key: cord-0280105-gd463nrq
authors: Ridenhour, B. J.; Sarathchandra, D.; Seamon, E.; Brown, H.; Leung, F.-Y.; Johnson-Leon, M.; Megheib, M.; Miller, C. R.; Johnson-Leung, J.
title: Effects of trust, risk perception, and health behavior on COVID-19 disease burden: Evidence from a multi-state US survey
date: 2021-11-19
journal: nan
DOI: 10.1101/2021.11.17.21266481
sha: 6b408ba6d79f04fe7c3e699588450e183c338f1f
doc_id: 280105
cord_uid: gd463nrq

Early public health strategies to prevent the spread of COVID-19 in the United States relied on non-pharmaceutical interventions (NPIs) as vaccines and therapeutic treatments were not yet available. Implementation of NPIs, primarily social distancing and mask wearing, varied widely between communities within the US due to variable government mandates, as well as differences in attitudes and opinions. To understand the interplay of trust, risk perception, behavioral intention, and disease burden, we developed a survey instrument to study attitudes concerning COVID-19 and pandemic behavioral change in three states: Idaho, Texas, and Vermont. We designed our survey (n = 1034) to detect whether these relationships were significantly different in rural populations. The best fitting structural equation models show that trust indirectly affects protective pandemic behaviors via health and economic risk perception. We explore two different variations of this social cognitive model: the first assumes behavioral intention affects future disease burden while the second assumes that observed disease burden affects behavioral intention. In our models we include several exogenous variables to control for demographic and geographic effects. Notably, political ideology is the only exogenous variable which significantly affects all aspects of the social cognitive model (trust, risk perception, and behavioral intention). While there is a direct negative effect associated with rurality on disease burden, likely due to the protective effect of low population density in the early pandemic waves, we found a marginally significant, positive, indirect effect of rurality on disease burden via decreased trust (p = 0.095). This trust deficit creates additional vulnerabilities to COVID-19 in rural communities which also have reduced healthcare capacity. Increasing trust by methods such as in-group messaging could potentially remove some of the disparities inferred by our models and increase NPI effectiveness.

Rural Americans face increased risk of severe illness and death from COVID-19 due 27 to health disparities, health care shortages, and social inequities [17, 18] . On average, 28 rural Americans are older, are more likely to live in poverty, have higher rates of chronic 29 disease and disability, and are less likely to be insured than urban dwellers [19, 20] . 30 Studies have consistently shown less compliance with NPIs in rural areas, particularly 31 among rural Americans identifying as conservative. These associations were less strong 32 among older rural individuals [17, 18] . The lack of healthcare resources due to hospital 33 closures, limited numbers of health professionals, and low critical-care capacity in rural 34 communities poses an additional risk in the face of a surge of patients with 35 COVID-19 [21] . 36 In this study, we use a survey instrument distributed in three socially and 37 demographically diverse US states (Idaho, Texas, and Vermont) during October and 38 November 2020 in order to examine the differences among rural and urban Americans in 39 their attitudes towards, and uptake of, NPIs. To advance health behavior theory, we 40 tested various causal relationships between trust in public health guidance, health and 41 economic risk perception, and resistance to pandemic behavioral change using structural 42 equation modeling. Secondarily, we also explore the relationship between disease burden 43 and behavior with models of our survey data. From the best-supported models, we 44 determine how rurality-along with other exogenous variables such as political 45 ideology-factors into behavior during the early portion of the COVID-19 pandemic. 46 Our work is important because it incorporates human attitudes, perceptions, and 47 behavioral intention into infectious disease models, which extends our ability to predict 48 expected differences in disease outcomes across the United States. Materials and methods 50 Survey development and data collection 51 Data for this research come from a sequential mixed-mode survey distributed to a 52 disproportionate stratified sample of households in Idaho, Texas, and Vermont. The 53 specific survey design, employing both an online and a paper survey option, as well as 54 English and Spanish translations, was selected in order to reach communities that are 55 typically harder to reach via online surveys (for example, rural and elderly populations, 56 individuals who lack access to reliable internet connections, and non-English 57 speakers) [22] . Following standard survey design principles, the survey design includes 58 several steps: pre-testing, field testing, pilot testing, and validation [23, 24] . We first 59 pre-tested the survey with a convenience sample of college students (n = 55) recruited 60 from the University of Idaho via the online survey platform Qualtrics (Provo, UT, 61 USA). Pre-testing enabled us to measure pertinent factors such as time for completion, 62 satisfaction, and level of difficulty. Subsequently, we field tested the survey 63 questionnaire by sharing it with 10 state and regional public health experts and one 64 community based organization serving Hispanic populations in Idaho (Community 65 Council of Idaho). This organization helped us to verify the Spanish translation and 66 determine its cultural resonance. Feedback from these experts was used to revise and 67 refine the survey questions to help ensure their validity and reliability. Lastly, we pilot 68 tested the survey using Qualtrics by distributing the survey to 50 respondents each from 69 ID, TX, and VT (n = 150) between August-September 2020. For each state, we 70 obtained equal proportions of rural and urban/suburban respondents. We conducted 71 consistency analysis using the pilot data and examined other factors such as time for 72 completion and any inexplicable patterns in the pilot data. The finalized survey covers 73 topics including: worry about COVID-19, social distancing, mask wearing, economic 74 impacts, contact tracing, vaccination intention, trust, information sources, and 75 demography. Our questions are theory driven, tapping into constructs from common 76 health behavior theories such as Social Cognitive Theory [25] and the Health Belief 77 Model [26] . We also rely on CDC's Behavioral Risk Factor Surveillance System 78 (BRFSS) and other published survey studies, e.g., Jamieson and Albarracín [27] , to 79 determine the consistency and validity of survey questions.

