This article examines the effects of negative political advertising in American elections. Studies of negative advertising have become unproductively bogged down on the question of whether or not more or less exposure to negative advertising leads to higher or lower turnout. I take a step back in the causal chain by looking at its effect on the kinds of information individuals bring to the voting decision. I present and test a theory that individuals lower in political sophistication gain little or no information from negative advertising. In contrast, higher sophisticates gain a great deal of information. The theory is tested in an experiment and through analysis of American National Election Study (ANES) data. The results are largely confirmatory. Analysis of the ANES data also illustrates that the effects of exposure to negative advertising on information levels differ from, and are more normatively troubling than, those of positive advertising.
Campaigns are the principal arena in which public officials and ordinary citizens interact over matters of public policy. In American elections political advertising is perhaps the central medium of campaign communication from rulers to ruled: candidates and their supporters spend vast amounts of money on advertising, and political advertising provides more televised information about candidates than does news (Kern 1989; Patterson and McClure 1976). The fact that political advertising is also frequently derided as shallow, manipulative, and ultimately detrimental to informed choice (e.g., Kamber 1997), particularly as according to conventional wisdom it has become more negative, has produced concern about its effects, not only in the United States but in other nations where elections are increasingly "Americanized" (Aronoff 2000; Lee, Tak, and Kaid 1998; Schoenbach 1996).
Disquiet over a potential connection between negative political advertising, the modern campaign, and public disillusionment with politics and elections in America has prompted a wealth of research.1 Much of it has drawn conclusions that are far more sanguine than Kambers (Finkel and Geer 1998; Freedman and Goldstein 1999; Goldstein and Freedman 2002; Kahn and Kenney 1999a), but some has not (Ansolabehere and Iyengar 1995; Houston and Roskos-Ewoldsen 1999). Most of this research explores the direct behavioral impact of negative political advertising, especially on turnout, while there has been far less examination of the informational effects that precede behavior. Indeed, Lau et al.s (1999) exhaustive meta-analysis of the most prominent academic studies of negative advertising includes only three unpublished works that looked at its impact on information about candidates.
This is surprising for two reasons. First, one of the fundamental roles of campaigns is in helping citizens to make an informed choice. Campaigns should teach citizens the "attitudes, temperaments, and competence of candidates; their policy commitments and intentions; their past actions in both public and private life; their party and other affiliations" (Kelley 1960, 12). More informed citizens are better able both to identify and articulate their true interests and are more likely to participate in elections (Delli Carpini and Keeter 1996; Downs 1957; Verba and Nie 1972); both are fundamental to democracy. Second, given that there is so much contention about the size and direction of negative advertising's behavioral effects one might expect an enhanced focus on possible explanations for the conflicting findings. The premise of this study is that rather than continuing to engage in more studies of the direct behavioral impact of negative advertising, the potential for greater understanding lies in going further back in the causal chain. If negative advertising affects political behavior such as turnout, it is surely the product of a multitude of reactions and responses to exposure in which changes in cognitive and affective information about candidates-leading to or from indifference or alienation (Downs 1957), for example-precede the behavioral decision. …