Trumpism, climate and COVID: Social bases of the new science rejection

and one negative coefficient, the indirect effect is negative. Where the sequence includes either two positive or two negative coefficients, the indirect effect is positive. For ACC rejection, indi- rect effects through Trumpism significantly modify (strengthen or weaken) conclusions that might be drawn from the direct effects of most background characteristics (age, white, evangel- ical, ideology, party, same-party friends and conspiracist beliefs). For example, the total effect of evangelical identity on ACC rejection is significantly stronger if we consider not only its positive direct effect, but also its positive indirect effect through Trumpism: evangelicals are more likely to support Trump, and (even controlling for party, ideology, race etc.) Trump sup- porters are more likely to reject ACC. Both party and conspiricism show significant indirect effects through Trumpism on vaccine rejection as well. Indirect effects on vaccine rejection from age, identification as white or evangelical, ideology and same-party friends fall just short of significance (all 0.10 > p > 0.05). Because the predictors education, ideology, friends and party appear also in interaction terms, their main effects in Table 2 can be interpreted as the effect of that variable when its interacting counterpart equals zero. For example, the main effects shown for education on Trumpism, NoACC and NoVax represent the effects of education when ideology = 0; that is, the effects of education among ideological moderates. Among moderates, education has little

effect on Trump support. Even among moderates, however, education decreases the probabil- ity of rejecting ACC or vaccination—as visualized by the down-sloping “moderate” curves in Fig 2A and 2B. Similarly, the main effects of having same-party friends correspond to the friends effect when party = 0; that is, the effect of same-party friends among independents. Having mostly same-party friends increases the probability that self-identified independents will approve of Trump (Fig 3A) but has little effect on their ACC or vaccination rejection (Fig 3B and 3C). Fig 2. Education×ideology interaction effects on (a) probability respondent rejects reality of anthropogenic climate change, or (b) does not intend to get vaccinated for COVID-19. Marginal probabilities adjusted for other covariates, calculated from the GSEM model in Table 2. https://doi.org/10.1371/journal.pone.0293059.g002 PLOS ONE Trumpism, climate and COVID PLOS ONE | https://doi.org/10.1371/journal.pone.0293059 January 10, 2024 9 / 20