To examine the association between criteria and priority scores, we regressed priority scores on criteria scores (Table 2). Approach scores predicted priority scores for all PIs’ applications (b = 0.62, SE = 0.06, P < .001), suggesting that having a strong (low) approach score was most important for earning a strong (low) priority score. Analyses within application type showed that the weight of the approach score in predicting the priority score was significantly larger for female than male PIs’ Type 2 applications (P < .05; Table 2). Only for female PIs’ Type 1 applications did significance scores also predict priority scores (b = 0.38, SE = 0.12, P < .01), but there were no scoring disparities for male and female PIs’ Type 1 applications.
Quantitative text analysis
Models showed a main effect of PI sex for positive evaluation words (b = −0.10, SE = 0.04, P < .05) and standout adjectives (b = −0.20, SE = 0.07, P < .01), indicating a significantly higher percentage of words from these categories in critiques of male PIs’ applications (Supplemental Digital Appendix 1, http://links.lww.com/ACADMED/A369). Interaction effects between PI sex and application type for standout adjectives (b = 1.24, SE = 0.50, P < .05) and ability words (b = 1.21, SE = 0.47, P < .05; Supplemental Digital Appendix 2, http://links.lww.com/ACADMED/A369) showed different patterns in critiques of Type 1 compared with Type 2 applications: Relative to the slight differences in critiques of male and female PIs’ Type 1 applications (Figure 3), markedly more critiques of female than male PIs’ Type 2 applications contained words from the ability and standout adjectives categories (Supplemental Digital Appendix 2, http://links.lww.com/ACADMED/A369).
We explored which subsections contributed to linguistic differences in whole critiques. Results showed that differences originated from the strengths subsections of the approach and significance criteria in critiques of funded applications.
Tests for nonresponse bias
The proportion of male and female PI participants in our sample closely resembles their proportions in our population: Males make up 68% (188/278) of UW-Madison R01 PIs from 2010 to 2014 and 61% (76/125) of participants, and females make up 32% (90/278) of R01 PIs and 39% (49/125) of participants. However, because a slightly higher percentage of female than male PIs participated, we tested for the possibility of nonresponse bias by modeling our data using propensity scores (i.e., the probability of a given PI’s participation conditional on observed baseline characteristics), and inverse probability weights with additional auxiliary variables (which penalizes oversampled groups).43–47 Results from these models did not substantially differ from our reported findings. Taken together, these analyses provide evidence to suggest that our findings are not attributable to different patterns of participation for male and female PIs.
In line with our hypothesis, we identified subtle but significant differences in reviewers’ scores and critiques for male and female PIs’ R01 applications, with the greatest differences for Type 2 applications. Results point to three major findings.
First, female PIs applying for Type 2 R01s may be disadvantaged in scoring: Results showed worse (higher) priority, approach, and significance scores for female than male PIs’ Type 2 applications (Table 1, Figure 2). Although approach scores predicted priority scores for both male and female PIs’ Type 1 and Type 2 applications (Table 2), regression weights were highest for female PIs’ Type 2 applications. This suggests that receiving an uncompetitive (high) approach score may have been more detrimental for these female PIs than for any other group.
Next, text analysis results from whole critiques suggest that reviewers may have held male and female PIs of Type 2 applications to different evaluative standards: Markedly more critiques of female than male PIs’ Type 2 applications contained words from the standout adjectives and ability categories (Figure 3).
Finally, text analyses of subsections showed that linguistic differences in whole critiques originated in the strengths subsections of the approach and significance criteria for funded applications, which suggests that these criteria may be most important for determining funding outcomes, particularly for renewals.
Findings from our study showing inconsistencies in scoring and critiques for male and female PIs’ R01 applications, particularly for renewals, may be reflective of objective differences in the quality of the work applicants proposed. However, if male PIs with Type 2 applications had outperformed female PIs, as their stronger priority, approach, and significance scores would suggest, one might expect to see evidence of this across all forms of evaluation. This was not the case. Critiques of female PIs’ Type 2 applications were linguistically stronger, more often containing standout adjectives and words about ability.
