Study finds feminine-coded occupations are valued less
09/08/2026
Decades of labor market data demonstrate that occupations predominantly held by women tend to pay less than those held predominantly by men. What’s been harder to prove is whether it is those occupations’ association with women that is causing their lower pay, or whether something else explains the pattern. A new study by IDS affiliate Chloe Hart of UW-Madison’s Department of Sociology offers some of the clearest causal evidence yet. The study shows that when an occupation is coded as feminine, people assign it a lower salary, even when nothing about the actual work has changed.
Hart ran three preregistered online survey experiments between 2023 and 2024, drawing on responses from more than 2,500 U.S. participants recruited through the research platform Prolific. In each study, participants read a description of an occupation and were then asked what salary they thought would be appropriate for someone in that role. Each participant saw a description that was modified in two ways: first, the occupation was described as being held mostly by women or mostly by men, and second, workers in the occupation were described using either “communal” traits associated with women (such as being patient and good with people) or “agentic” traits associated with men (such as being demanding and forceful). Participants were randomly assigned to see just one version of the description of the occupation, and regardless of the gender ratio or gendered traits they saw, they always read that the occupation required the same traits. Therefore, any differences in the salaries participants proposed between conditions could be attributed to these two experimentally manipulated aspects of the occupation, rather than to their own assumptions about the occupation.
Hart tested this setup across three different occupations to see whether the pattern would hold up. The first was a fictional occupation she invented for the study, a “remote work orchestrator,” chosen specifically because participants wouldn’t already have preconceptions about it. The other two were real occupations that carry different cultural associations: quality assurance engineer, a role embedded in the masculine-typed field of engineering, and, and health services manager, a role embedded in the feminine-typed field of healthcare. (Although Hart told participants that each occupation was held mostly by women or mostly by men, in reality, each occupation has a roughly equal split of women and men).
Occupations described as majority-women – averaging together those described as requiring feminine-typed communal traits and masculine-typed agentic traits – were judged worthy of $2,100 less in salary than the same occupation described as majority-men. Occupations described with communal traits – averaging together those described as majority-women and those described as majority-men – were also judged worthy of a lower salary than the same occupation described with agentic traits, by $2,600. But when Hart contrasted the proposed salary for an occupation where both signals of femininity were present at once (i.e., an occupation described as both majority-women and communal) against an occupation where both signals of masculinity were present (i.e., an occupation described as both majority-men and agentic) she found a large and consistent effect. Averaging data from across all three occupations, a doubly feminine-typed occupation was assigned a salary $4,700 lower than the identical “doubly masculine-typed” version. This effect remained consistent whether the occupation was fictional or real, and regardless of the gender identity of the participants. The magnitude of the effect was also strikingly consistent across occupations: in each case, the doubly feminine occupation was attributed a salary that was 6 to 7 percent lower than doubly masculine version of the occupation.
This finding is noteworthy, because it moves beyond correlation. Sociologists have long observed that occupations tend to lose ground in pay as they become more female-dominated, but it’s been difficult to rule out competing explanations.Perhaps those fields simply reward different skills, or maybe something about the timing of feminization coincides with other economic shifts. By experimentally manipulating gender composition and job traits while holding everything else about the occupation description constant, Hart’s study isolates the effect of an occupation’s association with women from these alternative explanations, providing direct evidence that cultural bias shapes how labor is valued.
The study also points to what Hart describes as a bind for women navigating the labor market. Prior research has shown that women tend to face less individual bias in hiring and in how their work is judged when they work in fields dominated by women. Put differently, gender bias against women is typically strongest in male-typed domains and weakest or reversed in female-typed ones. However, Hart’s findings suggest those same fields carry a collective penalty that operates independently: the feminine-typed domains where women can most easily avoid experiencing bias as individuals are also the domains most likely to be undervalued as a whole. As Hart puts it, women may simply trade the risk of bias faced as individuals for bias faced as members of a devalued occupational category.
Hart’s findings add to a growing body of research on why care work occupations are persistently paid less than their skill and difficulty would seem to justify. The study suggests that closing gender pay gaps may require more than just moving more women into male-dominated fields; it may also require confronting the cultural devaluation that follows women into the fields they already occupy.