In October 2004, we conducted a telephone survey of 643 adults living in Tennessee. In addition to routine demographic measures, the survey included a question on respondents' voting preferences for the upcoming presidential election, as well as items in four categories:

•  Respondents' religious identity, belief, and practice. Evangelicalism was measured with a single question asking respondents whether they considered themselves evangelical or "born-again" Christians. To help describe the kinds of evangelicals singled out by the measure, additional questions were posed, asking respondents about the nature of the Bible and whether they had ever "witnessed," an evangelical term for encouraging someone to believe in Jesus Christ or accept Jesus Christ as his or her savior. Frequency of worship service attendance was also measured.

•  Respondents' opinions about abortion and gay marital rights. A split-sample item was used to measure opinions toward homosexual marital rights. One of two question asked whether respondents thought marriages between homosexuals should or should not be recognized as valid, with the same rights as traditional marriages, while the other asked whether homosexual couples should be allowed to form legally recognized civil unions, giving them the same legal rights as married couples. Also included was a question regarding respondents' positions on abortion rights.

•  Respondents' opinions about each candidate's morality and intelligence. Subjects also were asked which, if either, of the candidates--John Kerry or George W. Bush--they perceived as having better morals. A similar question asked which candidate, if either, was more intelligent.

•  Respondents' level of ideological constraint. A measure of respondents' ideological constraint was constructed from responses to questions about five key domestic issues. Each issue chosen was one on which Bush and Kerry had articulated differing positions: taxation, health care, school vouchers, Social Security, and pollution control measures for plants and factories. Additionally, each issue reflected the same underlying question of whether the federal government, in keeping with conservative political ideology, should take a laissez-faire approach to economic matters or, in keeping with liberal political ideology, impose government control. Responses were used to create a scale ranging from -5 for an individual whose answers all coincided with Kerry's stated positions to +5 for one whose answers all coincided with Bush's stated positions. A respondent with a score of zero, meanwhile, either repeatedly answered "don't know" or "refused" to the questions posed, or gave a pro-Kerry answer as many times as he or she gave a pro-Bush answer.

Respondents' level of issue knowledge. Finally, after obtaining respondents' opinions on the five issues described above, the survey questionnaire described one view on each and asked which candidate, Bush or Kerry, supported the view. An index of issue knowledge ranging from 0 to 5 was then computed by giving one point for each time the respondent correctly named the candidate who favored the view described.