Extended Data Table 5 Effect of the Consensus condition on respondents' vaccination uptake: additional results
From: Communicating doctors’ consensus persistently increases COVID-19 vaccinations
- OLS coefficients. Standard errors clustered at the respondent level in parentheses. Randomization inference p-values in square brackets (ritest command in Stata). The dependent variable in all columns is an indicator for vaccination uptake, equal to 1 if the respondent reported having obtained at least one dose of a vaccine against Covid-19. Wave 6–11 sample used. Column 1 uses the full sample: Consensus condition n = 5,145 (981 clusters=respondents); Control n = 5,137 (983 clusters=respondents. Column 2 uses a sample of respondents participating in all 11 waves: Consensus n = 3,684 (614 clusters=respondents); Control n = 3,588 (598 clusters=respondents). Column 3 imputes missing vaccination uptake data by using the latest vaccination status in an earlier wave for each missing wave. Column 4 imputes missing vaccination uptake data by using the first reported vaccination status in a non-missing subsequent wave. Column 5 restricts the full sample to respondents who passed all attention checks embedded in the survey. Columns 6 and 7 restrict the sample to respondents underestimating and overestimating trust in the Covid-19 vaccines, respectively. Columns 8 and 9 restrict the sample to respondents underestimating and overestimating doctors’ intentions to get vaccinated, respectively. Columns 10 and 11 restrict the sample to respondents without and with intentions to get vaccinated prior to wave 0, respectively. In all columns we use the pre-registered set of controls. All columns include wave fixed effects. Estimated coefficients from a double-selection LASSO linear regression (dsregress command in Stata 17) selecting from a set of covariates in Extended Data Table 1 are reported in the bottom part of the panel. Rows titled “Comparison” in each panel report a chi-square statistic and a p-value for a test of equivalence of coefficients across two respective models estimated using seemingly unrelated regressions (suest command in Stata 17). For LASSO selected controls, we use OLS models with controls selected by LASSO. T-test (two-sided) p-values reported as *p < 0.10; **p < 0.05; ***p < 0.01. No adjustments for multiple comparisons.
