that is, irrespective of the number of pos- the last three weeks of the 2004 presidential itive ads that have also been shown. This makes Annu. Rev. Polit. Sci. 2009.12:285-306. Downloaded from www.annualreviews.org election (see also Hillygus & Monson 2006). little sense. We recommend devising separate But this is only one election, and the focus is on measures of the nature or tone of the campaign the presidential campaign. Today we must also (e.g., the proportion of all ads shown in a par- speak of the “cyber war,” the candidates’ offi- ticular market that are negative) and of its vol- cial web pages and the less official pages of their ume (the total number of ads shown in that market). Even if we have a very good idea about how 3 The Wisconsin Political Advertising Project also has much a candidate advertises in a particular me- CMAG data from the 1996 presidential election, but for no other races that election year. CMAG also records radio ads dia market, we still do not know how many and the text of many local and cable news programs (to the campaign messages any individual in that me- extent their content has been transcribed for the hearing im- dia market has been exposed to. There is a large paired), but these data are not stored at Wisconsin and are thus much more difficult for political scientists to obtain. We random element to individual exposure levels, have never seen anyone utilize these other CMAG data. but there is a larger nonrandom component to

www.annualreviews.org • Negative Campaigning 301 ANRV377-PL12-15 ARI 2 May 2009 12:49

exposure because most of us are creatures of general this result will prove to be is still an habit. Some of us watch three or four hours open question. of television almost every day whereas oth- We have already discussed the problem of ers watch an hour or less. Some of us watch the endogeneity of the decision of how to cam- television in the morning while getting ready paign with the outcome of the election, which for the day to start, others in the afternoon makes causal statements about the likely effects before the kids come home from school, and of those campaign strategies much more diffi- others exclusively during the prime-time hours cult to make. Holding that problem aside, there in the evening. Freedman & Goldstein (1999) is also the issue that every effort by one can- have devised survey-based measures of typi- didate to convince citizens to vote for her is cal viewing habits that can be combined with typically offset by the efforts of another can- objective measures of the number and mix of didate trying to convince those same citizens ads that were shown at various times of day. to vote for him. Cast in this light, it may seem This is analogous to, although more compli- miraculous whenever we can detect any effect of cated than, asking survey respondents which (if a campaign on its outcome. Fortunately, there any) newspaper they typically read before es- is still a great deal of variance in the conduct timating the impact of editorial endorsements of campaigns that gives us leverage to esti- on the vote choice. Best practice now dictates mate their effects. Even if candidates had the trying to combine some measure(s) of the con- same amount of money to spend (as presiden- tent of a media message with measures of its tial candidates do in the United States, if they likelihood of being heard, either through global accept public financing), they do not have to measures of volume or, ideally, with the addi- spend it in the same way, nor in the same lo- tion of more idiosyncratic measures of individ- calities. Competing candidates typically do not ual viewing habits. try to deliver the same message or focus on the same issues in their advertising, nor are indi- vidual candidates required to deliver the same Statistical Assumptions speeches or use the same advertisements in ev- Whenever we study a random sample of vot- ery locale within their constituency. Such vari- ers but measure media content at some ag- ance is great for analytic purposes, but we must gregated level (such as the media market), we remember to take it into account in our data- introduce another source of potential bias to gathering efforts. The Federal Election Com- our studies: data clustering. Inevitably with mission, for example, has since its inception such research designs, individual respondents kept very good records on how much money by Rutgers University Libraries on 11/01/10. For personal use only. are nested within media markets, which vio- candidates and parties raise, who they raise it lates basic assumptions of our most common from, and how much they spend. But there is Annu. Rev. Polit. Sci. 2009.12:285-306. Downloaded from www.annualreviews.org multivariate statistical techniques (Steenbergen almost no reporting on where the money is & Jones 2002). Unfortunately, following “best spent. practices” in media research does not yet in- The study of political campaigns is made clude explicitly modeling the multilevel struc- even more difficult by the fact that in most ture of our data, and the conclusions from that elections there are multiple races on the bal- research must consequently be somewhat sus- lot. Every two years all citizens face an elec- pect. One recent study found that for the most tion selecting their member of the House of part, the results of multilevel analysis converge Representatives, but usually there is at least with those of conventional methods (Lau & one and sometimes more simultaneously oc- Steenbergen 2006), but this result could be due curring statewide races for senator, governor, to the small intraclass correlations observed in lieutenant governor, and so on. And every four the three datasets examined in that paper. How years the presidential election is also on the

