Involvement as a Polarizing Factor? A Comprehensive Multi-Method Analysis across Representative Datasets

Abstract

Complex system models of attitudes, such as the Hierarchical Ising Opinion Model (HIOM), suggest that a person's involvement in an attitude object could be linked to attitude extremity and polarization. Despite its potential to integrate various theories of attitude change and despite the implications it could hold for attitude research, this assumption has not yet been studied systematically. We investigate the role of involvement in five large-scale, representative surveys on general political orientation and attitudes towards the EU and COVID-19 vaccines, conducted in 79 different countries over the last 8 years. We propose criteria to classify the degree of ideological divergence and introduce a modality detection measure suited for ordinal data and large sample sizes. We find that involvement is linked to attitude extremity and that predictions of HIOM are validated in a topic-specific dataset on COVID-19 vaccines. Results on political orientation and general attitudes towards the EU show either no effect of involvement or patterns that contradict HIOM's predictions. We discuss implications for the measurement of involvement, complex system models of attitudes and polarization research.