|Year : 2022 | Volume
| Issue : 1 | Page : 3-9
Individual characteristics and demographics associated with mask wearing during the COVID-19 pandemic in the United States
Echu Liu, Samantha Arledge
Department of Health Management and Policy, Saint Louis University, Saint Louis, MO, USA
|Date of Submission||17-Nov-2021|
|Date of Decision||07-Dec-2021|
|Date of Acceptance||28-Jan-2022|
|Date of Web Publication||25-Feb-2022|
Ph.D., 3545 Lafayette Ave, Saint Louis, MO 63104
Source of Support: None, Conflict of Interest: None
Introduction: Many scientific studies provide evidence of mask wearing as an effective strategy to reduce the spread of the COVID-19 virus. However, US citizens do not adhere to this prevention practice universally. Although past studies have shown disparities in mask wearing by age, gender, ethnicity, and location, the literature lacks a work that uses large-scale national survey data to understand the mask-wearing resistors' characteristics and demographics. This study's purpose is to fill this gap. Methods: We obtained this study's data from the COVID-19 Impact Survey, a nationally representative survey conducted by NORC at the University of Chicago. This survey aims at generating national and regional statistics by surveying representative regional and national samples in three time periods: April 20–26, 2020, May 4–10, 2020, and June 1–8, 2020. Data for our analysis are from the public-use files of these three waves. We performed logistic regressions to estimate the adjusted risk ratio (ARR) of not wearing masks for several personal characteristics and demographics. Results: Our results suggest that younger (average ARR = 1.66) and lower-income (average ARR = 1.51) adults are more likely not to wear a face mask to prevent the coronavirus spread. On the other hand, unhealthy (average ARR = 0.81), female (average ARR = 0.68), and minority (average ARR = 0.65) adults are less likely not to wear a mask. Furthermore, residents in the Northeast region (average ARR = 0.34) and urban residents (average ARR = 0.54) are less likely not to wear a face mask. Conclusion: Mask-wearing behavior differs by age, income, health status, gender, race, region, and geographical residence in the US.
Keywords: COVID-19, logistics models, masks, pandemics, preventive behavior, risk ratios, social-economic, United States
|How to cite this article:|
Liu E, Arledge S. Individual characteristics and demographics associated with mask wearing during the COVID-19 pandemic in the United States. Asian J Soc Health Behav 2022;5:3-9
|How to cite this URL:|
Liu E, Arledge S. Individual characteristics and demographics associated with mask wearing during the COVID-19 pandemic in the United States. Asian J Soc Health Behav [serial online] 2022 [cited 2023 Dec 3];5:3-9. Available from: http://www.healthandbehavior.com/text.asp?2022/5/1/3/338376
| Introduction|| |
There have been 265 million coronaviruses (COVID-19) cases and approximately 5.3 million deaths caused by it around the globe since the pandemic started in late 2019. This pandemic has created many health, social, and economic repercussions worldwide for families and communities. As has been previously reported in the literature, the impact of COVID-19 varies by socioeconomic status. For instance, several studies suggested that neighbors with the following features are associated with the higher transmission, hospitalization, and death rates of COVID-19: higher-income inequality, a higher percentage of the racial minority, lower median family income, and higher unemployment rate.,,, Prior research also suggests that the same connection exists at the individual level.
With more and more individuals are fully vaccinated, the growth rate of COVID-19 cases decreases overall, but the total number still increases in many countries. Other than expanding the vaccination uptake, most countries use several nonpharmaceutical public health measures to control the spread of COVID-19. Previous research suggests that wearing a face mask is the most effective among these measures, but its practice is visually noticeable.
The effectiveness of mask wearing to prevent the COVID-19 infection was a topic of fierce debate at the early stages of the pandemic. The doubt of this practice's effect becomes less as more scientific evidence supports the strong association between face mask use and the reduced transmission of COVID-19 reveal.,,, However, many individuals still refuse to wear a face mask, whether it is recommended or mandated. As prior research suggested masks are a powerful and cost-effective tool for combating COVID-19, it is imperative to understand what factors are associated with anti-mask.
