______
The Gendered Politics of Shutting Down
CAMMIE JO BOLIN AND KELLY ROLFES-HAASE
As political leaders began to craft policy responses to COVID-19, media outlets quickly identified a global trend in national pandemic responses: many of those countries that had taken early, decisive actions to reduce transmission of the virus, track cases, and build public trust in institutions were led by women (see, e.g., Wittenberg-Cox 2020; Zalis 2020). These actions have included shutdowns of nonessential businesses and other gathering spaces that can facilitate the spread of COVID-19, such as in-person religious services. In the United States, religious leaders in many communities have had the responsibility of deciding how their own congregations would conduct services during the coronavirus pandemic. Given the apparent difference in response between men and women world leaders, this project considers whether women religious leaders have also been more likely than religious leaders who are men to take actions motivated by the public health interests of their constituencies.
In this chapter, we explore the relationship between women’s leadership and public health concerns in America’s places of worship. Looking to political science literature on women’s political leadership, policy priorities, and role model effects, we examine the extent to which these theories of political representation are applicable in a religious context. In particular, we analyze the relationship between the presence of women’s leadership within a congregation and the cancellation of in-person worship due to COVID-19. We expect that places of worship with women serving in leadership roles will be more likely than places of worship that lack leadership by women to cancel in-person services. Relatedly, we are interested in the relationship between women’s leadership and congregant trust in their clergyperson to have their best health interests at heart. Using survey data from Paul A. Djupe, Ryan P. Burge, and Andrew R. Lewis, fielded March 23–26, 2020, and linear probability models using multivariate ordinary least squares regression, we find support for both of our hypotheses.
Our analyses suggest that congregants who witnessed women leading in their place of worship were about 5 percentage points more likely to have also experienced cancellation of their in-person services because of the coronavirus. Similarly, we find that congregants who reported seeing women in leadership positions within their congregations were about 5 percentage points more likely to trust clergy to have their best health interests at heart in comparison to congregants who had not seen women leading within their congregations. Though we cannot make causal claims, we find a positive and statistically significant relationship between women’s leadership and the prioritization of public health concerns in religious contexts. These findings build on existing research to suggest that gendered trends in women’s political leadership may be generalizable to religious settings.
This chapter proceeds as follows: we first discuss literature on women’s representation and role model effects as well as initial studies on gender and COVID-19 responses. Next, we outline our hypotheses, describe our data and methodology, and present our results. We end with a discussion of how our findings contribute to our understanding of the relationship between gender and leadership in the case of public health during the coronavirus pandemic.
Political Representation: Women’s Issues and Role Model Effects
Are women in positions of political leadership more likely than their peers who are men to make decisions that protect the health of their constituents? The answer is a qualified yes. Women officeholders tend to prioritize issues of health and social welfare, although the stances they take often differ according to partisan pressures. In a political context, women officeholders tend to prioritize issues and policies related to health care, welfare, and women’s health and safety (Frederick 2011, 193). Women officeholders’ emphasis on these so-called women’s issues is seen in their legislative priorities (e.g., Berkman and O’Connor 1993; Kreitzer 2015; Reingold 2000; Reingold et al. 2020; Williamson and Carnes 2013), bill introduction (Thomas and Welch 1991; Vega and Firestone 1995), bill cosponsorship (Swers 2005, 2013), speeches (Osborn and Mendez 2010; Pearson and Dancey 2011), and budgetary allocation (Holman 2014). Although it is important to note that women legislators are not monolithic and their policy priorities are influenced by racial, ethnic, and other identities (Brown 2014; Dittmar, Sanbonmatsu, and Carroll 2018), research consistently suggests that women legislators tend to view themselves as representatives of women—which often means pursuing initiatives that advance the health and well-being of those they represent.
