Underrepresented minority faculty in the USA face a double standard in promotion and tenure decisions

The paper investigates whether racial disparities in academia extend to promotion and tenure (P&T) decisions and under what conditions these disparities are most pronounced. Motivated by persistent underrepresentation of Black and Hispanic faculty in US higher education and evidence that diversity benefits scientific innovation and student outcomes, the authors test four hypotheses: (H1) URM status is negatively related to P&T voting outcomes; (H2) scholarly productivity moderates this relationship such that disparities are greatest for candidates with lower productivity (double standard); (H3) the moderation by productivity is most pronounced for URM women (intersectionality); and (H4) emphasizing candidates’ scholarship in external review letters (ERLs) mitigates disparities. The study situates P&T as a critical bottleneck for faculty careers and institutional diversity and leverages multi-institutional quantitative data to evaluate these hypotheses.

Prior work documents disadvantages for URM academics across the academic pipeline, including admissions, doctoral completion, peer review, grant funding, hiring, and patenting, alongside stereotype-driven perceptions of competence and hireability. Theoretical grounding comes from the shifting standards model, which posits that minority group members are often held to higher performance standards for equivalent outcomes. In academia, research productivity is central to evaluation, typically assessed via outputs/impact (e.g., h-index) and ERLs. Although gender disparities have been extensively studied, fewer quantitative studies address racial and especially intersectional (race × gender) disparities in academic reward systems. Guidance has emerged for anti-racist ERLs, but empirical evidence on ERL content’s role in P&T is limited. This study addresses these gaps by testing whether URM and particularly URM women face higher standards tied to productivity and whether ERL emphasis on scholarship can attenuate these disparities.

Design and sample: Archival, multi-institutional study of 1,571 P&T cases (promotion to associate professor with tenure or to full professor) from five US research-intensive universities, 2015–2022. Two HBCUs were excluded due to URMs not being a minority at those institutions. Voting outcomes were recorded at department and college levels (university-level voting available for a small subset and excluded from main analyses). ERLs were collected for each candidate, totaling 9,032 letters; ERL analyses used multilevel models with 5,191 letters matched to 935 candidates after exclusions.

Data collection and privacy: Institutional coders (provost’s offices) with routine access extracted de-identified candidate, voting, and ERL linguistic data following standardized protocols and weekly cross-institution meetings. LIWC2015 was used to compute dictionary-based linguistic features. Candidate identifiers used for scraping public productivity metrics (Google Scholar, Academic Analytics) were replaced with random IDs; personally identifiable information and institution identifiers were removed or dummy-coded. IRB deemed the study non-human-subjects research and reliance agreements were established.

Measures:

• URM status: 1 = Black/African American or Hispanic; 0 = White/Caucasian or Asian/Asian American (self-identified). Very small counts for Native American/Native Hawaiian/Other excluded from URM coding.
• Negative voting percentage: percentage of “no” votes at department and college levels; abstentions excluded.
• Unanimous vote: 1 if all votes were “yes,” else 0.
• Scholarly productivity: h-index at time of P&T (scraped from public sources).
• ERL scholarship language: custom LIWC dictionary capturing scholarship-related terms (e.g., top journal, funding, citation, impact); aggregated to candidate level (mean ≈ 0.26% of words per ERL).
• Controls: institution, discipline (CIP codes), university ranking (US News), rank (associate vs full), number of external grants as PI, STEM field (DHS list), gender (1 = woman), tenure in rank (years). Where h-index was not a moderator, it was included as a control.

Analytic strategy: Two-tailed tests. For continuous outcomes (negative vote percentage), OLS regressions; for binary outcomes (unanimous vote, provost vote), logistic regressions. Main effects of URM status assessed with and without controls. Moderation tests: (a) URM × h-index; (b) intersectional categories (non-URM man [reference], URM woman, URM man, non-URM woman) × h-index; (c) URM woman (1) vs others (0) × ERL scholarship language using multilevel models for ERL analyses. Indirect effects of URM on provost vote via college-level voting examined with logistic models. Robustness checks included controlling for LIWC tone, clout, authenticity, and analytic language.

