Scholars Valerie Bostwick (Kansas State University) and Bruce Weinberg (The Ohio State University) recently published a study measuring the effects of year-to-year variations in Ph.D. student cohort gender composition on persistence and degree completion. Their research utilizes AARC faculty rosters across Ohio-based STEM Ph.D. programs at public universities. The paper was published in the Journal of Labor Economics (10.1086/714921); supplementary material in the form of a data file is also available (

The Study in Context

While qualitative reports have highlighted the effects of a ‘female-unfriendly climate’ on students (Settles et al., 2016), this study quantified the effect of cohort gender composition within these programs. By estimating the effect of gender composition on persistence and on-time graduation rates, Bostwick and Weinberg sought to expand on the understanding of the gender seniority gap (where women struggle to progress at each step from junior to senior roles due to successive structural barriers). Also measured were the relationships between cohort gender composition and the likelihood of receiving research funding.

Ph.D. students were chosen due to their relatively high dropout rates (30% within six years of enrolment) and because they are likely to be highly committed to a career in STEM, differentiating them from undergraduates. In undertaking longitudinal observation at each institution and comparing these differences internally, this enabled the researchers to better control for inter-institutional variables (i.e., year-to-year fluctuations in cohort gender-composition at the same-course at the same institution are likely to be dictated to some extent by external factors). The average cohort consisted of 38% women.

Summary of Results

Women were 11.7% less likely to graduate on-time in comparison to their male peers in cohorts with no female peers, but a 1.0 standard deviation increase in the proportion of female students (a mean of around 21% in this study) increased the likelihood of on-time graduation for women relative to men by 4.4%. This effect is felt predominantly within the first year of a Ph.D., which is primarily during the coursework section of a doctorate program.

In typically male programs such as Math, Engineering, and Physics this effect is even greater: in cohorts no female peers, women were 18% less-likely than men to graduate on-time. Also of note is that male Ph.D. students were more likely to be funded in each year of the first four years of study. The effect of more female students on graduation time for men is negative but statistically insignificant.

Explaining Differences in Gender Persistence

Bostwick and Weinberg put forward several potential explanations for these differences which they go on to test. These are: ‘intangible effects on the climate surrounding the students’, improvements in the likelihood of women receiving research grants in cohorts which are more female, improved performance in first-year classes with a greater proportion of women, faculty gender composition and greater mentoring opportunities from the cohort above.

When controlling for these variables, the effect of gender-composition on GPA was found to be small (around a quarter of the total difference as an upper-bound) and there was no effect of gender composition on research-funding or mentoring. Faculty gender composition was also not found to have any effect on persistence.

Their results are most consistent with a climate-based mechanism, where a greater-number of women generates a more welcoming environment for women, which in turn encourages persistence among female peers.


Where there is an absence of university-specific studies on peer cohort gender-composition, this new study contributes to a growing body of literature across different institutional settings, which find that increasing the number of women increases the retention and success of other women (e.g., Huntington-Klein & Rose, 2018). Previous studies (e.g., Hill, 2017) found that women were less likely to choose programs and even universities with male-dominated cohorts, indicating that mostly-male cohorts act as a barrier both pre- and post-entry. Bostwick and Weinberg go on to suggest that ‘peer gender composition may be a useful proxy for climate.’

With this in mind, we believe that university faculties should consider monitoring the year-to-year gender composition of cohorts in STEM doctorate programs closely, providing additional support where necessary to women in male-dominated programs. This could include inter-disciplinary support groups for women Ph.D. students in STEM subjects and greater staff-support. Similarly, we believe this is indicative of a wider need to change attitudes and behaviors among students and faculty members regarding women in STEM, offering a more comprehensive range of gender-sensitivity and bias training. Furthermore, the difference in research funding between male and female students is an area where future research may prove rewarding.


Bostwick, V. K., & Weinberg, B. A. (2022). Nevertheless She Persisted? Gender Peer Effects in Doctoral STEM Programs. Journal of Labor Economics, 000–000.

Hill, Andrew J. 2017. The positive influence of female college students on their male peers. Labour Economics 44:151–60.

Huntington-Klein, N., & Rose, E. (2018) Gender peer effects in a predominantly male environment: Evidence from West Point. American Economic Review: Papers and Proceedings 108:1–4.

Settles, I.H., O’Connor, R.C., & Yap, S.C.Y. (2016) Climate Perceptions and Identity Interference Among Undergraduate Women in STEM: The Protective Role of Gender Identity. Psychology of Women Quarterly.