Previous studies showed mask wearing is closely related to moral beliefs. Some authors have also suggested that socioeconomic factors significantly predict moral beliefs. Therefore, it is reasonable to hypothesize that mask wearing is associated with socioeconomic factors. Several studies in the broader literature have examined this association. Pakpour et al. studied older adults in Taiwan and Iran. They found that fear of COVID-19 is significantly and positively related to the possibility of practicing COVID-19 prevention behavior, including handwashing and mouth covering when sneezing. Haischer et al. used data collected from shoppers in southwestern Wisconsin and found gender, age, and location (urban, suburban, and rural) are significant determinants of the odds of individuals wearing masks voluntarily. Khubachandani et al. surveyed 835 adult US citizens nationwide. They found that gender, age, ethnicity, marital status, and living arrangement are critical in determining whether an individual wears a mask. Papageorge et al. surveyed 1000 US individuals during the 3rd week of April 2020. They found that income, work arrangements, lost income, and beliefs about the effectiveness of social distancing are significantly associated with the likelihood of wearing a mask. Rieger studied approximately 200 employees and students of the University of Trier in Germany in 2020 and found demographic factors (age, gender, and university student) are not significantly associated with the likelihood of wearing a mask. Instead, worries about the pandemic, aversion against wearing a mask, the perception of others' judgments, self-protection, and the protection of others are significant predictors. Hearne and Niño used 4688 samples from COVID-19 Impact Survey to study the racial and gender differences in the possibility of wearing a face mask during the pandemic in the US. They found a significant disparity in the mask-wearing behavior across gender and ethnicity.
Although these studies have shown the connection between mask-wearing and individual characteristics and sociodemographic factors, they cannot be considered conclusive. For instance, Haischer et al. and Khubachandani et al. concluded that age and gender are significantly associated with the likelihood of mask wearing. At the same time, Rieger did not find an identical association. A closer look at this literature reveals two shortcomings. First, most of them use data from a small-scale survey (Khubachandani et al., Papageorge et al., and Rieger). Second, even they use data from an extensive scale survey, they use data from a regional survey (Haischer et al.), or they do not consider the sampling design of a national survey when analyzing the data (Hearne and Niño). As far as we know, no previous research has addressed these two issues to investigate the association between individual demographics and mask wearing. Our study aims to fill this gap and answer this question: What personal characteristics and demographics are associated with the likelihood of “not putting a mask on” in the United States, which has the highest number of COVID-19 cases?
| Methods|| |
We collected data for this analysis from the COVID Impact Survey that NORC at the University of Chicago conducted. The COVID Impact Survey intends to provide national and regional statistics about physical health, mental health, economic security, and social dynamics during the pandemic in the United States. Its results offer weekly estimates of the US adult household population nationwide and in eighteen regional areas, including ten states (California, Colorado, Florida, Louisiana, Minnesota, Missouri, Montana, New York, Oregon, and Texas) and eight metropolitan areas (Atlanta, Baltimore, Birmingham, Chicago, Cleveland, Columbus, Phoenix, and Pittsburgh).
Data collection from the COVID Impact Survey occurred over a week, and respondents received a small monetary incentive for completing the survey. The COVID Impact Survey's conductors released data from three specific weeks for public use: April 20–26, 2020, May 4–10, 2020, and June 1–8, 2020. We used data from these three waves for analysis in this work.
Sampling methods of the COVID impact survey
As mentioned previously, NORC collected COVID Impact Survey data to generate national and regional estimates. The conductors interviewed adults aged 18 and over to represent the 50 states and the District of Columbia to create national estimates, with a goal of 2,000 interviews each week. NORC conductors selected the interviewees from the AmeriSpeak Panel, a national and probability-based panel of household representatives of the US population. Selected individuals may complete the COVID Impact Survey online or by telephone with a NORC telephone interviewer. Each respondent for generating national estimates in the COVID Impact Survey has a probability weight, which we will use to analyze national data.
The NORC conductors collected data for regional estimates using a multi-mode, address-based approach that allows each area's residents to complete an interview online or with a telephone interviewer. Addresses are randomly selected within each region using stratified sampling. All sampled households received a postcard inviting them to complete the survey either online using a unique PIN or through telephone by calling a toll-free number. Interviews are conducted with adults aged 18 and over to reach approximately 400 interviews in each region each week. Identical to the sample for national estimates, each survey respondent for generating regional estimates has a probability weight, which we will use for the regional data analysis.
