Religious Participation and Women’s Health: A Reply to Homan and Burdette 

The Washington Post recently ran a story with the striking headline, “Barring women as leaders in church may be bad for their health, new study finds.” The study in question, “When Religion Hurts: Structural Sexism and Health in Religious Denominations,” was authored by sociologists Patricia Homan and Amy Burdette, and appears in the current issue of the American Sociological Review. They examine the effect of “structural sexism” in religious communities on women’s health in particular, and conclude that, while “women who attend inclusive congregations have better self-rated health than non-attenders and women who attend sexist congregations,” “there is no statistically significant difference in self-rated health between non-attenders and women who attend sexist congregations.”¹ While the evidence Homan and Burdette marshal does indeed indicate an association between women’s representation in church leadership and their self-rated health, several issues in the paper’s analysis and interpretation suggest that their conclusions ought to be qualified.

Homan and Burdette contrast two bodies of research, one documenting the associations of religious participation with positive health and well-being outcomes, and the other documenting the associations of “structural sexism” in various social settings with negative health and well-being outcomes, particularly for women. Their paper is an attempt to coordinate these two literatures, by assessing how structural sexism might modify the association of religious participation with health.² They pursued this investigation using data on self-rated health³ and opportunities for female leadership in religious congregations, drawn from the General Social Survey and the National Congregations Survey.

“Structural Sexism” or Gender-Leadership Disparities? 

The key organizing concept in this paper is of course “structural sexism,” which Homan and Burdette define as “the degree of systematic gender inequality in power and resources characterizing a given gender structure.”⁴ They assess a congregation’s structural sexism in three complementary ways: 1) barring women from serving on the governing board; 2) barring women from serving as the congregation’s “head clergyperson or primary religious leader”; and 3) a composite “sexism scale,” which aggregates the prior two conditions with bans on women’s preaching or teaching in co-ed religious education settings.⁵ The “sexism” assessment thus pertains to what are often doctrinal restrictions on the leadership structure, regardless of the actual treatment of, respect for, or other roles of women in those congregations. Understandably, many within congregations of this sort might object to being called sexist solely for those doctrinal commitments.

While the authors do reflect briefly on the possible influences of their notion of congregational structural sexism on disparities in domestic life,⁶ each of their measures of structural sexism in fact provides a different perspective on what one might call the “gender-leadership disparity” (GLD) in a given congregation, i.e., the gap between male and female representation in the community’s leadership. For the purposes of their paper, congregations are “structurally sexist” if and only if they have a high GLD, and so, outside of quotations, we’ll use this more neutral terminology in the remainder of the present discussion. 

As noted above, Homan and Burdette concluded that women who attended “inclusive congregations,” in which leadership roles are open to them, had better self-rated health than either non-attenders or attenders of high-GLD congregations, while the outcomes of attenders of high-GLD congregations were not notably different from those of non-attenders. As the Post reported about the study, “Women in inclusive churches had an average self-reported health score of 3.03. Women in sexist congregations had an average score of 2.79.”

In our view, the authors overinterpret their results in light of the totality of the prior evidence. There is good reason to think that women and men alike would derive substantial health and well-being benefits even from attending services at some congregations that would be classed as “structurally sexist” on one or more of Homan and Burdette’s measures. 

Clergy vs. Governing Boards

The gap between high-GLD and “inclusive” congregations varied depending on the measure used. It was highest for congregations with all-male governing boards, which represented only 14% of congregations in the study. The gap was somewhat narrower for congregations scoring 3 or 4 on the aggregate sexism scale, which represented 41% of the congregations studied, and narrower still (an approximately .1 mean difference on a scale of 1-4) for the 59% of congregations that restricted the priesthood or pastorate to men.⁷

Issues of Analysis and Interpretation

Additionally, several analytical decisions and issues of interpretative framing likely affect the investigators’ effect-size estimates. First, somewhat prominently, Homan and Burdette do not control for geographic region, but rather only a rural vs. urban indicator. Geographic region may well be operating here as a confounder, since, for instance, both high-GLD congregations and relatively lower self-rated health may both cluster in the South. This may not explain away the entirety of the effect of high GLD on health, but it might well somewhat mute it, and ought to have been controlled for. 

