People say they don't like negative political ads, but do they work?

—Matthew Robison, Contoocook, N.H.

Donald Green, professor of political science at Columbia University, answers:

For 20 years political scientists have investigated whether negative campaign ads, those that criticize the opponent, are more effective than positive ads, which extol the sponsoring candidate. Yet the jury is out on whether “going negative” pays off.

A comprehensive literature analysis published in 2007 in the Journal of Politics examined the effects of political ads. The authors reported that negative ads tended to be more memorable than positive ones but that they did not affect voter choice. People were no less likely to turn out to the polls or to decide against voting for a candidate who was attacked in an ad.

Though noteworthy, this study did not settle the debate. The research analyzed was limited to surveys and laboratory experiments, both of which have drawbacks. The typical survey looks at the correlation between television ad exposure and public opinion, yet that TV advertising is neither targeted nor received randomly, so the apparent correlation between perceptions of a candidate and exposure to negative campaigns may be misleading. In the lab, although exposure to TV advertising can be randomly assigned, participants are exposed to ads in a contrived setting, and their candidate preferences are usually measured shortly after. Thus it's unclear whether the effects of ads persist after participants leave the lab. After all, in an actual campaign, people seldom vote immediately after viewing TV ads.

To overcome these limitations, it is important to study the effectiveness of TV ads on voter preferences during a campaign. In a study that my colleagues and I conducted in collaboration with the 2006 election campaign of Governor Rick Perry of Texas, 18 media markets in the state were randomly assigned to receive different levels of pro-Perry TV ads, and daily tracking polls gauged whether Perry's numbers improved as a function of increased advertising. The results suggested that advertising effects are short-lived. Perhaps the effects would have lasted longer than a week had the ads revealed memorable damning information about the opponent. But no field experiment has done a head-to-head comparison of TV advertising tone.

Although evidence on the effectiveness of negative political ads is inconclusive, campaign consultants clearly believe in their power, which explains why negative ads are so often used.