Party and ideology are also important in mitigating and shaping the role that gender plays in legislators’ policy priorities and legislative behaviors (Osborn 2012). On the one hand, women officeholders are often seen as more liberal by voters (Sanbonmatsu and Dolan 2009), and there is evidence that women from both parties may hold more liberal policy preferences on so-called women’s issues (J. Clark 1998; I. Diamond 1977; Poggione 2004; but see also Osborn 2012). Such work suggests that substantive gendered differences are often more pronounced among Republicans, with Republican women being less conservative, on average, than their colleagues who are men (Poggione 2004). On the other hand, partisanship also influences the kinds of policy approaches championed by women legislators (Osborn and Kreitzer 2014). On abortion legislation, for example, women officeholders tend to lead on both sides of the party divide (Reingold et al. 2020). A recent analysis of congressional voting on abortion-related legislation over a twenty-five-year period between 1993 and 2018 finds gender differences between Republican men and women to attenuate over time (Rolfes-Haase and Swers 2021).
Women’s leadership in politics is not only associated with the types of policies officeholders pursue but also with changes in how women in the public perceive and participate in the political arena. Research suggests that the presence of women candidates or officeholders increases women’s interest in politics (Campbell and Wolbrecht 2006), political discussion among young women (Wolbrecht and Campbell 2017), levels of political trust (Ulbig 2007), and feelings of political efficacy (Mansbridge 1999; but see also Broockman 2014). These role model effects may be contingent on such factors as the competitiveness of the election (Atkeson 2003), candidates and constituents sharing a partisan identity (Dolan 2006), or the novelty of women candidates (Wolbrecht and Campbell 2017).
Clergywomen: Women’s Issues and Role Model Effects
There is reason to suspect that women religious leaders may act similarly to women political leaders on issues related to health and social welfare. Scholars have established a precedent for comparing women leaders in religious and political contexts (e.g., Djupe and Olson 2013). Both religious leaders and political leaders, for example, must maintain a degree of responsiveness to those that they lead to maintain their leadership position (see Djupe and Olson 2013; Djupe and Gilbert 2003; and Olson, Crawford, and Deckman 2005 for more information). Clergywomen tend to be more ideologically liberal on a number of policy issues and are more politically active than clergymen (Deckman et al. 2003; Djupe and Gilbert 2008; Djupe and Olson 2013). These ideological divisions are most striking on abortion, but clergywomen also tend to be more liberal than clergymen on an array of social welfare issues (Deckman et al. 2003; Finlay 1996). These studies echo findings in political contexts of differences in opinion and action between women and men political leaders on so-called women’s issues.
When exploring role model effects in a religious context, there is further support for the similarities between women’s leadership in political and religious contexts. Knoll and Bolin (2018) find that women’s leadership in congregations can have an empowering effect (albeit a modest one) for women congregants in terms of their levels of religiosity, spirituality, and efficacy in their congregations. In both political and religious contexts, diverse leadership can make a difference in the lives of those who are represented. Seeing leadership that reflects one’s own identities can affect how someone views an institution and their role within it.
Women’s Leadership: COVID-19 Response
Scholars have begun to research the relationship between women’s leadership and COVID-19 responses in politics (e.g., Aldrich and Lotito 2020; Bauer, Kim, and Kweon 2020; Coscieme et al. 2020; Johnson and Williams 2020; Piscopo 2020) and in “the workplace” (Brooks and Saad 2020). Although media outlets emphasize the effectiveness of women political leaders’ COVID-19 responses (Wittenberg-Cox 2020; Zalis 2020), there is mixed empirical evidence of the relationship between women’s executive political leadership and a country’s pandemic response.
While Coscieme et al. (2020) find evidence of lower COVID-19 mortality rates among countries led by women, Piscopo (2020) finds that factors such as the state’s governing capacity are more likely to drive national pandemic responses. Aldrich and Lotito (2020), similarly, find no relationship between the gender of a nation’s leader and the timing of policy responses to COVID-19, but they do find that countries with higher percentages of women in their legislatures closed schools more quickly than did countries with lower levels of women’s representation.
