Causal Inference in Conjoint Analysis: Understanding Multidimensional Choices Via Stated Preference Experiments

Abstract

Edited by R. Michael AlvarezSurvey experiments are a core tool for causal inference. Yet, the design of classical survey experiments prevents them from identifying which components of a multidimensional treatment are influential. Here, we show how conjoint analysis, an experimental design yet to be widely applied in political science, enables researchers to estimate the causal effects of multiple treatment components and assess several causal hypotheses simultaneously. In conjoint analysis, respondents score a set of alternatives, where each has randomly varied attributes. Here, we undertake a formal identification analysis to integrate conjoint analysis with the potential outcomes framework for causal inference. We propose a new causal estimand and show that it can be nonparametrically identified and easily estimated from conjoint data using a fully randomized design. The analysis enables us to propose diagnostic checks for the identification assumptions. We then demonstrate the value of these techniques through empirical applications to voter decision making and attitudes toward immigrants.

Voting in a Multi-dimensional Space: A Conjoint Analysis Employing Valence and Ideology Attributes of Candidates

“…It has been applied only very recently to research questions in political science (Hainmueller and Hopkins 2012;Hainmueller, Hopkins, and Yamamoto 2014 Adams et al 2011;Enelow and Hinich 1982;Fiorina 1981;Groseclose 2001;Londregan and Romer 1993;Stone and Simas 2010). These factors are not particularly meaningful or useful in pairwise comparisons between generically labelled candidates.…”

“…They are either context-specific or instrumental -and the latter are not valued intrinsically by voters. In light 4 Like Hainmueller, Hopkins and Yamamoto (2014), we exclude party labels because the opinions participants have with regard to a given party may either be correlated with existing attributes or be proxies for omitted ones, therefore confounding our analysis of how respondents trade between policy and valence. With generic labels, the unobserved components of the choice function are less likely to be cross-correlated and more likely to have the same distribution (Hensher, Rose, and Greene 2005, 112-3).…”

How economic, humanitarian, and religious concerns shape European attitudes toward asylum seekers

“…Conjoint experiments ask subjects to evaluate hypothetical profiles with multiple, randomly varied attributes and are widely used in marketing and increasingly in other social science fields to measure preferences and the relative importance of structural determinants of multi-dimensional decision-making (26,27). Specifically, we used a conjoint experiment to ask 18,000 European eligible voters to evaluate 180,000 profiles of asylum-seekers that randomly varied on nine attributes that asylum experts and the previous literature have identified as potentially important (28).…”

Mass support for global climate agreements depends on institutional design

“…The core of our analysis draws on respondent choices between alternative global climate agreements presented within an experimental conjoint framework (15,16). Conjoint analysis involves having respondents rank or rate two or more hypothetical choices that have multiple attributes with the objective of estimating the influence of each attribute on respondents' choices or ratings.…”

“…Because we are interested in the features' marginal effects, we can use a linear probability model to estimate these elasticities. Specifically, we regress the variable Agreement Support on dummy variables for values of the agreement dimensions to nonparametrically estimate the effect of variation in any given attribute of an agreement on support for an agreement (16). We also reestimated the effects using a probit model, and the results remain unchanged (SI Appendix).…”