Table 1.
Predictor wave 1 | Direct predictors of vaccination intention time 1 | Direct predictors of vaccination intention change | Correlates of vaccination intention time 1 |
Trust in health authorities | 0.246 (0.183, 0.307) | 0.460 (0.426, 0.488) | |
COVID-specific misinformation | −0.461 (−0.493, −0.431) | ||
COVID-specific conspiracy beliefs | −0.177 (−0.221, −0.131) | −0.427 (−0.452, −0.398) | |
General conspiracy thinking/beliefs | −0.365 (−0.412, −0.320) | ||
Vaccination knowledge | 0.408 (0.363, 0.459) | 0.253 (0.205, 0.301) | 0.656 (0.636, 0.677) |
Worry about COVID | 0.042 (0.010, 0.074) | 0.286* | |
Taken flu vaccine once | 0.051 (0.027, 0.075) | 0.032 (0.009, 0.053) | −0.001 |
Taken flu vaccine more than once | 0.156 (0.127, 0.184) | 0.085 (0.059, 0.109) | 0.376* |
Reliance on conservative media | −0.136* | ||
Reliance on very conservative media | −0.045 (−0.071, −0.020) | −0.175* | |
Reliance on mainstream media | 0.036 (0.012, 0.060) | 0.256* | |
Reliance on social media | −0.165* | ||
Strong Republican | 0.038 (0.001, 0.074) | −0.108* | |
Not very strong Republican | −0.036 | ||
Closer to Republican | −0.051* | ||
Closer to Democrat | −0.033 (−0.064, −0.001) | 0.075* | |
Not very strong Democrat | 0.030 (0.001, 0.053) | 0.033 | |
Strong Democrat | −0.078 (−0.118, −0.044) | 0.123* | |
Female | −0.089 (−0.109, −0.067) | −0.157* | |
Age 25 y to 34 y | −0.063 (−0.087, −0.035) | −0.121* | |
Age 35 y to 44 | −0.053 (−0.076, −0.029) | −0.095* | |
Age 45 y to 54 y | −0.050 (−0.072, −0.029) | −0.059* | |
Age 55 y to 64 y | 0.027 | ||
Age 65+ y | 0.080 (0.063, 0.098) | 0.210* | |
Black identity | −0.062 (−0.091, −0.038) | −0.175* | |
Evangelical status | −0.021 (−0.040, −0.002) | −0.184* | |
Education level | 0.017 (0.002, 0.031) | 0.196* | |
Vaccination intention time 1 | 0.373 (0.342, 0.402) | ||
Variance explained | 0.475 | 0.610 |
Predictors that were fixed at zero in the model have no weight. Correlations with wave 1 intentions in column 3 were the observed values (with those at P < 0.01 noted with an asterisk), while correlations with latent variables were obtained from the measurement model for the SEM (SI Appendix, Table S1). Paths that only predicted within the 95% CI are italicized.