Data

This paper employs the Household Pulse Survey (HPS) Data published by the U.S. Census Bureau. The HPS is a biweekly cross-sectional survey of US households designed to deploy quickly and efficiently collecting data to measure household experiences during the coronavirus pandemic and recovery. The HPS is a 20-minute online survey studying how the coronavirus pandemic and other emergent issues are impacting American households across the country from a socioeconomic and health perspective. Sampling was drawn from the Census Bureau Master Address File and the Census Bureau Contract Frame, containing approximately 140 million housing units with matched phone or email contacts. Households were contacted by email and text message, and data were collected via the online survey. The HPS is a continuing project, and additional phases have been scheduled. For this analysis, we use the Phase 3.5: June 1, 2022 –August 8, 2022 dataset [20].

Statistical method

Probit regression is used to model dichotomous or binary outcome variables. The purpose of the model is to estimate the probability that an observation with particular characteristics will fall into a specific one of the categories. Here, the response variable Y is binary, that is it can have only two possible outcomes which we will denote as 1 and 0. Those who received at least one dose of the Covid-19 vaccine are designated as 1 and those who did not receive any dose of the Covid-19 vaccine are designated as 0. The probability of being vaccinated (P(Y = 1)) is here a function of demographic, socio-economic and medical-psychological covariates (X^T).

where P is the probability and ϕ is the cumulative distribution function of the standard normal distribution. The parameters, β, are estimated by the maximum likelihood estimator. X^T includes demographic factors like age, sex at birth (male or female), sexual orientation (straight or LGBTQIA+), race (White, Black, Hispanic, Asian or Mixed), marital status (married, widowed, divorced, separated or single), socio-economic factors like level of education, total household income, the form of employment (inperson or online), housing and living conditions and whether or not received pandemic stimulus benefit as well as medical-psychological factors like chronic physical illness, mental illness, Covid-19 illness, health insurance, trust on Covid-19 vaccine (I don’t trust Covid vaccine) and beliefs about COVID-19 vaccine (Covid vaccine wouldn’t protect me or I don’t believe I need a COVID-19 vaccine).

Summary statistics of the variables

Table 1 presents the frequency and percentage statistics of the variables under the study. The analytical sample consists of 69,114 participants of whom 60,326 (87.28%) received at least one dose of the COVID-19 vaccine and 8,788 (12.72%) did not receive the COVID-19 vaccine. COVID-19 vaccine hesitancy is measured by the response from the survey participants who were not fully vaccinated and did not express definite intentions of getting the COVID-19 vaccine and answered the following questions: 1) I don’t trust COVID-19 vaccines; 2). I don’t know if a COVID-19 vaccine will protect me; 3). I don’t believe I need a COVID-19 vaccine. Answers to the questions are coded as 1 for yes and 0 for no. The detailed analysis of summary statistics is exhausted for brevity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Summary statistics.

https://doi.org/10.1371/journal.pgph.0001927.t001

Result of probit analysis

Table 2 reports the results of the Probit models analysed in STATA software. The null hypothesis of the Wald chi-square test is that at least one of the predictors’ regression coefficients is not equal to zero. The Wald chi-square statistics with a p-value of 0.000 tells us that our models as a whole are statistically significant, that is, they fit significantly better than models with no predictors. Pseudo R-square reports the goodness of the fits of the models.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Result of probit models.

Dependent variable is the Covid-19 vaccination status (1 = vaccinated, 0 = unvaccinated).

https://doi.org/10.1371/journal.pgph.0001927.t002

The coefficient of age is negative and the coefficient of age squared is positive which implies that there exists a U-shaped relationship between age and the predicted probability of Covid-19 vaccination. This means that an increase in age increases the predicted probability of Covid-19 vaccination after a threshold point of age. This is because of the fact that as for age differences, confidence was found to decrease with age whereas perceived needs were found to increase with age [12]. These age differences are consistent with prior research on vaccine hesitancy as younger age cohorts are more likely to be healthy and have fewer underlying health conditions, as a result, they may not perceive the need of taking the vaccine. On the other hand, older age people are more likely to suffer from chronic diseases and more likely to be vaccinated.

The coefficient of sex is positive which indicates that females are more likely to be vaccinated than the reference group of males at <0.01 level of significance. These findings are generally in line with evidence revealed in prior research. Compared to men, women, in general, are more likely to engage in preventive medicine and use health services [21]. During the COVID-19 pandemic, women are more likely to perceive it as a serious health problem and abide by public health restrictions than men. The coefficient of sexual orientation is positive which implies that LGBTQIA+ people are more likely to be vaccinated than the reference group of straights at <0.01 level of significance. This may be the case due to the fact that LGBTQIA+ people are more prone to infectious diseases like HIV and are more likely to believe in the safety and protection of vaccines than heterosexual adults.

