J. Mark Ramseyer & Eric B. Rasmusen, Measuring Voter Ideology: Descriptive Regression Measurement of the Left-Right Spectrum (June 9, 2015).
Abstract: For scholars studying the political attitudes of the general public, someone’s position on the ideological spectrum is a good place to start. Typically, scholars identify that position through factor analysis on survey questions, making the assumption that the most important artificially constructed factor indicates the person’s position on the liberal-conservative spectrum. The leading attitudinal surveys, however— the GSS, the CCES, and the ANES— include a variable giving a respondent’s self-identified ideology, a variable given no special prominence by factor analysis. We suggest a new ideology measure: the individual’s fitted value from a regression of self-identified ideology on other variables. We describe various ways to choose those other variables. This approach gives proper priority to the usefulness of self-reported ideology. It lets us test whether voters identify their own ideology through identity-group variables; avoids the bias introduced in choosing which issue variables to include in factor analysis; and shows which positions the average American— as opposed to the analyst— thinks define “liberal” and “conservative”.