The authors also report relatively modest positive effects for frequency of religious service attendance in their study. However, they code service attendance continuously (on a scale of 1-9), whereas the most substantial effects are typically seen when comparing weekly or more attendance with less. The way they handle this variable in the analysis will therefore make effect sizes for service attendance look smaller than they really are. However, even with this coding, contrary to what seems to be suggested in their paper, their own analyses in the paper’s Table 2 provides evidence that frequency of service attendance is associated with better self-rated health, both overall and also for men.⁸

Religion and Public Health in Better-Designed Studies

Finally, there is the matter of the relation of Homan and Burdette’s findings to the broader literature on religion and health, including a number of papers produced by our team at the Human Flourishing Program at Harvard University. Their paper uses cross-sectional data (n=2,234), which is a relatively weak study design, on a single subjective outcome (self-rated health). More rigorous studies using longitudinal data over time, of much larger cohorts and with objective health outcome measures, have found substantial average effects of service attendance on various outcomes, such as diagnosed depression, substance abuse, suicide, and overall mortality. 

In fact, well-designed studies conducted by scholars from the Human Flourishing Program along with other collaborators have found:

• 25% reductions in depression among women who attend services weekly compared with non-attenders;
•  5- to 6-fold reductions in suicide;
• a 33% lower hazard of death from despair among men and a 68% lower hazard among women compared with non-attenders;
• and a 26% reduction in all-cause mortality (another study found a 33% reduction in 16-year all-cause mortality for more than weekly attendance in a cohort of 74,534 women). 

All these effects appear to be at least as strong for women as for men. These analyses also provide evidence for comparable effect sizes in Catholic communities as well, in which ordination and preaching are restricted to men. Indeed, with lower rates of suicide, the effects seem to be even larger in Catholic communities than in Protestant communities. 

In our view, then, the authors overinterpret their results in light of the totality of the prior evidence. There is good reason to think that women and men alike would derive substantial health and well-being benefits even from attending services at some congregations that would be classed as “structurally sexist” on one or more of Homan and Burdette’s measures. 

Important Lessons from Homan and Burdette

Religious communities can, however, still draw important lessons from their analysis. While religious leaders and believers will rightly insist that their faith’s primary aim is not boosting self-rated health, but rather attaining some form of communion with God, that does not render Homan and Burdette’s findings irrelevant. The major world religions all ultimately aim at the full flourishing of each person. We are to love God, yes, but also our neighbor, an understanding shared by Christians, Jews, and Muslims alike. Evidence that some members of their community—in this case women in some congregations—are not flourishing should be deeply concerning to every person of faith. 

Homan and Burdette do provide some modest evidence that restricting opportunities for women to exercise leadership within a congregation may diminish the positive health effects of service attendance for them, and perhaps especially so in communities with the highest gender-leadership disparities. Such leadership structures may, but need not necessarily, lead to contexts that are problematic for women, and these issues should be taken seriously. Every effort ought to be made for women to thrive. Even without revisiting age-old and doctrinally-freighted conditions for ordination or denying that there are typical differences of personality or preferences between men and women, these communities might receive this research as an invitation to reflect anew on how they can provide the fullest possible scope for their female members to exercise their God-given gifts for teaching or administration in the service of God and neighbor. 

Brendan W. Case, Th.D. is the Associate Director for Research of the Human Flourishing Program at Harvard University and the author The Accountable Animal (T&T Clark, 2021). Tyler VanderWeele, Ph.D., is John L. Loeb and Frances Lehman Loeb Professor of Epidemiology at Harvard’s T.H. Chan School of Public Health, and Director of the Human Flourishing Program at Harvard University. 

1. Patricia Homan & Amy Burdette, “When Religion Hurts: Structural Sexism and Health in Religious Denominations,” The American Sociological Review 2021, vol. 86 (2): 234-55, here 248. 

2.Ibid., 239. 

3. Self-rated health is the respondent’s subjective assessment of his or her health (ranging from “poor” to “excellent”) at the time of the survey (Ibid., 241). 

4. Ibid., 235. 

5. A congregation which bars women from three or all of these roles (scoring a 3 or 4 on the 1-4 scale) is treated as sexist on this third measure (ibid., 242). 

6. Ibid., 239. 

7.Ibid., 249.

8. The frequency of attendance variable in the bottom row of Table 2 indicates effect sizes of 0.058-0.059 for men, and 0.037-0.038 overall, both of which pass p-value thresholds of <0.001. The effect size in this table for frequency of service attendance when restricted to women is 0.025-0.026, but the confidence intervals for this estimate extend from about -0.002 to 0.053.