Fig. 1.

Patriotism message effects relative to the control group, by partisanship. Marginal effects of the patriotism treatment relative to the control group by the respondent’s partisan identification; 90% confidence intervals (two-tailed) are illustrated. The marginal effects are from a set of analyses in which the treatment conditions were interacted with partisan identification. Ordinary least squares was used for Masks as Helpful, Masks as Important, and Trump Mask, while Logit was used for Mask Mandate.