Investigating What Demographics Predict Adamant Vaccine Holdouts
by Zachary L. Hertz
Introduction
As the Covid-19 vaccine rollout reaches its fifth month, the pace of vaccine administration slows, and experts now calculate that as much as 80 percent of the population may need to be vaccinated in order to reach the herd immunity threshold, vaccine hesitancy has received increased scrutiny.
An earlier Data for Progress blog post showed that even Americans who said they were somewhat unlikely to get a coronavirus vaccination appear to be getting the shot. But the Data for Progress COVID-19 tracking poll displays a troubling trend. Examining Americans' likelihood to get a coronavirus vaccination over the past four months shows that while all other groups have significantly decreased in size as vaccinations have increased, the share of Americans who say they are very unlikely to get vaccinated has decreased at a much lower rate and remains significant at 19 percent.
Given their persistence, it is imperative to investigate what the adamantly vaccine resistant look like in order to clarify what groups remain opposed to taking the coronavirus vaccine and understand what might convince holdouts to get the vaccine.
Some have pointed to race, religion, and partisanship as potentially salient factors. Particular focus has been given to vaccination rates among white evangelicals as a source of low vaccination rates. Additional evidence suggests that vaccine hesitancy is greater among racial and ethnic minorities, and some say hesitancy is particularly high among Black Americans. And despite former President Donald Trump encouraging his supporters to get the vaccine, counties with deep support for Trump have lower vaccination rates.
But these findings are hardly incontrovertible; an NPR/PBS NewsHour/Marist survey found very little difference in willingness to get the vaccine among Black and white Americans, and additional research suggested that white evangelicals were actually more likely to get the vaccine than the general public.
How do we make sense of these potentially irreconcilable findings, and in what demographics, if any, do anti-vaxxers differ from the rest of the general public? To begin to answer these questions, I examined the last five waves of the Data For Progress tracking poll that asked respondents how likely they are to get the coronavirus vaccine when they are eligible. Using post-stratification weights, the sample is nationally representative of American adults by gender, age, region, education, race, the interaction of education and race, and 2020 presidential vote. Pooling the sample provides 1359 respondents who have said they are “very unlikely” to get the vaccine.
Nonwhite Americans are relatively larger shares of the vaccine hesitant, though these differences are slight
First, I turned to the possible effects of race on vaccine hesitancy. Broken down by racial demographics, the "very unlikely" group seems generally similar to the larger sample. A majority of those "very unlikely" to get a coronavirus vaccine are white (60.6 percent), though this is a decrease of 2 percentage points from their share of the general population. In contrast, there was about a one percentage point increase in both Black and Hispanic Americans' share of the "very unlikely" group, compared to the American public.
Another way to visualize this is to break racial groups down by their vaccine likelihood. Compared to all Americans, we can see that white and Hispanic Americans had slightly lower shares of respondents “very unlikely” to take the Covid-19 vaccine, while Black Americans had slightly higher numbers. Neither of these differences, however, were greater than one percentage point. Overall, I find limited evidence to suggest a substantively large difference in vaccine hesitancy between white and nonwhite Americans.
“Very Unlikely” vaxxers are slightly more Evangelical than the public as a whole
With an increase of 4.4 percentage points, those who identify as “born-again” make up a slightly larger share of the very unlikely to get vaccinated than the public at large. This difference may be driven, in part, by a deeper split among white people who identify as Evangelical and white people who do not.
While a larger share of white evangelicals are “very unlikely” to get the vaccine, the “very unlikely” actually make up a slightly smaller share of white non-evangelicals than the general public, creating an almost 7 percentage point split between white evangelicals and non-evangelicals. Demographic splits between the "very unlikely" to get vaccinated and the general public run deeper on Evangelicalism than race, though both differences are relatively small, with no differences greater than 5 percentage points.
Just 10 percent of those “very unlikely” to get vaccinated are over 65 years of age
Slightly larger differences emerge when comparing those "very unlikely" to get the vaccine and the general public on age. While the "very unlikely" have a similar share of 50-64 year-olds as the public at large, the younger cohorts of 18-29 year-olds and 30-49 year-olds are a greater share of the "very unlikely" by about 5 percentage points each.
Meanwhile, the share of people "very unlikely" to get vaccinated who are over the age of 65, 10 percent, is half the share of the general population who are older than 65. In short, we see considerable and larger differences between the "very unlikely" to get vaccinated and the general public among age groups than on either race or religion.
An overwhelming majority of the vaccine hesitant are Trump voters or did not vote
Examining presidential vote reveals stark splits with the "very unlikely" to get vaccinated. 87 percent of those "very unlikely" to get vaccinated voted for Donald Trump or did not vote. At 46 percent, nonvoters were 13 percentage points more of the "very unlikely" to get vaccinated than the American public, and 41 percent of those unlikely to get the shot voted for Donald Trump, nearly 10 percentage points greater than their share of the general population. Meanwhile, Biden voters made up a significantly reduced but still noticeable share, nearly 12 percent, of the "very unlikely" to get vaccinated, though this was a nearly 23 percentage point decrease in their share of American adults.
