From Psychological States to Survey Responses: A Theory of Latent States and Response Style

Le résumé fourni par la source

Understanding human behaviour, thoughts, and experiences often relies on self-report measures, yet responses frequently conflate multiple sources of information. We propose a theory in which self-reported responses reflect both the constructs being assessed (latent states) and how respondents choose to communicate that information (response style). We apply this theory to Likert scale data through the Latent State and Response Style (LASARS) model, an ordinal-probit approach that separates these sources of variance, yielding cleaner latent state estimates and interpretable measures of response behaviour. LASARS represents response styles as stable, three-dimensional signatures describing how individuals map internal experiences onto response options. Our theory enables quantitative tests of previously difficult-to-evaluate predictions. Across a series of empirical studies, we show that response styles can be selectively adjusted to task demands without corresponding changes in latent state expression, and that they are relatively stable across time and survey instruments, often matching or exceeding the stability of latent psychological states. We also support a novel prediction: response style information from one survey can be used to deconfound response style from substantive latent state signal in a separate survey, even when that survey cannot identify the two. Together, these findings support a unified theory of self-report responding in which response styles are meaningful, stable, and informative individual differences rather than mere measurement error. LASARS, therefore, provides a principled framework for improving the precision, interpretability, and validity of survey-based measurement.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.