Negative Ads Turnoff Voters, Enthrall News Media

Researcher says attack ads help candidates dominate press coverage

The 2006 election cycle will set a record for the amount of money spent on campaign advertising in an off-year election. By most accounts, the tone of the ad campaigns reached new lows, with candidates and the parties relying heavily on harsh and often personal attacks.

Just how did the 2006 ad campaign play out in the key Senate races? Were the candidates in fact better off relying on negative rather than positive messages? And what role does the media play in fueling the cycle of negative advertising?

To find out, Stanford's Political Communication Lab designed an online experiment in which voters from the seven battleground states of Ohio, Pennsylvania, Maryland, Tennessee, Missouri, Montana and Virginia watched a pair of either negative or positive ads (one from each candidate). While viewing each of the two ads, participants indicated how they felt about the content of the ad by continuously moving a slider from 0 (suggesting strong negative reactions) to 100 (indicating strong positive feelings).

The slider procedure allows us to compare how voters felt about the various positive and negative ads they encountered in these races. Because we also asked participants to indicate how they intended to vote in each senate race, and to indicate how warmly or coldly they felt about the various candidates, we can trace the relationship between voters' reactions to the ad campaign and their behavior at the polls.

Our results include a few surprises. Even though negative ads did reliably accomplish their immediate objective -- making the opposing candidate appear less attractive -- positive ads had the stronger impact on vote choice. More in keeping with the conventional wisdom, both negative and positive ads polarized the electorate along party lines, with Democrats appreciating the content of their candidates' ads while detesting just about everything in Republican ads, and vice-versa. While positive and negative ads both elicit cheers from strong partisans, negative campaigns tend to turn off independents and weak partisans from the political process.

Methodology

Approximately 1,900 registered voters from the national research panel maintained by the research firm of Polimetrix agreed to participate in the study. Participants were recruited from the seven battleground states of interest noted above. The size of the participant group from each state varied from a low of 80 in Montana to a high of 530 in Pennsylvania. (Of course, each participant was assigned to his or her home state condition, i.e. Montana residents only watched ads from the Montana Senate race.)

Democrats outnumbered Republicans in the sample by about five percentage points. The group was split evenly by gender, was 88 percent white, and the median age was 49. Twenty percent of the participants had a high school education, 35% had some college, and 45% were college graduates.

A novel feature of this experiment was the use of an online "dial" to monitor voters' reactions to advertising. Instead of asking for a summary evaluation of the ad, participants reacted continuously while the ad was playing. They were instructed (and given a practice task) on how to move a slider located immediately below the video in accordance with their feelings about the content of the ad. ("If what you see or hear makes you feel good, or you agree with the speaker, indicate this by moving the slider towards the green end. If, however, your reaction is negative, and you dislike what you see or hear, then move the slider to the red zone.") By recording position of the slider (0n a scale from 0 to 100) as the ad played, we can monitor voters' reactions from beginning to end.

After completing the practice trial with the slider, participants were asked to react to a pair of Senate ads. They were assigned at random to either the positive or negative tone condition. The former featured two positive ads, the latter two negative ads (in each case, they reacted to one ad from each candidate). The ads themselves were sponsored mainly by the individual candidates, but in some instances we used ads from the Democratic or Republican Senate campaign committees.

After watching and evaluating the two ads, participants then completed a brief survey including questions about their intended vote in the senate election. They also rated the two candidates on a "feeling thermometer" from 0 (cold) to 100 (warm) and responded to a series of seven questions measuring their sense of political cynicism.

Results

We began by examining voters' evaluations of the ads. We pooled across the seven states to compute an average evaluation of Democratic and Republican negative and positive advertising. (The full set of voter reactions to the twenty-eight ads used in this study can be viewed at the PCL Web site).

Overall, voters reacted only lukewarmly to negative ads and somewhat more warmly to positive ads. However, in both the positive and negative tone conditions, Republican ads elicited significantly lower ratings than Democratic ads (41 versus 46 in the case of negative ads, 51 versus 54 for the positive ads), suggesting that voters were generally more responsive to Democratic over Republican appeals.