Who Said What?

Findings

In general, people who approve of the president's job performance and who are upbeat over the state of the economy vote for candidates of the president's party. Midterm elections are also opportunities for voters to signal their displeasure over the course of public policy. We would naturally expect proponents of the war in Iraq to vote Republican. Figure 1 shows the degree to which these factors shaped vote choice in both samples.

Clearly, voting choices are inextricably linked with evaluations of President Bush and his policies. Among the staunchest supporters of the Iraq war and Bush's performance in office, virtually 100 percent intend to vote Republican. At the opposite end of the scale, 95 percent of those who feel the war has made us less safe and who strongly disapprove of Bush's performance intend to vote Democratic. Assessments of the state of the economy are almost as potent voting cues; the Republican share of the "pessimist" vote is 13 percent. Among readers of The Washington Post, the results were consistent, albeit at a reduced level of Republican voting.

Elections are not just referenda on the performance of the incumbent administration, but also reflections of the issues people consider important. Threats of an impending terrorist attack make people more concerned about terrorism and more inclined to vote for the candidate they see as better equipped to deal with terrorism. Since most voters are poorly informed about individual candidates, they typically extrapolate from their images of the political parties. Heightened concern over terrorism thus boosts Republican candidates because their party is viewed as better able to deal with the problem than Democrats. In this sense, campaigns are struggles to control the issue agenda; success depends on being able to get the public to pay attention to the issues on which the candidate's party enjoys the edge

As a rough barometer of the state of the campaign agenda, we asked participants to identify the single most important issue in the upcoming election. As shown in Figure 2, the same four issues emerged in both samples. washingtonpost.com readers ranked the war foremost, followed by the economy, with immigration and terrorism tied for a distant third. In the national sample, terrorism was cited most frequently, flowed closely by Iraq, the economy, and immigration.

The relevance of the issue agenda to voting is apparent in Figure 3. Terrorism and immigration both work to advantage Republicans. The efforts of the Bush administration to entice the news media into discussions of terrorism are well-founded: more than 90 percent of the national sample participants who identify terrorism as the major issue intend to vote Republican. Even among washingtonpost.com readers, Republicans win a clear majority of those who care about terrorism. The Republican edge among people preoccupied with immigration is also significant, but on a smaller scale.

If immigration and terrorism are "Republican" issues, then Iraq and the economy clearly belong to the Democrats: 85 and 77 percent of the national respondents who cited the economy and Iraq respectively intended to vote Democratic. The steady trickle of "bad news" out of Iraq thus works to advantage Democratic candidates.

Given the strength of the connections between voters' political predispositions and issue concerns on the one hand, and their congressional vote choice on the other, it comes as no surprise that a pair of issue ads contributed little of added value. In fact, as shown in Figure 4, observable fluctuations in voting choice across the 16 cells representing the different combinations of issues were few and far between. (The results from the washingtonpost.com sample are excluded because there were substantially fewer Republicans than Democrats in the various conditions and because there were no effects of advertising whatsoever.)

Contrary to the conventional wisdom, Democrats are credible on national security; the ad pairing characterized by the lowest level of Republican voting (26 percent) consisted of a Democratic attack on "staying the course" and a Republican ad arguing that Iraq is a key arena in the war on terror. The Republican share of the vote also fell significantly (to 32 percent) when both parties addressed national security. We cannot be confident that these observed differences are attributable solely to issue content because the ads in question differ in various subtle respects other than content. The sound track in the Republican ad on terrorism, for instance, features continuous gunfire in the background. Nonetheless, the data suggest that Democrats need not shy away from national security and that they can successfully "dialogue" with Republicans on this issue.

For Republicans, highlighting their tough stance on immigration and enticing pro-immigration groups to weigh in may both be effective strategies. The election becomes a close contest in two of the four cells featuring the Republican "we need a fence" ad. And Republicans secured their best outcome (winning 53 percent of the vote) when the "immigrants have rights" ad was coupled with an ad supporting the war in Iraq. None of these differences, however, were statistically reliable.

The negligible impact of issue ads on voting preference does not necessarily mean that issue ads are a waste of resources. Advertising is also designed to fan the flames of partisanship and push Republicans and Democrats further into their respective camps. We looked at the reinforcing effects of issue content by comparing both party-line voting and partisans' evaluations of their respective parties. In terms of voting, there was a significant decrease in party defection when partisans watched ads in which their party addressed an "owned" issue. Democrats were less apt to defect when their party advertised on the economy and Iraq while Republicans were more loyal after viewing Republican ads on immigration or terrorism. In terms of party images, we found -- as shown in Figure 5 -- that the selection of issues by the parties did in fact move partisans' enthusiasm for their party.

Among Democrats nationwide, exposure to a Democratic ad on the economy resulted in a significant improvement in ratings of the party. Conversely, advertising on terrorism was a downer for Democrats. Among Democrats in the washingtonpost.com sample, the subject of the Democratic ad made no difference. On the other side, Republican advertising on terrorism elicited the most favorable response from partisans, while the war in Iraq made Republicans noticeably less enthused about their party.

All told, the main implication of our results is that the great majority of Americans are already "locked in" to their congressional vote. Campaign advertising over the next few weeks may still make a difference, but primarily through differential mobilization of party supporters. Of course, this study ignores an important ingredient of congressional voting -- voters' impressions of individual candidates. Can an attractive and telegenic Republican candidate attract support from voters disillusioned with the state of the country? That is the subject of our next experiment which addresses the interplay between candidate ads and public opinion in five of the hottest Senate races.

Stay tuned.

Shanto Iyengar is Director of the Political Communication Lab at Stanford University and author of the forthcoming book/DVD "Media Politics: A Citizen's Guide."