[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
Figure 1.
Inclusion and Exclusion Criteria and Sources of Missing Data
Inclusion and Exclusion Criteria and Sources of Missing Data

ASJC indicates All Science Journal Classification; ICC, intraciting commentary.

Figure 2.
Odds of Authoring an Invited Commentary for Women vs Men by Percentile of Years Active
Odds of Authoring an Invited Commentary for Women vs Men by Percentile of Years Active

Results are adjusted for field of expertise, h-index percentile, and total number of publications percentile. The 10th percentile indicates approximately 8 years; the 50th percentile, approximately 19 years (the median number); 90th percentile, approximately 38 years; line, odds ratio (OR); shading, 95% CI.

Figure 3.
Subgroup Analyses by Journal Topic
Subgroup Analyses by Journal Topic

Journals may have multiple All Science Journal Classification (ASJC) codes; thus, the topics overlap with respect to journals included. Model 1 (orange squares) controls for authors’ fields of expertise through matching. Model 2 (blue squares) includes the adjustments in model 1 and also adjusts for years active percentile, h-index percentile, and total number of publications percentile. Years active indicates years since first publication in Scopus; OR, odds ratio; error bars, 95% CIs.

Figure 4.
Journal-Specific Odds Ratios (ORs) Plotted Against CiteScore
Journal-Specific Odds Ratios (ORs) Plotted Against CiteScore

The line and shading represent the OR and 95% CI, respectively, for women vs men of authoring invited commentaries as a function of the CiteScore of the journal in which the commentary was published. Each circle represents the OR estimated for a single journal, and circle diameter is inversely proportional to the SE of the log OR estimate. Only journals with more than 50 matched sets are shown.

Table.  
Author-Level Variables by Case-control Status for Unique Authors Included in Analyses
Author-Level Variables by Case-control Status for Unique Authors Included in Analyses
Supplement.

eAppendix 1. Supplementary Methods

eAppendix 2. Supplementary Results

eFigure 1. Example of Concepts (Noun Phrases) Identified in an Abstract

eFigure 2. Articles With Ranked Concepts (Noun Phrases) and Author Profile for Dr Francesca Dominici

eTable 1. Gender by Author-Level Variables for Unique Authors, Including Authors With Unknown Gender

eTable 2. Results From Pooled Conditional Logistic Regression Models

eFigure 3. Odds Ratio for Invited Commentary Authorship as a Function of Years Active, H-Index, and Number of Publications

eTable 3. Results From Conditional Logistic Regression Model Including Interaction Between Gender and Years Active

eFigure 4. Results From Conditional Logistic Regression Model Including Interaction Between Gender and Years Active

eTable 4. Results From Conditional Logistic Regression Model Including Interaction Between Gender and H-Index

eFigure 5. Results From Conditional Logistic Regression Model Including Interaction Between Gender and H-Index

eTable 5. Results From Conditional Logistic Regression Model Including Interaction Term Between Gender and Number of Publications

eFigure 6. Results From Conditional Logistic Regression Model Including Interaction Term Between Gender and Number of Publications

eFigure 7. Association of Gender With Authoring an Invited Commentary

eAppendix 3. Estimated Journal-Specific Odds Ratios vs Journal CiteScore

eTable 6. Association of Gender With Invited Commentary Authorship Esimated via 2-Stage Random Effects Meta-analysis

eFigure 8. 2-Stage Meta-analysis Results by Journal Topic

eFigure 9. Meta-regression of Journal-Specific Effect Sizes Against CiteScore

eTable 7. Association of Gender With Invited Commentary Authorship Accounting for Missing Gender Data via Multiple Imputation

eTable 8. Association of Gender on Invited Commentary Authorship After Excluding Duplicates for Case Authors Who Appear in Multiple Journals, Excluding Articles That May Be Replies to Other Articles, and Keeping Only Top 2 Controls Per Case

