Investigating Social Pressure, Altruism, Free-Riding, and Non-Compliance in Mask Wearing by U.S. Residents in Response to COVID-19 Pandemic

Human behavior, including making decisions about personal protective behaviors such as wearing a mask in public, impacts the trajectory of the COVID-19 pandemic and associated reopening/recovery efforts. A nationally representative survey of 1,198 U.S. residents was used to study demographics, perceptions, and stated beliefs of residents who indicated they believes that masks have a role in society in response to COVID-19 and whether they self-reported wearing masks in the seven public locations studied. Of those who believed masks had a role, a lower percentage of respondents voluntarily wore a mask in their workplace (42%) or gym (49%) when compared to other locations studied. A higher percentage of women who believed masks have a role in society voluntarily wore a mask at a big box or specialty grocery store, home improvement store, school, clothing retailer or other retailer when compared to the percentage of men. Individuals who believed that wearing masks protected others were more likely to report voluntarily wearing them, showing evidence of altruism. Social pressure was found to negatively impact the probability of voluntary mask wearing amongst those who believed masks have a role. This suggests that social shaming won’t increase compliance among these individuals and that bandwagoning was not evident in encouraging mask wearing behavior, for at least some segments of the population. Free-riding is one possible explanation for why an individual respondent may self-report belief that mask wearing has a role in society and simultaneously self-report not voluntarily wearing a mask in public locations, although it is admittedly not the only plausible explanation. Promotion of altruism, rather than social shaming, is likely to increase mask wearing based on the data collected in June of 2020. Tactics to improve public health initiative compliance and participation may change as the duration of the pandemic’s impacts increase and/or may differ in segments of the population prone to greater levels of individualism or with varying levels of knowledge about viral disease spread.


Introduction
COVID-19 spread in the human population is an epidemiological question, focusing on the study of health in a population, in addition to concerns about 'how' and 'why' the disease progresses through that population [1]. One could delve into the differentiation between descriptive epidemiology [2] versus analytical epidemiology [3] to identify factors associated with disease and inform targeting of public health prevention and control [3]. The medical and public health aspects of the pandemic remain at the frontier of the knowledge gap necessary to advance societies in combatting the COVID-19 pandemic. But, while understanding the virus and disease may inform better control and treatment plans, the behavior of individual people within communities determines the spread and ultimate course of the pandemic within a population.
Social and behavioral sciences relevant during a pandemic include work related to threat perception, leadership, individual and collective interests, science communication, social context, and stress and coping [4]. Human behavior, including the allocation of finite resources such as time, money, and our own emotional or mental attention is a critical component of COVID-19 response and recovery. There are a variety of human behavior explanations for sub-optimal individual efforts or behaviors/practices in response to COVID-19, especially after many months of coping and confusion in response to public health messaging. Free-riding, altruism, and bandwagoning behaviors have been studied in health-related practices prior to the COVID-19 pandemic, including notably in vaccination decisions [5]. Free-riding means fundamentally that an individual is taking advantage of the efforts by others to establish some collective good without actually contributing oneself, which is often used in the contexts of economics, psychology, and political science to refer to the negative impacts of this behavioral problem [6].
Altruism, the selfless concern for others or general caregiving for others beyond oneself, is a powerful psychological factor or trait that has been studied in great depth with regard to how it can influence human behavior and decision making [7,8,9]. Framed in the context of game theory and vaccination for influenza, Shim et al. (2012) have found that contrary to the assumption that individuals maximize their personal payoffs when making decisions and act according to self-interest, that altruism indeed plays a role in vaccination decisions [9]. Altruism bas been referenced with respect to mask wearing in response to COVID-19 [10] although there are undoubtedly a number of frameworks for understanding such behavior, of which altruism is only one. Bandwagoning behavior reflects an activity or action that is currently fashionable or socially supported, often recognized as peer pressure or some degree of societal inertia.