Our disproportionate stratified sample purchased from Dynata (Shelton, CT, USA) 81 consists of 2000 rural and 2000 urban or suburban addresses from each of Idaho, Texas, 82 and Vermont (12000 in total). Dynata classifies addresses as rural if they fall outside of 83 a metropolitan statistical area (MSA) as defined by the US Office of Management and 84 Budget. We employed the services of Washington State University's Social & Economic 85 Sciences Research Center to distribute the survey. All household addresses within the 86 sampling frame were sent an initial invitation letter-which included a $1 USD 87 incentive-directing respondents to a URL where they were asked to enter their unique 88 response ID and complete the survey online. Non-respondents were sent a reminder 89 postcard one week later, and two weeks after that a final reminder letter was mailed. 90 We offered a phone number and an email address with the option to reach out to us to 91 request a paper copy of the survey for those preferring the paper option. Online survey 92 data collection occurred during October and November, 2020. Requested paper surveys 93 were mailed in mid-November, and the data collection was completed in December, 94 2020.

Our survey questionnaire (Appendix S1) was approved by the University of Idaho . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 19, 2021. ;  clicking a button to proceed to the full survey. The participants who took the paper 101 survey read a consent form and voluntarily mailed back their completed surveys. This 102 study does not include any retrospective medical records or archived samples.

Our demographic variables are comprised of direct measures of five attributes. Political 105 ideology is coded as an unordered factor with levels: liberal, moderate, conservative, 106 libertarian, non-political, and other; moderate is designated as the reference level for 107 statistical analyses. The remaining measures are recorded as Boolean variables 108 measuring race (white = 1), gender (female = 1), age (over 64 years = 1), and 109 geography (rural = 1). See Table S3 .1 for a detailed breakdown of demographic 110 characteristics. Except for geography, which is determined by our de-identified provides a more detailed geographic structure for urban and rural delineation [29] .

For our analyses, we geographically mapped all survey respondents (n = 1034) for all 120 three states (ID, TX, and VT), and associated RUCA codes based on the respondents' 121 de-identified addresses. We used ArcGIS software from Environmental Systems 122 Research Institute, Inc. (ESRI; West Redlands, CA, USA) to perform this spatial 123 association. We then designated respondents whose RUCA primary code was 1, 2, 3, or 124 4 as urban, and all other RUCA codes as rural (see supplemental Table S3 .2 for a full 125 list of all RUCA code designations). This stricter classification, as opposed to MSA 126 classifications used in the sampling frame, ensures that rural-classified responses would 127 reflect rural attitudes and experience [29] . 128 We use two different measures of disease burden. For models where behavioral 129 intention is hypothesized to affect disease burden, we consider cumulative cases per 100 130 people from the beginning of the pandemic in January 2020 through 30 April 2021, at 131 the county level, as reported by the New York Times [30] . Choosing a date after the 132 survey period enables observation of delayed consequences of behavior on disease burden. 133 The chosen sample date captures the main wave of the pandemic in the US prior to 134 widespread availability of the vaccines. Exploratory analyses showed that the exact 135 choice of date has little-to-no effect on model results, which is to be expected given the 136 auto-correlative structure of spatiotemporally-distributed cumulative disease data.

