Fig 1.
Belief networks in each group detected by RCA.
In left-side network plots, each node corresponds to one variable, and each line connecting two nodes to the correlation between them, only if significant at a p ≤ 0.05. Line shades and widths are proportional to the strength of the correlation. Given the complexity of the graph, only positive correlation lines are drawn. Networks are drawn using Fruchterman-Reingold algorithm, so that distances between the nodes inversely correspond to the edge weights connecting them. On the right side, the respective correlation matrices are reported. Light blue squares represent positive correlations and red-shaded squares negative correlations. The correlation coefficient is proportional to the intensity of the respective color.
Fig 2.
Average indexes’ levels by RCA groups.
Average levels of confidence and risk perception with and without vaccination in each group detected by RCA. All standardized coefficients, 95% CIs.
Fig 3.
Multinomial logistic regression predicting the probability to belong to each RCA group as a function of individuals’ sociodemographic characteristics.
Predictor variables include: educational level, age, gender, whether respondents have children, religious affiliation, geographic area of residence and rural or urban area of residence. Relative Risk Rations, 95% CIs, weighted coefficients.
Fig 4.
Multivariate binomial logistic regression models predicting the probability to be vaccine hesitant (0 = no hesitancy; 1 = hesitancy).
Models are controlled for individuals’ sociodemographic characteristics. Average Marginal Effects. All standardized coefficients. 95% CIs, weighted coefficients.