Table 2. Predicting trust in science and research before and during the Covid-19 pandemic.
| 09/2019 | 04/2020 | |||||||
| b | p | 95% CI | SE | b | p | 95% CI | SE | |
| Intercept | 2.86 | <0.001 | [2.50, 3.21] | 0.18 | 2.63 | <0.001 | [1.933, 3.27] | 0.34 |
| Gender (1 = female) | -0.16 | 0.062 | [-0.34, 0.00] | 0.09 | -0.06 | 0.556 | [-0.27, 0.15] | 0.11 |
| Age (1 = 60 years or older) | -0.11 | 0.258 | [-0.28, 0.11] | 0.10 | -0.15 | 0.228 | [-0.36, 0.11] | 0.12 |
| Education (1 = A-level) | 0.24 | 0.008 | [0.06, 0.42] | 0.09 | 0.29 | 0.009 | [0.08, 0.51] | 0.11 |
| Children aged < 14 years in household (1 = yes) | 0.00 | 0.968 | [-0.19, 0.23] | 0.11 | -0.17 | 0.219 | [-0.45, 0.08] | 0.14 |
| Populist party preference (1 = AfD) | 0.11 | 0.353 | [-0.13, 0.34] | 0.12 | -0.57 | 0.057 | [-1.17, 0.04] | 0.30 |
| Political decisions should be based on scientific evidence.a | 0.17 | <0.001 | [0.07, 0.25] | 0.05 | 0.34 | <0.001 | [0.21, 0.47] | 0.07 |
| Adj. R2 | .09 | .15 | ||||||
| F value | F(6, 894) = 3.85, p < .001 | F(6, 943) = 10.32, p < .001 | ||||||
| N | 901 | 928 | ||||||
| 05/2020 | 11/2020 | |||||||
| b | p | 95% CI | SE | b | p | 95% CI | SE | |
| Intercept | 2.65 | <0.001 | [2.24, 3.24] | 0.25 | 2.43 | <0.001 | [2.05, 2.93] | 0.22 |
| Gender (1 = female) | -0.20 | 0.069 | [-0.43, 0.00] | 0.11 | -0.17 | 0.060 | [-0.35, 0.00] | 0.09 |
| Age (1 = 60 years or older) | -0.13 | 0.218 | [-0.35, 0.07] | 0.11 | -0.02 | 0.839 | [-0.22, 0.16] | 0.09 |
| Education (1 = A-level) | 0.25 | 0.019 | [0.03, 0.46] | 0.11 | 0.55 | <0.001 | [0.39, 0.74] | 0.09 |
| Children aged < 14 years in household (1 = yes) | -0.02 | 0.928 | [-0.33, 0.47] | 0.21 | -0.22 | 0.133 | [-0.55, 0.02] | 0.14 |
| Populist party preference (1 = AfD) | -0.34 | 0.307 | [-1.23, 0.19] | 0.34 | -0.47 | 0.045 | [-0.93, 0.01] | 0.23 |
| Political decisions should be based on scientific evidence.a | 0.32 | <0.001 | [0.20, 0.42] | 0.06 | 0.29 | <0.001 | [0.18, 0.38] | 0.05 |
| Adj. R2 | .22 | .24 | ||||||
| F value | F(6, 943) = 12.57, p < .001 | F(6, 923) = 26.39, p < .001 | ||||||
| N | 950 | 930 | ||||||
Note. Analyses used survey weights and were computed using the R package survey v4.0 [34]. In all regression models, the assumption of normality of the residuals was violated (which can be retraced by running the R syntax we share, see Methods section); therefore, standard errors and confidence interval bounds (95%, two-sided) of b coefficients were bootstrapped. Bootstrapping was done with the R package boot v1.3–25 [35] using the bias-corrected and accelerated method (BCa; [37]), which accounts for the skewness and lack of symmetry in the observed data [36]. Boldface = p < .05.
a In the 09/19 wave, this item was introduced in a reference to climate change research and policy-making; in the 04/20, 05/20 and 11/20 waves it was introduced in a reference to the COVID-19 pandemic.