For models where disease burden is hypothesized to affect behavioral intention, we 138 use the cumulative case count from January 2020 to the recorded response date of the 139 observation. Using this measure is consistent with the idea that previous, personal . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 19, 2021. ;  of simultaneous linear regression equations. Latent variables are those for which there is 150 no direct measurement, but rather the variables are inferred indirectly via indicators.

All analyses were performed in R v4.1.0 (see Appendix S2 for R code). The four central 152 latent variables in our attitudinal framework are inferred as follows.

"Health risk perception" and "economic risk perception" are derived from survey 154 questions which asked directly about respondents' concern for their own and community 155 health and economic security. For these two measures, higher values of the latent 156 variable indicate higher perceived risk. "Trust" is derived from similar questions which 157 probe the degree of trust in COVID-19 guidance from governmental public health, 158 medical, and scientific authorities. For our trust measure, higher values indicate higher 159 trust in selected sources. Prior studies have considered various causal relationships between trust and risk 176 perception [32] . To explore these relationships in the context of pandemic response, we 177 test six different competing hypotheses using SEM. In particular, we consider the causal 178 relationship between disease burden and behavioral intention as well as three causal 179 configurations between trust and risk perceptions. For three of the models, we assume 180 that behavioral intention affects disease burden (models 1A, 2A, and 3A); the 181 remaining three models assume the converse, i.e., that disease burden affected 182 behavioral intention (models 1B, 2B, and 3B). Regardless of the disease-behavior 183 relationship, the models consist of either: trust influencing risk perceptions (models 1A 184 and 1B), risk perceptions influencing trust (models 2A and 2B), or no relationship 185 between risk perception and trust (models 3A and 3B). Hypothesized conceptual models. We tested several hypotheses about the interplay between trust, risk perception, behavioral intention, and COVID-19 disease burden. Each path diagram shows the hypothesized causal relationships between our measured variables. Latent variables are shown in ovals; exogenous variables are shown in rectangles. Structural equation modeling (SEM) was used to assess which model was best supported by our survey data.

Overall, we received 1087 responses, a majority online. 57 people chose to receive a 189 paper copy of the survey, and 44 of those respondents mailed completed surveys back to 190 us. Our overall response rate is 9.98%, excluding the 1110 addresses that were not 191 November 16, 2021 5/16

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint Christian (17%), followed by Catholic (16%), Mainline Christian (15%) and Agnostic 207 (14%). Table S3 .1 has a full breakdown of our demographic variables. Overall, our 208 survey sample is disproportionately white, has higher levels of education and income, 209 and is older than the national and state distributions, which has been observed 210 elsewhere in mail surveys [33] . 211 We tested whether demographic distribution of our responses is dependent on the is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this this version posted November 19, 2021. ; https://doi.org/10.1101/2021.11.17.21266481 doi: medRxiv preprint perceived health risk (p A , p B < 0.001). Elderly individuals perceive significantly higher 288 health risk (p A , p B < 0.001).

The most significant exogenous factor included from our survey data is political 290 ideology. Compared to respondents self-identifying as moderates, self-identified liberals 291 communicate more trust (p A , p B = 0.001) and self-identified conservatives communicate 292 the least trust (p A , p B < 0.001). Those self-identifying as non-political or libertarian 293 also express significantly less trust than self-identified moderates (p A , p B < 0.001 for 294 both). In terms of risk to health from COVID-19, self-identified liberals are significantly 295 more concerned (p A , p B = 0.013), while self-identified libertarians are less concerned indicating that the barrier to public health engagement is stronger in these regions.

Importantly, our research suggests cultivating trust in authorities tasked with 316 communicating public health information would be the optimal way to increase 317 adoption of NPIs to slow the spread of future pandemics.

