Unprofessional peer reviews disproportionately harm underrepresented groups in STEM

Abstract

Background: Peer reviewed research is paramount to the advancement of science. Ideally, the peer review process is an unbiased, fair assessment of the scientific merit and credibility of a study; however, well-documented biases arise in all methods of peer review. Systemic biases have been shown to directly impact the outcomes of peer review, yet little is known about the downstream impacts of unprofessional reviewer comments that are shared with authors.

Methods: In an anonymous survey of international participants in science, technology, engineering, and mathematics (STEM) fields, we investigated the pervasiveness and author perceptions of long-term implications of receiving of unprofessional comments. Specifically, we assessed authors' perceptions of scientific aptitude, productivity, and career trajectory after receiving an unprofessional peer review.

Results: We show that survey respondents across four intersecting categories of gender and race/ethnicity received unprofessional peer review comments equally. However, traditionally underrepresented groups in STEM fields were most likely to perceive negative impacts on scientific aptitude, productivity, and career advancement after receiving an unprofessional peer review.

Discussion: Studies show that a negative perception of aptitude leads to lowered self-confidence, short-term disruptions in success and productivity and delays in career advancement. Therefore, our results indicate that unprofessional reviews likely have and will continue to perpetuate the gap in STEM fields for traditionally underrepresented groups in the sciences.

Keywords: Intersectionality; Peer review; STEM; Underrepresented minorities.

Figure 1. Examples of unprofessional peer reviews from survey respondents.

Permission to publish these comments was explicitly given by respondents who certified the comments were reported accurately.

Figure 2. Survey demographics.

(A) Representative career stages (N = 11), (B) scientific disciplines (N = 14) and (C) countries (N = 46) from survey participants. Color in subset (C) represents number of surveys from each country where white is 0.

Figure 3. Results from Bayesian ordinal logistic regression.

Figure shows the probability of selecting a 1–5 for (A) doubting scientific aptitude (N = 617), (B) delayed productivity (N = 620) and (C) delayed career advancement (N = 618) across intersectional groups after receiving an unprofessional peer review. Data are medians and two-tailed 95% BCI. Colors represent level of impact with the lightest (1) as no perceived impact and the darkest (5) as the highest impact. Women and non-binary people were grouped for the statistical analysis to represent marginalized genders in STEM fields.