Table 1 shows the distribution of the dependent variable of our study and the distribution of the answers to its component questions. Based on our multi-item measurement, three quarters (76%) of the adult Hungarians were vaccine confident, while one quarter (26%) of Hungarians were vaccine hesitant in March 2021. If we were to calculate vaccine hesitancy based on only one of the component questions, the ratio of vaccine confidents would be 8–25% points (pp) lower. Two thirds (68%) answered that they would probably or definitely get vaccinated or they were already vaccinated with “one of the vaccines” and with “the European/American vaccines”. The acceptance ratios of Russian and Chinese vaccines were much lower (52%).
The distribution of COVID-19 and fear-related indices is presented in the box plots of Fig. 3. It is visible that COVID-19 fear index and Precautiousness index are significantly higher among the vaccine confident group than the vaccine hesitant group of our sample (two-tailed t-tests, p < 0.001). However, the personal fear index is not significantly different in the two groups.
The distribution of the categorical independent variables of our analysis are presented in Table 2. We separately presented the categories’ ratios within the total sample, vaccine confident and vaccine hesitant subgroups. We ran Pearson chi-squared tests for independence to see whether the two subgroups differ significantly regarding the given variables. There were significant difference in relation to the following variables: political preference (Chi2 = 36.9, p = 0.000), age groups (Chi2 = 39.6, p = 0.000), gender (Chi2 = 6.4, p = 0.011), family status (Chi2 = 17.1, p = 0.001), fear from partner (Chi2 = 5.6, p = 0.018), static financial status (Chi2 = 23.6, p = 0.000) and dynamic financial status (Chi2 = 17.1, p = 0.000), religiousness (Chi2 = 18.5, p = 0.000) and class identification (Chi2 = 13.3, p = 0.01).
Because our research applies two operationalizations of conspiratorial beliefs, we separately present these results. Table 3 shows the distribution of the level of beliefs in conspiracy theories, as a categorical variable. “China theory” is the most accepted: more than half of our respondents were uncertain or tend to believe this theory. The answers broke down along similar lines when it came to the “Pharma theory”. The other theories are rejected to a much larger extent, although the ratio of those who could not or did not want to answer the “Population control theory” was quite high compared to the other theories.
The distribution of conspiratorial beliefs, as a continuous variable is visualized in Fig. 4. The level of these beliefs is higher amongst vaccine hesitant people in relation to all examined theories (two-tailed t-tests, p < 0.001).
Table 4 explores the political, demographic, social, economic roots of vaccine hesitancy. Political preference, age and gender were significant in Model 1. The probability of being vaccine hesitant was 10% points higher amongst opposition voters (p < 0.001), and 19 pp higher amongst undecided voters (p < 0.001) compared to government supporters. People older than 60 were 22 pp less likely (p < 0.001), people aged 50–59 were 11 pp less likely (p < 0.05) to be hesitant compared to adults under 30. Women were 6 pp less likely to be hesitant than men (p < 0.05).
Involving COVID-19 experience in Model 2 does not lead to substantial changes, but once fear and precautious behavior indices are involved (Model 3-Model 6), the effects of being woman, opposition voter and 50–59 years old lose significance. The effect size of both being an undecided voter and being older than 60 halves in these models and their significance also reduces with involving more variables. COVID-19 survivors were 13 pp less likely to be hesitant (p < 0.05) in Model 2, but it loses significance in Model 3–Model 6.
Model 3 shows that 0.1 points higher COVID-19-fear index decreases the probability of being vaccine hesitant by 5.1 pp on average. 0.1 points higher precautious behavior index decreases this probability by 1.7 pp on average. These effects keep significance and effect sizes only slightly change in Model 4–Model 6. Personal fear index’s effect is insignificant in Model 3, but after involving personal relationship variables, it became significant at p < 0.05 (Model 4–Model 6). 0.1 point higher personal fear index results in 1.5–1.7 pp higher probability of vaccine hesitancy on average.
Model 4 reveals that fear from partner’s aggressive behavior during the lockdown is negatively associated with vaccine hesitancy, those who are concerned about it were 9 pp less likely to be hesitant about vaccination (p < 0.01). Effect size and significance of this variable slightly increased in Model 5 and Model 6. Relationship status only had significant effect after involving economic variables in the analysis (Model 5 and Model 6). Model 6 shows that childless people living in a relationship were 8 pp more likely to be hesitant than childless, single people (p < 0.05).
Model 5 reveals that economic stability increases vaccine acceptance. Regarding dynamic economic evaluation, those who perceived that their financial situation worsened were 6 pp more (p < 0.05), while those who experienced financial improvements were 15 pp less likely to be hesitant (p < 0.01), both groups compared to those experiencing no change. Amongst groups based on static economic evaluation, those groups who were better off were more likely to accept vaccination compared to those whose income barely covered living expenditures. Those who had no financial problems, but could not save up were 8 pp, those who could save up a little were 13 pp and those who could save much were 19 pp less likely to be vaccine hesitant. Model 6 showed that religiousness and class identity did not significantly influence willingness to vaccinate.
Table 5 shows that all kinds of pandemic-related conspiratorial beliefs significantly correlate with vaccine hesitancy. In the first set of models, belief in “No virus theory” has the strongest effect, 0.1 point increase results in 3.8 pp higher probability of vaccine hesitancy. The effect size of “Population control theory” and “Microchip theory” are somewhat smaller (3.2 pp and 2.7 pp). “China theory” and “Pharma theory” have the weakest effects (2 pp, 2.3 pp). All conspiracy measures’ effects were significant at p < 0.001 in these models.
In the second set of models, the groups of non-believers of given theories serve as baselines. People hesitant about “No virus theory” and “Population control theory” were 20 pp and 15 pp more likely to be vaccine hesitant (p < 0.001). Respondents hesitant about “Pharma theory” and “Microchip theory” were 7 pp and 8 pp more likely to be vaccine hesitant (p < 0.05), but people hesitant about “China theory” did not differ significantly. Believers of all theories had lower probability to accept vaccination (p < 0.001). The effects of being believer of “No virus theory” and “Microchip theory” were particularly strong (33 pp and 30 pp higher probability of vaccine hesitancy). Believers of Population control, China and Pharma theories are 26, 20 and 19 pp more likely to be hesitant than non-believers. Respondents not expressing their views were also significantly more likely to be vaccine hesitant regarding all theories (p < 0.001).