Assessing the impact of novelty and conformity on hesitancy towards COVID-19 vaccines using mRNA technology.

Background: Public hesitancy towards Covid-19 vaccines remains a major hurdle for mass vaccination programs today. While mRNA vaccines are more efficacious than conventional vaccines, it is unknown how much the novelty of this technology increases hesitancy.
Methods: We quantify this "novelty penalty" in a large online experiment with 35,173 adults in nine countries. Subjects were randomly selected and assigned to one of two vaccine groups (conventional or mRNA), and one of five hypothetical inoculation rate groups (0%, 20%, 40%, 60%, or 80%). Subjects reported their willingness to accept the Covid-19 vaccine on a five-point Likert scale.
Results: The novelty of the mRNA vaccine technology reduces the odds of a higher level of vaccine acceptance by 14.2% (odds ratio 0.858; p < 0.001). On the other hand, we find that social conformity reduces vaccine hesitancy. At a 0% inoculation rate, 31.7% report that they are "very likely" to get a mRNA vaccine while at a 20% inoculation rate, willingness jumps to 49.6%. Conclusions: The novelty of the mRNA vaccine increases hesitancy, but social conformity reduces it. A small group of early adopters can provide momentum for vaccination.

Introduction

Covid-19 has infected more than 235 million people and caused 4.8 million deaths worldwide[1]. The dominant public policy response has been to encourage mass vaccinations to allow healthcare systems and economies to return to normalcy. At efficacy rates of between 60% and 95%[2-4], the vaccination rate required is estimated to be as high as 84% to 90% of the population[5].

New vaccines have always encountered some degree of hesitancy among members of the public – past research has demonstrated this effect for vaccines for HPV and the influenza A (H1N1) virus, for example[6-10]. In the current pandemic, hesitancy towards vaccines for Covid-19 has been mapped across the world and remains a seemingly intractable obstacle in the fight against the pandemic[11-13]. In a survey of 13,426 people in nineteen countries, only 47% reported that they “completely agree” with getting a Covid-19 vaccine[14]. Indeed, in most countries, these levels of support were insufficient to meet the requirements for herd immunity[15].

Recent studies have identified several determinants for Covid-19 vaccine acceptance, for example, acceptance was lower among females, people with lower levels of education, and in Black communities[16-18]. Vaccine acceptance is also influenced by the characteristics of the vaccine, such as efficacy, risk of serious side effects, the manufacturer, and the place of administration[16].

Several recent studies investigated if people were more hesitant to get vaccines, which use novel, mRNA-based technology[19-21]. Results have been mixed, with a few studies finding some level of hesitancy towards new vaccine technologies among health care workers[19,20], while one other study found no significant difference in preference between conventional (weakened viral) and mRNA-based vaccines for Covid-19[21].

However, these studies were conducted prior to the approval and roll-out of Covid-19 vaccines to the public (which began with the approval of Pfizer-BioNTech’s mRNA-based Covid-19 vaccine for emergency use in the United States of America on December 11, 2020). Public interest in mRNA vaccine technology jumped around the second week of December when the web search volume for “mRNA vaccine” increased more than 30-fold[22]. The rise in interest was accompanied by substantial misinformation regarding the new vaccine technology[23], which in turn significantly increased vaccine hesitancy[24].

It is both urgent and important to add empirical heft to our understanding of public hesitancy towards conventional and mRNA vaccines for Covid-19 at a crucial moment when information about vaccines is spreading rapidly and many people have experienced receiving the vaccines.

This study has two main goals. The first is to determine the increased hesitancy towards novel mRNA vaccines after the vaccines were approved for public use. We do this by conducting a large-scale, global survey to investigate if there are differences in willingness to accept conventional and mRNA-based vaccines. We anticipate an increase in vaccine hesitancy due to the novelty of the mRNA vaccine technology, which we call a “novelty penalty.” This term first coined by psychologists to refer to people’s preference for familiar experiences over novel ones[25].

Our second goal is to investigate forces that may reduce vaccine hesitancy. Previous research has shown that social conformity is useful in overcoming hesitancy towards novel science[26]. Hence, we examine how an individual’s decision to accept the vaccine may be influenced by the decisions of others in the community. Specifically, would a person’s willingness to get vaccinated increase dramatically if the person found out that many members of the community had already been vaccinated? What level of vaccination would be required to trigger this dramatic increase? Such bandwagon effects, if proven, will provide governments with a powerful policy instrument against the pandemic.

We find that the novelty of the mRNA vaccine technology reduces the odds of a higher level of vaccine acceptance by 14.2% (odds ratio 0.858; p < 0.001). The magnitude of this “novelty penalty” varies across countries. We also find that social conformity reduces vaccine hesitancy. When no one in the country is vaccinated, only 31.7% of people responded that they are very likely to get an mRNA vaccine for Covid-19 and 35.1% a conventional vaccine. Upon learning that 20% of their peers have been vaccinated, the proportion jumps to 49.6% for an mRNA vaccine and 52.4% for a conventional vaccine. Above 20%, the proportion of people responding that they are very likely to receive the vaccine continues to increase, although more slowly. Our findings highlight the importance of early adopters to create momentum for vaccinations.

