Appendix 1 ________________________________________________________

Question Wording and Coding

2004 Blair Center Survey

The 2004 Blair Center Survey and the 2000 Knowledge Networks Election Study were conducted by the survey firm Knowledge Networks. Knowledge Networks conducts their surveys online, but remains a probability-based sample because panel members are recruited using random digit dialing sampling methods and then provided with an Internet connection if they do not have one. All telephone numbers have an equal probability of selection, and the sampling is done without replacement. In exchange for the Internet connection, Knowledge Network panelists are asked to complete surveys three to four times per month. Participants are sent an email informing them that their next survey is ready to be taken and individuals then complete the survey at their own convenience.

The 2004 Blair Center Survey was a national postelection survey. The sample plan consisted of 2,800 interviews from three strata: 1,150 from southern states, 1,150 from other states, and 500 from the general adult population. The overall survey completion rate was 68.1 percent. Poststratification weights that adjust for nonresponse and the survey sampling design were provided by Knowledge Networks.

When classifying individuals as incongruent or congruent, we created a conservative measure. If the question-response format included a middle, neutral, or don’t-know category, individuals selecting these responses were coded as being congruent with their political party, even if they might consider their moderate position incongruent with their party’s more extreme position. Thus, to be classified as a cross-pressured partisan, an individual must not only disagree with the position taken by her own party but also agree with the position of the opposition party. For the 2004 measure from the Blair Center Survey, we also included only those issues that respondents indicated were “extremely” or “very” important to them personally. The 2004 cross-pressures measure was constructed from the following ten policy questions.

Social Security: “Some people have proposed allowing individuals to invest portions of their social security taxes in the stock market, which might allow them to make more money for their retirement, but would involve greater risk than the current government-run system. Do you favor or oppose allowing individuals to invest a portion of their social security taxes in the stock market? [completely favor, somewhat favor, neither favor nor oppose, somewhat oppose, completely oppose]”

Income Equality: “Do you favor or oppose the federal government in Washington trying to reduce the income differences between the richest and poorest Americans? [completely favor, somewhat favor, neither favor nor oppose, somewhat oppose, completely oppose]”

Taxes: “Which of the following do you think should be emphasized more in an economic recovery plan: an increase in minimum wage or tax cuts for businesses? Where would you place yourself on the scale below? [7-point scale with minimum wage increase more important on one end, tax cuts more important on the other]”

Health Care: “In thinking about health care reform, which of the following do you think should be emphasized more in a reform plan: expanding coverage for low-income adults through federal and state health care programs or aiding small businesses in offering health care to their employees? Where would you place yourself on the scale below? [7-point scale with more important to expand coverage through government programs on one end, more important to aid small businesses in offering coverage on the other end]”

Gay Marriage: “Would you favor or oppose an amendment to the U.S. Constitution that would define marriage as being between a man and a woman, thus barring marriages between gay or lesbian couples? [completely favor, somewhat favor, neither favor nor oppose, somewhat oppose, completely oppose]”

Stem Cell Research: “On the whole, do you favor or oppose the use of stem cells taken from human embryos in medical research? [completely favor, somewhat favor, neither favor nor oppose, somewhat oppose, completely oppose]”

Gun Control: “Do you favor or oppose the registration and licensing of all new handguns sold in America? [completely favor, somewhat favor, neither favor nor oppose, somewhat oppose, completely oppose]”

Affirmative Action: “Some people say that affirmative action programs are still needed to counteract the effects of discrimination against minorities, and are a good idea as long as there are no rigid quotas. Other people say that affirmative action programs have gone too far in favoring minorities, and should be ended because they unfairly discriminate against whites. Still others have opinions somewhere in between these two. Where would you place yourself on the scale below? [7-point scale with affirmative action is necessary on one end, affirmative action should be ended on other]”

Environment: “Do you favor or oppose relaxing some environmental standards to increase oil and gas production in the United States? [completely favor, somewhat favor, neither favor nor oppose, somewhat oppose, completely oppose]”

Abortion: “Which of the following opinions comes closer to your view about abortion? [It should never be permitted; it should only be permitted when the woman’s life is in danger; it should only be permitted if the woman’s health or life is in danger; by law, a woman should always be able to obtain an abortion as a matter of personal choice]”