302 Lau · Rovner ANRV377-PL12-15 ARI 2 May 2009 12:49

ballot. How much information are we losing and their political consultants during an actual by the almost universal practice of focusing political campaign, however. If this is one’sgoal, on only one of these campaigns and ignoring then longitudinal designs, which gather infor- every other race on the ballot? If our goal is mation (e.g., relative standing in the polls) and to study the effectiveness of a particular cam- measure behavior (the volume and tone of cam- paign strategy, the practice might be justified. paigning in different media markets) at multi- If instead we are interested in the indirect and ple points in time across an entire campaign, are presumably unintentional effects of negative crucial. We as scientists can wait to analyze the campaigns on the political system as a whole success of any advertising campaign until after (e.g., political efficacy, turnout), the practice it is over, but campaigners must make their de- seems much harder to justify. Most voters do cisions during the heat of battle, so to speak, not make race-by-race decisions of whether to and undoubtedly under conditions of consider- vote—they make one universal decision to go to able uncertainty. Fortunately the CMAG data the polls or stay home, and to fully understand are collected with very precise time stamps, so that universal decision, we must consider all of we know exactly when and where different ads the political stimuli to which they have been are shown. It is typically much harder to get subjected. comparably precise measures of the public’s re- sponse to any advertising campaign, but many campaigns conduct tracking polls that provide Research Designs exactly those data—if we can get a hold of them. All of the methodological issues raised so far Developing personal relationships with polit- apply to the problems of studying real-world ical consultants who often manage high-level political campaigns. Most of these problems political campaigns can prove invaluable to such are solved by experiments, although now the an enterprise [see Sides (2006) for a recent study natural focus becomes studying the effects of utilizing this type of data]. particular campaign events (e.g., individual campaign ads) rather than the effects of more extended campaigns. This review is not the CONCLUSION: ARE NEGATIVE place to discuss the relative merits of experi- CAMPAIGNS BAD FOR mental and nonexperimental designs, or even DEMOCRACY? the desirability of conducting “realistic” as The literature on the use and utility of negative opposed to more artificial or “ideal setting” campaigns is extensive, including descriptive experiments. Some degree of artificiality is in- accounts, systematic experimentation, survey by Rutgers University Libraries on 11/01/10. For personal use only. evitable in almost any controlled environment, research analysis, and formal theoretical pre- but particularly in a discipline such as political scriptions. Although we have not had the space Annu. Rev. Polit. Sci. 2009.12:285-306. Downloaded from www.annualreviews.org science, many of whose practitioners have not to present a comprehensive review of all of this been trained as experimental scientists, the ex- research, we do believe we have presented a perimental methodology developed by Iyengar thorough one. Descriptive accounts, many of & Kinder (1987) is about as good as it gets, which are by political consultants rather than providing high levels of internal and external academics, address the strategic concerns guid- validity. In the Iyengar & Kinder protocol, a ing the decision to attack. Empirical investi- carefully controlled experimental manipulation gations include both systematic manipulations is unobtrusively delivered in a very familiar of hypothesized causal factors, analysis of me- setting—for example, an experimental ad is dia content—real, as well as fictitious—and sur- inserted into the normal commercial break in vey analysis of voter reactions to actual political a tape of an actual television program. campaigns. Experiments may not be as useful in trying Yet there is nothing in what we have pre- to understand the decisions made by candidates sented so far that would allow us to say that