To find the individual characteristics and socioeconomic factors associated with not wearing a face mask, we created a binary indicator that equals 1 if a participant of the COVID impact survey responds “Yes” to the question “Are you wearing a face mask in response to the coronavirus?” Otherwise, it equals 0.
Following the approach Sribany proposed, covariates in our analysis of the COVID Impact Survey data for generating national estimates (“national sample” hereafter) include the following variables: age, having a chronic disease, gender, race (whether an individual respondent is non-white), household income, educational attainment, census region, and current home area. Age is a variable with the following four categories: 18–29, 30–44, 45–59, and 60 or older. Having a chronic disease is a binary indicator equal to 1 if a survey respondent said they have one of the following conditions: diabetes, hypertension, heart disease, heart attack or stroke, asthma, chronic lung diseases and COPD, bronchitis, emphysema, a mental health condition, cystic fibrosis, liver disease or end-stage liver disease, cancer, or a compromised immune system. Gender is a dummy variable that equals 1 if the respondent is female and 0 otherwise. Other than gender, the COVID Impact Survey also solicits information about the respondents' racial backgrounds by categorizing them into 1) Caucasian, non-Hispanic; 2) African American, non-Hispanic; 3) Hispanic; and 4) Other, non-Hispanic. Since non-White respondents account for a smaller percentage (less than 30%) of the national sample for our analysis, we created a binary indicator of race that equals 1 if an individual respondent is nonwhite and 0 otherwise.
To consider the confounding effect of income, we also included household income as one of the independent variables. Based on the survey responses, we used five binary indicators to categorize our national sample's household income: less than $10,000, $10,000 to $29,999, $30,000 to $49,999, $50,000 to $99,999, and higher than $100,000. As educational attainment is a critical determinant of income, we also controlled for the respondent's highest level of education by including the following four dichotomous variables in our analysis: less than high school, high school, some college, and bachelor's degree or higher. To control for the regional differences in our analysis, we created four dummies, which equal 1 if a survey respondent's household is located in the Northeast, Midwest, South, or West region. Since demographic indicators and measures vary according to the type and nature of people's residence, we created three binary indicators to define the geographical category of residence based on the survey responses: rural, suburban, and urban.
We used the same covariates to analyze the COVID Impact Survey data for regional estimates (“regional sample” hereafter), excluding family income because Sribany's approach suggested that this variable is an insignificant predictor of face-mask-wearing behavior for the regional sample.
Given the nature of the survey design of the COVID Impact Survey, we performed our analysis for the national and regional samples, respectively. Descriptive statistics describe these two analysis samples, and we used logistic regressions to study the association between sociodemographic and mask-wearing factors for these two samples. Also, we considered the survey's regional and national sampling weights when reporting descriptive statistics and performing logistic regressions. Since the coefficients in logistic regressions are hard to interpret, we report adjusted risk ratios (ARR) for covariates that Norton and Miller proposed. The statistical significance of ARR estimates is determined based on the threshold of P value smaller than 5%.
Since data used in this study are from public-use files, in which all records are not individually identifiable, of the COVID-19 Impact Survey, our analysis would not involve human subjects. Therefore, this research does not require the review and approval of the Institutional Review Board of authors' affiliation.
| Results|| |
As [Table 1] shows, the national sample has a higher percentage of respondents not wearing masks than the regional sample. [Table 1] also shows that the age compositions of our national and regional samples look similar, and the individuals aged over 60 account for the highest percentage of both samples. Most individuals in the national and regional samples had a chronic disease, and the gender compositions of the national and regional samples, according to [Table 1], are almost identical (with approximately 50% being females). The regional sample has a higher rate of nonwhite respondents, and the household income distributions are similar in the national and regional samples. Most national and regional samples are from households with incomes between $50,000 and $99,999. The distribution of educational attainment is also similar for the national and regional samples. Most national and regional samples have a college or above degree. [Table 1] also shows that most of our analysis sample lived in urban areas and the South.