Bandwagoning is rooted in conformity and group think in social psychology, fundamentally suggesting that the rate of acceptance of behavior or belief goes up the more that those behaviors or beliefs have already been adopted by others, irrespective of the individual's own views or opinions [11,12]. Bandwagoning in medicine has been described by Cohen and Rothschild (1979) as "the overwhelming acceptance of unproven but popular ideas" that are often disproved, abandoned, and replaced by another bandwagon (or sometimes proven valid and justified, albeit after the fact) [13]. Indeed, bandwagoning and the want to conform to social pressures was found to impact nursing personnel decisions in an experimental survey conducted in-the-field by Canterelli et al. (2018), which also included other factors such as denominator neglect, zero-risk effects, halo effects, and anchoring [12].
Individual decision making about vaccination may be affected by the choices of others.
Vaccination produces the externality of reducing transmission of a disease, and can thus provide incentives for others to free-ride on the benefits while not incurring the costs of vaccination themselves [14]. Evidence has been found that altruism, free riding, and bandwagoning were significant motivators for vaccination acceptance against a contagious disease in a hypothetical research study setting [5]. In contrast, empirical evidence of vaccination creating peer-pressure rather than free-riding has been found by other researchers in a discrete choice experiment setting [15]. Vaccination is not the only health practice to which free-riding, altruism, and bandwogoning behaviors can be hypothesized; washing of hands, various hygiene practices, wearing of facial coverings (masks), and isolating oneself from others when ill can all be considered through these lenses. Adherence to a vaccination protocol is hard to observe socially, but handwashing is more easily casually observed, albeit only while in the restroom or near handwashing facilities. Whether one chooses to isolate themselves when ill to prevent spread to others may be too extreme, as the individual has knowledge of the potential consequences, as they are verifiably ill. But not washing one's hands properly or not wearing a mask when one feels well may indeed be subject to interpretation as to why one would seek to avoid such personal costs when the potential for personal and societal consequences are known. Dating back to 1847 with Dr. Ignaz Semmelweis in Vienna [16] handwashing is a known essential component of infection control [17,18]. Social pressure applied to hand washing behaviors in individuals have demonstrated varying influence, while organizational culture interventions have shown positive results [18].
Considering the potential for free-riding, altruism, peer-pressure (i.e. bandwagoning behavior), and protest/angry resistance driving mask wearing behaviors by individuals in the U.S., a survey was designed and administered to a nationally representative sample of respondents who were asked about their beliefs and behaviors with regard to facial masks in response to COVID-19 in June 2020. This analysis fundamentally seeks to gain understanding of the demographics and stated perceptions of a specific segment of the population, those who direct-stated agreement that masks have a role in the U.S. response to the COVID-19 pandemic, but also report not wearing a mask in public locations where they have visited during the pandemic time period. Stated beliefs by those who wear/do not wear masks in various public locations (including in-person religious services, big box grocery store/supermarket, specialty grocery store, gym, home improvement store, restaurant, workplace, school, clothing store, and retail store other than grocery clothing or home improvement) are summarized to offer insights into what beliefs were prevalent among those wearing masks voluntarily versus those not. The potential for externalities in one's behaviors protecting or threatening others, in addition to the possibility of legitimate misunderstandings about masks and/or perceived risks based on geographical location are discussed.

Materials and Methods
The demographics of the survey respondents were targeted to be representative of the U.S. population (U.S. Census) for the demographic categories of gender, age, income, education, and region of residence. Region of residence was as defined by the U.S. census [19]. The survey questions, which were designed to gain a better understanding of the impact of COVID-19 as well as the beliefs surrounding and usage of masks, were developed and distribute using Qulatrics [20]. Data collection took place during the beginning of the relaxation of social distancing in many regions of the U.S., from June 12, 2020 to June 20, 2020. Kantar, a company which hosts an opt-in online panel of potential respondents was used to recruit and contact respondents [21]. A total of 1,198 completed responses were obtained and analyzed. Nevada [23].
