A Field Experiment on the Effects of Negative Campaign Mail on Voter Turnout in a Municipal Election

Article excerpt

This field experiment is used to expose a random sample of voters in a 2003 mayoral race to various pieces of negative direct mail advertising. Exposure to the negative advertising stimulus improved turnout overall about 6 percent over that of the control group. Results show that different topics and amounts of negative advertising had different effects on turnout. The results suggest that alarm bells sounded by some previous research and by public officials may be overheated, because the effects of campaign negativity may not be monolithic, and it would appear political negativity can have a positive effect on turnout.

Is voter turnout subject to the effects of negative advertising? Political science research answers alternatively yes, no, or maybe. This study uses a field experiment in which voters in a mayoral contest were randomly exposed to negative campaign mail to assess the effects of negativity and move toward a better understanding of what has become a thoroughly confusing line of scholarship.

Indeed two of the most prominent studies on campaign advertising offer quite differing views on the effects of negativity. Ansolabehere and lyengar (1995) conclude that negative ads directly result in lower voter turnout. Far from qualifying their results, Ansolabehere and lyengar (1995:12) assert the evidence is definitive that negative campaign messages "pose a serious threat to democracy" and are "the single biggest cause" of public disdain for politics (2). By contrast, Green and Gerber (2004: 59) describe the effect of campaign advertising negativity as "slight." Depending on the circumstances, Green and Gerber find negativity modestly nudging turnout upwards or downwards. Far from labeling their results conclusive, however, Green and Gerber suggest much more work needs to be done to better understand negativity's effect.

While this study addresses Green and Gerbers call for continuing research on this question, studying the effects of campaign negativity is of value beyond simply satisfying an academic curiosity. Understanding the effects of negativity obviously has implications for how candidates, parties, and interest groups conduct campaigns. Moreover, various government bodies have expressed interest in some form of negative ad regulation. Legislative proposals have been introduced at the local, state, and national level to limit negative campaigning with measures such as forcing candidates to appear in their ads or subjecting political advertising copy to some form of official scrutiny. Indeed, "1 would ban negative ads," says Senator John McCain (R-Arizona) of the legislation he would create if he could find a constitutional procedure to accomplish the task.1 Thus, to understand negativity and its effects better is to become better armed to participate in a debate which pits the First Amendment against the very popular notion of cleaning up campaigns.

NEGATIVITY AND ITS EFFECTS

While there is no consensus definition of negative advertising, most researchers start with the notion that negativity involves the invoking of an opponent by a candidate (for example, Djupe and Peterson 2002). That is, a negative ad suggests the opponent should not be elected rather than that the sponsoring candidate should be elected. West (2001) defines a negative campaign ad as advertising that focuses at least 50 percent of its attention on the opponent rather than the sponsor of the ad. Such negativity may be focused on any aspect of the opponents record, statements, campaign, or background.

Precise estimates vary, but there is no doubt that negativity occupies a significant place in the modern campaign advertising arsenal. In the 2000 presidential election, for example, content analyses of television commercials from the two parties' nominees found between half and 70 percent were negative (Benoit et al. 2003; West 2001). Other forms of communication, such as radio ads, were even more negatively oriented (Benoit et al. …