The coefficients of race reveal that Black Americans are 14% less likely to get Covid-19 vaccination than White Americans at <0.01 level of significance. Moreover, people of mixed race are 3.4% less likely to get Covid-19 vaccination than White Americans at <0.10 level of significance. However, Hispanic and people of Asian origin are more likely to be vaccinated than the reference group of White Americans <0.01 level of significance. Institutional racism and historical inequities in health care may also play a role in vaccine hesitancy among African Americans and other people of color [22].

The coefficients of marital status disclose that widowed and divorced Americans are less likely to get Covid-19 vaccination than married Americans at the conventional level of significance. This may be due to the fact that married people are more concerned about their children health and feel the necessity of being vaccinated in order to protect their offspring. However, bachelor Americans are 12% more likely to get Covid-19 vaccination than married Americans at <0.01 level of significance.

The coefficients of educational status expose that people with a formal educational degree are more likely to get Covid-19 vaccination than those with less than high school or no formal degree at the conventional level of significance. It is consistent with the prior research that educational levels are positively related to safety concerns and thus people with higher education levels are more likely to be vaccinated [23].

The coefficients of household income imply that households with higher income are more likely to get Covid-19 vaccination than those with lower income at the conventional level of significance. People with high income are more concerned about their health care and physical fitness and have better access to medications and vaccines than the low-income people. Moreover, this study finds that persons doing inperson work are 7% more likely to get Covid-19 vaccination than those doing work virtually at <0.05 level of significance. Persons working outside inperson are more vulnerable to catching Covid-19 than persons doing online work from home and thus they have higher propensity to be vaccinated.

The coefficients of housing and living conditions indicate that homeless people are 22.3% less likely to get Covid-19 vaccination than those living in one-family house detached from any other house at <0.01 level of significance. Vaccine access for persons experiencing homelessness can be enhanced by using multiple strategies, including pop-up vaccination clinics in convenient locations, mobile clinics in partnership with trusted providers, and street outreach teams. Additionally, households that received pandemic stimulus benefits from the government are 2% more likely to get Covid-19 vaccination than those who did not receive any pandemic stimulus benefits from the government at <0.05 level of significance. Fiscal stimulus provides incentives to the people and motivated them to take vaccines in order to get future monetary benefits.

Medical and psychological factors reveal that persons with a chronic physical illness like diabetes, asthma, arthritis, chronic heart, lung or kidney disease are 12% more likely to get Covid-19 vaccination than those who did not any chronic physical illness at <0.01 level of significance. Moreover, persons with a mental illness are 28% more likely to get Covid-19 vaccination than those who did not any mental illness at <0.01 level of significance. Respondents with chronic respiratory disease and those with autoimmune diseases were more likely to want to be vaccinated to protect themselves from COVID-19 [24]. Persons who tested Covid-19 positive before are 48% more likely to get Covid-19 vaccination than those who did not detect such illness at <0.01 level of significance. Though prior Covid-19 infection had offered protection against reinfection before the participants were vaccinated, this protection had waned. Vaccination provided increased, longer-lasting protection to previously infected participants, and it gave them greater than 90% protection against reinfection more than 18 months after the primary infection [25]. However, this paper finds no statistically significant relationship between Covid-19 vaccination and health insurance coverage which implies that since Covid-19 vaccines are publicly provided free of charge, health insurance does not matter.

Moreover, people who don’t have trust in the Covid-19 vaccine are 60% less likely to be vaccinated than those who have trust in the Covid-19 vaccine at <0.01 level of significance (Model 1). Trust is a core predictor, with distrust in vaccines raising vaccine hesitancy. The most given reasons for the general lack of trust in vaccines are doubts about the efficiency of the vaccine and concerns about the side effects of the vaccine. People who believe that the Covid-19 vaccine wouldn’t protect them are 63% less likely to be vaccinated than those who do not have such belief at <0.01 level of significance (Model 2). People who believe that they don’t need the Covid-19 vaccine are 48% less likely to be vaccinated than those who do not have such belief at <0.01 level of significance (Model 3). What is more, COVID-19 is not only a pandemic, but an infodemic [26]. The growth in internet use and reliance on social media sources such as YouTube, Facebook, Twitter, and TikTok has changed the landscape of information gathering [27]. Low trust in vaccines and anti-vax beliefs are associated with a greater reliance on social media for health information generation [28]. Thus, along with removing barriers to vaccination and educational campaigns about vaccine safety and the societal benefits of high vaccination uptake, mandates could be introduced for increasing vaccination uptake to the required levels.