In order to gauge general perceptions of facial coverings in response to the pandemic, respondents were asked Do you agree that masks (meaning any face covering that covers your nose and mouth) have any role in U.S. society related to the spread of viral disease, especially COVID-19, in the June -December 2020 time frame? Answer choices provided included NOthey have absolutely no role whatsoever in U.S. society or YES -they have some potential role in U.S. society. The test of proportions, conducted using STATA/SE16 [24], was used to compare the demographics of the respondents who selected yes, and those that selected no. The test of the difference of two proportions 1 and 2 , was calculated as: given: where x1 and x2 are the total number of successes in the two populations [25].
Respondents that indicated they believed masks had some potential role in U.S. society were also asked to indicate the locations they visited and their mask wearing status while at that location. Locations included in-person religious services, big box grocery store/supermarket, specialty grocery store, gym, home improvement store, restaurant, workplace, school, clothing store, and retail store other than grocery clothing or home improvement. The percentage of respondents that visited a location and voluntarily wore a mask was statistically compared among locations using the test of proportions (Eq 1-2). Whether the respondent visited a location and voluntarily wore a mask was further broken down and statistically compared by gender, income, education, child status, and state COVID-19 classification. Income was condensed to higher income (an income of $50,000 or higher) and lower income (an income of $49,999 or lower). Education was condensed to college or more and less than college education. The test of proportions (Eq 1-2) was used to compare demographics and voluntary mask wearing. For example, the percentage of women versus men who voluntarily wore a mask at an in person religious service.
Respondents were asked to indicate on a scale from 1(not impacted) to 5(impacted) the level of COVID-19 related impact they experienced for four different activities outside of work/school, specifically respondent's daily activities outside of work/school, ability to buy paper products (e.g. toilet paper, paper towels), ability to find meat, milk, and perishable grocery items, and activities related to the respondents work/school. Respondents were also permitted to select this activity does not apply to me. For example, some respondents may not attend work or school, and those responses were not included in this analysis. Activities included: the respondent's daily activities outside of work/school, ability to buy paper products (e.g. toilet paper, paper towels), ability to find meat, milk, and perishable grocery items, and activities related to the respondent's work/school. The mean score between those who voluntarily wore a mask and did not voluntarily wear a mask were statistically compared using a t-test for those locations that pertained to a particular activity. For example, the mean responses to the impact COVID-19 had on the respondent's daily activities outside of work/school were statistically compared between those who went to and voluntarily wore a mask at an in-person religious service and those that did not voluntarily wear a mask. The test for µx (sample x) =µy (sample y) for unknown σx (standard deviation) and σy and σx≠ σy is [26]: where ̅ is the mean of sample x, ̅ is the mean for sample y, s is the standard deviation and n is the sample size. The result of Equation 3 has a Student's t distribution with v degrees of freedom given by [27]: Respondents were asked to indicate if they agreed with a series of 7 statements regarding mask usage. A series of logit models of the probability a respondent visited a location and voluntarily wore a mask were estimated. Logit models were chosen because the probability the person visited and voluntarily wore a mask took on the form of either 1 or 0. The latent utility (Vin) for location i and respondent n can be represented by the equation: indicates the respondent agreed that masks will prevent them from spreading COVID-19, LockDownin indicates the respondent agreed that masks will help prevent future lockdowns, Pressurein indicates the respondents agrees there is social pressure in their community to wear a mask, and NoPreventin indicates the respondent agrees that masks do not prevent the spread of COVID-19. The unobserved error term which is assumed independently identically, distributed extreme value is represented by ein [28]. The logit probability (Pin) for location i and respondent n becomes: Because coefficients from logit models can be difficult to interpret, marginal effects were estimated.

Results
Out of the 1,198 respondents obtained, 996 (83%) indicated that masks have a role in U.S. society to prevent the spread of viral disease, including COVID-19, and 202 (17%) did not (Table 1). A lower percentage of respondents who believed masks had a role were 25-34 (17%) and a higher percentage were 65+ when compared to those who did not believe 24% and 9%, respectively. A lower percentage who believed masks had a role had an income of $0-24,999 (22%) when compared to those who did not believe (33%). Conversely, a higher percentage of respondents who said yes had an income of $100,000 and higher (21%) when compared to those who said no (9%). Of those who said they believed masks had a role in society to prevent viral spread, a lower percentage graduated from high school but did not attend college (27%), when compared to those who did not believe (37%). A higher percentage of those who said yes attended college, associates or Bachelor's degree earned (33%) or attended college, graduate or professional degree earned (14%) when compared to those who said no (22% and 9%, respectively). Of those who said they believed masks have a role, a higher percentage were from the northeast (20%) when compared to those who did not believe (11%). Finally, a lower percentage of respondents who said yes, that masks have a role, were from high increase in COVID-19 case states (21%) when compared to those who said no (28%).