www.annualreviews.org • Negative Campaigning 303 ANRV377-PL12-15 ARI 2 May 2009 12:49

negative campaigns are inherently good or bad. to tell us its vacuum cleaners sometimes fail to Many recognize that voters are capable of sep- suck. That is the job of the opposition, in pol- arating vicious attacks from legitimate compar- itics as well as in commerce. Geer argues that isons but also appreciate that even justifiable democracy requires negative campaigning if cit- attacks are not always effective, nor are they izens are to become fully informed about the always desirable. The standards for making choices they face. such judgments should be the public’s inter- Although Ansolabehere & Iyengar (1995) ests, not the candidates’. Elections are, after conclude their book with policy recommenda- all, about choices, not courtesy (Lau & Pomper tions aimed at limiting the extent of negative 2004), so questions about the effectiveness of campaigning, we see no need to impose any fur- negative campaigns become irrelevant. ther restrictions on negative advertising. Voters The literature does reflect an overall in- are clearly capable of distinguishing legitimate dictment of baseless attacks and unwarranted criticism from unfair attacks and punish can- character assaults. Beyond that, however, exists didates who are out of line. We do approve of a spectrum of views. The widespread concern recent legislation requiring federal candidates that negative campaigns can reduce turnout ap- to appear in and publicly approve of the ads pears unwarranted. There is less concern but they sponsor, as this increases accountability in stronger evidence that negative campaigns can election campaigns and exposes those individu- deleteriously affect the political system in other als who would prefer to shield themselves from ways, for example, by reducing political efficacy boomerang effects by allowing surrogates to do and decreasing trust in government (Lau et al. the attacking. Of course what this change has 2007). But negative campaigns also stimulate really done is push most of the attacking off to interest and learning about the candidates and party-sponsored ads, but as long as voters can issues in an election. hold the parties accountable for their actions, Is it unfair to objectively point out an oppo- that too is fine. nent’s shortcomings? Conversely, are so-called Although it is difficult to convincingly docu- “positive” ads that boast of a candidate’s ment the widely presumed increase in negative strength to the point of exaggerated self- campaigning over recent decades, no one is sug- aggrandizement fair, simply because they are gesting the practice is on the decline, and there not negative? Mayer (1996) points out that no is little doubt that negative campaigns are an candidate can speak honestly and substantively important part of the current political scene. about what he intends to do upon election with- Thus we are sure that negative campaigning out drawing a contrast to his opponent and ex- will remain fertile ground for political science by Rutgers University Libraries on 11/01/10. For personal use only. plaining why he is a superior choice over the research for years to come. Future research other candidate. If we assume that voters want should explore the impact of negative ads that Annu. Rev. Polit. Sci. 2009.12:285-306. Downloaded from www.annualreviews.org to know why a given candidate is better suited are sponsored by “independent” (504) groups or more qualified than his or her opponent, we and individuals other than the candidates them- cannot legitimately fault a candidate for tact- selves. It would be good to examine the ex- fully juxtaposing him- or herself against that tent to which the messenger matters, and how opponent. attacks from third parties are weighted rela- Geer (2006) takes this argument even far- tive to those launched personally by candidates. ther. All advertisements, by their very nature, How does the timing of attacks during a cam- exaggerate the truth. But this point applies paign, or the medium through which attacks equally to positive campaigns as well as to neg- are delivered, influence their effectiveness? Is ative. We cannot expect candidates to discuss the internet a medium where almost anything their own policy failures and shortcomings, any goes, and candidates can “get away with” more more than we can expect Chevrolet to tell us vicious attacks of their opponents? What is the about the problems with its cars, or Hoover role of music, production quality, or particular

304 Lau · Rovner ANRV377-PL12-15 ARI 2 May 2009 12:49

types of images in the effectiveness or effects of attack and its impact on political behavior, a negative advertisements? Despite the volume myriad of questions remain for future research of research already available on the decision to to address.