eReferences

1.
West  JD, Jacquet  J, King  MM, Correll  SJ, Bergstrom  CT.  The role of gender in scholarly authorship.  PLoS One. 2013;8(7):e66212. doi:10.1371/journal.pone.0066212PubMedGoogle Scholar
2.
Zeng  XHT, Duch  J, Sales-Pardo  M,  et al.  Differences in collaboration patterns across discipline, career stage, and gender.  PLoS Biol. 2016;14(11):e1002573. doi:10.1371/journal.pbio.1002573PubMedGoogle Scholar
3.
Ahmadi  M, Khurshid  K, Sanelli  PC,  et al.  Influences for gender disparity in academic neuroradiology.  AJNR Am J Neuroradiol. 2018;39(1):18-23. doi:10.3174/ajnr.A5443PubMedGoogle ScholarCrossref
4.
Chauvin  S, Mulsant  BH, Sockalingam  S, Stergiopoulos  V, Taylor  VH, Vigod  SN.  Gender differences in research productivity among academic psychiatrists in Canada.  Can J Psychiatry. 2019;64(6):415-422. doi:10.1177/0706743718802798PubMedGoogle ScholarCrossref
5.
Diamond  SJ, Thomas  CR  Jr, Desai  S,  et al.  Gender differences in publication productivity, academic rank, and career duration among US academic gastroenterology faculty.  Acad Med. 2016;91(8):1158-1163. doi:10.1097/ACM.0000000000001219PubMedGoogle ScholarCrossref
6.
Mayer  EN, Lenherr  SM, Hanson  HA, Jessop  TC, Lowrance  WT.  Gender differences in publication productivity among academic urologists in the United States.  Urology. 2017;103:39-46. doi:10.1016/j.urology.2016.12.064PubMedGoogle ScholarCrossref
7.
Nguyen  V, Marmor  RA, Ramamoorthy  SL, Blair  SL, Clary  BM, Sicklick  JK.  Academic surgical oncologists’ productivity correlates with gender, grant funding, and institutional NCI comprehensive cancer center affiliation.  Ann Surg Oncol. 2018;25(7):1852-1859. doi:10.1245/s10434-018-6450-5PubMedGoogle ScholarCrossref
8.
Pfeiffer  M, Fischer  MR, Bauer  D.  Publication activities of German junior researchers in academic medicine: which factors impact impact factors?  BMC Med Educ. 2016;16(1):190. doi:10.1186/s12909-016-0712-3PubMedGoogle ScholarCrossref
9.
Hsiehchen  D, Hsieh  A, Espinoza  M.  Prevalence of female authors in case reports published in the medical literature.  JAMA Netw Open. 2019;2(5):e195000-e195000. doi:10.1001/jamanetworkopen.2019.5000PubMedGoogle ScholarCrossref
10.
Holman  L, Stuart-Fox  D, Hauser  CE.  The gender gap in science: how long until women are equally represented?  PLoS Biol. 2018;16(4):e2004956. doi:10.1371/journal.pbio.2004956PubMedGoogle Scholar
11.
Bendels  MHK, Dietz  MC, Brüggmann  D, Oremek  GM, Schöffel  N, Groneberg  DA.  Gender disparities in high-quality dermatology research: a descriptive bibliometric study on scientific authorships.  BMJ Open. 2018;8(4):e020089. doi:10.1136/bmjopen-2017-020089PubMedGoogle Scholar
12.
Bendels  MHK, Müller  R, Brueggmann  D, Groneberg  DA.  Gender disparities in high-quality research revealed by Nature Index journals.  PLoS One. 2018;13(1):e0189136. doi:10.1371/journal.pone.0189136PubMedGoogle Scholar
13.
Helmer  M, Schottdorf  M, Neef  A, Battaglia  D.  Gender bias in scholarly peer review.  Elife. 2017;6:e21718. doi:10.7554/eLife.21718PubMedGoogle Scholar
14.
Filardo  G, da Graca  B, Sass  DM, Pollock  BD, Smith  EB, Martinez  MA-M.  Trends and comparison of female first authorship in high impact medical journals: observational study (1994-2014).  BMJ. 2016;352:i847. doi:10.1136/bmj.i847PubMedGoogle ScholarCrossref
15.
Lerback  J, Hanson  B.  Journals invite too few women to referee.  Nature. 2017;541(7638):455-457. doi:10.1038/541455aPubMedGoogle ScholarCrossref
16.
Larivière  V, Ni  C, Gingras  Y, Cronin  B, Sugimoto  CR.  Bibliometrics: global gender disparities in science.  Nature. 2013;504(7479):211-213. doi:10.1038/504211aPubMedGoogle ScholarCrossref
17.
Ioannidou  E, Rosania  A.  Under-representation of women on dental journal editorial boards.  PLoS One. 2015;10(1):e0116630. doi:10.1371/journal.pone.0116630PubMedGoogle Scholar
18.
Topaz  CM, Sen  S.  Gender representation on journal editorial boards in the mathematical sciences.  PLoS One. 2016;11(8):e0161357. doi:10.1371/journal.pone.0161357PubMedGoogle Scholar
19.
Buckley  HL, Sciligo  AR, Adair  KL, Case  BS, Monks  JM.  Is there gender bias in reviewer selection and publication success rates for the New Zealand Journal of Ecology N Z J Ecol. 2014;38(2):335-339.Google Scholar
20.
Edwards  HA, Schroeder  J, Dugdale  HL.  Gender differences in authorships are not associated with publication bias in an evolutionary journal.  PLoS One. 2018;13(8):e0201725. doi:10.1371/journal.pone.0201725PubMedGoogle Scholar
21.
Fox  CW, Burns  CS, Muncy  AD, Meyer  JA.  Gender differences in patterns of authorship do not affect peer review outcomes at an ecology journal.  Funct Ecol. 2016;30(1):126-139. doi:10.1111/1365-2435.12587Google ScholarCrossref
22.
Heckenberg  A, Druml  C.  Gender aspects in medical publication: the Wiener klinische Wochenschrift.  Wien Klin Wochenschr. 2010;122(5-6):141-145. doi:10.1007/s00508-010-1302-8PubMedGoogle ScholarCrossref
23.
Lane  J, Linden  DJ.  Is there gender bias in the peer review process at Journal of Neurophysiology J Neurophysiol. 2009;101(5). doi:10.1152/jn.00196.2009Google Scholar
24.
Primack  RB, Ellwood  E, Miller-Rushing  AJ, Marrs  R, Mulligan  A.  Do gender, nationality, or academic age affect review decisions: an analysis of submissions to the journal Biological Conservation Biol Conserv. 2009;142(11):2415-2418. doi:10.1016/j.biocon.2009.06.021Google ScholarCrossref
25.
Berg  J.  Examining author gender data.  Science. 2019;363(6422):7. doi:10.1126/science.aaw4633PubMedGoogle ScholarCrossref
26.
Murray  D, Siler  K, Larivière  V,  et al.  Gender and international diversity improves equity in peer review.  Preprint. Posted online August 29, 2019. bioRxiv:400515. doi:10.1101/400515
27.
Conley  D, Stadmark  J.  Gender matters: a call to commission more women writers.  Nature. 2012;488(7413):590. doi:10.1038/488590aPubMedGoogle ScholarCrossref
28.
Süßenbacher  S, Amering  M, Gmeiner  A, Schrank  B.  Gender-gaps and glass ceilings: a survey of gender-specific publication trends in Psychiatry between 1994 and 2014.  Eur Psychiatry. 2017;44:90-95. doi:10.1016/j.eurpsy.2017.03.008PubMedGoogle ScholarCrossref
29.
Qureshi  R, Lê  J, Li  T, Ibrahim  M, Dickersin  K.  Gender and editorial authorship in high impact epidemiology journals [published online April 17, 2019].  Am J Epidemiol. doi:10.1093/aje/kwz094PubMedGoogle Scholar
30.
Silver  JK, Poorman  JA, Reilly  JM, Spector  ND, Goldstein  R, Zafonte  RD.  Assessment of women physicians among authors of perspective-type articles published in high-impact pediatric journals.  JAMA Netw Open. 2018;1(3):e180802-e180802. doi:10.1001/jamanetworkopen.2018.0802PubMedGoogle ScholarCrossref
31.
Carr  PL, Raj  A, Kaplan  SE, Terrin  N, Breeze  JL, Freund  KM.  Gender differences in academic medicine: retention, rank, and leadership comparisons from the National Faculty Survey.  Acad Med. 2018;93(11):1694-1699. doi:10.1097/ACM.0000000000002146PubMedGoogle ScholarCrossref
32.
Elsevier.  What is Scopus Preview?https://service.elsevier.com/app/answers/detail/a_id/15534/supporthub/scopus/#tips. Accessed March 25, 2019.
33.
Harzing  A-W, Alakangas  S.  Google Scholar, Scopus and the Web of Science: a longitudinal and cross-disciplinary comparison.  Scientometrics. 2016;106(2):787-804. doi:10.1007/s11192-015-1798-9Google ScholarCrossref
34.
genderize.io.  Determine the gender of a first name. https://genderize.io/. Accessed March 28, 2019.
35.
La Gamba  F, Corrao  G, Romio  S,  et al.  Combining evidence from multiple electronic health care databases: performances of one-stage and two-stage meta-analysis in matched case-control studies.  Pharmacoepidemiol Drug Saf. 2017;26(10):1213-1219. doi:10.1002/pds.4280PubMedGoogle ScholarCrossref
36.
 Data set for investigation of gender disparities in invited commentary authorship for medical journals. https://github.com/emgthomas/gender_and_invited_commentaries. Accessed September 9, 2019.
37.
Westring  A, McDonald  JM, Carr  P, Grisso  JA.  An integrated framework for gender equity in academic medicine.  Acad Med. 2016;91(8):1041-1044. doi:10.1097/ACM.0000000000001275PubMedGoogle ScholarCrossref
38.
Ellinas  EH, Kaljo  K, Patitucci  TN, Novalija  J, Byars-Winston  A, Fouad  NA.  No room to “lean in”: a qualitative study on gendered barriers to promotion and leadership.  J Womens Health (Larchmt). 2019;28(3):393-402. doi:10.1089/jwh.2018.7252PubMedGoogle ScholarCrossref
39.
Babcock  L, Recalde  MP, Vesterlund  L, Weingart  L.  Gender differences in accepting and receiving requests for tasks with low promotability.  Am Econ Rev. 2017;107(3):714-747. doi:10.1257/aer.20141734Google ScholarCrossref
40.
Guarino  CM, Borden  VM.  Faculty service loads and gender: are women taking care of the academic family?  Res High Educ. 2017;58(6):672-694. doi:10.1007/s11162-017-9454-2Google ScholarCrossref
41.
Gupta  N, Kemelgor  C, Fuchs  S, Etzkowitz  H.  The “Triple Burden”: a cross-cultural analysis of the consequences of discrimination for women in science.  Curr Sci. 2004;89(8):1382-1386.Google Scholar
42.
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(41):16474-16479. doi:10.1073/pnas.1211286109PubMedGoogle ScholarCrossref
43.
Quadlin  N.  The mark of a woman’s record: gender and academic performance in hiring.  Am Sociol Rev. 2018;83(2):331-360. doi:10.1177/0003122418762291Google ScholarCrossref
44.
Knobloch-Westerwick  S, Glynn  CJ, Huge  M.  The Matilda effect in science communication: an experiment on gender bias in publication quality perceptions and collaboration interest.  Sci Commun. 2013;35(5):603-625. doi:10.1177/1075547012472684Google ScholarCrossref
45.
Krawczyk  M, Smyk  M.  Author’s gender affects rating of academic articles: evidence from an incentivized, deception-free laboratory experiment.  Eur Econ Rev. 2016;90:326-335. doi:10.1016/j.euroecorev.2016.02.017Google ScholarCrossref
46.
Hengel  E.  Two Essays on Bankruptcy and One Essay on Gender Differences in Academic Publishing [doctoral thesis]. Cambridge, UK: University of Cambridge; 2015.
47.
Merton  RK.  The Matthew effect in science: the reward and communication systems of science are considered.  Science. 1968;159(3810):56-63. doi:10.1126/science.159.3810.56Google ScholarCrossref
48.
Bol  T, de Vaan  M, van de Rijt  A.  The Matthew effect in science funding.  Proc Natl Acad Sci U S A. 2018;115(19):4887-4890. doi:10.1073/pnas.1719557115PubMedGoogle ScholarCrossref
49.
Rossiter  MW.  The Matthew Matilda effect in science.  Soc Stud Sci. 1993;23(2):325-341. doi:10.1177/030631293023002004Google ScholarCrossref
50.
Lerchenmueller  MJ, Sorenson  O.  The gender gap in early career transitions in the life sciences.  Res Policy. 2018;47(6):1007-1017. doi:10.1016/j.respol.2018.02.009Google ScholarCrossref
51.
Santamaría  L, Mihaljević  H.  Comparison and benchmark of name-to-gender inference services.  PeerJ Comput Sci. 2018;4:e156. doi:10.7717/peerj-cs.156Google Scholar
52.
Zhang  C, Bu  Y, Ding  Y, Xu  J.  Understanding scientific collaboration: homophily, transitivity, and preferential attachment.  J Assoc Inf Sci Technol. 2018;69(1):72-86. doi:10.1002/asi.23916Google ScholarCrossref
53.
Bates  C, Gordon  L, Travis  E,  et al.  Striving for gender equity in academic medicine careers: a call to action.  Acad Med. 2016;91(8):1050-1052. doi:10.1097/ACM.0000000000001283PubMedGoogle ScholarCrossref
54.
Zou  J, Schiebinger  L.  AI can be sexist and racist: it’s time to make it fair.  Nature. 2018;559(7714):324-326. doi:10.1038/d41586-018-05707-8PubMedGoogle ScholarCrossref
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    Original Investigation
    Medical Journals and Publishing
    October 23, 2019

    Gender Disparities in Invited Commentary Authorship in 2459 Medical Journals

    Author Affiliations
    • 1Harvard T.H. Chan School of Public Health, Harvard University, Boston, Massachusetts
    • 2Elsevier Inc, New York, New York
    • 3Elsevier BV, Amsterdam, the Netherlands
    • 4Harvard Data Science Initiative, Harvard University, Cambridge, Massachusetts
    JAMA Netw Open. 2019;2(10):e1913682. doi:10.1001/jamanetworkopen.2019.13682
    Key Points español 中文 (chinese)

    Question  Is gender associated with authorship of invited commentaries in medical journals among authors with comparable scientific credentials?

    Findings  In this case-control study of invited commentaries published in 2459 journals from January 1, 2013, through December 31, 2017, the odds of authoring an invited commentary were 21% lower for women compared with men who had similar fields of expertise and publication metrics among researchers who had been actively publishing for the median of 19 years.

    Meaning  Women had lower odds of authoring invited commentaries in medical journals compared with men with similar scientific expertise, seniority, and publication metrics.

    Abstract

    Importance  In peer-reviewed medical journals, authoring an invited commentary on an original article is a recognition of expertise. It has been documented that women author fewer invited publications than men do. However, it is unknown whether this disparity is due to gender differences in characteristics that are associated with invitations, such as field of expertise, seniority, and scientific output.

    Objective  To estimate the odds ratio (OR) of authoring an invited commentary for women compared with men who had similar expertise, seniority, and publication metrics.

    Design, Setting, and Participants  This matched case-control study included all medical invited commentaries published from January 1, 2013, through December 31, 2017, in English-language medical journals and multidisciplinary journals. Invited commentaries were defined as publications that cite another publication within the same journal volume and issue. Bibliometric data were obtained from Scopus. Cases were defined as corresponding authors of invited commentaries in a given journal during the study period. Controls were matched to cases based on scientific expertise by calculating a similarity index for abstracts published during the same period using natural language processing. Data analyses were conducted from March 13, 2019, through May 3, 2019.

    Exposure  Corresponding or sole author gender was predicted from author first name and country of origin using genderize.io.

    Main Outcomes and Measures  The OR for gender was estimated after adjusting for field of expertise, publication output, citation impact, and years active (ie, years since first publication), with an interaction between gender and years active.

    Results  The final data set included 43 235 cases across 2549 journals; there were 34 047 unique intraciting commentary authors, among whom 9072 (26.6%) were women. For researchers who had been active for the median of 19 years, the odds of invited commentary authorship were 21% lower for women (OR, 0.79 [95% CI, 0.77-0.81]; P < .001) compared with men who had similar scientific expertise, number of publications, and citation impact. For every decile increase in years active, the OR decreased by a factor of 0.97 (95% CI, 0.96-0.98; P < .001).

    Conclusions and Relevance  In this case-control study, women had lower odds of authoring invited commentaries than their male peers. This disparity was larger for senior researchers. Journal editors could use natural language processing of published research to widen and diversify the pool of experts considered for commentary invitations.

    Introduction

    Women are underrepresented as authors of peer-reviewed publications across a wide variety of scientific fields,1,2 including many areas of medical research3-8 and nonresearch medical publications.9 According to a 2018 study by Holman et al,10 most disciplines are years or decades from achieving gender parity. The gender gap is widest at top-tier journals and for prestigious authorship positions.10-14 Bibliometric analyses suggest that women review papers less frequently than male peers,15 are cited less often,11-13,16 and are underrepresented on editorial boards.13,17,18 Single-journal analyses of acceptance rates after peer review have often found no evidence for gender differences,19-24 although imbalances favoring men have been observed in larger data sets.25,26

    In medical journals, publications that have been solicited by the editors are a recognition of expertise and can raise the author’s profile. Common types of solicited publications include commentaries, editorials, and reviews. The proportion of women authoring invited articles is low22,25,27-30 even when compared with the proportion of female authors of all articles.10 The probability of authoring an invited publication depends on scientific field, seniority, and publication record, factors that are also associated with gender10,31; these associations may explain the disparity in solicited publication authorship. However, to our knowledge, previous studies of gender and invited article authorship did not control for these factors.10,22,25,27-30 This raises the question of whether women author fewer invited articles than men with similar scientific expertise and credentials.

    In this case-control study, we estimated the association of gender with authorship of medical invited commentary articles in peer-reviewed medical and interdisciplinary journals, controlling for field of expertise, author seniority, and publication metrics. This was not a descriptive analysis of gender disparities; we investigated whether gender differences in invited commentary authorship were reduced or eliminated after adjustment for key author characteristics. We obtained journal-specific and pooled estimates and investigated modifiers of the association of gender with invited commentary authorship, including author seniority, journal topic, and journal impact.

    Methods

    This study did not meet the definition of human subjects research under US federal regulation 45 CFR§46.102(e); therefore, institutional review board oversight was not required. We observed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines for case-control studies.

    Data

    We extracted bibliometric data from Scopus,32 a comprehensive online database of peer-reviewed abstracts and citations. Scopus has been shown to be a reliable source for studying life sciences publication patterns.33

    Outcome

    The outcome was corresponding authorship on an invited commentary article. To identify invited commentaries, we searched for intraciting commentary (ICC) articles, defined as publications that cite 1 or more other publications (the focal article) in the same journal volume and issue. Articles with missing volume or issue number were excluded. We restricted our analysis to ICCs for 3 reasons: (1) ICCs are a common type of invited commentary in medical journals; (2) ICCs are almost certainly solicited by journal editors, since citing another publication in the same journal volume and issue requires advance knowledge of the focal article and its publication timeline; and (3) ICCs can be identified through automated search in Scopus. Although other types of solicited publications exist (eg, invited reviews, opinion articles), no consistent scheme exists for identifying these publications across journals, to our knowledge.

    We considered corresponding authors rather than first or last authors under the assumption that the corresponding author received the original commentary invitation. Sole authors were assumed to be the corresponding author. Multiauthor articles with no corresponding author were excluded.

    Exposure

    The exposure was author gender, treated for the purpose of this analysis as binary (ie, male or female). Gender was inferred from author first name and country of origin using the software service genderize.io34 and several supplementary steps (eAppendix 1 in the Supplement). This process returned 1 of 3 results: male, female, or unknown.

    Journal and Article Inclusion Criteria

    Inclusion and exclusion criteria are shown in Figure 1. We included journals that publish in English and were listed as active in Scopus at data extraction on December 5, 2018. To ensure our analysis reflected current practice, we included only ICCs published from January 1, 2013, through December 31, 2017. Data analyses were conducted from March 13, 2019, to May 3, 2019.

    We restricted our analysis to medical journals (All Science Journal Classification [ASJC] codes starting with 27) and multidisciplinary journals (ASJC code 1000). For medical journals, we included all ICC articles. For multidisciplinary journals, we included only medical ICC articles (article ASJC codes starting with 27); these publications were included to capture medical research published in high-profile general science journals.

    Sample size was determined by the number of journals and ICCs in those journals meeting our inclusion criteria (Figure 1).

    Study Design

    We used a matched case-control study design to estimate the odds ratio (OR), γj, of authoring an ICC in journal j for women compared with men. Cases were defined as corresponding or sole authors of 1 or more ICC article published in j from January 1, 2013, through December 31, 2017. Potential controls were authors of at least 5 articles (any authorship position) published in any Scopus-indexed journal from January 1, 2013, through December 31, 2017, who were not also cases for journal j. Case authors could act as controls for ICC authors published in a different journal but could not act as controls for other case authors within the same journal. Thus, each matched set corresponded to a unique corresponding author-journal pair, so that authors of ICCs in m > 1 distinct journals during the study period contributed m cases to the data set.

