Factors underlying COVID-19 vaccine and booster hesitancy and refusal, and incentivizing vaccine adoption

Abstract

The paper investigates the factors underlying COVID-19 vaccine and booster hesitancy in the United States, and the efficacy of various incentives or disincentives to expand uptake. We use cross-sectional, national survey data on 3,497 U.S. adults collected online from September 10, 2021 to October 20, 2021 through the Qualtrics platform. Results from a multinomial logistic regression reveal that hesitancy and refusal were greatest among those who expressed a lack of trust either in government or in the vaccine's efficacy (hesitancy relative risk ratio, or RRR: 2.86, 95% CI: 2.13-3.83, p<0.001). Hesitancy and refusal were lowest among those who typically get a flu vaccine (hesitancy RRR: 0.28, 95% CI: 0.21-0.36, p<0.001; refusal RRR: 0.08, 95% CI: 0.05-0.13, p<0.001). Similar results hold for the intention to get a booster shot among the fully vaccinated. Monetary rewards (i.e., lottery ticket and gift cards) fared poorly in moving people toward vaccination. In contrast, the prospect of job loss or increased health insurance premiums was found to significantly increase vaccine uptake, by 8.7 percentage points (p<0.001) and 9.4 percentage points (p<0.001), respectively. We also show that the motivations underlying individuals' hesitancy or refusal to get vaccinated vary, which, in turn, suggests that messaging must be refined and directed accordingly. Also, moving forward, it may be fruitful to more deeply study the intriguing possibility that expanding flu vaccine uptake may also enhance willingness to vaccinate in times of pandemics. Last, disincentives such as work-based vaccination mandates that would result in job loss or higher health insurance premiums for those who refuse vaccination should be strongly considered to improve vaccine uptake in the effort to address the common good.

Fig 1. Factors associated with vaccine hesitancy and refusal among individuals aged 25 and older.

Notes: Number of partially or fully vaccinated individuals = 2,232; number of individuals unwilling to get vaccinated = 350; number of individuals undecided = 519. Relative risk ratios (RRR) from multinomial logistic regression model predicting vaccine hesitancy (“undecided”) and refusal (“unwilling”) relative to those partially or fully vaccinated. The x-axis adopts a logarithmic scale. Weighted model. Inclusion of a dummy variable for respondents with missing income information (4.48% of the 25+ sample has missing income information) and of regional, labor force participation, and income covariates. Table 5 in S1 File shows regression results. The lines identify the 95% confidence intervals.

Fig 2. Vaccine hesitancy/acceptance by type of reason and vaccination status (%).

Notes: For each vaccination category (partially or fully vaccinated, hesitant or undecided, and unwilling), the height of the bars represents the (weighted) percentage of respondents who select a particular reason. Table 6 in S1 File displays the results.

Fig 3. Factors associated with hesitancy/refusal to take the booster shot among fully vaccinated individuals aged 25 and older.

Notes: Number of fully vaccinated individuals who are willing to take the booster shot = 1,280; number of fully vaccinated individuals who are hesitant or refuse the booster = 752. Odds ratios from logistic model predicting booster shot hesitancy/refusal among fully vaccinated individuals age 25 and older. The x-axis adopts a logarithmic scale. Weighted model. Inclusion of a dummy variable for respondents with missing income information (4.38% of the 25+ fully vaccinated sample has missing income information) and of regional and income covariates. Table 7 in S1 File shows regression results. The lines identify the 95% confidence intervals. Results are robust to the inclusion of the younger age group (18–24) and to the use of weights (Table 7 in S1 File).

Fig 4. Change in willingness to be vaccinated compared with the baseline (%) among unvaccinated individuals.

Notes: Each bar denotes the percentage of unvaccinated individuals who are more or less likely to get vaccinated in the incentive scenario compared with the baseline. To avoid response randomness, we excluded unvaccinated individuals who were in the bottom 10% and in the top 10% of the distribution of time to take the survey. That is, we excluded those who may not have taken the survey seriously enough (short duration) or who may have suffered from survey fatigue (long duration). Number of observations = 773. The values of the lottery ticket and of the gift card were randomly assigned across the respondents (although respondents were exposed to the same amount in the lottery question as in the gift card one). Number of observations with $100 gift card/lottery ticket = 385; number of observations with $200 gift card/lottery ticket = 388. Weighted results. Table 8 in S1 File shows figures. Table 9 in S1 File shows the results of t-tests that assess whether net movement (i.e., the proportion toward vaccination minus the proportion away from vaccination) differs significantly from zero. **: The significance test is accepted at the 1% level. *: The significant test is accepted at the 5% level.