In the case of COVID-19, trust in the message and the messenger has been 319 undermined by several factors. Namely, there was a lack of uniform national, state and 320 local strategies; inadequate reach, accessibility, and consistency of public health 321 information; and widespread misinformation and disinformation that was not 322 adequately refuted [34, 35] . Studies suggest that misinformation not only erodes trust in 323 public health authorities, but also decreases the motivation to seek and adopt correct 324 information [36] . The influence of social media on information consumption exacerbates 325 the impact of misinformation [27] . News partisanship further impacts trust in public 326 health authorities' message of risk and the reduction of risk through social distancing 327 and other actions [37, 38] .

COVID-19 pandemic response protocols ask individuals, families, schools, and 329 communities to adopt life-altering precautions and behavioral changes. To adopt these 330 practices individuals must perceive the risk of COVID-19 to themselves, their families, 331 and communities. Furthermore, they must trust public health authorities to accurately 332 identify and communicate protective disease intervention protocols [39] . One 333 consequence of the request by authorities for social distancing and mask wearing was 334 increased uncertainty and skepticism [40, 41] . Individuals with more trust in public 335 health authorities are less likely to characterize such requests as a result of incompetence 336 or malfeasance and comply [35] . The result of increased trust leading to increased 337 November 16, 2021 8/16

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 19, 2021. ; https://doi.org/10.1101/2021.11.17.21266481 doi: medRxiv preprint pandemic protective behavior, as measured by decreasing day-to-day activities and 338 increasing mask wearing, is borne out in both of our best supported models (Fig. 2) .

Observed early support for NPIs in the US was notably absent in rural communities 340 and essential workers [42] . Our analyses similarly shows lower levels of institutional 341 trust, lower levels of intention to comply with public health measures, and decreased 342 risk perception in rural areas (Fig. 2) . Nonetheless, our analyses also shows that disease 343 burden was significantly lower among rural persons. This suggests that, at least in the 344 earlier stages of the pandemic, rurality had a protective effect. This was most likely due 345 to reduced population density and time-of-onset of epidemic waves in those areas.

However, not all rural residents were at low exposure risk to SARS-CoV-2. Some rural 347 residents working in meat, poultry, food processing, and agricultural industries face 348 additional COVID-19 risks as these industries involve working and/or traveling in 349 enclosed spaces closer than the recommended 6-foot distance. These industries were 350 deemed essential and were not closed, even in cases of high community transmission. As 351 a result, outbreaks of COVID-19 disproportionately impacted workers and their families 352 in such industries [43] . Decreased levels of trust in rural areas likely worsened the issues 353 stemming from these outbreaks among essential workers.

Our best supported models propose a role for behavioral intention in influencing [44, 45] . Surprisingly in model 1B, we find that increased observed prior 363 disease burden actually leads to reduced prophylactic behavior. While this result is 364 counter intuitive, it is perhaps not without precedent. Recent research [46, 47] shows 365 that perceived disease severity is influenced by various ideological and social factors.

Therefore, one potential explanation for the predictions of model 1B would be a 367 disconnect between perceived and actual disease burden in a county. If individuals are 368 being told by their in-group that disease burden is not severe, then they may continue 369 engaging in behaviors that increase their chances of contracting COVID-19, even in the 370 face of high case counts. These effects may have been worsened by the fact that a 371 majority of COVID-19 cases are mild and deaths are concentrated in the elderly [48, 49] . 372 In the United States, adoption and approvals of public health interventions for 373 COVID-19 fall along political lines. Specifically, other research finds people identifying 374 as democrats favor publicly mandated disease interventions and practice protective 375 health recommendations more than people identifying as republicans [37, [50] [51] [52] .

Political ideology similarly influences every aspect (trust, risk perception, and 377 behavioral intention) of our social cognitive model results. The finding that political 378 ideology affects trust and compliance with NPIs (i.e., behavioral intention in our 379 models) has been reported in several other studies [34, 37, 41, 53] . Furthermore, our 380 findings are consistent with other recent work in which partisan differences were found 381 to be more significant than other factors in determining social distancing behavior, and 382 with results of disparate health outcomes based on party identity [38, 54] . Thus, our 383 work adds to the body of evidence for the consequences of political ideology on 384 behavioral changes in response to the pandemic.