Regression results: Factors associated with not wearing masks
[Table 2] and [Table 3] demonstrate that ARRs with the three age dummies (aged between 18 and 29, between 30 and 44, aged between 45 and 59) are more extensive than 1 and statistically significant. This result indicates that younger adults, on average, are more likely not to wear a face mask than older adults (aged 60 and over), holding all else constant. The ARRs on the chronic disease variable indicated that, all things being equal, individuals with a chronic disease are less likely not to wear a face mask, on average. The less than one and statistically significant ARRs on the female variable reported in [Table 2] and [Table 3], imply that, on average, women are less likely not to wear a face mask when other factors are equal. The same conclusion holds for the minority sample, with ARRs of 0.64 and 0.66; both estimates are statistically significant on the minority variable for the national and regional samples. The ARRs on the household income dummies are more extensive than one and statistically significant in [Table 2]. These results mean that compared to survey respondents with higher household income (more than $100,000), typically the individuals in the national sample with lower household income (less than $10,000, between $10,000 and $29,999, between $30,000 and $49,999, and between $50,000 and $99,999) are more likely not to wear a face mask when things being equal.
[Table 2] and [Table 3] reveal that education is not a significant predictor of the risk of not wearing face masks. However, the ARR estimate on the variable “high school” shows that compared to those with less education, on average, high school graduates in the national sample are significantly likely not to wear a face mask when other factors are equal. In addition, the statistically significant 0.69 ARR estimate on “college or above” reported in [Table 3] indicated that generally respondents with a college or above degree in the regional sample have a significantly lower likelihood of not wearing a face mask.
The less than one ARRs for the three dummies of census regions in [Table 2] show that, compared to those who live in the South, typically respondents in the national sample who live in the North, Midwest, and West are less likely not to wear a face mask, holding other things constant. However, this estimated risk is statistically insignificant for the Midwesterners. ARRs on the three regional dummies reported in [Table 3] for the regional sample do not lead to a similar conclusion. The ARR for the “Northeast” variable implies that ceteris paribus, on average Northeasterners, have 0.38 times the risk of Southerners not wearing a face mask in the regional sample. This estimate is statistically significant (P = 0.00). Furthermore, according to [Table 3], Midwesterners had on average 1.41 times the risk of Southerners not wearing a face mask in the regional sample when all else being equal. This estimate is also statistically significant. However, according to the ARR on the variable “West” reported in [Table 3], the Westerners in the regional sample had the almost identical risk of not wearing a face mask to Southerners (ARR = 1.00), and this estimate is statistically insignificant.
The ARR estimates on the “urban” variable in [Table 2] and [Table 3] show that urban residents, on average, have a significantly lower risk of not wearing a face mask than rural residents when other factors are equal. However, the ARR estimate on the suburban variable equals 1 in [Table 2], meaning that suburban residents had an identical risk of not wearing a face mask to that of rural residents in the national sample, but this estimate is statistically insignificant. Conversely, [Table 3] demonstrates that generally, suburban residents, compared to rural residents, have a lower risk (ARR = 0.64) of not wearing a face mask holding other things constant, and the risk estimate is statistically significant.
| Discussion|| |
Since the start of the COVID-19 pandemic, the United States remained the country with the most cases of COVID-19 and resulting deaths. The US Centers for Disease Control and Prevention recommends that vaccinated and unvaccinated people wear masks in crowded outdoor settings and significant community and indoor public places to slow the spread of COVID-19, especially its more infectious variants. Unfortunately, wearing a mask in public is still a controversial and politicized issue in the United States. Many Americans refuse to wear a face mask to prevent the virus's spread and claim that wearing it or not is their choice, even if they reside in a state or locality whose authorities have mandated mask use. Given that masks are cost-effective options that help improve the public health outcome during the pandemic, it is essential to understand the demographic factors associated with mask-wearing behavior.
This study shows mask-wearing behavior differs by age, health status (whether having a chronic disease), gender, ethnicity, and income. Studies have shown that age is associated with a higher likelihood of COVID-19 infection and that older adults are more likely to develop severe illness from COVID-19., Therefore, a higher prevalence of mask-wearing among older adults (60+) from our analysis is understandable. A past study also shows that preexisting chronic diseases are essential risk factors of mortality associated with COVID-19. Therefore, it is natural and more possible for individuals with chronic diseases to be more compliant with mask-wearing recommendations, which is what our analysis shows. One study showed that more men have negative emotions toward wearing masks and are less likely to believe that COVID-19 is a threat. This result may explain why our analysis shows that females are more likely to wear a face mask in response to the COVID-19 pandemic. Another past study shows that racial minorities are more likely to wear a facial mask to prevent coronavirus transmission. This result could occur due to the comparatively higher percentage of minorities not working from home during the pandemic.