Of respondents who visited the following locations, greater than half reported wearing a mask voluntarily in a big box grocery store/supermarket, a specialty grocery store, a home improvement store, a school, a clothing store, or a retail store other than a grocery, clothing, or home improvement store (Table 2). A higher percentage of females voluntarily wore a mask when compared to men at big box grocery stores (69% vs 57%), specialty grocery stores (66% vs 53%), home improvement stores (65% vs 55%), school (65% vs 50%), clothing stores (64% vs 54%) and other retail stores (67% vs 58%). A higher percentage of respondents with a lower income voluntarily wore a mask at all locations studied with the exception of the gym, workplace and school when compared to those with a higher income. A higher percentage of respondents without college degrees voluntarily wore a mask at an in person religious service (57%) or a clothing store (63%) when compared to those with a college degree or higher 46% and 54%, respectively. A higher percentage of respondents without children voluntarily wore a mask at a retail store other than grocery clothing or home improvement store (65%) when compared to those with children (57%). For all locations with the exception of the workplace, a higher percentage of people who were not in the high case/population states voluntarily wore a mask when compared to the high case/population states. When compared to the not high increase in COVID-19 states, a higher percentage of respondents in the high increase in cases states voluntarily wore a mask at the big box grocery store/supermarket (74% vs 60%), at the specialty grocery store (70% vs 57%), at the home improvement store (73% vs 56%), and at other retail stores (73% vs 60%) when compared to the not high increase in COVID-19 cases states.
Those who voluntarily wore a mask in a big box grocery store/supermarket indicated a higher mean level of impact of COVID-19 on their daily activities outside of work (3.90), their ability to buy paper products (3.66), and their ability to find meat, milk, and perishable grocery items when compared to those who did not voluntarily wear a mask, 3.65, 3.66, and 2.96 respectively (Table 3). A higher mean impact score for their daily activities outside of work/school and ability to buy paper products was found for those who voluntarily wore a mask at a specialty grocery store (4.05 and 3.69) when compared to those who did not (3.71 and 3.47).
Those who voluntarily wore a mask at the gym, a restaurant, a clothing store, or other retail store all had higher COVID-19 impact scores for their daily activities outside of work/school. The impact score for ability to buy paper products was higher for those who voluntarily wore a mask at a home improvement store (3.63) when compared to those who did not (3.39). For those who voluntarily wore a mask at their workplace, the COVID-19 impact score for activities related to their work/school (4.07) was higher than those who did not voluntarily wear a mask (3.74).
A higher percentage of those who voluntarily wore a mask at a big box grocery store (85%), a specialty grocery store/supermarket (83%), a home improvement store (83%) or other retail store (82%) also selected that they agreed with the statement masks help prevent me from getting COVID-19 when compared to those who did not voluntarily wear a mask, 73%, 75%, 74%, 82%, and 72% respectively (

Discussion
Voluntary mask wearing is socially and culturally complicated, and a variety of measurement or reporting issues arise that further complicate analysis of mask wearing behaviors. Taken together, the demographics of voluntary mask-wearers are somewhat noisy and vary depending the specific location in question. However, in order to voluntary wear a mask in a given location, the respondent had to visit that location, which was accounted for explicitly in the data collection and analysis. Yet, the potential exists that particularly concerned citizens and/or those with high-risk family members or young children who may be more likely to wear a mask did not venture to various public places studied, even after restrictions were lessened or eliminated.