Our model, however, offers a more nuanced view of where partisanship plays a role 386 in affecting various aspects of cognition. In particular, the social construct of trust in 387 public health guidance seems to be affected by all of the political categories we analyzed 388 (i.e., liberal, moderate, conservative, libertarian, non-political). For the cognitive 389 November 16, 2021 9/16 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted November 19, 2021. ;  constructs, only libertarian identity is significant for both health and economic risk 390 perception. In addition, health risk perception is also significantly affected by liberal 391 identity, while economic risk perception is significantly affected by conservative identity. 392 Lastly, self-identified liberals expressed willingness to reduce their day-to-day activities 393 as the risk of SARS-CoV-2 infection increased, while conservative and libertarian 394 identities were significantly associated with reluctance to reduce activity. Therefore, 395 public health strategies appealing to certain cognitive constructs might be better 396 focused toward particular partisan groups. For example, advertising health risks of a 397 disease may impact liberals and libertarians more effectively than other groups. Still, 398 trust has the strongest effect on both types of risk perception, there we suggest 399 maintaining trusting relationships with all groups is the most vital action.

Our findings related to gender are also in-line with other studies that report women 401 as more concerned about the health consequences of COVID-19 [55] [56] [57] . These results 402 are somewhat surprising given that men are more likely to contract severe COVID-19 403 cases resulting in hospitalization or death [58] . However, our findings that women 404 engage in higher levels of activity that could expose them to SARS-CoV-2 differ from 405 other studies [56] . This might be explained partially by the increased household 406 responsibilities of women resulting in higher activity levels [59] . 64% of the women in 407 our survey indicated that they are married and therefore may feel increased pressure to 408 perform some the day-to-day activities about which we asked. Finally, women also 409 perceived more economic risk to themselves and their community from COVID- 19, 410 which is consistent with women having generally higher risk perception [60] .

Our study has several limitations. First, survey instruments are subject to response 412 bias. Our respondents tend to be older, wealthier, more-educated individuals compared 413 to the population as a whole. This is typical of many survey-based studies [33] . We heterogeneous biases in under-reporting of cases. Fourth, we are limited by the fact that 423 COVID-19 cases are reported at the county level within the US. We may have been able 424 to achieve greater resolution in our study had we been able to associate case counts 425 with census tracts, the geographic level at which the geographic analysis was conducted. 426 Related to this, in determining whether a zip code is rural or urban, we use the RUCA 427 classification system. This system offers a finer level of granularity of which locations Understanding how individuals process and respond to threats in their environment is 437 critical to optimizing public health messaging and policy. Using structural equation our survey results best support the hypothesis that building trust in government 440 organizations can be used to influence behavioral intentions indirectly via risk 441 perception. Higher risk perception leads to reduced behavioral intention, and model 1A 442 predicts reduced behavioral intentions leads to reduced disease burden. We therefore 443 propose decision makers focus efforts on trust building to increase NPI effectiveness in 444 future pandemics. Our work is novel in its attempt to reach and understand individuals 445 living in rural areas. Rural populations indicate less trust and reduced risk perception 446 compared to urban populations, making them vulnerable to higher disease burden and a 447 possible focus area for public health. Lack of trust in rural communities combined with 448 increased risked to essential workers could have negative synergy; this issue is beyond 449 the scope of this work but merits future study. In agreement with other COVID-19 450 studies, political ideology seems to be an overwhelming factor influencing the 

Cryptic transmission of SARS-CoV-2 and the first COVID-19 wave

Timing of Community Mitigation and Changes in Reported COVID-19 and Community Mobility -Four

Timing of State and Territorial COVID-19 Stay-at-Home Orders and Changes in Population Movement -United States

Infection prevention behaviour and infectious disease modelling: a review of the literature and recommendations for the future

Sociodemographic Predictors of Health Risk Perception, Attitude and Behavior Practices Associated with Health-Emergency Disaster Risk Management for Biological Hazards: The Case of COVID-19 Pandemic in Hong Kong, SAR China

Use of Rapid Online Surveys to Assess People's Perceptions During Infectious Disease Outbreaks: A Cross-sectional Survey on COVID-19

Understanding face mask use to prevent coronavirus and other illnesses: Development of a multidimensional face mask perceptions scale