Policy Importance Measure: “Regardless of where you stand on the following political issues, please indicate how important each issue or policy is to you personally (abortion policy; gay marriage policy; gun control policy; health care policy; stem cell research policy; education policy; social security policy; environmental policy; welfare policy; military issues like global terrorism; employment issues like job security; business issues like regulation; legal issues like tort reform; racial issues like affirmative action; policies about prescription drugs; policies about school prayer) [extremely important, very important, moderately important, slightly important, not at all important]”

Open-ended Question: “We find that many people disagree with their political party on a variety of different political issues. You said that you consider yourself to be a (Democrat or Republican). Can you think of any political issues about which you disagree with the (Democratic or Republican) Party?” The coding of open-ended questions started with the fifty issue categories used by the American National Election Studies in classifying the open-ended likes/dislikes responses. Additional categories were added as needed.

As measures of campaign exposure in the 2004 presidential election, we used the number of presidential visits to the state in which a respondent lived, whether or not the state was defined as a battleground state by the presidential campaigns (using data provided by Daron Shaw) and measures of political attention and political awareness from 2004 Blair Center Survey. The measure of political attention was based on the question, “Some people seem to follow what’s going on in government and public affairs most of the time, whether there’s an election going on or not. Others aren’t that interested. Would you say you follow what’s going on in government and public affairs … [most of the time, some of the time, only now and then, never]?” To measure political awareness we created an additive scale of the number of eight political issues—abortion, gay marriage, gun control, stem cell research, environment, welfare, business regulation, and affirmative action—for which the respondent was able to correctly identify the Republican Party as more conservative than the Democratic Party. Respondents correct on seven or more issues were coded as politically aware, while those who were correct on less than seven issues were coded as unaware.

As a control variable, we coded a respondent’s general level of political knowledge as the number of correct answers from the following questions. “Who is the current attorney general of the United States? [Donald Rumsfeld, John Ashcroft, Condoleezza Rice, or Colin Powell]”; “What is the term of office of a U.S. senator? [two years, four years, six years, eight years]”; “Who is responsible for nominating judges to federal courts? [the president, Congress, the Supreme Court]”

2000 Knowledge Networks Election Study

The 2000 Knowledge Networks Election Study was created from repeated interviews of 28,000 panelists. Over the course of the campaign, seventy-five randomly assigned surveys (with widely varying sample sizes) were sampled from the Knowledge Networks panel. The political questionnaires varied from a single vote choice question added to the end of a market survey to a 10–15 minute survey about political attitudes, behaviors, and experiences. Following the election, a postelection survey of 12,000 respondents was conducted.

The resulting data set is a two-way unbalanced panel in that the number of observations are not the same for every respondent and the intervals between observations are not equal. The modal number of interviews per respondent was three and the average was five. Given that the intermittent missing values are random (i.e., individuals have missing observations because they were not part of the random sample selected for a given survey), it is reasonable to assume that the analysis should give the relevant inferences. Data were weighted to independent population estimates based on the 2000 Current Population Survey. These weights take into account age, gender, race, region of residence, and MSA.

Because of the structure of the data, the number of issue questions asked of each respondent varied. Among partisans in the postelection sample, more than one third were previously asked no issue questions and 50 percent were asked three or more issue questions. Analysis is restricted to the subset of respondents who answered at least three policy issues. In the event that respondents were asked more than one question in the same policy domain, a combined measure of the policy attitude was created. A respondent was not coded as cross-pressured on the issue if they gave conflicting responses to multiple questions in the same policy domain. The following policy questions were used to create the 2000 cross-pressures measure.

Gun Control: “Do you support or oppose making it illegal to buy or sell handguns anywhere in the United States? [support strongly, support somewhat, oppose somewhat, oppose strongly]”; “How well does the phrase ‘an NRA supporter’ describe you? [very well, somewhat well, not very well, not well at all]”

Gay Rights: “Do you support or oppose making gay and lesbian marriages legal? [support strongly, support somewhat, oppose somewhat, oppose strongly]”; “Do you think marriages between homosexuals should or should not be legal, with the same rights as traditional marriages? [not legal, legal, don’t know]”

School Prayer: “Do you support or oppose allowing public schools to start each day with a prayer in the classrooms? [support strongly, support somewhat, oppose somewhat, oppose strongly]”; “Would you favor or oppose an amendment to the U.S. Constitution that would permit prayers to be said in public schools? [favor, oppose, don’t know]”