    We controlled for key author-level factors, such as field of expertise, seniority, number of publications, and citation record, that may have been associated with gender and ICC authorship. Gender ratios can vary across specific fields of expertise within medical disciplines,1 and we hypothesized that some fields would attract more ICCs. We further hypothesized that ICCs were most likely to be authored by senior researchers with strong publication records, factors also associated with gender.

    Control for Author-Level Factors

    We matched controls to cases on field of expertise. This was achieved by generating an expertise profile for each author via natural language processing of all abstracts published January 1, 2013, through December 31, 2017, in Scopus-indexed journals (eFigure 1 and eFigure 2 in the Supplement). We compared the expertise profiles of all Scopus-indexed authors with each case author and generated a similarity index for each pair. Full details on our matching method are presented in the eAppendix 1 in the Supplement. The top 50 matches based on this index were potential controls for each case author. To ensure a minimum level of match quality, we pooled potential controls for all cases, ranked on expertise similarity to their respective cases, and excluded the overall lowest quartile. The top 10 remaining controls per case were included in our final data set.

    We also controlled for years active (a measure of seniority, defined as years since first publication), h-index (a measure of author citation impact), and number of publications. Percentile ranks of these variables were computed and included in our regression models. Percentile ranks were used to reduce the effect of outliers on our data.

    Statistical Analysis

    First, we concatenated the data for all journals, including those with sample sizes too small for journal-specific estimation, and estimated the overall OR using conditional logistic regression. Authors whose gender could not be inferred were excluded from all analyses. We ran 5 conditional logistic regression models including the following independent variables: (1) only gender; (2) gender and percentiles of years active, h-index, and number of publications, allowing for nonlinearity using natural cubic splines with internal knots at 0.25, 0.50, and 0.75; and (3) 3 separate models adding an interaction between gender and the linear term for each continuous variable. All models controlled for field of expertise through matching.

    Next, we estimated γj, the OR for journal j, separately for each journal with sufficient data for an individual estimate. Owing to smaller sample sizes, we controlled for continuous variables using polynomial terms rather than splines (eAppendix 1 in the Supplement).

    To investigate the robustness of our overall OR estimate to potential sources of error, we performed 3 sensitivity analyses (eAppendix 1 in the Supplement). First, we used random effects meta-analysis to account for possible between-journal heterogeneity in the OR.35 Second, we used multiple imputation to account for missing gender. Third, we increased the stringency of criteria for matching on field expertise, excluded ICCs that may have been replies to other articles, and excluded duplicate records for authors of ICCs in multiple journals.

    To examine possible sources of heterogeneity in the magnitude of the association, we performed 2 secondary analyses. First, to investigate whether γj was associated with the journal citation impact as measured by 2016 CiteScore (mean number of citations received from 2014 through 2016 for all items published from 2014 through 2016), we included an interaction between gender and CiteScore in our pooled conditional logistic regression models. We allowed for nonlinearity using natural cubic splines with external knots at the maximum and minimum CiteScore in our data set and 2 internal knots at the tertiles of ICC source journal CiteScore. We tested the null hypothesis of no interaction between gender and CiteScore using a likelihood ratio test.

    Second, we investigated whether log(γj) varied across medical journal topics as defined by ASJC codes. Scopus-indexed journals are assigned 1 or more ASJC code according to the scientific content of a typical volume. We concatenated the data for sets of journals with each ASJC code to obtain a topic-specific estimate. Topics were overlapping because many journals have multiple codes; we were therefore unable to perform statistical tests for differences between topics. All analyses were performed in R statistical software version 3.4.2 (R Project for Statistical Computing). P values were 2-sided, and statistical significance was set at .05.

    Results

    Deidentified matched data and code are available elsewhere.36 We identified 72 038 ICCs that met our inclusion criteria (Figure 1). Gender could be inferred for corresponding authors of 62 567 ICCs (86.9%), 14 140 of whom were female (22.6%). eTable 1 in the Supplement presents characteristics of male, female, and unknown gender authors.

    After removing authors with missing gender and other processing steps, our analysis included 2549 journals with a total of 46 235 matched sets (Figure 1). The median (interquartile range [IQR]) number of controls per matched set was 8 (5-9) articles. For 11 047 matched sets with female case authors, the median (IQR) percentage of female controls was 40.0% (22.2%-55.6%). For 35 188 matched sets with male case authors, the median (IQR) percentage of female controls was 25.0% (11.1%-40.0%).

    Our data set included 34 047 unique ICC authors, among whom 9072 (26.6%) were female. There were 7721 ICC authors (22.7%) who had published in more than 1 journal. Compared with authors who served only as controls, authors who served only as cases or as both cases and controls had higher median (IQR) years since first publication (control only: 17 [11-27] years; case only: 22 [14-32] years; case and control: 24 [16-33] years), numbers of publications (control only: 60 [30-119] publications; case only: 77 [41-142] publications; case and control: 147 [81-257] publications), and h-indices (case only: 17 [10-29]; control only: 22 [12-36]; case and control: 33 [21-50]) (Table). We had missing data for 1 or more of these variables for 134 controls (0.1%) and no cases.

    Association of Gender With ICC Authorship

    Controlling for field of expertise only, the estimated OR for women compared with men was 0.70 (95% CI, 0.68-0.72; P < .001). After adjustment for percentiles of years active, h-index, and number of publications, the OR estimate was 0.78 (95% CI, 0.76-0.80; P < .001). Full results of these analyses are presented in eTable 2 and eFigure 3 in the Supplement.

    After including an interaction between gender and years active, an increase of 1 decile in years active was associated with a decrease in OR by a factor of 0.97 (95% CI, 0.96-0.98; P < .001). Full results of these analyses are presented in eTable 3 and eFigure 4 in the Supplement. The estimated OR among researchers who had been actively publishing for approximately 8 years (ie, 10th percentile) was 0.90 (95% CI, 0.86-0.94; P < .001) (Figure 2). At 19 years of actively publishing (50th percentile), the OR was reduced to 0.79 (95% CI, 0.77-0.81; P < .001). At 38 years of actively publishing (90th percentile), the OR was 0.69 (95% CI, 0.66-0.72; P < .001) (Figure 2).

    Results were similar in models including interactions for gender with h-index and gender with number of publications (eTable 4, eTable 5, and eFigures 5-7 in the Supplement). This result is likely associated with the high pairwise correlation between years active, h-index, and number of publications (Spearman correlation coefficient: years active and h-index, 0.72; years active and number of publications, 0.70; h-index and number of publications, 0.91).

    There were 1410 journals (55.3%) with sample sizes large enough for journal-specific OR estimates. These analyses are presented in eAppendix 3 in the Supplement, including yearly ICC numbers alongside journal-specific ORs.

    Sensitivity Analyses

    Point estimates and 95% CIs from sensitivity analyses were close to those from the main analyses. These analyses are presented in the eAppendix 2, eTables 6-8, eFigure 8, and eFigure 9 in the Supplement.

    Interactions With Journal Topic and CiteScore

    Subgroup analyses by journal topic yielded statistically significant adjusted ORs favoring male authors for 38 of 48 topics (79%) (Figure 3). The multidisciplinary journals (ASJC code 1000), which had a pooled OR of 0.96 (95% CI, 0.73-1.27), were a notable exception.

    The association of OR with journal CiteScore was nonmonotonic (Figure 4) and statistically significant (adjusted for field of expertise: χ24, 35.1; P < .001; fully adjusted: χ24, 35.2; P < .001). However, the odds favored men at all CiteScores (Figure 4). One journal with an outlier CiteScore was excluded (CA: A Cancer Journal for Clinicians; CiteScore = 89.23).

    Discussion

    In this matched case-control study, we estimated the odds ratio of authoring an invited commentary in a medical journal for women compared with men who had similar scientific expertise, seniority, and publication metrics. We observed a statistically significant association in the odds for female researchers, which were approximately 21% lower than for their male peers. This result challenges the common assumption that gender disparities in invited article authorship can be explained by greater publication success, seniority, or self-selection into competitive fields among male scientists. The gender disparity in our study was larger for researchers who had been actively publishing for longer, increasing by appoximately 3% for every 1-decile increase in years active (Figure 2). The odds of invited commentary authorship consistently favored men for journals across the spectrum of citation impact (Figure 4) and for most medical subdisciplines (Figure 3). These results may be generalizable to English-language medical journals, but gender disparities can vary geographically.16 We obtained individual estimates for 1410 journals. Deidentified, matched data, and code for analyses are available elsewhere.36

    Only 23% of invited commentaries in our data set had women corresponding authors, which was consistent with the literature on gender disparities in solicited article authorship.10,22,25,27,28 Our study suggested that this gap may be explained by gender-dependent factors unrelated to field of expertise, seniority, or publication impact, characteristics we expected to motivate commentary invitations. One possible explanation is that women declined invitations more frequently. Previous studies have suggested that women academics have lower access to resources,37 experience greater role overload,38 and take on a higher burden of nonresearch tasks.39,40 These factors could reduce their capacity or willingness to accept commentary invitations. Women scientists may also decline invitations more than men owing to lower self-perceived expertise. For future research, journals could keep records of invitations to establish whether there are gender differences in acceptance rates.

    Another possible explanation for our results was the presence of gender bias in commentary invitations from publishing gatekeepers. Overt discrimination against women scientists persists in some settings,41 and its effect cannot be ruled out. However, gender bias often takes subtler forms. Unconscious gender bias favoring men has been shown to affect evaluations of scientific competency42,43 and manuscript quality44-46 in some contexts. Men editors may have favored men authors owing to same-gender preferences in professional networks.13,26 Combined with the overrepresentation of men editors,13,17,18 this could have boosted invited authorship by men. Additionally, some journals invite reviewers of research articles to write commentaries. Since women review research articles less frequently,15 this mechanism may also have favored men. Many large journals dedicate an editor to commentaries and editorials, while smaller journals often assign this role to editors with less time to devote to the task. Time-pressured editors may rely more on professional contacts, reviewers, or instinctive impressions to select commentary authors, increasing vulnerability to bias. This difference could help explain the slightly smaller gender disparity for large, high-impact journals observed in our study (Figure 4).

    In our study, gender disparities in invited commentary authorship increased with author seniority and publication metrics. We offer 2 interpretations for this finding. First, commentary authors may have been more likely to be solicited for future articles, causing gender differences among junior researchers to increase through time. This phenomenon, whereby success begets success, is known as the Matthew effect47 and has been observed in science funding outcomes.48 Second, the benefits of seniority and high-impact publications may have been greater for men than for women. The systematic underrecognition of women scientists’ work, known as the Matilda effect,49 has also manifested in lower transition rates to principal investigator status for women compared with men who have similar publication records.50

    Strengths and Limitations

    Our study has several strengths. First, we controlled for field of expertise through matching, an approach that is insensitive to modeling assumptions, and for years active and publication metrics using flexible nonlinear methods. This adjustment allowed us to establish that gender disparities in invited commentary authorship could not be explained by gender differences in these variables. To our knowledge, this is the first study of gender in academic publishing to control for field of expertise using natural language processing. Most previous studies of the association of gender with authorship in medical journals, either invited or contributed, did not control for author-level covariates.8,10-14,22,28,29 Some medical specialty–specific studies applied basic controls for a small number of author-level variables.3-7 Second, using ICC articles as our outcome allowed us to study a particular type of solicited article for a large number of medical journals and with a large sample size. Third, to our knowledge, our study is the first to provide journal-specific estimates of gender disparities for a large number of medical journals using consistent methods alongside all-journal averages. Journal-specific estimates are important for editorial boards seeking to address gender bias, and consistent methods are needed for between-journal comparisons. Overall estimates are required to obtain a broad view of the field.

    Our study also had some limitations. First, we used genderize.io34 to infer the corresponding author’s gender from first name and country of origin. Genderize.io performs well compared with similar software,51 although all such software is subject to error. Gender could not be inferred for 21% of unique authors. However, multiple imputation analyses did not change our conclusions. We analyzed gender for corresponding authors rather than other authorship positions, assuming that the corresponding author received the commentary invitation. This assumption may not hold for all articles.

    Second, we used published abstracts to match fields of expertise. Publications are not the only measure of expertise, and this matching process was imperfect. Anecdotally, controls were often repeat coauthors of cases. Due to gender homophily in coauthorship,52 this pattern would likely artificially lower gender diversity within matched sets, attenuating our OR estimates toward the null. We also controlled for publication metrics and years since first publication as a measure of seniority. Publications and citations may have accrued in an environment in which women were systematically disadvantaged.53 Women scientists may therefore have been more qualified in general to provide commentary in their respective field than men who were similar on these measures, biasing the estimated OR in favor of women.

    Third, our outcome included only ICCs; we did not expect these to be systematically different from other invited commentaries. However, ICCs may have been (1) written by in-house writers, including editors; (2) volunteered by reviewers of research articles; or (3) replies to other articles. While authors of these article types were preselected by the journal, we could not know whether they were invited. We could not ascertain the prevalence of types 1 and 2 with the available data. Removing possible reply articles (type 3) had minimal effect on our results. We excluded articles with missing corresponding author, journal volume, or journal issue information. This administrative missingness was unlikely to bias our overall results. However, it may have rendered some journal-specific estimates unreliable, as missingness varied by year for some journals, likely due to changes at the journal or in Scopus recording methods.

    Fourth, our statistical analysis did not account for correlation due to repeated observations from the same author or journal. However, sensitivity analyses suggested this affected our results minimally.

    Conclusions

    This study suggests that women had lower odds of authoring invited commentaries than men with similar expertise and scientific standing. Invited commentaries confer career advantages on the author by providing exposure and fostering professional connections with editors. These benefits may accumulate if invited article authors are more likely to be solicited for future pieces. Extending article invitations to researchers primarily based on seniority, perceived prestige, and professional connections may contribute to male scientists’ entrenched advantage and compound gender inequity. Our results suggested that female talent was underused. As such, increasing gender diversity in solicited article authorship may increase the quality of these articles.

    Our study indicated one possible way forward: journals could use natural language processing of bibliographic databases to identify experts on invited paper topics. This approach could alert editors to highly qualified author candidates they otherwise might not have considered. Our study provides proof-of-concept for this proposal because we used natural language processing to identify thousands of women with relevant expertise on previous commentary topics and showed that they were underrepresented as commentary authors compared with their male peers. Future work should focus on developing fair, data-driven approaches to expert-finding that avoid propagating biases in bibliographic data.54 If successfully implemented, this strategy could diversify professional networks, strengthen the author talent pool, and provide opportunities to a wider group of scientists.

    Back to top
    Article Information

    Accepted for Publication: September 2, 2019

    Published: October 23, 2019. doi:10.1001/jamanetworkopen.2019.13682

    Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2019 Thomas EG et al. JAMA Network Open.

    Corresponding Author: Francesca Dominici, PhD, Harvard Data Science Initiative, Harvard University, 8 Story St, Ste 380, Cambridge, MA 02138 (fdominic@hsph.harvard.edu).

    Author Contributions: Ms Thomas and Dr Dominici had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

    Concept and design: Thomas, Jayabalasingham, Collins, Dominici.

    Acquisition, analysis, or interpretation of data: All authors.

    Drafting of the manuscript: Thomas, Jayabalasingham, Collins, Dominici.

    Critical revision of the manuscript for important intellectual content: All authors.

    Statistical analysis: Thomas, Dominici.

    Administrative, technical, or material support: Jayabalasingham, Collins, Dominici.

    Supervision: Dominici.

    Conflict of Interest Disclosures: Drs Jayabalsingham, Collins, Geertzen, and Bui are affiliated with Elsevier, the publisher of some of the journals included in this study. Dr Dominici reported that Elsevier has given a gift to the Harvard Data Science Initiative of which Dr Dominici is a codirector. No other disclosures were reported.

    Meeting Presentation: This article was presented at the Women in Statistics and Data Science Conference of the American Statistical Association; October 4, 2019; Bellevue, Washington.

    Additional Contributions: Elsevier provided access to their data but was not compensated for access. Lena Goodwin, MA (freelance editor), edited and refined drafts of this work and was compensated by the Harvard Data Science Initiative.