Other research analyzes differences in how women and men experience the pandemic, finding that women are less likely than men to believe COVID-19 conspiracy theories (Cassese, Farhart, and Miller 2020), more likely to support government action related to the pandemic (Algara, Fuller, and Hare 2020), and are disproportionately burdened with increased domestic responsibilities during the pandemic (Power 2020). Research by Smothers et al. (2020) explores gender and COVID-19 in a religious context, finding differences in responses to the virus between men and women.
While Smothers et al. (2020) uncover differences between women and men congregants’ attitudes and religious behaviors in light of COVID-19, our study explores the relationship between women and men’s leadership within a church and the church’s response to COVID-19 as well as congregants’ trust in their clergypersons to have their best health interests at heart. In studying the relationship between women’s leadership in a religious context and a congregation’s COVID-19 precautions, we offer insight into how religious congregations are responding to the pandemic as well as offering further evidence for the comparability of women’s leadership in political and religious contexts.
Hypotheses
Following scholarship that finds women political leaders to prioritize women’s issues (including health care) during their tenure in office (e.g., Swers 2013, 2005) and scholarship that describes the comparability of political and religious leadership (e.g., Djupe and Gilbert 2003), we expect to find greater concern for public health in congregations with women’s religious leadership.
Hypothesis 1: Women’s Issues
Respondents who have witnessed women exercising leadership in their religious congregations in the past year will be more likely than respondents who have not to report having had their in-person worship canceled because of the coronavirus.
We look to political science literature on role model effects when developing our second hypothesis. In a political context, the presence of women’s leadership is often associated with increases in constituent trust in the political system (e.g., Atkeson and Carrillo 2007). In a religious context, we expect to find a similar relationship between women’s leadership and congregant trust.
Hypothesis 2: Role Model Effects
Respondents who have witnessed women exercising leadership in their religious congregations in the past year will be more likely to report that they “trust clergy to have [their] best health interests at heart.”
Data and Methodology
To evaluate the relationships between leadership by women in religious settings and (i) the likelihood that a place of worship canceled in-person services because of the coronavirus and (ii) that congregants trust clergy to have their best health interests at heart, we rely on survey data fielded March 23–26, 2020 (Djupe, Burge, and Lewis 2020). The survey included a battery of questions related to respondents’ demographic characteristics, partisan and ideological preferences, religious beliefs, and attendance at religious services. To measure our key independent variable of interest, women’s leadership in religious settings, we relied on responses to the following question: “Thinking about any group or organization that you have seen personally in your community in the past year, have you seen women exercising leadership?” and coded the variable 1 for those who responded that they had witnessed women exercising leadership in their religious congregation (i.e., “organizing a small group, activity or serving as clergy”).1 Our two dependent variables are also binary. The first is coded 1 for those who affirmed that “in-person worship has been canceled for now because of the coronavirus.” Our second dependent variable is coded 1 for those who either strongly agreed or agreed that they “trust clergy to have [their] best health interests at heart.” Descriptively, the survey results align with our hypotheses. Those who reported seeing women exercising leadership in their religious congregation were more likely to have had religious services canceled because of the coronavirus (about 5 percentage points more likely) and to trust clergy to have their best health interests at heart (about 15 percentage points more likely). To control for other factors that may be driving the apparent relationship between women’s leadership and our two key dependent variables, we estimate linear probability models using multivariate ordinary least squares regression. This approach allows us to account for a comprehensive set of demographic, political, and religious factors that could be related to both a given respondent’s likelihood of having seen women lead in a religious setting and the likelihood that (i) their regular in-person religious services were canceled or (ii) their levels of trust in clergy.2 In addition to the gender of the respondent, we control for other standard demographic characteristics, including their race, ethnicity, and age. Patterns in both religious affiliation and church attendance by socioeconomic status indicate that family income and educational attainment should be controlled for, as well (Masci 2016; Schwadel et al. 2009).
