Georgios Melios
London School of Economics



(Thursday, 18th May 2023)

Title : From words to actions: Connecting beliefs, identities and behaviour in economics and political science

How Do We Choose Our Identity?

Are identities fungible?

Do people accurately report their identities and beliefs in survey data?

The answer to these questions is crucially important for research on polarisation, trust, and social cohesion, as well as the study of income inequality, attitudes towards immigration, and preferences for taxation. Most research on these topics relies on surveys to elicit people’s attitudes and beliefs. Prior findings suggest, however, that people often misreport their true beliefs because they lack incentives to deeply reflect on the problem, fall victim to social desirability bias, and engage in party cheer-leading by saying things that are congenial to their partisan identity without necessarily believing those answers. Hence, conclusions across a large body of research across the social sciences may be invalid.

Survey research is one of the most used research methods across the social sciences. However, critics raise several concerns about the ability of survey questionnaires to accurately measure people’s preference and beliefs. First, because survey questions are typically unincentivized, participants may be insufficiently motivated to deeply reflect on the problem at hand. Second, people routinely report socially desirable behaviour, even when their actual choices are less virtuous. Third, for politically contentious questions, partisans often provide answers that are congenial to their partisan identity. Insofar as answers to survey questions indeed do not correspond to actual preferences and beliefs, the validity of a large body of research across the social sciences may be called into question.

To examine the validity of these critiques, in this workshop we will review recent work on whether self-reported beliefs and identities correspond to important real-world decisions. Following the adage that actions speak louder than words, we aim to provide real world evidence on the congruence (or lack thereof) between what people Say and what people Do.