DISCLOSURE STATEMENT The authors are not aware of any affiliations, memberships, funding, or financial holdings that might be perceived as affecting the objectivity of this review.

LITERATURE CITED Ansolabehere S, Iyengar S. 1995. Going Negative. New York: Free Press Ansolabehere S, Iyengar S, Simon A. 1999. Replicating experiments using aggregate and survey data: the case of negative advertising and turnout. Am. Polit. Sci. Rev. 93:901–9 Ansolabehere S, Iyengar S, Simon A, Valentino N. 1994. Does attack advertising demobilize the electorate? Am. Polit. Sci. Rev. 88:829–38 Bartels L. 1991. Instrumental and “quasi-instrumental” variables. Am. J. Polit. Sci. 35( July):777–800 Brader T. 2005. Striking a responsive chord: how political ads motivate and persuade voters by appealing to emotions. Am. J. Polit. Sci. 49(Apr.):388–405 Brader T. 2006. Campaigning for Hearts and Minds. Chicago: Univ. Chicago Press Brooks D, Geer J. 2007. Beyond negativity: the effects of incivility on the electorate. Am. J. Polit. Sci. 51( Jan.):1– 16 Buell E, Sigelman L. 2008. Attack Politics: Negativity in Presidential Campaigns since 1960. Lawrence: Univ. Press Kansas Cornfield M. 2004. Politics Moves Online: Campaigning and the Internet. New York: Century Found. Press Dwyre D, Kolodny R. 2003. National parties after BCRA. In Life after Reform: When the Bipartisan Campaign Reform Act Meets Politics, ed. M Malbin, pp. 95–112. Lanham, MD: Rowman & Littlefield Finkel S, Geer J. 1998. A spot check: casting doubt on the demobilizing effect of attack advertising. Am. J. Polit. Sci. 42:573–95 Freedman P, Goldstein K. 1999. Measuring media exposure and the effects of negative campaign ads. Am. J. Polit. Sci. 43:1189–208 Freedman P, Wood W, Lawton D. 1999. Do’s and don’ts of negative ads: what voters say. Campaigns Elections 20:20–25 Garramone G. 1984. Voter response to negative political ads. J. Q. 61:250–59 Geer J. 2006. In Defense of Negativity: Attack Ads in Presidential Campaigns. Chicago: Univ. Chicago Press

by Rutgers University Libraries on 11/01/10. For personal use only. Geer J, Lau R. 2006. A new approach for estimating campaign effects. Br. J. Polit. Sci. 35:269–90 Goodman A. 1995. Producing TV: a survival guide. Campaigns Elections 16:22–24 Harrington J, Hess G. 1996. A spatial theory of positive and negative campaigning. Games Econ. Behav. 17:209– Annu. Rev. Polit. Sci. 2009.12:285-306. Downloaded from www.annualreviews.org 29 Haynes A, Rhine S. 1998. Attack politics in presidential nomination campaigns: an examination of the frequency and determinants of intermediated negative messages against opponents. Polit. Res. Q. 51:691–721 Hillygus S, Monson J. 2006. Campaign microtargeting and presidential voting in 2004. Presented at Annu. Meet. Am. Polit. Sci. Assoc., 102nd, Philadelphia Iyengar S, Kinder D. 1987. News that Matters. Chicago: Univ. Chicago Press Jackson R, Sides J. 2006. Revisiting the influence of campaign tone in senate elections. Polit. Anal. 14(Spring):206–18 Jacobson G. 2004. The Politics of Congressional Elections. New York: Longman. 6th ed. Jamieson KH, Campbell C. 1983. The Interplay of Influence: Mass Media and their Publics in News, Advertising, Politics. Belmont, CA: Wadsworth Jamieson KH, Waldman P, Sherr S. 2000. Eliminate the negative? Categories of analysis for political adver- tisements. In Crowded Airwaves, ed. JA Thurber, CJ Nelson, DA Dulio, pp. 44–64. Washington, DC: Brookings Inst.