The Role of Risk Perceptions and Affective Consequences

Protective Behaviors

Modeling compliance with COVID-19 prevention guidelines: the critical role of trust in science

Social Influence: Compliance and Conformity

Motivations, barriers, and communication recommendations for promoting face coverings during the COVID-19 pandemic: Survey findings from a diverse sample

Collective Action as the Material Expression of Opinion-Based Group Membership

The Origins and Consequences of Affective Polarization in the United States

Partisans Without Constraint: Political Polarization and Trends in American Public Opinion

Ideologues without Issues: The Polarizing Consequences of Ideological Identities

Using social and behavioural science to support COVID-19 pandemic response

Quality of Life in Rural and Urban Adults 65 Years and Older: Findings From the National Health and Nutrition Examination Survey. The Journal of Rural Health

Rural areas at risk during COVID-19 pandemic. The Lancet Infectious Diseases

Potentially Excess Deaths from the Five Leading Causes of Death in Metropolitan and Nonmetropolitan Counties -United States

Prevalence of Disability and Disability Types by Urban-Rural County Classification

Community Susceptibility and Resiliency to COVID-19 Across the Rural-Urban Continuum in the United States

Survey Methods for Hard-to-Reach Populations: Introduction to the Special Issue

Internet, Phone, Mail and Mixed-Mode Surveys: The Tailored Design Method

Human agency in social cognitive theory

US) National Cancer Institute. Theory at a glance : a guide for health promotion practice

of Health and Human Services, National Institutes of Health, National Cancer Institute

The Relation between Media Consumption and Misinformation at the Outset of the SARS-CoV-2 Pandemic in the US

Rural-urban commuting area codes

Metropolitan Urban and Rural Commuting Areas: Toward a Better Depiction of the United States Settlement System

Covid-19) Data in the United States

lavaan: An R Package for Structural Equation Modeling

Trust and Risk Perception: A Critical Review of the Literature

Whose experience is measured?: A pilot study of patient satisfaction demographics in pediatric otolaryngology

Stay-at-home orders, social distancing, and trust

Profile of a Conspiracy Theorist: The Role of Government Trust and Technology on Misinformation during an Epidemic

Effects of COVID-19 Misinformation on Information Seeking, Avoidance, and Processing: A Multicountry Comparative Study

Polarization and public health: Partisan differences in social distancing during the coronavirus pandemic

Partisan differences in physical distancing are linked to health outcomes during the COVID-19 pandemic

Public health and public trust: Survey evidence from the Ebola Virus Disease epidemic in Liberia

COVID-19 pandemic in the United States. Health Policy and Technology

Unmasking Partisanship: How Polarization Influences Public Responses to Collective Risk

Public Attitudes, Behaviors, and Beliefs Related to COVID-19, Stay-at-Home Orders, Nonessential Business Closures, and Public Health Guidance -United States

COVID-19 Among Workers in Meat and Poultry Processing Facilities 19

The effect of public health measures on the 1918 influenza pandemic in U.S. cities

Nonpharmaceutical Interventions Implemented by US Cities During the 1918-1919 Influenza Pandemic

Worldviews, trust, and risk perceptions shape public acceptance of COVID-19 public health measures

Beyond political affiliation: an examination of the relationships between social factors and perceptions of and responses to COVID-19

Age-specific mortality and immunity patterns of SARS-CoV-2

Burden and characteristics of COVID-19 in the United States during 2020

Elusive consensus: Polarization in elite communication on the COVID-19 pandemic

Toxic Mask-ulinity: The Link between Masculine Toughness and Affective Reactions to Mask Wearing in the COVID-19 Era

Democrats move even further apart in coronavirus concerns

Staying at home: Mobility effects of COVID-19

Partisanship, health behavior, and policy attitudes in the early stages of the COVID-19 pandemic

Gender differences in COVID-19 attitudes and behavior: Panel evidence from eight countries

Gender Differences in Fear and Risk Perception During the COVID-19 Pandemic

Gender Differences in the COVID-19 Pandemic Risk Perception, Psychology, and Behaviors of Spanish University Students

Male gender is a predictor of higher mortality in hospitalized adults with COVID-19

A multicountry perspective on gender differences in time use during COVID-19

Gender, race, and perceived risk: The 'white male' effect

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint