Benjamin E Lauderdale, “Studying Individual-Level Attitudes Towards Important Problems Using Pairwise Comparisons”
How can we measure citizens’ attitudes towards political problems in a way that generates data which are comparable across individuals, across countries, and across time? There are many kinds of societal problems that may become the subject of politics, there are many potential categorizations of these, and their scale of assessment is difficult to specify. In this paper, I show how to measure these attitudes using pairwise comparisons, such that the latent assessments can be used as either individual-level dependent or independent variables for subsequent analyses. Even where it is infeasible for respondents to answer sufficient pairwise comparisons to recover a respondent-level ranked order, model-based multiple imputation enables the analyses that would be possible if all respondents responded to all possible pairs of categories. I validate and illustrate this approach using assessments of political problems in the UK according to the categories of the Comparative Agendas Project. By enabling individual-level analysis on researcher-defined categories with accessible questions, these methods make supplementing or replacing the typical most important problem/issue questions asked on national election studies with pairwise comparisons broadly advantageous for future research.