Research in sociology draws parallels between participation in secular and nonsecular societal institutions, such as workplaces, marriage, and religious organizations, that all integrate individuals into social life (Mueller and Johnson 1975; Schwadel et al. 2009). Therefore, we account for whether a respondent works full-time and their marital status. We also control for region, partisan affiliation, ideological preferences, and the level of attention that respondents pay to politics. Finally, we include their current religious affiliation, attendance, the size of the church they attend, and whether they consider themselves born-again Christians and biblical literalists. We find correlations in our data between respondents’ religious behaviors and seeing women engaging in leadership and think it likely that such religious characteristics are related to our outcomes of interest.3 We use linear probability models because they tend to produce similar estimates as those produced by maximum likelihood estimation with the benefit of being more easily interpretable (Angrist and Pischke 2008).4 We use robust standard errors.
Analyses and Results
The coefficient estimates shown in Figures 11.1 and 11.2 are for our complete models estimated, including all demographic, political, and religious control variables. Omitted categories for binary variables that occur in a set (e.g., region) are noted below each figure.
The estimates presented in Figure 11.1 evaluate the likelihood that a respondent reported in-person religious services were canceled because of the coronavirus. We find evidence that respondents who reported seeing women exercise leadership in their religious congregations were more likely to have their in-person religious services canceled due to the pandemic even after controlling for our robust battery of potentially confounding factors. More specifically, we find respondents who saw women leading in their congregations to be about 4.8 percentage points more likely than those who did not to report in-person services being canceled. Though the effect appears small, the rates of closure were relatively high at the time of the survey (just under 90 percent of those surveyed in our sample reported in-person services being canceled), so it is revealing to estimate a coefficient of this magnitude given the amount of variation left to explain. We also find non-Hispanic Black respondents to be less likely than white respondents to have seen their in-person religious services canceled, those who reported not having a religious affiliation to be less likely than evangelical Christians to have seen in-person services canceled, and church size to be positively related to the cancellation of services.5
Figure 11.1 Coefficient Estimates for whether In-Person Services Were Canceled. Note: Dependent variable is binary and coded 1 if in-person worship services were canceled due to COVID; confidence intervals for estimates are 95 percent using robust standard errors; omitted race/ethnicity is white; omitted region is South; omitted partisan affiliation is Democrat; and omitted religion is evangelical Christian. N = 1,845. (Source: March 2020 survey.)
Figure 11.2 Coefficient Estimates for Trust in Clergy to Have Their Best Health Interests at Heart. Note: Dependent variable is binary and coded 1 if respondent trusts that clergy have their best health interests at heart; confidence intervals for estimates are 95 percent using robust standard errors; omitted race/ethnicity is white; omitted region is South; omitted partisan affiliation is Democrat; and omitted religion is evangelical Christian. N = 2,068. (Source: March 2020 survey.)
The estimates presented in Figure 11.2 evaluate the likelihood that a respondent said that they agreed or strongly agreed with the statement “I trust clergy to have my best health interests at heart.” Seeing women exercise leadership in their religious congregations is again positively linked to trust in their clergy even after controlling for a variety of demographic, political, and religious factors. In line with our second hypothesis, these estimates suggest that respondents who saw women leading in a religious setting are about 4.7 percentage points more likely to trust their clergy to have their best health interests at heart. This estimate is approximately equivalent to the difference in trust levels reported between Republicans and Democrats, the former being about 4.6 percentage points more likely than the latter to trust clergy to have their best health interests at heart (we find independents to be about 6.3 percentage points less likely than Democrats to trust clergy in this regard).
Our religious control variables also tell an interesting story. Holding constant the other characteristics included in our model, respondents who attend church more regularly, consider themselves to be born-again Christians, and are biblical literalists are more likely to trust clergy to have their best health interests at heart. We find that attending church several times per month or more is associated with an increase in the likelihood that respondents trust clergy to have their best health interests at heart of about 11 percentage points, over twice the magnitude of the relationship we estimate for seeing women lead. The difference between being a born-again Christian and not is also similar to the difference between having seen women lead and not on trusting clergy. Our analyses also suggest that a one-standard-deviation increase on the literalism measure is associated with about an 8-percentage-point increase in the likelihood that a respondent expressed trusting clergy to have their best health interests at heart. Of course, these relationships could be operating in the opposite direction, with individuals who are more likely to trust clergy to protect their health and safety also being more likely to attend church more regularly, consider themselves to be born-again Christians, and take the Bible more literally.
