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

cult flavored by grievances, authoritarianism and white nationalism. In the future, Trump’s political style and the devotion of his base might spread to other leaders. His Republican rivals have been described as trying to “out-Trump Trump” [87,88] or “claim the Trumpist mantle” [89]; retrospective accounts refer to his “historical dominance” and a “Trump era” that did not necessarily end with his presidency. Such descriptions foreshadow Trumpism as a phenome- non persisting beyond Trump. Currently, however, his personality cult appears central, while his targets and positions may shift—as in past reversals regarding COVID scientists and vac- cines [18]. It seems best at this point to operationalize Trumpism simply in terms of expressed Trump support, as previous researchers have done, while recognizing that more elaborate con- tent-based measures could be developed and validated in future research. As an emergent and currently important sociopolitical identity, Trumpism or Trump sup- port can be depicted as an intervening variable—partly descending from older identity mark- ers (such as ideology and party) and conspiracism, while also contributing direct effects of its own. Intervening-variable hypotheses invoke a structural equation modeling approach [90]. Fig 1 schematically depicts the model applied in this study. Because our endogenous variables (Trumpism, NoACC and NoVax) are ordinal or dichotomous, we use a variation called gener- alized structural equation modeling (GSEM). By incorporating generalized linear modeling

into the structural equation framework, GSEM permits multi-equation models with nonlinear link functions and non-Gaussian distribution families, such as ordered logistic regression (logistic link function, ordinal distribution family)—equivalent to binomial logit, if the depen- dent variables are dichotomies. Data management and calculation of GSEM models, with diag- nostic tests and graphing of interaction effects, were carried out using Stata 16.1. 5. Results 5.1 GSEM modeling Results from GSEM analysis modeling climate-change and vaccination rejection as functions of demographic and identity/belief characteristics, with Trump support as an intervening vari- able, appear in Table 2. GSEM applies generalized linear modeling (GLM) within a structural equation framework. In this case, the GLM aspect involves ordered logistic regression. Table 2 shows logit coefficients with their robust standard errors, the latter calculated by a generalized Fig 1. Schematic representation of the generalized structural equation model (GSEM). https://doi.org/10.1371/journal.pone.0293059.g001 PLOS ONE Trumpism, climate and COVID PLOS ONE | https://doi.org/10.1371/journal.pone.0293059 January 10, 2024 7 / 20