The "Right" Way to Use Polling Data: What Democrats Need to Do to Win Elections

By Amy R. Gershkoff

With less than a hundred days until the midterm congressional elections, political consultants on both sides of the aisle are hard at work putting together a strategy for the campaign season. With Democrats eager to take control of Congress, many of the party’s political strategists have suggested that the secret to winning in 2006 may be to be more like the Republicans—that is, to move towards the center in key policy areas.

But Democrats can win in 2006 without compromising their principles. The page they need to steal from the Republican playbook is the one that describes how to use polling data. As the Republicans have long recognized, not all voters care about the same issues to the same degree. They have applied this very simple observation in three ways, and it has led them to win countless local, state, and national elections.

Using data from the University of Michigan’s American National Election Studies (ANES), fielded during the 2004 election season, we can illuminate the ways Republicans use polling data to inform campaign strategy. Here is how the Republican playbook reads:

The most important piece of information to learn from a poll is what issues are most important to voters.

When Republicans segment voters into groups, they typically use the “most important issue” questions to do it, whereas Democrats typically use demographics, speculating, for instance, on how married women or African Americans will vote and then tailoring their campaign strategies to the various groups. While demographics offer insights into who is thinking what, it is much more useful to be able to identify the single issue that will determine someone’s vote.

The most important issue questions can be used to distinguish between issues that are important and those that are vote-deciding. With increasing numbers of voters being pulled in different directions by their views on various issues, this information is critical, and yet the questions needed to make these determinations are often not asked on Democrats’ surveys.

As Figure 1 shows, the issue the respondent considers the single most important one correlates highly with vote choice. Figure 1 breaks down the vote by the issue respondents considered most important, with the blue bars showing the percentage of such voters who chose Kerry and the purple bars showing the percentage who chose Bush.

Those who considered terrorism, taxes, or abortion to be the single most important issue voted for Bush by margins of three to one in the case of terrorism, and two to one in the cases of taxes and abortion. One hundred percent of those who considered immigration to be the single most important issue voted for Bush (although caution is needed in interpreting this finding, since only a small percentage of voters considered this to be the most important issue).

The findings are just as striking for Kerry: Those who considered jobs, health care, or education to be the single most important issue voted for Kerry by margins of nine to one, while those who chose Iraq, rights for homosexuals, and the environment voted for Kerry by margins of three to one.

By contrast (though there are, of course, notable exceptions), we rarely see nine to one or three to one margins when analyzing the impact of demographic characteristics on the vote. Thus, segmenting the voters based on their vote-deciding issue provides much more insight into their political behavior than sorting them according to demographics.