Retrieving True Preference under Authoritarianism

• In Event: Advances in Survey Design

Abstract

In this paper, we address preference falsification, where survey respondents under authoritarianism render untruthful answers to sensitive questions due to political pressure. We first provide a theoretical framework where we classify survey respondents under authoritarianism into true regime supporters, candid non-supporters, and preference-falsifying sub-populations. Then, using the quasi-experimental setting from the Chinese General Social Survey data, we explain how latent profile analysis (LPA) – a tool to analyze survey respondents based on their answer patterns – can be used to infer the preference-falsifying subpopulation under authoritarianism and the survey questions that are likely to induce them to exaggerate support for the regime. We further test the application of LPA to preference falsification studies through data simulations and the World Value Survey data.

While the recent development of survey methodology has been focusing on designing survey experiments, this research puts equal emphasis on post-survey adjustments. This allows scholars to make better use of the existing survey data, rather than to implement new ones. It is particularly timely, with the increasing political tensions and hostile survey environment in countries like Russia and China.

Authors

• Jongyoon Baik, University of Pennsylvania

• Xiaoxiao Shen, Princeton University