Tuesday, September 12
Measuring Policy Preferences
Chris Tausanovitch, assistant professor of Political Science at UCLA, will be discussing his research.
For more than a half-century, scholars have grappled with the question of whether Americans’ responses to policy questions reflect a shared belief system, or indeed anything more than a spur-of-the-moment inclination. However, surveys typically ask only a few questions about policy preferences. The low correlations between these responses has been taken as evidence of very little structure in public opinion. Using data from an Amazon Mechanical Turk Survey with a panel structure and a very long questionnaire, I estimate an item response model to measure preferences in three different policy domains. I show that this method can separate attitudes from non-attitudes and provide a clear picture of the degree of shared structure in policy preferences. Respondents are shown to have stable beliefs in distinct dimensions, and very few are genuinely cross-pressured.