Fig 3. Effects of suppressor variables on socially conservative attitudes–COVID-19 precautions relationship.

Studies 1 and 2 conditional effects of moderated linear regressions, in which the previously-identified suppressor variables were added to the models specified in Fig 1. These (centered) suppressors were economic conservatism, trust in scientists as information sources, trust in liberal and moderate figures as information sources, and liberal media consumption. Further, each of these suppressor variables interacted with political party affiliation in the model, because they had different effects on precautionary behavior as a function of party identification (see Fig 2 and S3 Appendix for details). Bands around regression lines are 95% confidence intervals.