It is also possible that our findings are a consequence of male and female PIs working in different research areas, but we found no evidence to support this. Similar proportions of male and female PIs proposed clinical research, and applications were reviewed across 103 study sections and funded by 21 NIH ICs, with no systematic patterns for male and female PIs. Thus, both male and female PIs in our sample were engaged in a similarly diverse range of clinical, basic, and behavioral research projects spanning multiple fields. We did not find evidence of nonresponse bias, suggesting that our findings are reflective of R01 PIs at UW-Madison. Because there are similar criteria for hiring, promotion, and productivity for all UW-Madison faculty, male and female PIs would be expected to have similar background qualifications. Further support that male and female PIs in our sample had similar qualifications and productivity levels comes from the absence of any significant differences in scoring and critiques of the investigator criterion section of their proposals in which reviewers evaluate the PI’s qualifications, productivity, and achievements.48 For such a relatively homogenous sample of male and female applicants, what could explain contrasting scores and critiques for Type 2 applications?
Our findings most strongly align with a large body of work spanning the past 30 years regarding the impact of gender stereotypes on evaluative judgments. This broad array of theoretically grounded experimental and observational studies show that stereotype-based beliefs that women lack the agentic traits (e.g., independence, leadership ability, logic, strength) associated with ability in male-typed domains like science can lead reviewers to doubt women’s competence.9,10,14,15,19 This type of bias is often unconscious, occurs despite explicitly held egalitarian beliefs, and most directly impacts those with a strong belief in their own objectivity (e.g., scientists).19,21,31,32,49 Such bias is most likely to occur when a review is for a high-status position or award (as leadership and mastery are highly agentic),11,15,50 and is tenacious. For example, Kawakami et al51 found that pro-male bias in leader selection persisted even with counterstereotype training.
Compared with first R01s, renewals are higher-status awards in a male-typed field (science52), and applicants are judged by criteria that align with leadership as they are required to have “an ongoing record of accomplishments that have advanced their field(s).”48 Taken together, experimental studies would predict that these factors would heighten the salience of an applicant’s sex and lead reviewers, however unconsciously and inadvertently, to more easily judge female PIs as less competent than male PIs to lead Type 2 R01s.
Depending on the nature of a review process, and the type of criteria used to evaluate applicants, such stereotype-based gender bias can surface in different ways.9,15 For example, experiments in the realm of status characteristics theory by Biernat and Kobrynowicz,10 Foschi,14 and Heilman and Haynes15 have shown that assumptions that women are less competent than men can lead reviewers to hold women to higher ability standards by requiring them to have higher-quality work or more prior achievements.14,15,53 This research would suggest that more laudatory commentary in critiques of female PIs’ Type 2 applications in our sample could be evidence that female PIs needed higher-quality applications than male PIs to earn scores in the fundable range.53 Another possible explanation for our results comes from studies by Glick and Fiske54,55 showing that implicit beliefs that women are less competent than men are confounded by perceptions that women are weak and need to be protected from negative experiences. Consequently, reviewers may give women worse numerical ratings, but “soften the blow” with faint praise and positive remarks. Although this could explain worse scores and stronger critiques for female than male PIs’ Type 2 applications in our sample, this interpretation is unlikely because so-called “ambivalent sexism” is most likely to occur when raters hold explicit personal beliefs about gender stereotypes, which is increasingly uncommon.54–56
More consistent with our findings of worse scores and stronger critiques for female than male PIs’ Type 2 applications is a body of research that documents the co-occurrence of more positive linguistic comments and poorer numerical rankings for women than men in male-typed roles.10,12,13,17,20,25,28,57,58 One such study by Biernat et al57 analyzed performance evaluations for attorneys in the high-status male-typed field of finance law: Women received more praise in written evaluations but worse numerical ratings, which mattered most for promotion to partner. Broadly, this body of research shows a pattern of in-group bias where members of a positively stereotyped in-group (e.g., men, whites) receive favorable ratings on criteria (i.e., scores) that matter most for obtaining tangible rewards (e.g., raises, awards), and members of a negatively stereotyped out-group (e.g., women, ethnic/racial minorities) receive favorable ratings on criteria that matter least (e.g., written commentary, verbal praise).10,12,13,17,20,25,28,57,58 With respect to gender, such bias is most likely to operate when women make up less than 25% of applicants.59–61 Although women make up over 25% of applicants for Type 1 R01s, they are under 25% for Type 2 R01s.3,6,7 Taken together, this research would predict that the conditions under which male and female PIs’ Type 2 R01s are evaluated could disadvantage female R01 renewal applicants in scoring—which matters most for determining funding outcomes—despite strong critiques, which are less consequential.9,10,14,15,22,24,58,62
Linguistic findings from this study closely replicate results from our study of R01 outcomes from applications reviewed at NIH in 2008, which showed significantly higher levels of standout adjectives and greater reference to ability and competence in critiques of female than male PIs’ Type 2 applications.16 In that study, however, we found no scoring disparities, raising the possibility that the new review format may somehow contribute to scoring disparities for male and female R01 renewal applicants. Overall, our current findings suggest that despite the changes implemented in 2009, gender bias may continue to operate in NIH’s peer review process to disadvantage female R01 renewal applicants, and that text analysis may be an effective way to probe for this bias.16
Our study has limitations. The observational design limits any assertion of causality: Even though we attempted to rule out nonresponder bias and selection bias in several ways, it is possible that our findings relate to unidentified differences between male and female PIs apart from applicant sex that could account for the observed differences in scores and critiques. In addition, even though our participants were reviewed by 103 NIH study sections, funded by 21 NIH institutes, and represent 30 departments, they were all faculty at a single institution which may limit the generalizability of our findings to PIs at other institutions. Another possible limitation is that we used only seven word categories and counts of single words. Although extending the analyses to other text analysis procedures or additional word categories might yield different results, the seven word categories we chose were previously validated,16 and single word counts are an effective and widely used text-analytic technique for detecting evaluator sentiment, particularly in large corpora that cannot be feasibly hand annotated.63 Another limitation of our study is that data represent only PIs’ applications that were either funded as first submissions or as revisions; we do not have outcomes from terminally unfunded applications. The NIH keeps the identity of these applicants confidential, which prevented our access to a full range of R01 applicants to invite to participate in our study. However, our findings of disparities in critiques and scoring for male and female PIs’ applications within the fundable range provide compelling evidence for the need to examine critiques and scores for unscored applications, where bias would be very consequential. As a final limitation, the “clinical” meaning of effect sizes that are statistically significant is important to consider. In our study, effect sizes for differences in scores and critiques ranged from ~0.1 to 0.45 (Cohen d). Such effect sizes have proven meaningful in social/behavioral science research.64,65
In spite of its limitations, our study has important implications. Women remain underrepresented in high ranks and leadership in academic medicine and biomedical research—positions that depend on strong records of NIH funding.1,2 If, as our study suggests, stereotype-based gender bias contributes to disparities in reviewers’ ratings of male and female PIs’ Type 2 applications, the impact could be highly consequential. For example, a simulation study by Martell et al66 found that slight pro-male bias in performance ratings (e.g., 1%–5%) significantly impacted promotion rates and left female employees underrepresented in high ranks after only a few cycles of evaluation. If disparities in NIH peer reviewers’ ratings similarly contribute to the lower R01 renewal award rates observed for female PIs nationally,7 the magnitude of the effect on women’s representation in academic medicine could be equally detrimental. To estimate this impact, we applied award rates from Ley and Hamilton’s3 study to the period between 1998 and 2014 showing lower R01 renewal award rates for female PIs.7 We estimated that ~2,000 female PIs went without renewal funding and potentially had to close their labs during that time.3,7 The Association of American Medical Colleges’ (AAMC’s) most recent report shows that although women make up just 21% of professors, they are overrepresented (56%) as instructors.1 These AAMC data show just one of the potential impacts of women’s lower R01 success rates: Women fail to advance at equivalent rates to men and are more likely to teach than remain in research careers. Because women are more likely than men to study issues within the realm of women’s health, women’s attrition from research careers perpetuates health disparities.21,67 This loss also limits the pool of research mentors for early career scientists, particularly for women who derive benefit from mentoring in multiple role management that senior women can provide.21,67,68
If future studies with experimental designs or national datasets confirm that gender bias disadvantages female R01 applicants in NIH peer review, bias-reducing interventions69,70 may be useful as a part of NIH reviewer training. However, such strategies must be carefully constructed and evaluated. Simply increasing awareness of the ubiquity of stereotype-based bias has been shown experimentally to exacerbate the application of age, gender, and body weight stereotype-based bias.71 Conversely, either informing participants that the prevalence of stereotype-based bias is low or that most people are trying to overcome the influence of stereotypes on their evaluations of others reduced the application of gender bias compared with no message or the message about the high prevalence of stereotype-based bias.71 Building on this research, a simple intervention to study might involve randomly including such a message (e.g., “most NIH scientific peer reviewers are working hard to reduce the influence of stereotypes in their evaluation of R01s”) in the materials sent to a random sample of R01 reviewers. Analysis of the critique text and scores could be compared for reviewers in the experimental and control groups. Other strategies to reduce the salience of gender in the NIH peer review process might include replacing abstract descriptors that reinforce male stereotypes (e.g., high-risk, independent) with more concrete and less gender-valenced language (e.g., “research with the potential to change the direction of current investigation,” “an investigator who has been the PI on a grant proposal or supervised graduate students”).11,50,72
Findings from this study raise the possibility that despite the NIH’s alterations to its peer review system in 2009, stereotype-based gender bias may continue to operate in the review process. Because female applicants for R01 renewals may be particularly disadvantaged, future research should target reasons for applicant sex disparities in Type 2 R01 award rates.