Papageorge et al. indicated that higher-income individuals need to adopt more self-protective behavior than lower-income people during the pandemic because lower-income individuals may only need to adjust slightly after the COVID-19 pandemic starts. Our analysis of the national sample, with results displayed in [Table 2], is consistent with the findings in Papageorge et al. According to our study, residents in the Northeast have a significantly lower possibility of not wearing a mask. Determining the exact reason for this result is challenging, but a higher number of COVID-19 cases per capita in this area may explain it. Our results also show that urban residents have a significantly lower chance of not wearing a face mask. The possible explanation is they are more discerning of COVID-19 risks than suburban and rural residents when other factors are equal because of higher population density.
This study has several limitations. First, NORC only released data from three 2020 waves of the COVID Impact Survey. Therefore, this study cannot fully identify the sociodemographic differences in mask wearing after the COVID-19 vaccine is available to the public, and an increasing number of US citizens are fully vaccinated. Second, the literature showed that state mandates increase mask wearing, but the COVID Impact Survey does not provide the state of a respondent's residence. We, therefore, cannot control this critical confounding factor in the analysis. Third, the household income may be subject to self-reported bias, an issue that cannot address using information from the COVID-19 Impact Survey. Last, the COVID Impact Survey is not a longitudinal or repeated cross-sectional survey. Therefore, we cannot interpret our analysis as causal inference.
| Conclusion|| |
This study shows Americans' variability in mask-wearing behavior by individual characteristics and demographics. Policymakers should develop communication strategies targeting specific groups, such as younger and healthier individuals, to increase their awareness of the effectiveness of mask wearing in decreasing the coronavirus's spread and their compliance with this protective behavior. Consequently, the pandemic can end sooner, and the burden on the medical care system will be lessened.
We thank two anonymous referees for their valuable suggestions and comments.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
Omer SB, Benjamin RM, Brewer NT, Buttenheim AM, Callaghan T, Caplan A, et al
. Promoting COVID-19 vaccine acceptance: Recommendations from the Lancet Commission on vaccine refusal, acceptance, and demand in the USA. Lancet 2021;398:2186-92.
Abedi V, Olulana O, Avula V, Chaudhary D, Khan A, Shahjouei S, et al
. Racial, economic and health inequality and COVID-19 infection in the United States. J Racial Ethn Health Disparities 2021;8:732-42.
Bhowmik T, Tirtha SD, Iraganaboina NC, Eluru N. A comprehensive analysis of covid-19 transmission and fatality rates at the county level in the united states considering socio-demographics, health indicators, mobility trends and health care infrastructure attributes. PLoS One 2021;16:e0249133.
Jalali AM, Khoury SG, See J, Gulsvig AM, Peterson BM, Gunasekera RS, et al
. Delayed interventions, low compliance, and health disparities amplified the early spread of COVID-19. medRxiv 2020;2020.07.31.20165654. [doi: 10.1101/2020. 07.31.20165654].
Robertson LS. Predictors of COVID-19-Confirmed cases and fatalities in 883 US counties with a population of 50,000 or More: Estimated effect of initial prevention policies. J Urban Health 2021;98:205-210.
Rozenfeld Y, Beam J, Maier H, Haggerson W, Boudreau K, Carlson J, et al
. A model of disparities: Risk factors associated with COVID-19 infection. Int J Equity Health 2020;19:126.
Talic S, Shah S, Wild H, Gasevic D, Maharaj A, Ademi Z, et al
. Effectiveness of public health measures in reducing the incidence of covid-19, SARS-CoV-2 transmission, and COVID-19 mortality: Systematic review and meta-analysis. BMJ 2021;375:e068302.
Martinelli L, Kopilaš V, Vidmar M, Heavin C, Machado H, Todorović Z, et al
. Face masks during the COVID-19 pandemic: A simple protection tool with many meanings. Front Public Health 2021;8:606635.