www.annualreviews.org • Negative Campaigning 305 ANRV377-PL12-15 ARI 2 May 2009 12:49

Kahn KF. 1993. Gender differences in campaign messages: the political advertisements of men and women candidates for U.S. Senate. Polit. Res. Q. 46:481–502 Kahn KF, Kenney P. 1999. Do negative campaigns mobilize or suppress turnout? Clarifying the relationship between negativity and participation. Am. Polit. Sci. Rev. 93:877–89 Kahn KF, Kenney P. 2000. How negative campaigning enhances knowledge of senate elections. In Crowded Airwaves, ed JA Thurber, CJ Nelson, DA Dulio, pp. 65–95. Washington, DC: Brookings Inst. Kahn KF, Kenney P. 2004. No Holds Barred: Negativity in U.S. Senate Campaigns. Upper Saddle River, NJ: Pearson Educ. Kamber V. 1997. Poison Politics: Are Negative Campaigns Destroying Democracy? Cambridge, MA: Perseus Kanouse D, Hanson L Jr. 1971. Negativity in evaluations. In Attribution: Perceiving the Causes of Behavior, ed. EE Jones, DE Kanouse, HH Kelley, RE Nisbett, S Valins, B Weiner, pp. 47–62. Morristown, NJ: General Learning Kern M. 1989. Thirty Second Spots: Political Advertising in the Eighties. New York: Praeger Kernell S. 1977. Presidential popularity and negative voting: an alternative explanation of the midterm con- gressional decline of the president’s party. Am. Polit. Sci. Rev. 71(Mar.):44–66 King E, Hendersen R, Chen H. 1998. Viewer response to positive vs negative ads in the 1996 presidential campaign. Presented at Annu. Meet. Midwest Polit. Sci. Assoc., Chicago Lau R. 1982. Negativity in political perception. Polit. Behav. 4(Dec.):353–78 Lau R. 1985. Two explanations for negativity effects in political behavior. Am. J. Polit. Sci. 29(Feb.):119–38 Lau R, Pomper G. 2001. Negative campaigning by U.S. Senate candidates. Party Polit. 7( Jan.):69–87 Lau R, Pomper G. 2002. Effectiveness of negative campaigning in U.S. Senate elections. Am. J. Polit. Sci. 46:47–66 Lau R, Pomper G. 2004. Negative Campaigning: An Analysis of U.S. Senate Elections. New York: Rowman & Littlefield Lau R, Sigelman L, Heldman C, Babbitt P. 1999. The effects of negative political advertisements: a meta- analytic assessment. Am. Polit. Sci. Rev. 93:851–75 Lau R, Sigelman L, Rovner IB. 2007. The effects of negative political campaigns: a meta-analytic reassessment. J. Polit. 69(Nov.):1176–209 Lau R, Steenbergen M. 2006. Reconsidering “best practices” in political communications research. Presented at Annu. Meet. Midwest Polit. Sci. Assoc., 64th, Chicago Magelby D, Monson JQ, Patterson K. 2006. Mail communications in political campaigns: the 2004 campaign communications survey. Presented at Annu. Meet. Midwest Polit. Sci. Assoc., Chicago Mansnerus L. 2005. Forrester tosses his “Hail Mary” ad. NY Times, Oct. 23, p. NJ14 Marcus G, Neuman WR, MacKuen M. 2000. Affective Intelligence and Political Judgement. Chicago: Univ. Chicago Press Mayer W. 1996. In defense of negative campaigning. Polit. Sci. Q. 111:437–55 Proctor D, Schenck-Hamlin W, Haase K. 1994. Exploring the role of gender in the development of negative political advertisements. 14(2):1–22 by Rutgers University Libraries on 11/01/10. For personal use only. Women Polit. Roese N, Sande G. 1993. Backlash effects in attack politics. J. Appl. Soc. Psychol. 23(Aug.):632–53 Shaw D. 1999. The effect of TV ads and candidate appearances on statewide presidential votes, 1988–1996.

