Gender, Benevolent Sexism, and Public Health Compliance

Abstract Growing research shows a correlation between gender, benevolent sexism, partisanship, and COVID-19 public health compliance. We show first that women are more likely than men to engage in protective behaviors to slow the spread of COVID-19. We also find that while Republicans and Independents are less likely to comply with these measures, benevolent sexism moderates the effect of partisanship and can increase compliance. These results suggest that framing public health directives in terms of chivalry and protection activates benevolent sexism, potentially offsetting patterns of noncompliance associated with partisanship. We discuss the negative consequences of these results and posit a need for bipartisan messages to reduce reliance on benevolent sexism in the future.

hygiene and public health behaviors (Anderson et al. 2008), in addition to being more likely to identify as Democratic (Barnes and Cassese 2017).
We also highlight the importance of sexist beliefs, particularly the concept of benevolent sexism. Glick and Fiske (1996) note that sexist beliefs are more about ambivalence than aversion. That is, while we often think of prejudice in terms of dislike of a group, sexism is defined more by ambivalence toward women. This ambivalence (broadly termed "ambivalent sexism") is defined by two constructs. First, hostile sexism is a set of beliefs generally thought to justify a hierarchy placing men above women and lacking in positive affect toward women (Glick and Fiske 2001). Second, benevolent sexism, in contrast with hostile sexism, is defined by paternalistic or chivalrous positive affect toward women combined with support for prosocial behaviors. For our purposes, benevolent sexism is particularly interesting, as it couples positive affect toward women and support for stereotypical views of women with a tendency toward prosocial behaviors. It is this propensity for socially helpful behaviors that makes benevolent sexism particularly relevant to the COVID-19 pandemic.
Prior evidence suggests that sexism, especially benevolent sexism, supports engagement in preventive health measures, including mask wearing (Hesse 2020). For example, benevolent sexism predicts support for proscriptive rules regarding pregnancy (Murphy et al. 2011), restricting the choices of pregnant women (Sutton, Douglas, and McClelland 2011), and restrictive abortion attitudes (Duerksen and Lawson 2017). Yet sexism exhibits a complicated partisan pattern. While ambivalent sexism generally predicts support for Republican candidates (Frasure-Yokley 2018;Valentino, Wayne, and Oceno 2018), given the right framing, benevolent sexism can increase support for candidates of all parties (Cassese and Holman 2019).
Given the countervailing partisan and prosocial narratives around COVID-19, we believe an examination of partisanship must also consider gender and benevolent sexism. First, in line with Allcott et al. (2020), we expect women to be more likely to comply with protective measures than men (H 1 ). Second, we argue that benevolent sexism may help offset partisan differences in individual protective behaviors. That is, while Republicans (and, to a lesser extent, Independents) are less committed to social distancing practices, higher levels of benevolent sexism may increase compliance with these prosocial protective measures (H 2 ).

DATA AND METHODS
We test these expectations using a quota-sampled survey of Americans conducted through Lucid Theorem (LT) from May 25 to 26, 2020. LT matches samples to census demographics to approximate national representativeness. The demographics of our sample (N = 1,000) are presented in the Appendix A in the supplementary materials online.
Our dependent variables consist of four Likert-style questions asking individuals whether they would get a vaccine if one becomes available, as well as their tendency to wear a face mask, practice physical distancing, and shop online instead of in store. We also use an index of eight items asking whether individuals have engaged in the following activities: regularly washed their hands, avoided dining in at restaurants and bars, sanitized their home or workspace, engaged in no-touch greetings, changed travel plans, worked from home, canceled social engagements, and used curbside pickup or delivery (alpha = .70). Each of the five dependent variables is analyzed using ordinary least squares regression with demographic and political controls. 1 Importantly, we include measures of both benevolent and hostile sexism, using four items from the Ambivalent Sexism Inventory (Glick and Fiske 1996), and interact benevolent sexism with partisanship. Table 1, we find consistent support for H 1 that women are more likely than men to comply with COVID-19 protective measures. 2 The first four dependent variable are scaled to run from 0 to 1, suggesting that women are 6 points more likely to wear a face mask and shop online and 4 points more likely to maintain physical distancing. However, there are no gender differences on the likelihood of getting the vaccine. The social distancing index represents total actions taken, so the coefficients are directly interpretable as the number of actions. Women are likely to engage in approximately 0.34 more protective actions than 1. Full question wording is available in Appendix B. Beyond demographic controls, we include hostile sexism, which represents the second component of ambivalent sexism, and authoritarianism. We include authoritarianism because it captures deference to authority figures and correlates with Republican identification, like ambivalent sexism. Thus, we use it to control for respect for health authorities and to isolate the effects of benevolent sexism on prosocial behaviors.

