Modeling Individual Differences in Beliefs and Opinions Using Thurstonian Models
Michael Lee, Michelle Y. Ke · Cambridge University Press eBooks · 2022
One common and informative way that people express their beliefs, preferences, and opinions is by providing rankings. We use Thurstonian cognitive models to explore individual differences in naturally occurring ranking data for a variety of political, lifestyle, and sporting topics. After demonstrating that the standard Thurstonian model does not capture individual differences, we develop two extended models. The first allows for subgroups of people with different beliefs and opinions about all of the stimuli. The second allows for just a subset of polarized stimuli for which some people have different beliefs or opinions. We apply these two models, using Bayesian methods of inference, and demonstrate how they provide intuitive and useful accounts of the individual differences. We discuss the benefits of incorporating theory about individual differences into the processing assumptions of cognitive models, rather than through the statistical extensions that are currently often used in cognitive modeling.