Acknowledgments: The authors would like to acknowledge Dr. Jennifer Sheridan, executive and research director, Women in Science and Engineering Leadership Institute, University of Wisconsin–Madison (UW-Madison), for her help in accessing data on background characteristics for principal investigators in the study sample.
1. Lautenberger O, Dandar V, Raezel C, Sloane R. The State of Women in Academic Medicine: The Pipeline and Pathways to Leadership 2013–2014. 2014.Washington, DC: Association of American Medical Colleges.
2. Committee on Bridges to Independence: Identifying Opportunities for and Challenges to Fostering the Independence of Young Investigators in the Life Sciences; Board on Life Sciences; Division on Earth and Life Studies; National Research Council. Bridges to Independence: Fostering the Independence of New Investigators in Biomedical Research. 2005.Washington, DC: National Academies Press.
3. Ley TJ, Hamilton BH. Sociology. The gender gap in NIH grant applications. Science. 2008;322:1472–1474.
6. Pohlhaus JR, Jiang H, Wagner RM, Schaffer WT, Pinn VW. Sex differences in application, success, and funding rates for NIH extramural programs. Acad Med. 2011;86:759–767.
9. Biernat M. Devine P, Plant A. Stereotypes and shifting standards: Forming, communicating and translating person impressions. In: Advances in Experimental Social Psychology. Vol 2012.45. San Diego, Calif: Academic Press.
10. Biernat M, Kobrynowicz D. Gender- and race-based standards of competence: Lower minimum standards but higher ability standards for devalued groups. J Pers Soc Psychol. 1997;72:544–557.
11. Carnes M, Geller S, Fine E, Sheridan J, Handelsman J. NIH Director’s Pioneer Awards: Could the selection process be biased against women? J Womens Health (Larchmt). 2005;14:684–691.
12. Castilla EJ. Gender, race, and meritocracy in organizational careers. AJS. 2008;113:1479–1526.
13. Cleveland J, Stockdale MS, Murphy KR. Women and Men in Organizations: Sex and Gender Issues at Work. 2000.Mahwah, NJ: Lawrence Erlbaum Associates.
14. Foschi M. Double standards in the evaluation of men and women. Soc Psychol Q. August 1996:237–254.
15. Heilman ME, Haynes MC. Borgida E, Fiske ST. Subjectivity in the appraisal process: A facilitator of gender bias in work settings. In: Beyond Common Sense: Psychological Science in the Courtroom. 2008.Malden, Mass: Blackwell Publishers.
16. Kaatz A, Magua W, Zimmerman DR, Carnes M. A quantitative linguistic analysis of National Institutes of Health R01 application critiques from investigators at one institution. Acad Med. 2015;90:69–75.
17. Lyness KS, Heilman ME. When fit is fundamental: Performance evaluations and promotions of upper-level female and male managers. J Appl Psychol. 2006;91:777–785.
18. Madera JM, Hebl MR, Martin RC. Gender and letters of recommendation for academia: Agentic and communal differences. J Appl Psychol. 2009;94:1591–1599.
19. Moss-Racusin CA, Dovidio JF, Brescoll VL, Graham MJ, Handelsman J. Science faculty’s subtle gender biases favor male students. Proc Natl Acad Sci U S A. 2012;109:16474–16479.
20. Murphy KR, Cleveland JN. Understanding Performance Appraisal: Social, Organizational, and Goal-Based Perspectives. 1995.Thousand Oaks, Calif: Sage Publications, Inc.