Assistance to Minorities: “How much do you agree that the government in Washington should make every effort to improve the social and economic position of minority groups? [7-point scale with agree strongly on one end, disagree strongly on the other]”

Abortion: “Do you agree or disagree that it should be legal for a woman to have an abortion? [agree strongly, agree somewhat, disagree somewhat, oppose strongly]”; “How well does the phrase” pro-life“ describe you? [very well, somewhat well, not very well, not well at all]”; “Under what circumstances should abortion be legal? [always; only in cases of rape, incest, or to save mother’s life; never]”

Environmental Protection: “How urgent do you think the problem of pollution and environmental damage is? [5-point scale with not very urgent on one end, extremely urgent one the other]”; “How well does the phrase ‘an environmentalist’ describe you?” [very well, somewhat well, not very well, not well at all]“;” How about government spending on the environment? [spending too much, about right, spending too little]“

Tax Cut Priority: “As you may know, the federal government is currently running a budget surplus, meaning it is taking in more money than it spends. Which is the most important thing to do with the surplus? [tax cuts, reduce debt, increase spending on important programs and services]”

Privatization of Social Security: “Some people have suggested allowing individuals to invest portions of their social security taxes on their own, which might result in more money for their retirement but would also involve greater risk. What do you think of this idea? [good idea, bad idea]”; “Of the following two statements by candidates about social security, which do you agree with more? Candidate A: People should be allowed to invest some of their social security money in the stock market. It’s their money, so they should be allowed to invest it themselves; Candidate B: Investing social security money in the stock market is too risky. Some people would end up getting much less than what they get now with Social Security. [candidate A, candidate B, don’t know]”

Defense Spending: “How about government spending on defense? [spending too much, about right, spending too little]”

Political awareness is an indicator variable constructed from two different measures. For roughly 2,700 respondents, an individual’s awareness was estimated based on whether she could correctly identified the party of the incumbent candidate in the Senate race in her state (or whether the race was an open seat). For those respondents not asked this question, an individual was coded as aware if she had a college degree.

Design and Questions for Survey Experiment

The survey experiment was fielded in November 2006 as part of the 2006 Cooperative Congressional Election Study (CCES). The CCES was a collaborative survey project conducted by Polimetrix, an online survey company. The CCES was coordinated through the MIT Political Science Department and Polimetrix by Stephen Ansolabehere, Doug Rivers, and Lynn Vavreck. Research teams at more than thirty universities participated, each designing a 1,000-person survey for a total of 30,000 cases in the common content survey. The common content always preceded each university’s module. The common content questionnaire was developed by Stephen Ansolabehere, Robert Erikson, Elisabeth Gerber, Donald Kinder, Wendy Rahn, Jeremy Pope, and John Sides. Our sample of 1,000 respondents was randomly assigned to three different treatments. All respondents received the information presented to the control group, but those assigned to an experimental vignette received an additional screen of information. All respondents were then asked a series of survey questions about their attitudes and behaviors.

Control Group: “We are interested in people’s opinions about potential candidates for the 2008 presidential election. Currently, neither party has a clear front-runner for the nomination. Some party leaders have suggested it would be best to choose a new face—someone with less political baggage than the high-profile politicians. Among the names being discussed are Senators Tim Johnson and Larry Craig. Senator Johnson is a mainstream Democrat and a popular two-term senator from the Midwest. Senator Craig is a mainstream Republican recently elected to his third term as senator of a large midwestern state. Both candidates are thought to have strong records on national security and economic policies.”

Stem Cell Vignette: “Political analysts predict that one of the key issues of the campaign will be stem cell research. According to a leading analyst, ‘By 2008, many Americans will agree on the direction the country should take on foreign policy, so attention will likely shift to focus on the growing debate over stem cell research.’ Senator Johnson supports federal funding for stem cell research while Senator Craig opposes it. Given the distance between these candidates on this issue, the future of stem cell research policy could be one of the biggest issues at stake in the 2008 presidential election.”

Social Security Vignette: “Political analysts predict that one of the key issues of the campaign will be social security reform. According to a leading analyst, ‘By 2008, many Americans will agree on the direction the country should take on foreign policy, so attention will likely shift to focus on the growing debate over social security privatization.’ Senator Craig supports a proposal that has been made that would allow people to put a portion of their social security payroll taxes into personal retirement accounts that would be invested in stocks and bonds. Senator Johnson opposes this proposal. Given the distance between these candidates on this issue, the future of social security policy could be one of the biggest issues at stake in the 2008 presidential election.”