As shown in
2. As proof of concept, we include main effects models in Appendix C, which shows that women are generally more likely to practice protective measures and that higher levels of benevolent sexism increase likelihood of compliance. men on average. These results confirm that women are more likely than men to engage in behaviors designed to slow the spread of COVID-19 (with the exception of vaccination likelihood).
Regarding H 2 , benevolent sexism appears to have little effect on Democrats except for vaccination likelihood (b = -0.16, p < .05). Conversely, we see significant effects for Republicans and Independents. Focusing on Republicans, we see significant interactions for vaccination intention and the social distancing scale (b = 0.31 and b = 1.15, p < .05), as well as a marginally significant interaction for wearing face masks (b = 0.15, p < .10). These interactions must be viewed in light of significant negative main effect coefficients for Republicans and Independents, which indicate that they are less likely, all else being equal, to comply with social distancing than Democrats. Figure 1 presents predicted values for non-Democratic identifiers at 1 standard deviation above and below the mean level of benevolent sexism among their partisan group. These results demonstrate that benevolent sexism may mitigate tendencies among non-Democrats to ignore COVID-19 public health practices. A shift of 2 standard deviations in benevolent sexism increases Republican vaccination intent by 7.5 points and face mask wearing by more than 9 points. It also increases the number of Republican social distancing practices by almost 0.75.
While less consistent, for Independents, there is a marginally significant interaction for wearing face masks (b = -0.29, p < .10) and a significant interaction for social distancing (b = 0.26, p < .05). For Independents, moving from 1 standard deviation below the mean to 1 standard deviation above on benevolent sexism increases likelihood of wearing a face mask by more than 14 percentage points and increases physical distancing likelihood by nearly 13 points. Thus, Republicans (and, to a lesser extent, Independents) who are high in benevolent sexism consistently engaged in more prosocial COVID-19 protective behaviors than less benevolently sexist copartisans.
The patterns that emerge demonstrate the importance of the prosocial aspect of benevolent sexism. The actions that are most likely to prevent the social spread of the coronavirus (vaccination, face masks, social and physical distancing) are influenced by an individual's level of benevolent sexism (along with their partisanship). In contrast, shopping online, which focuses on convenience and has an aspect of individualistic selfpreservation in addition to social protection, is much less influenced by benevolent sexism. As we continue to investigate the role of benevolent sexism, we should expect to see its influence emerge more often for socially protective measures than for self-protective measures.

DISCUSSION AND CONCLUSION
These results point to the importance of gender and stereotypes in the realm of public health. Our gender results align with prior research suggesting that women are more inclined to engage in public health measures than men. Yet our results on benevolent sexism reveal a more complicated picture.
What we see, unfortunately, is that discussing COVID-19 protective behaviors in prosocial terms appears to activate benevolent sexism, especially among Republicans and Independents. Indeed, we are already seeing mainstream publications discussing framing behavior in benevolent sexist terms to increase compliance as well as elites using social media tactics such as the hashtag #RealMenWearMasks (Hesse 2020;Tschorn 2020). While frames around protection and chivalry may FIGURE 1. Effect of benevolent sexism on COVID-19 public health compliance, among self-identified Republicans and Independents. lead to greater public health compliance, they come at the cost of reinforcing and encouraging benevolent sexism. Were benevolent sexism purely about protection, we might be less concerned about this. But benevolent sexism also endorses traditional, hierarchical gender roles. Thus, a strategy based around benevolent sexist appeals must grapple with the very real consequence of reifying sexism. Thus, a public health strategy based on benevolent sexist appeals may legitimize these beliefs and grant greater social acceptance to ideas about the "proper" roles for men and women in society.
More work is needed to understand the complex relationship between benevolent sexism and engagement with protective health measures, as well as the relationship between political messaging, sexism, and compliance. Regardless, we should not consider benevolent sexist frames a viable long-term strategy. A unified, bipartisan message of compliance would alleviate the need for harmful sexist framing. As we see, Democrats (whose elites are largely unified in support of these measures) are relatively unphased by benevolent sexism. We would expect the same for Republicans and Independents were there consistent political messages from elites, but that has largely not been the case. While prosocial, protective frames may be necessary in our current partisan public health environment, we should push for a unified political front for public health in the future. With a bipartisan effort, we can reduce the need to frame public health in unnecessary and harmful sexist terms.

SUPPLEMENTARY MATERIAL
To view supplementary material for this article, please visit https://doi.org/ 10.1017/S1743923X20000495 Philip Chen is an Assistant Professor in the Department of Political Science at Beloit College: chenp@beloit.edu; Christina Farhart is an Assistant Professor of Political Science at Carleton College: cfarhart@carleton.edu.