Does Conjoint Analysis Mitigate Social Desirability Bias?

• In Event: The Methodology of Experiments

Abstract

How can we elicit truthful responses in survey research? Scholars are often concerned about systematic mismeasurement in survey questions asking about sensitive topics due to social desirability bias (SDB). Recently, conjoint analysis has become a popular tool for measuring preferences when SDB is a concern, despite the lack of systematic evidence. In this paper, we employ a novel experimental design to investigate whether a standard fully randomized conjoint design can reduce bias in self-reported preferences about a socially sensitive behavior. We hypothesize that conjoint analysis mitigates SDB through two mechanisms, which we call imperception and rationalization. Our experiment isolates the SDB reduction through these mechanisms by comparing a standard conjoint design against a partially randomized design where only the socially sensitive attribute is randomly varied between the two profiles in each paired evaluation task. Our experiment also includes control conditions that are designed to remove confounding due to the increased attention to the varying attribute under the partial design. We implement the proposed experiment in an online survey about eco-friendly materials used in shoes. We find suggestive evidence that conjoint analysis does, indeed, mitigate SDB.

Authors

• Yusaku Horiuchi, Dartmouth College

• Teppei Yamamoto, Massachusetts Institute of Technology

• Zachary Markovich, MIT