Stem Cell Research: “Do you favor or oppose the use of stem cells taken from human embryos in medical research? [completely favor, somewhat favor, neither favor nor oppose, somewhat oppose, completely oppose]”

Social Security: “A proposal had been made that would allow individuals to invest portions of their social security taxes in the stock market. Some say that this proposal would allow individuals to make more money for their retirement, while others say the proposal involves too much risk. Do you favor or oppose the proposal? [completely favor, somewhat favor, neither favor nor oppose, somewhat oppose, completely oppose]”

Questions Included in Cross-pressure Measures from the American National Election Studies Cumulative File

Detailed information about question wording and methodology of the American National Election Study can be found on their website at www.electionstudies.org.

The following issues were used in creation of the general cross pressures measures for the temporal analysis in chapter 3.

1972: civil rights; segregation; busing; aid to blacks; health insurance; guaranteed jobs

1976: civil rights; segregation; busing; aid to blacks; health insurance; guaranteed jobs

1980: civil rights; busing; aid to blacks; guaranteed jobs; abortion; defense spending

1984: civil rights; busing; aid to blacks; guaranteed jobs; abortion; defense spending; assistance for blacks; food stamps

1988: civil rights; aid to blacks; guaranteed jobs; abortion; defense spending; assistance for blacks; food stamps; homeless; discrimination against homosexuals

1992: civil rights; aid to blacks; guaranteed jobs; abortion; defense spending; assistance for blacks; food stamps; homeless; discrimination against homosexuals; affirmative action; homosexuals in the military

1996: aid to blacks; guaranteed jobs; abortion; defense spending; food stamps; homeless; discrimination against homosexuals; affirmative action; homosexuals in the military

2000: aid to blacks; guaranteed jobs; abortion; defense spending; assistance for blacks; food stamps; discrimination against homosexuals; affirmative action; homosexuals in the military

2004: aid to blacks; guaranteed jobs; abortion; defense spending; assistance for blacks; food stamps; discrimination against homosexuals; affirmative action; homosexuals in the military

The following issues were used in creating the racial cross-pressures measure for the temporal analysis in chapter 5.

1964–1968: civil rights; segregation

1972–1980: civil rights; segregation; busing; aid to blacks

1984: civil rights; busing; aid to blacks; government assistance for blacks

1988: civil rights; aid to blacks; government assistance for blacks; discrimination makes it difficult for blacks; blacks should have special favors; blacks should try harder; blacks have gotten less than they deserve

1992: civil rights; aid to blacks; government assistance for blacks; affirmative action; discrimination makes it difficult for blacks; blacks should have special favors; blacks should try harder; blacks have gotten less than they deserve

1996: aid to blacks; affirmative action

2000: civil rights; integration; aid to blacks; government assistance for blacks; affirmative action; discrimination makes it difficult for blacks; blacks should have special favors; blacks should try harder; blacks have gotten less than they deserve

2004: aid to blacks; affirmative action; discrimination makes it difficult for blacks; blacks should have special favors; blacks should try harder; blacks have gotten less than they deserve

The following issues were used in creation of the moral cross-pressures measure for the temporal analysis in chapter 5.

1980–1984: abortion

1988: abortion; moral views should change with society; traditional family ties should be emphasized more; new lifestyles contribute to the breakdown of society; tolerate new moral standards; protect homosexuals from discrimination; spending for AIDS

1992: abortion; homosexuals in the military; moral views should change with society; traditional family ties should be emphasized more; new lifestyles contribute to the breakdown of society; tolerate new moral standards; protect homosexuals from discrimination; spending for AIDS; homosexual adoptions

1996: abortion; homosexuals in the military; moral views should change with society; traditional family ties should be emphasized more; new lifestyles contribute to the breakdown of society; tolerate new moral standards; protect homosexuals from discrimination; spending for AIDS

2000: abortion; homosexuals in the military; moral views should change with society; traditional family ties should be emphasized more; new lifestyles contribute to the breakdown of society; tolerate new moral standards; protect homosexuals from discrimination; spending for AIDS; homosexual adoptions

2004: abortion; homosexuals in the military; moral views should change with society; traditional family ties should be emphasized more; new lifestyles contribute to the breakdown of society; tolerate new moral standards; protect homosexuals from discrimination; homosexual adoptions