• College-level disparities (H1): URM status predicted worse college-level voting outcomes. Without controls, more negative votes (β ≈ 0.33; P < 0.001; 95% CI: 0.15, 0.51) and fewer unanimous votes (OR = 0.56; P = 0.004; 95% CI: 0.38, 0.83). With controls, more negative votes (β = 0.31; P = 0.002; 95% CI: 0.12, 0.51) and fewer unanimous votes (OR = 0.56; P = 0.017; 95% CI: 0.34, 0.90). Unstandardized estimate: URM faculty received 7% more negative votes at the college level (b = 0.07; P = 0.002; 95% CI: 0.03, 0.12) and were 44% less likely to receive a unanimous vote. No reliable disparities at the department level.
• Provost vote: College-level voting strongly predicted provost outcomes. Indirect effects showed URM status reduced likelihood of a positive provost vote via higher college negative vote percentage (indirect OR = 0.87; P < 0.001) and via lower college unanimous vote (indirect OR = 0.90; P < 0.001). No direct URM effect on provost vote (OR = 1.12; P = 0.835).
• Double standard by productivity (H2): h-index moderated URM effects at the college level. For negative vote percentage, URM × h-index β = -0.34; P = 0.044 (95% CI: -0.915, -0.010). For unanimous votes, URM × h-index OR = 4.24; P = 0.003 (95% CI: 1.62, 11.06). URM candidates with lower h-indexes received more negative votes and fewer unanimous votes than comparable non-URM candidates; disparities diminished at higher h-index levels.
• Intersectionality (H3): Compared with non-URM men, URM women showed stronger moderation by productivity. Negative vote percentage: URM woman × h-index β = -0.97; P = 0.001 (95% CI: -1.547, -0.384). Unanimous votes: OR = 6.59; P = 0.019 (95% CI: 1.37, 31.73). No significant moderation for URM men or non-URM women relative to non-URM men.
• ERL scholarship language mitigation (H4): For URM women, scholarship language in ERLs mitigated disparities. Negative vote percentage interaction: B = -0.22; P = 0.007 (95% CI: -0.385, -0.061). Unanimous vote interaction: OR = 15.67; P = 0.005 (95% CI: 2.33, 105.54). Results robust controlling for LIWC tone, clout, authenticity, and analytic language.
• Exploratory split-sample: Disparities significant for promotion to associate professor (with tenure) but not for promotion to full professor.

Findings support the double standard hypothesis: URM faculty, particularly URM women, are held to higher standards on scholarly productivity during college-level P&T voting. Disparities manifest as both increased negative votes and reduced unanimity, which in turn indirectly reduce positive provost decisions and may have broader reputational and support consequences. The productivity interaction indicates biased evaluation beyond resource disparities; even accounting for grants and h-index, URM status affects evaluation at lower productivity levels. ERLs emerge as a lever for equity: emphasizing candidates’ scholarship substantially improves outcomes for URM women, suggesting structured guidance to reviewers could mitigate bias. Effects appearing primarily at the college level may reflect evaluators’ greater social distance from candidates, increasing reliance on stereotypes. The results clarify how P&T processes contribute to attrition in URM representation from assistant to associate to full professor.

This multi-institutional study shows that P&T processes contribute to URM underrepresentation through biased college-level evaluations: URM faculty receive about 7% more negative votes and are 44% less likely to garner unanimous support, with the strongest penalties at lower productivity levels and for URM women. Emphasizing scholarship in ERLs can mitigate these disparities for URM women. Contributions include quantitative evidence of racial and intersectional double standards in P&T and identification of ERL content as a practical mitigation pathway. Policy implications include reconsidering heavy reliance on h-index-like metrics and implementing ERL guidelines that foreground candidates’ scholarship, particularly for URM women. Future work should test interventions (e.g., structured ERLs, reviewer training), examine broader evaluative settings (annual reviews, merit raises), expand to diverse institution types and disciplines, and develop richer, fairer metrics for productivity and grants.

• Generalizability: Five US research-intensive universities; two HBCUs excluded; results may differ at elite top-50, teaching-focused, and community colleges.
• Scope: Focus on P&T; related evaluative contexts (annual/merit reviews) not examined.
• Measures: h-index has known limitations as a productivity proxy; grant measure captures only count of PI awards; LIWC scholarship dictionary counts word instances without contextual nuance.
• Discipline-specific nuances: Broad, cross-disciplinary approach precludes deep discipline-level analyses.
• Data constraints: Some voting levels (university) had limited data; ERL analyses rely on aggregated dictionary scores; formal normality/homoscedasticity tests not conducted.
• Coding and privacy: Centralized institutional coding limits traditional interrater reliability, though multiple safeguards and checks were used.
• Intersectional cell sizes: Fewer URM candidates at high h-index levels limit power to test performance at the upper tail.