This cut at the data provides additional evidence to suggest that even as Trump and other Republican leaders encourage vaccination, resistance to the vaccine remains among their voters.
Which demographic factors predict vaccine hesitancy?
These demographic splits, while informative, seem difficult to reconcile with previous research. Additionally, splitting the groups on single demographics at a time fails to consider how they might interact in shaping vaccine hesitancy. And as vaccination rates drop ahead of the estimated threshold for herd immunity, there is a growing need to understand those who are vaccine resistant as well as what demographic factors might best predict vaccine hesitancy, to better target and decrease it.
To parse through the potential predictors of vaccine hesitancy and their importance, I constructed a linear regression model controlling for race, presidential vote choice, age, education, geography, Evangelicalism, and gender. I find that the demographics with the largest and statistically significant effects on being very unlikely to get a coronavirus vaccine are Donald Trump voters, Black Americans, those without a college degree, and those living in rural areas.
Controlling for these variables reveals several interesting findings that are not immediately clear from examining the demographic splits on their own. Looking at presidential vote choice, we find that being a nonvoter does not have a statistically significant effect on the odds of being very unlikely to get a Covid-19 vaccination. Voting for one of the two major candidates, however, is a significant predictor: among Donald Trump voters, an individual's odds of being strongly vaccine hesitant increase by 73 percent (p=0.008) and voting for Joe Biden leads to a similarly strong 73 percent decrease in an individual’s odds of being strongly vaccine hesitant (p<0.001).
When considering race, Americans who identified as Hispanic or Other Race did not significantly differ from White Americans in their likelihood to be strongly hesitant to get a Covid-19 vaccine. In contrast, a stark difference exists between Black Americans and White Americans: in comparison to those who identify as White, there was an increase of 106 percent (p<0.001) in an individual's odds of being very vaccine hesitant among Black Americans.
Educational attainment also had a strong effect on vaccine hesitancy. In comparison to those without a college degree, individuals with a college degree or higher saw a 47 percent decrease (p<0.001) in their odds of being very vaccine hesitant.
While those who were "very unlikely" to get vaccinated appeared to have slightly higher shares of those who identified as "born-again", the effects of evangelicalism on strong vaccine hesitancy disappear entirely when controlling for the other demographic characteristics.
Geography, too, appears to be salient when considering vaccine holdouts. In comparison to suburban residents, an urban resident's odds of being strongly vaccine hesitant decrease 18.4 percent (p=0.015). Rural residents, meanwhile, see an increase of nearly 37 percent (p<0.001) in their odds of being "very unlikely" to get the coronavirus vaccination.
Additionally, gender seems to affect one’s odds of being reticent on vaccination. In comparison to women, when controlling for presidential vote, race, education, evangelicalism, geography, and age, the odds of being "very unlikely to get the coronavirus vaccine drop by one-third (p<0.001) among men.
Finally, in comparison to being between the age of 30 and 44, there does not appear to be a statistically significant relationship between being between 18 to 29 years old and vaccine hesitancy. Older groups did have a statistically significant decrease in likelihood to be vaccine hesitant; those age 50 to 64 years old had a 27 percent decrease (p<0.001) in their odds of being "very unlikely" to get a Covid-19 vaccination and those over 65 years old saw a considerable drop of 67 percent (p<0.001) in their odds of being "very unlikely" to get vaccinated.
Conclusion
While there are demographic splits between the "very unlikely" to get vaccinated and the American public as a whole, particularly on age and 2020 presidential vote, these differences do not tell the full picture; educational attainment, geography and gender also have significant effects on vaccine hesitancy.
We can see that the largest increases on one's odds of being "very unlikely" to get the coronavirus vaccine are among Black Americans, those who voted for Donald Trump for President, and those with less than a college degree as their highest educational attainment. Similarly large decreases exist among those who voted for Joe Biden for president, those with a college degree, and those over the age of 65.
These findings are not necessarily surprising. Black Americans have faced a long history of medical abuse which may erode trust in the medical system and, even in December, had high levels of vaccine hesitancy. And as age increases risk for severe cases of coronavirus it makes sense that adults over 65 are incredibly less likely to be vaccine hesitant.
Still, if the Biden administration truly hopes to have 70 percent of American adults vaccinated by July 4, it will be critical for the administration to recognize that vaccine hesitancy remains significant among these groups and to tailor its public health responses accordingly.
Zachary L. Hertz (@zacharylhertz) is a Research Associate at the Tufts Public Opinion Lab and rising MA student at the University of Chicago.