21. National Academy of Sciences, National Academy of Engineering, and Institute of Medicine Committee on Maximizing the Potential of Women in Academic Science and Engineering. Beyond Bias and Barriers: Fulfilling the Potential of Women in Academic Science and Engineering. 2007.Washington, DC: National Academies Press.
22. Schmader T, Whitehead J, Wysocki VH. A linguistic comparison of letters of recommendation for male and female chemistry and biochemistry job applicants. Sex Roles. 2007;57:509–514.
23. Steinpreis RE, Anders KA. The impact of gender on the review of the curricula vitae of job applicants and tenure candidates: A national empirical study. Sex Roles. 1999;41:509–528.
24. Trix F, Psenka C. Exploring the color of glass: Letters of recommendation for female and male medical faculty. Discourse Soc. 2003;14:191–220.
25. Vescio TK, Gervais SJ, Snyder M, Hoover A. Power and the creation of patronizing environments: The stereotype-based behaviors of the powerful and their effects on female performance in masculine domains. J Pers Soc Psychol. 2005;88:658–672.
26. Waisbren SE, Bowles H, Hasan T, et al. Gender differences in research grant applications and funding outcomes for medical school faculty. J Womens Health (Larchmt). 2008;17:207–214.
27. Wenneras C, Wold A. Nepotism and sexism in peer-review. Nature. 1997;387:341–343.
28. Wilson KY. An analysis of bias in supervisor narrative comments in performance appraisal. Hum Relat. 2010;63:1903–1933.
29. Heilman ME. Description and prescription: How gender stereotypes prevent women’s ascent up the organizational ladder. J Soc Issues. 2001;57:657–674.
30. Heilman ME, Haynes MC. No credit where credit is due: Attributional rationalization of women’s success in male–female teams. J Appl Psychol. 2005;90:905–916.
31. Devine PG. Stereotypes and prejudice: Their automatic and controlled components. J Per Soc Psychol. 1989;56:5–18.
32. Valian V. Why So Slow? The Advancement of Women. 1999.Cambridge, Mass: MIT Press.
33. Nosek BA, Smyth FL, Hansen JJ, et al. Pervasiveness and correlates of implicit attitudes and stereotypes. Eur Rev Soc Psychol. 2007;18:36–88.
37. Jagsi R, Motomura AR, Griffith KA, Rangarajan S, Ubel PA. Sex differences in attainment of independent funding by career development awardees. Ann Intern Med. 2009;151:804–811.
38. Pennebaker J, Chung C, Ireland M, Gonzales A, Booth R. The Development and Psychometric Properties of LIWC2007. 2007.Austin, Tex: LIWC Net.
39. Isaac C, Chertoff J, Lee B, Carnes M. Do students’ and authors’ genders affect evaluations? A linguistic analysis of medical student performance evaluations. Acad Med. 2011;86:59–66.
40. Belotti F, Deb P, Manning WG, Norton EC. Twopm: Two-part models. Stata J. 2015;15:3–20.
41. Fletcher D, Mackenzie D, Villouta E. Modelling skewed data with many zeros: A simple approach combining ordinary and logistic regression. Env Ecol Stat. 2005;12:45–54.
42. Papke LE, Wooldridge JM. Econometric methods for fractional response variables with an application to 401 (K) plan participation rates. J Appl Econ. 1996;11:619–632.
44. Cuddeback G, Wilson E, Orme JG, Combs-Orme T. Detecting and statistically correcting sample selection bias. J Soc Ser Res. 2004;30:19–33.
45. Fitzgerald J, Gottschalk P, Moffitt RA. An Analysis of Sample Attrition in Panel Data: The Michigan Panel Study of Income Dynamics. 1998.Cambridge, Mass: National Bureau of Economic Research.
46. Foster EM, Fang GY; Conduct Problems Research Group. Alternative methods for handling attrition: An illustration using data from the Fast Track evaluation. Eval Rev. 2004;28:434–464.
47. Miller R, Hollist C, Neil J. Salkind N. Attrition bias. In: Encyclopedia of Measurement and Statistics. Vol. 1. 2007.Thousand Oaks, Calif: Sage.
49. Uhlmann EL, Cohen GL. “I think it, therefore it’s true”: Effects of self-perceived objectivity on hiring discrimination. Organ Behav Hum Decis Process. 2007;104:207–223.