Media stories have highlighted the lack of return to dining out, for example, even when restaurants are allowed to legally reopen in different geographical regions [29]. Thus, while mask wearing compliance and behavior was of primary focus of the analysis, other behaviors such as social distancing, staying home as much as possible, avoiding public places, limiting trips, and other more conservative practices are necessarily 'at odds' with mask wearing behavior since in order to wear a mask in public, the individual must have ventured into public.
Greater levels of self-reported impact on daily activities due to COVID-19 were reported among those who wore masks voluntarily in the public places studied. Fundamentally, but generically, it was to be expected that those who reported more directly negative consequences responded by taking actions themselves. Past studies have identified one's own experiences to influence probability of taking action to safeguard against illness. For example, experiencing influenza exposure in the past increased the likelihood of vaccination acceptance in an experimental study [14]. Josef Woodman, CEO of Patients Beyond Borders recently stated "It's much harder for Americans to grasp the widespread harm a pandemic can cause, making them less enthusiastic about group sacrifices that can curb the disease.". In the recent Politico article in which he pointed out the recent dodging of pandemics by the U.S. or relatively light impacts of those which did arrive on U.S. soil [30]. Lack of dire consequences seen first-hand in other nations, in particular Asian countries who now readily embrace mask wearing, may aid in explaining why U.S. residents do not subscribe as readily to taking individual actions to prevent societal harm.
Perhaps direct impacts or personal consequences being related to future protective measures are expected. However, the impacts of social pressure and/or voluntary mask wearing for the protection of others, and not in response to one's own personally incurred costs, is much more complicated. There was a decrease in the probability that a respondent wore a mask to a variety of public places if they agreed that there was social pressure to do so, fundamentally indicating that there was a willful 'pushing back' against social pressures to wear masks. This rebellion against mask wearing is fodder for debate in mass media. Masks are not worn for a variety of reasons in the U.S. such as seeing mask mandates as an attack on freedom, believing masks make them look weak, believing (incorrectly) masks cut off oxygen supply, or simply finding masks uncomfortable [30]. These viewpoints differ when compared particularly to Asian countries where mask wearing is more commonly believed to be part of civic obligation in public health [30].

Arguments about individual rights and unconstitutional restrictions during COVID-19
indeed point to the will of individuals to continue on with chosen practices or behaviors, unfettered by public health restrictions. The Supreme Court rejected, 5 to 4, a request from a church to block enforcement of restrictions on attendance at religious services by the state [31].
A Pew Research Center study found that 79% of Americans believed that religious houses of worship should be required to follow the same social distancing and gathering rules as other organizations or businesses in the same geography, whereas the other 19% believed that they should be offered more flexibility [32]. While the specific location, such as a church versus a grocery store, may impact views, the conversation about putting one's individual preferences ahead of public health needs remains heated and heavily rooted in cultural expectations.
Individualism is proposed as one of the reasons that the U.S. is among the few developed countries in the world without a universal health care system, proposed Josef Woodman, CEO of Patients Beyond Borders in a Politico article [30].
Regression analysis has provided evidence, in response to hypothetical scenarios presented to subjects, that altruism, free riding, and bandwagoning were significant motivators in the decision to undergo vaccination [5]. Interestingly, that same study found that "Frames stressing the opportunity to free ride increase free riding. Frames stressing altruism do not increase altruism. If generalizable to other settings, these results suggest that public health programs to increase vaccine usage should stress high vaccination rates." [5]. Given the finding that the probability of voluntary mask wearing decreased as respondents reported social pressure around mask wearing suggests similarities to framing and presentation of public health programs as those seen in studying vaccination. Social pressure, while working in other regions of the world with a more established mask wearing practice, appears counterproductive. Taken together with past findings about encouraging vaccination, perhaps presentation of high compliance rates in mask wearing would serve public health better than shaming or attempts to convince the public of altruistic aspects of the practice.