Annu. Rev. Polit. Sci. 2009.12:285-306. Downloaded from www.annualreviews.org Am. Polit. Sci. Rev. 93( June):345–61 Sides J. 2006. Swing your partner? The importance of interactions in campaigns. Presented at Annu. Meet. Am. Polit. Sci. Assoc., Washington, DC Sides J, Grossmann M, Trost C, Lipsitz K. 2003. Candidate attacks and voter aversion: the uncertain link between negativity and campaign satisfaction. Presented at Annu. Meet. Am. Polit. Sci. Assoc., Philadelphia Sigelman L, Kugler M. 2003. Why is research on the effects of negative campaigning so inconclusive? Un- derstanding citizens’ perception of negativity. J. Polit. 65:142–60 Skaperdas S, Grofman B. 1995. Modeling negative campaigning. Am. Polit. Sci. Rev. 89:49–61 Steenbergen M, Jones B. 2002. Modeling multilevel data structures. Am. J. Polit. Sci. 46( Jan.):218–37 Swint K. 1998. Political Consultants and Negative Campaigning: The Secrets of the Pros. Lanham, MD: Univ. Press Am. Theilmann J, Wilhite A. 1998. Campaign tactics and the decision to attack. J. Polit. 60:1050–62 West D. 1993. Air Wars: Television Advertising in Election Campaigns, 1952–1992. Washington, DC: CQ Press West D. 2005. Air Wars: Television Advertising in Election Campaigns, 1952–2004. Washington, DC: CQ Press. 4th ed.

306 Lau · Rovner ANRV377-PL12-15-Lau.qxd 4/7/09 12:21 Page 1

60 Positive 50 40 Negative 30

Televised 20 Contrast

political ads (%) 10

1998 2000 2002 Year

Figure 1 Positive, negative, and contrast televised political ads in recent U.S. federal elections.

250 a 200

150

100

50 negative campaigning Newspaper stories about 1980 1984 1988 1992 1996 2000 2004 Year

20 b Political science 15 Communications/marketing Psychology economics

by Rutgers University Libraries on 11/01/10. For personal use only. 10

5 Annu. Rev. Polit. Sci. 2009.12:285-306. Downloaded from www.annualreviews.org Academic studies of Academic

negative campaigning (%) 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 Year

Figure 2 (a) Number of newspaper stories about negative political campaigns. (b) Academic studies of negative campaigning, by discipline.

www.annualreviews.org ● Negative Campaigning C-1 ANRV377-PL12-15-Lau.qxd 4/7/09 12:21 Page 2

1.5000

1.0000

0.5000

0.0000

Effect size (Δ) –0.5000

S = Surveys –1.0000 X= Experiments A= Aggregate

–1.5000

SSSSSXXXXXXXXXXXXXXXAAAAAAAAAA SSSSSSSSSSSSSSSSSSSSS XXXXX A

Actual turnout Intended turnout Studies testing the demobilization hypothesis

Figure 3 Do negative campaigns demobilize the electorate? This figure displays the effect size associated with every study testing the demobi- lization hypothesis, adjusted for sampling error and measurement unreliability (Lau et al. 2007). Data have been sorted by whether the dependent variable is actual (reported) vote or intended vote, and within that distinction, by methodology. Negative effect sizes are consistent with the demobilization hypothesis. by Rutgers University Libraries on 11/01/10. For personal use only. Annu. Rev. Polit. Sci. 2009.12:285-306. Downloaded from www.annualreviews.org

