Using survey experiments for construct validation: “strong leader” questions and support for authoritarian leadership

Abstract

ABSTRACT Survey experiments are an increasingly popular tool for causal inference in political science. We argue that an under-utilized application for survey experiments is nomological/construct validation, where researchers evaluate whether indicators really measure systematized concepts. We demonstrate this approach by examining respondents’ preferences for autocratic leadership by asking whether those that say they want strong leaders who bend the rules or ignore parliament and elections really want undemocratic leadership in the context of an experimental task. While approaches that measure construct validity with observational data support the validity of these measures, our experimental data tell a different story. We find that respondents – even those who indicate a preference for “strong” leaders in survey questions – are less likely to choose hypothetical candidates who ignore democratic institutions and refuse to compromise with other parties. Our study contributes to the literatures on survey measurement and support for democracy and authoritarian values in established democracies.