Conclusion
We find evidence of a relationship between women’s leadership in a religious congregation and a congregation’s COVID-19 response. Respondents who reported seeing women’s leadership in their congregation were about 5 percentage points more likely to report that their in-person services had been canceled because of the coronavirus. Although modest in magnitude, it is one of the few covariates to remain statistically significant at conventional levels following the inclusion of multiple demographic, political, and religious controls in our first model. Moreover, this finding supports our women’s-issues hypothesis that the presence of women’s religious leadership in a congregation is associated with an increased likelihood that a congregation temporarily cancels religious services out of concern for public health.
Our analyses also reveal support for our second hypothesis (i.e., our role-model hypothesis). As expected, we find that respondents who reported seeing women’s leadership in their congregation were more likely to “trust in clergy to have [their] best health interests at heart.” Congregants who have seen women lead in their congregation in the past year were, on average, about 5 percentage points more likely than those who did not to trust their clergy regarding their health interests. This difference is robust to a comprehensive set of controls and approximately equivalent to the difference in trust levels reported between Republican and Democratic respondents.
These findings also support the overarching hypothesis of this study, that women’s leadership in religious contexts can be compared to women’s leadership in political contexts and that drawing on political science representation literature can inform research on religious leadership. In both religious and political contexts, seeing women in leadership positions can be associated with how individuals view a given institution and their relationship to it. The positive relationship we find between the presence of women’s leadership and cancellation of in-person religious services because of the coronavirus echoes the findings of studies in political contexts that suggest women’s political leadership is associated with the increased prioritization of social welfare issues, efforts to protect the health and well-being of seniors, generosity of family and medical leave policies, and public health expenditures (Courtemanche and Green 2017; Giles-Sims, Green, and Lockhart 2012; Holman 2014; Williamson and Carnes 2013). Likewise, our finding that congregants who see women’s leadership in a religious congregation are increasingly likely to trust their clergy to have their health interests at heart mirrors findings in a political context where constituents who see women’s leadership are increasingly likely to trust their government (Atkeson and Carillo 2007; Ulbig 2007; but see Lawless 2004). Although our results cannot identify causal stories about the impact of women’s religious leadership on the cancellation of in-person services and trust in clergy, our analyses support the possibility that women’s leadership may matter in relation to the prioritization of public health concerns as well as congregant trust. Future research should continue to explore the relationship between women’s religious leadership and congregational responses to other issues related to public health, welfare, education, and women’s health to determine whether women religious leaders behave similarly to women political leaders.
Future analysis of religious responses to COVID-19 should explore the relationship between women holding specific leadership positions within a congregation and a congregation’s COVID-19 response. While our study analyzed women holding any religious leadership role within a congregation (e.g., “organizing a small group, activity or serving as clergy”), it would be interesting to see if women’s presence in specific leadership roles is associated with varied COVID-19 responses. For example, within the same denomination or religious tradition, do we find that clergywomen differ from clergymen in their COVID-19 responses? While women remain vastly underrepresented among clergy in America’s places of worship (Knoll and Bolin 2018), anecdotal as well as empirical evidence identifies the importance of women serving in all levels of leadership in America’s places of worship. Previous research finds that women congregants who see women occupying a majority of lay leadership positions within their congregation report similar levels of trust and commitment to their congregation as women congregants who see a woman serving as their head pastor or priest (Knoll and Bolin 2018). This study, similarly, points to the importance of women’s religious leadership—in all forms—within a congregation. Religious responses to COVID-19 provide an additional example of the difference women’s leadership can make in the lives of their congregants.
____________
Material referencing an appendix in this chapter can be found online available here: https://dataverse.harvard.edu/dataverse/epidemic_among_my_people.