Interestingly, Mah et al. (2006) suggested that hand hygiene non-adherence is better addressed by social marketing than by education or policy [33]. Whitby et al. (2007) highlights the self-interest consideration to motivate hand-hygiene practices [34]. Wilson et al. (2011) summarize Rothschild (2000) with respect to management of social issue and public health behaviors as, "In Rothschild's view, the primary benefit to one's selfthe key motivator in most situationsis often vague, uncertain, and in the distant future with public health issues. Rather than education, a social marketing approach encompassing free choice, apathy, and inertia is necessary for managing public health behavior." [35,36]. Wilson et al. (2011) conclude, with respect to hand washing behavior, "Social marketing approaches that tailor intervention messaging to an audience's beliefs, values, and unique knowledge levels should also be considered." [35]. Thus, in order to craft meaningful social marketing we must first understand the audience's beliefs, values, and knowledge; arguably any given population will have multiple audiences to be considered, with varying starting points with respect to knowledge or varying viewpoints on the topic.

Conclusions and Implications
Overarching conceptually to this analysis is the idea of free-riding in public health practices to curb the spread of COVID-19. Free-riding behavior was observed quite readily in the sense that U.S. residents reported a belief that masks have a role in society in responding to the COVID-19 pandemic crisis, yet those same individuals reported not wearing masks in various public places.
Admittedly, free-riding is only one possible explanation for this finding, which applies in the sense that respondents believe that there is a role for masks, but that the role did not extend to them as individuals in all of the public places studied and/or at all times. Alternative explanations include incorrect or incomplete knowledge about suggested mask wearing in public which could lead to a mis-match in reporting that they indeed believe that masks have a role in the public health response but that they legitimately did not understand what that role was suggested to be at the time data was collected. Alternatively, it is possible that individuals viewed specific locations, such as religious gatherings, as exempt in some way and/or that they prefer to avoid mask waring in some locations (i.e. gyms) due to personal comfort or preference, although they still agree that masks have a role in other places.
Mask wearing (or lack thereof) in public is visually observable and thus a practice that is easily socially responded to through shaming, ostracizing, or positive recognition. In contrast to the readily observed mask wearing, hand washing after using the restroom is observable only to altruism into game-theoretic models of vaccination for influenza and conclude that promoting altruism could be a potential strategy to improve public health outcomes [9]. Given the negative finding surrounding the use of social pressure and positivity associated with altruism, we support the notion that altruism promotion may be a potential strategy to improve voluntary mask wearing. Indicates the percentage or mean of respondents who indicated there had children in their household and those that did not is statistically different at the <0.05 level 1 Indicated on a scale from 1 (strongly disagree) to 5 (strongly agree). Indicates the percentage who said yes is statistically different than the percentage that said no. 2 Matching lowercase letters indicates the percentage is the same down the column. For example the percentage who voluntarily wear a mask in an in person religious service is equal to the percentage who voluntarily wear a mask to the gym, but statistically different from the percentage who wear a mask in a big box grocery store/supermarket. Ψ Indicates the percentage of respondents within the category are statistically different for that location. For example, the percentage of woman who voluntarily wore a mask in a big box grocery store/supermarket is statistically different than the percentage of men. 3 Lower income is defined as $49,999 or less, higher income is defined as $50,000 or greater. Table 3. Mean reported level of impact from COVID-19 on activities compared between those who voluntarily wear a mask and do not voluntarily wear a mask at specific locations. Impact scale was from 1(not impacted) to 5(impacted). N given in table.

Voluntarily wears mask
Your daily activities outside of work/school Ability to buy paper products (e.g., toilet paper, paper towels) Ability to find meat, milk, and perishable grocery items Indicates the mean response for the statement is statistically different between those who voluntarily wear a mask and that location and those who do not. For example the mean response that COVID-19 impacted the respondent's daily activities outside of work/school was greater for those who voluntarily wore a mask at a big box grocery store/supermarket when compared to those who do not voluntarily wear a mask at that location. Masks have negative health consequences for the mask wearer Table 5. Estimated marginal effects (from logit models) of respondent demographics, self-reported COVID-19 impacts, and beliefs regarding masks on voluntary mask wearing in 10 public locations. N given in table and specific to each location based on the number of respondents voluntarily wearing masks to that location.

Explanatory Variables
Household income