Methods

Experimental Design

Between February 3 and March 5, 2021, we surveyed 35,180 adults in nine of the most populated countries in the Americas (Brazil, Mexico, and the United States of America), Asia (China, India, and Indonesia) and Europe (Germany, Russia, and the United Kingdom)[27]. Around 3900 subjects were recruited in each country. The survey implementation was administered by SurveyMonkey, a third-party survey company, and the recruitment of online subjects was conducted by one of SurveyMonkey’s global panel partners. SurveyMonkey implemented its proprietary method of quota sampling to meet the age and gender distributions of each country’s census. The only inclusion criterion was that the subjects had to be 18 years of age or older. The National University of Singapore’s Institutional Review Board (NUS-IRB) granted the study (NUS-IRB-2020-733) an exemption from IRB review and from the need for informed consent, as it was deemed to be of minimal risk and did not involve the collection or use of any potentially sensitive data. Following best practice in the behavioral sciences, the study was also pre-registered at ClinicalTrials.gov (NCT04693689).

The survey asked about people’s confidence in vaccines in general (based on four questions used in the Wellcome Global Monitor study[28]), and their willingness to receive a Covid-19 vaccine if it were provided for free. In addition, we examined the extent to which their willingness to receive a Covid-19 vaccine was influenced by two factors: the novelty of mRNA vaccine technology, and the hypothetical vaccine adoption rate in their country. All subjects were adults (aged 18 and above) and received a small amount of financial compensation for participating in the survey.

To study if Covid-19 vaccine acceptance is influenced first, by the type of technology, and second, by the vaccine adoption rate in the country, we implemented a 2 × 5, between-subjects randomized control trial (RCT) design. We provided a brief description of how conventional and mRNA-based vaccines work and then elicited the subjects’ willingness to get vaccinated using two technologies (conventional or mRNA) under five hypothetical vaccine adoption rates (0%, 20%, 40%, 60%, or 80%). The full survey instrument is available in Supplementary Methods.

The key outcome of interest was the response to the question, “Suppose the [conventional/RNA] COVID-19 vaccine is endorsed by your Government, free, [but no one/and 20%/and 40%/and 60%/and 80% of people] in your country [has/have] received the vaccine. How likely are you to accept the vaccination?” Five answer choices were given (“Very likely,” “Somewhat likely,” “Neither likely nor unlikely,” “Somewhat unlikely,” “Very unlikely”). Use of the 5-point Likert scale was consistent with previous large-scale research on vaccine acceptance[9,10,28]. Prior research suggests that people usually only carry out the specific behavior in question if their intentions are at least mildly positive[29]. We therefore focused on those “very likely” to accept the vaccine as a reliable predictor of actual vaccination rates. (In Supplementary Discussion, we report results using the proportion of people who reported that they were either “very likely” or “somewhat likely” to accept the vaccine. These results likely overstate the actual vaccination rate).

Statistics and Reproducibility

To test the “novelty penalty” hypothesis, we conducted a Kolmogorov-Smirnov test. This was done to detect whether there was a significant difference in the distribution of responses to the vaccine acceptance question regarding conventional and mRNA vaccines. In addition, we conducted a two-sample proportion test to check whether there is a difference in the proportion of respondents very likely to accept a conventional vaccine and those very likely to accept an mRNA vaccine.

To check whether there was an interaction between vaccine novelty and the adoption rate, we conducted a two-way analysis of variance (ANOVA). We also conducted a multivariate ordered logistic regression analysis to estimate the effects of vaccine novelty and the adoption rate on the odds of subjects having a higher level of vaccine hesitancy. All analyses were performed using Stata software version 13.0. The raw data is included as Supplementary Data 1[30]. The analytic code is included as Supplementary Data 2[30].

Table 1

Ordered Logistic Regression Results.

	(1) Dependent variable: Vaccine acceptance
mRNA	−0.153*** (0.021) [OR=0.858]
Adoption rate (Baseline: 0%)
20%	0.842*** (0.032) [OR=2.321]
40%	0.948*** (0.032) [OR=2.581]
60%	1.024*** (0.032) [OR=2.784]
80%	1.186*** (0.033) [OR=3.276]
Female	−0.278*** (0.021) [OR=0.758]
Age	0.015*** (0.001) [OR=1.016]
Years of schooling (Baseline: ≤6)
7–9	−0.056 (0.053) [OR=0.945]
10–12	−0.067 (0.044) [OR=0.935]
>12	0.177*** (0.041) [OR=1.193]
Country (Baseline: USA)
Brazil	1.024*** (0.047) [OR=2.784]
China	0.818*** (0.045) [OR=2.266]
Germany	0.038 (0.044) [OR=1.039]
India	0.810*** (0.044) [OR=2.247]
Indonesia	0.718*** (0.046) [OR=2.049]
Mexico	1.099*** (0.047) [OR=3.002]
Russia	−0.617*** (0.044) [OR=0.540]
UK	0.619*** (0.045) [OR=1.856]
Pseudo R2	0.051
Observations	35,173

Notes: Table reports estimates from multivariate ordered logistic regression model. The dependent variable is vaccine acceptance, which takes on five values (0 = “Very unlikely”, …, 4 = “Very likely”). Estimated coefficients are shown, with standard errors in parentheses and odds ratios in square brackets. * p < 0.005 ** p < 0.001 *** p < 0.0001.