C-2 Lau ● Rovner AR377-FM ARI 7 April 2009 11:25

Annual Review of Political Science Contents Volume 12, 2009

A Conversation with Robert A. Dahl Robert A. Dahl and Margaret Levi ppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp1 Neorepublicanism: A Normative and Institutional Research Program Frank Lovett and Philip Pettit ppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp11 Domestic Terrorism: The Hidden Side of Political Violence Ignacio Sánchez-Cuenca and Luis de la Calle pppppppppppppppppppppppppppppppppppppppppppppppp31 Women in Parliaments: Descriptive and Substantive Representation Lena Wängnerud pppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp51 Self-Government in Our Times Adam Przeworski pppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp71 Social Policy in Developing Countries Isabela Mares and Matthew E. Carnes ppppppppppppppppppppppppppppppppppppppppppppppppppppppp93 Variation in Institutional Strength Steven Levitsky and María Victoria Murillo pppppppppppppppppppppppppppppppppppppppppppppp115 Quality of Government: What You Get Sören Holmberg, Bo Rothstein, and Naghmeh Nasiritousi pppppppppppppppppppppppppppppppp135

by Rutgers University Libraries on 11/01/10. For personal use only. Democratization and Economic Globalization Helen V. Milner and Bumba Mukherjee ppppppppppppppppppppppppppppppppppppppppppppppppppp163

Annu. Rev. Polit. Sci. 2009.12:285-306. Downloaded from www.annualreviews.org Has the Study of Global Politics Found Religion? Daniel Philpott pppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp183 Redistricting: Reading Between the Lines Raymond La Raja ppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp203 Does Efficiency Shape the Territorial Structure of Government? Liesbet Hooghe and Gary Marks ppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp225 Bargaining Failures and Civil War Barbara F. Walter ppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp243 Hobbesian Hierarchy: The Political Economy of Political Organization David A. Lake ppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp263

v AR377-FM ARI 7 April 2009 11:25

Negative Campaigning Richard R. Lau and Ivy Brown Rovner pppppppppppppppppppppppppppppppppppppppppppppppppppp285 The Institutional Origins of Inequality in Sub-Saharan Africa Nicolas van de Walle ppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp307 Riots Steven I. Wilkinson pppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp329 Regimes and the Rule of Law: Judicial Independence in Comparative Perspective Gretchen Helmke and Frances Rosenbluth ppppppppppppppppppppppppppppppppppppppppppppppppp345 Field Experiments and the Political Economy of Development Macartan Humphreys and Jeremy M. Weinstein ppppppppppppppppppppppppppppppppppppppppp367 Laboratory Experiments in Political Economy Thomas R. Palfrey ppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp379 Field Experiments on Political Behavior and Collective Action Eline A. de Rooij, Donald P. Green, and Alan S. Gerber ppppppppppppppppppppppppppppppppp389 Experiments on Racial Priming in Political Campaigns Vincent L. Hutchings and Ashley E. Jardina pppppppppppppppppppppppppppppppppppppppppppppp397 Elections Under Authoritarianism Jennifer Gandhi and Ellen Lust-Okar ppppppppppppppppppppppppppppppppppppppppppppppppppppp403 On Assessing the Political Effects of Racial Prejudice Leonie Huddy and Stanley Feldman pppppppppppppppppppppppppppppppppppppppppppppppppppppppp423 A “Second Coming”? The Return of German Political Theory Dana Villa ppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp449 Group Membership, Group Identity, and Group Consciousness:

by Rutgers University Libraries on 11/01/10. For personal use only. Measures of Racial Identity in American Politics? Paula D. McClain, Jessica D. Johnson Carew, Eugene Walton, Jr.,

Annu. Rev. Polit. Sci. 2009.12:285-306. Downloaded from www.annualreviews.org and Candis S. Watts ppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp471 Opiates for the Matches: Matching Methods for Causal Inference Jasjeet Sekhon pppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppppp487

Indexes

Cumulative Index of Contributing Authors, Volumes 8–12 ppppppppppppppppppppppppppppp509 Cumulative Index of Chapter Titles, Volumes 8–12 ppppppppppppppppppppppppppppppppppppp511

Errata

An online log of corrections to Annual Review of Political Science articles may be found at http://polisci.annualreviews.org/

vi Contents