Pakpour AH, Liu CH, Hou WL, Chen YP, Li YP, Kuo YJ, et al
. Comparing fear of COVID-19 and preventive COVID-19 infection behaviors between iranian and taiwanese older people: Early reaction may be a key. Front Public Health 2021;9:740333.
Howard J, Huang A, Li Z, Tufekci Z, Zdimal V, van der Westhuizen HM, et al
. An evidence review of face masks against COVID-19. Proc Natl Acad Sci U S A 2021;118:e2014564118.
Zhang R, Li Y, Zhang AL, Wang Y, Molina MJ. Identifying airborne transmission as the dominant route for the spread of COVID-19. Proc Natl Acad Sci U S A 2020;117:14857-63.
Lyu W, Wehby GL. Community use of face masks and COVID-19: Evidence from a natural experiment of state mandates in the US. Health Aff (Millwood) 2020;39:1419-25.
Chen I, Ahorsu DW, Ko N, Yen F, Lin C, Griffiths MD, et al
. Adapting the motors influenza acceptance scale into the motors of COVID-19 vaccination acceptance scale: Psychometric evaluation among mainland Chinese university students. Vaccine 2021;39:4510-5.
Chan EY. Moral foundations underlying behavioral compliance during the COVID-19 pandemic. Pers Individ Dif 2021;171:110463.
Abbott, BW, Greenhalgh M, St. Clare SI, Bush J. Making Sense of the Research on COVID-19 and Masks; 2020. Available from: https://pws.byu.edu/covid-19-and-masks
. [Last accessed on 2021 Nov 16].
Caravita SC, Giardino S, Lenzi L, Salvaterra M, Antonietti A. Socio-economic factors related to moral reasoning in childhood and adolescene: The missing link between brain and behavior. Front Hum Neurosci 2012;6:262.
Haischer MH, Beilfuss R, Hart MR, Opielinski L, Wrucke D, Zirgaitis G, et al
. Who is wearing a mask? Gender-, age-, and location-related differences during the COVID-19 pandemic. PLoS One 2020;15:e0240785.
Khubchandani J, Saiki D, Kandiah J. Masks, gloves, and the COVID-19 pandemic: Rapid assessment of public behaviors in the United States. Epidemiologia 2020;1:16-22.
Papageorge N, Zahn MV, Belot M, van den Broek-Altenburg E., Choi S, Jamison JC, et al
. Sociodemographic factors associated with self-protecting behavior during the COVID-19 pandemic. J Pop Econ 2021;34:691-738.
Rieger MO. To wear or not to wear? Factors influencing wearing face masks in Germany during the COVID-19 pandemic. Soc Health Behav 2020;3:50-4. [Full text]
Herane BN, Niño MD. Understanding how race, ethnicity, and gender shape mask-wearing adherence during the COVID-19 pandemic: Evidence form the COVID impact survey. J Rac Ehnics Health Disparities 2021;19:1-8.
Norton, EC, Miller, MM. Computing adjusted risk ratios and risk differences in Stata. Stata J 2013;13:492-509.
Dong Y, Mo X, Hu Y, Qi X, Jiang F, Jiang Z, et al
. Epidemiological characteristics of 2,143 pediatric patients with 2019 coronavirus disease in China. J Emerg Med 2020;58:712-3.
Ma C, Gu J, Hou P, Zhang L, Bai Y, Guo Z, et al
. Incidence, clinical characteristics and prognostic factor of patients with COVID-19: A systematic review and meta-analysis. medRxiv 2020;2020.03.17.20037572. Availble from: http://medrxiv.org/lookup/doi/100.1101/2020.03.17.20037572
. [Last accessed on 2021 Nov 29].
Alyammahi SK, Abdin SM, Alhamad DW, Elgendy SM, Altell AT, Omar HA. The dynamic association between COVID-19 and chronic disorders: An updated insight into prevalence, mechanisms and therapeutic modalities. Infect Genet Evol 2021;87:104647.
Capraro V, Barcelo H. The effect of messaging and gender on intentions to wear a face-covering to slow down COVID-19 transmission. PsyArXiv 2020. [doi: 10.31234/osf.io/tg7vz].
[Table 1], [Table 2], [Table 3]