Dependent Preferences and Over-Time Instability in Survey Responses

Ohio State University Research Foundation -Do Not Use, Columbus OH

Investigators

Abstract

People frequently give different answers at different times when asked their opinions on important political issues. This tendency of the public to offer unstable survey responses has led many public opinion researchers to conclude that "Large portions of an electorate simply do not have meaningful beliefs, even on issues that have formed the basis for intense political controversy among elites for substantial periods of time" The reach of this conclusion extends beyond public opinion polls, market research, and election surveys. If so many people know so little about issues, the very foundations of democratic government are called into question. Survey responses may be unstable over time due to an inability of surveys as currently constructed to capture the complexity of people's preferences. The investigator has found in a set of limited surveys that many people have dependent preferences on a host of policy issues. A person has dependent preferences when her preference on one issue depends on the outcome of other issues. Yet most of the research in public opinion assumes that people have independent preferences, or that their preference on one issue is not linked to the outcome of other issues. The researcher proposes a national, three-wave panel survey of 1200 adult US citizens. The survey includes new questions to detect dependent preferences, revealing the connections that people see across issues. The investigator hypothesizes that respondents who have dependent preferences in the first wave of the survey will be the most likely to change their survey responses over-time as they react to a changing world. Two survey waves will occur before the November 2004 national elections; one survey wave, after. The survey data will reveal which people have dependent preferences, and for which issues. The investigator also works on new statistical models to classify changes in survey responses as random or systematic. As the researcher has shown in previous research, unstable survey responses may be due to dependent preferences among respondents. Even if dependent preferences are reasonable, well-formed, and stable, they may compel people to exhibit survey responses that appear irrational, ill-formed, and unstable. Third, uncovering dependent preferences opens a new set of questions for survey researchers to explore. Who has dependent preferences, and why? If this proposal is funded, the data gathered will be available to the general public and scientific community, along with data on respondents' demographic characteristics, information levels, attention to politics and the media, political predispositions, and vote intention and vote choice in a presidential election. These data may provide answers to many puzzles in the literature on voting behavior, public opinion, and election campaigns.