The conversation about gender gaps in STEM has spent two decades counting the same thing: how many women enter engineering, how many men enter nursing. A study published this week in Nature Human Behaviour shows that metric is incomplete — and ignoring the rest of the picture has a measurable cost for institutions that claim to have a DEI strategy.

What the evidence shows

A team led by Joseph R. Cimpian (NYU), with Zachary T. McDermott and Taek H. Kim, analyzed nationally representative U.S. data on gender and LGBTQ+ representation across college majors and achievement levels. The core finding: in fields where one gender numerically dominates — engineering and computer science on the male side, nursing and education on the female side — students who match the field’s dominant gender norm tend to stay and graduate even when their academic performance is lower than that of students who don’t fit the norm.

The accompanying analysis, by Bryce E. Hughes, spells out the direct implication for LGBTQ+ populations: the same “in-group advantage” that protects lower-performing students who fit the norm functions as an added barrier for LGBTQ+ students in male-dominated fields. This isn’t only a headcount problem. It’s a question of who is treated as legitimately belonging.

Why this is HR’s problem, not just Education’s

Corporate DEI functions and university diversity offices share the same design flaw: they measure inclusion by counting heads at entry — hiring, admissions — and rarely track what happens to retention once someone is already inside, cross-tabulated with sexual orientation and gender identity as variables independent of the binary gender count.

The result is predictable: generic DEI training that doesn’t distinguish between “not enough women in the major” and “LGBTQ+ students or employees leaving a specific field despite strong performance.” Those are two different problems with two different causes, and they need two different interventions.

The gap nobody has filled in Latin America

Cimpian et al.’s study works exclusively with U.S. data. No equivalent analysis exists yet using Latin American data that systematically cross-references academic or professional retention, field of specialization, and LGBTQ+ status. That absence of regional data isn’t a footnote — it means any university or company in Mexico, Spain, or Ecuador that wants to apply this framework first needs its own climate diagnostic, not an imported conclusion from a different context.

That diagnostic work — sitting at the intersection of organizational psychology, human capital management, and anti-discrimination law — is exactly what a climate assessment by major or department, rather than an aggregate survey, is built to do.

From evidence to action

Three concrete steps for universities and companies

1. Segmented climate audits by major, department, or business unit — not one organization-wide climate survey.
2. Retention and attrition tracking that cross-references performance, gender, and LGBTQ+ identity as separate variables, not collapsed into a single “diversity” category.
3. Redesigning the norm of belonging in the units identified as highest-attrition, with 12- and 24-month follow-up metrics — not a one-time training session.

What serious DEI compliance should require

This is precisely what a DEI compliance framework aligned with standards like ISO 26000 should require before certifying an organization as “inclusive”: evidence of segmented retention, not just evidence of diverse hiring.

Inclusion cannot be proven by who enters alone. It must also be measured by who stays, advances, and is treated as legitimately belonging.

SEO FAQ

DEI, LGBTQ+ talent and retention

Reference

Cimpian, J. R., McDermott, Z. T., & Kim, T. H. (2026). Gender and LGBTQ+ representation across college majors and achievement levels using nationally representative US data. Nature Human Behaviour. DOI: 10.1038/s41562-026-02510-8.

Companion analysis: Hughes, B. E. (2026). News & Views, Nature Human Behaviour, September 3.