50. Carnes M, Bland C. Viewpoint: A challenge to academic health centers and the National Institutes of Health to prevent unintended gender bias in the selection of clinical and translational science award leaders. Acad Med. 2007;82:202–206.
51. Kawakami K, Dovidio JF, van Kamp S. Kicking the habit: Effects of nonstereotypic association training and correction processes on hiring decisions. J Exp Soc Psychol. 2005;41:68–75.
52. Nosek BA, Smyth FL, Sriram N, et al. National differences in gender-science stereotypes predict national sex differences in science and math achievement. Proc Natl Acad Sci U S A. 2009;106:10593–10597.
53. Biernat M, Collins EC, Katzarska-Miller I, Thompson ER. Race-based shifting standards and racial discrimination. Pers Soc Psychol Bull. 2009;35:16–28.
54. Glick P, Fiske ST. The ambivalent sexism inventory: Differentiating hostile and benevolent sexism. J Person Soc Psychol. 1996;70:491–512.
55. Glick P, Fiske ST. Dixon J, Levine, M. An ambivalent alliance: Hostile and benevolent sexism as complementary justifications for gender inequality. In: Beyond Prejudice. 2012.Cambridge, Mass: Cambridge University Press.
56. King EB, Botsford W, Hebl MR, Kazama S, Dawson JF, Perkins A. Benevolent sexism at work: Gender differences in the distribution of challenging developmental experiences. J Manage. 2012;38:1835–1866.
57. Biernat M, Tocci MJ, Williams JC. The language of performance evaluations: Gender-based shifts in content and consistency of judgment. Soc Psychol Personal Sci. 2012;3:186–192.
58. Biernat M, Vescio TK. She swings, she hits, she’s great, she’s benched: Implications of gender-based shifting standards for judgment and behavior. Pers Soc Psychol Bull. 2002;28:66–77.
59. Heilman M. Sex bias in hiring women. Organ Behav Hum Perform. 1980;26:386–395.
60. van Ommeren J, de Vries RE, Russo G, van Ommeren M. Context in selection of men and women in hiring decisions: Gender composition of the applicant pool. Psychol Rep. 2005;96:349–360.
61. Isaac C, Lee B, Carnes M. Interventions that affect gender bias in hiring: A systematic review. Acad Med. 2009;84:1440–1446.
62. Madera JM, Hebl MR, Martin RC. Gender and letters of recommendation for academia: Agentic and communal differences. J Appl Psychol. 2009;94:1591–1599.
63. Pang B, Lee L. Opinion mining and sentiment analysis. Found Trend Inf Retr. 2008;2:1–35.
64. Glass G. Primary, secondary, and meta-analysis of research. Educ Res. 1976;5:3–5.
65. Lipsey MW, Wilson DB. The efficacy of psychological, educational, and behavioral treatment. Confirmation from meta-analysis. Am Psychol. 1993;48:1181–1209.
66. Martell R, Lane D, Emrich C. Male–female differences: A computer simulation. Am Psychol. 1996;51:157–158.
67. Carnes M, Morrissey C, Geller SE. Women’s health and women’s leadership in academic medicine: Hitting the same glass ceiling? J Womens Health (Larchmt). 2008;17:1453–1462.
68. Isaac C, Byars-Winston A, McSorley R, Schultz A, Kaatz A, Carnes ML. A qualitative study of work–life choices in academic internal medicine. Adv Health Sci Educ Theory Pract. 2014;19:29–41.
69. Carnes M, Devine PG, Baier Manwell L, et al. The effect of an intervention to break the gender bias habit for faculty at one institution: A cluster randomized, controlled trial. Acad Med. 2015;90:221–230.
70. Bauer CC, Baltes BB. Reducing the effects of gender stereoytpes on performance evaluations. Sex Roles. 2002;47:465–476.
71. Duguid MM, Thomas-Hunt MC. Condoning stereotyping? How awareness of stereotyping prevalence impacts expression of stereotypes. J Appl Psychol. 2015;100:343–359.
72. Kaatz A, Gutierrez B, Carnes M. Threats to objectivity in peer review: The case of gender. Trends Pharmacol Sci. 2014;35:371–373.
Supplemental Digital Content
© 2016 by the Association of American Medical Colleges