    References
    1.
    West  JD, Jacquet  J, King  MM, Correll  SJ, Bergstrom  CT.  The role of gender in scholarly authorship.  PLoS One. 2013;8(7):e66212. doi:10.1371/journal.pone.0066212PubMedGoogle Scholar
    2.
    Zeng  XHT, Duch  J, Sales-Pardo  M,  et al.  Differences in collaboration patterns across discipline, career stage, and gender.  PLoS Biol. 2016;14(11):e1002573. doi:10.1371/journal.pbio.1002573PubMedGoogle Scholar
    3.
    Ahmadi  M, Khurshid  K, Sanelli  PC,  et al.  Influences for gender disparity in academic neuroradiology.  AJNR Am J Neuroradiol. 2018;39(1):18-23. doi:10.3174/ajnr.A5443PubMedGoogle ScholarCrossref
    4.
    Chauvin  S, Mulsant  BH, Sockalingam  S, Stergiopoulos  V, Taylor  VH, Vigod  SN.  Gender differences in research productivity among academic psychiatrists in Canada.  Can J Psychiatry. 2019;64(6):415-422. doi:10.1177/0706743718802798PubMedGoogle ScholarCrossref
    5.
    Diamond  SJ, Thomas  CR  Jr, Desai  S,  et al.  Gender differences in publication productivity, academic rank, and career duration among US academic gastroenterology faculty.  Acad Med. 2016;91(8):1158-1163. doi:10.1097/ACM.0000000000001219PubMedGoogle ScholarCrossref
    6.
    Mayer  EN, Lenherr  SM, Hanson  HA, Jessop  TC, Lowrance  WT.  Gender differences in publication productivity among academic urologists in the United States.  Urology. 2017;103:39-46. doi:10.1016/j.urology.2016.12.064PubMedGoogle ScholarCrossref
    7.
    Nguyen  V, Marmor  RA, Ramamoorthy  SL, Blair  SL, Clary  BM, Sicklick  JK.  Academic surgical oncologists’ productivity correlates with gender, grant funding, and institutional NCI comprehensive cancer center affiliation.  Ann Surg Oncol. 2018;25(7):1852-1859. doi:10.1245/s10434-018-6450-5PubMedGoogle ScholarCrossref
    8.
    Pfeiffer  M, Fischer  MR, Bauer  D.  Publication activities of German junior researchers in academic medicine: which factors impact impact factors?  BMC Med Educ. 2016;16(1):190. doi:10.1186/s12909-016-0712-3PubMedGoogle ScholarCrossref
    9.
    Hsiehchen  D, Hsieh  A, Espinoza  M.  Prevalence of female authors in case reports published in the medical literature.  JAMA Netw Open. 2019;2(5):e195000-e195000. doi:10.1001/jamanetworkopen.2019.5000PubMedGoogle ScholarCrossref
    10.
    Holman  L, Stuart-Fox  D, Hauser  CE.  The gender gap in science: how long until women are equally represented?  PLoS Biol. 2018;16(4):e2004956. doi:10.1371/journal.pbio.2004956PubMedGoogle Scholar
    11.
    Bendels  MHK, Dietz  MC, Brüggmann  D, Oremek  GM, Schöffel  N, Groneberg  DA.  Gender disparities in high-quality dermatology research: a descriptive bibliometric study on scientific authorships.  BMJ Open. 2018;8(4):e020089. doi:10.1136/bmjopen-2017-020089PubMedGoogle Scholar
    12.
    Bendels  MHK, Müller  R, Brueggmann  D, Groneberg  DA.  Gender disparities in high-quality research revealed by Nature Index journals.  PLoS One. 2018;13(1):e0189136. doi:10.1371/journal.pone.0189136PubMedGoogle Scholar
    13.
    Helmer  M, Schottdorf  M, Neef  A, Battaglia  D.  Gender bias in scholarly peer review.  Elife. 2017;6:e21718. doi:10.7554/eLife.21718PubMedGoogle Scholar
    14.
    Filardo  G, da Graca  B, Sass  DM, Pollock  BD, Smith  EB, Martinez  MA-M.  Trends and comparison of female first authorship in high impact medical journals: observational study (1994-2014).  BMJ. 2016;352:i847. doi:10.1136/bmj.i847PubMedGoogle ScholarCrossref
    15.
    Lerback  J, Hanson  B.  Journals invite too few women to referee.  Nature. 2017;541(7638):455-457. doi:10.1038/541455aPubMedGoogle ScholarCrossref
    16.
    Larivière  V, Ni  C, Gingras  Y, Cronin  B, Sugimoto  CR.  Bibliometrics: global gender disparities in science.  Nature. 2013;504(7479):211-213. doi:10.1038/504211aPubMedGoogle ScholarCrossref
    17.
    Ioannidou  E, Rosania  A.  Under-representation of women on dental journal editorial boards.  PLoS One. 2015;10(1):e0116630. doi:10.1371/journal.pone.0116630PubMedGoogle Scholar
    18.
    Topaz  CM, Sen  S.  Gender representation on journal editorial boards in the mathematical sciences.  PLoS One. 2016;11(8):e0161357. doi:10.1371/journal.pone.0161357PubMedGoogle Scholar
    19.
    Buckley  HL, Sciligo  AR, Adair  KL, Case  BS, Monks  JM.  Is there gender bias in reviewer selection and publication success rates for the New Zealand Journal of Ecology N Z J Ecol. 2014;38(2):335-339.Google Scholar
    20.
    Edwards  HA, Schroeder  J, Dugdale  HL.  Gender differences in authorships are not associated with publication bias in an evolutionary journal.  PLoS One. 2018;13(8):e0201725. doi:10.1371/journal.pone.0201725PubMedGoogle Scholar
    21.
    Fox  CW, Burns  CS, Muncy  AD, Meyer  JA.  Gender differences in patterns of authorship do not affect peer review outcomes at an ecology journal.  Funct Ecol. 2016;30(1):126-139. doi:10.1111/1365-2435.12587Google ScholarCrossref
    22.
    Heckenberg  A, Druml  C.  Gender aspects in medical publication: the Wiener klinische Wochenschrift.  Wien Klin Wochenschr. 2010;122(5-6):141-145. doi:10.1007/s00508-010-1302-8PubMedGoogle ScholarCrossref
    23.
    Lane  J, Linden  DJ.  Is there gender bias in the peer review process at Journal of Neurophysiology J Neurophysiol. 2009;101(5). doi:10.1152/jn.00196.2009Google Scholar
    24.
    Primack  RB, Ellwood  E, Miller-Rushing  AJ, Marrs  R, Mulligan  A.  Do gender, nationality, or academic age affect review decisions: an analysis of submissions to the journal Biological Conservation Biol Conserv. 2009;142(11):2415-2418. doi:10.1016/j.biocon.2009.06.021Google ScholarCrossref
    25.
    Berg  J.  Examining author gender data.  Science. 2019;363(6422):7. doi:10.1126/science.aaw4633PubMedGoogle ScholarCrossref
    26.
    Murray  D, Siler  K, Larivière  V,  et al.  Gender and international diversity improves equity in peer review.  Preprint. Posted online August 29, 2019. bioRxiv:400515. doi:10.1101/400515
    27.
    Conley  D, Stadmark  J.  Gender matters: a call to commission more women writers.  Nature. 2012;488(7413):590. doi:10.1038/488590aPubMedGoogle ScholarCrossref
    28.
    Süßenbacher  S, Amering  M, Gmeiner  A, Schrank  B.  Gender-gaps and glass ceilings: a survey of gender-specific publication trends in Psychiatry between 1994 and 2014.  Eur Psychiatry. 2017;44:90-95. doi:10.1016/j.eurpsy.2017.03.008PubMedGoogle ScholarCrossref
    29.
    Qureshi  R, Lê  J, Li  T, Ibrahim  M, Dickersin  K.  Gender and editorial authorship in high impact epidemiology journals [published online April 17, 2019].  Am J Epidemiol. doi:10.1093/aje/kwz094PubMedGoogle Scholar
    30.
    Silver  JK, Poorman  JA, Reilly  JM, Spector  ND, Goldstein  R, Zafonte  RD.  Assessment of women physicians among authors of perspective-type articles published in high-impact pediatric journals.  JAMA Netw Open. 2018;1(3):e180802-e180802. doi:10.1001/jamanetworkopen.2018.0802PubMedGoogle ScholarCrossref
    31.
    Carr  PL, Raj  A, Kaplan  SE, Terrin  N, Breeze  JL, Freund  KM.  Gender differences in academic medicine: retention, rank, and leadership comparisons from the National Faculty Survey.  Acad Med. 2018;93(11):1694-1699. doi:10.1097/ACM.0000000000002146PubMedGoogle ScholarCrossref
    32.
    Elsevier.  What is Scopus Preview?https://service.elsevier.com/app/answers/detail/a_id/15534/supporthub/scopus/#tips. Accessed March 25, 2019.
    33.
    Harzing  A-W, Alakangas  S.  Google Scholar, Scopus and the Web of Science: a longitudinal and cross-disciplinary comparison.  Scientometrics. 2016;106(2):787-804. doi:10.1007/s11192-015-1798-9Google ScholarCrossref
    34.
    genderize.io.  Determine the gender of a first name. https://genderize.io/. Accessed March 28, 2019.
    35.
    La Gamba  F, Corrao  G, Romio  S,  et al.  Combining evidence from multiple electronic health care databases: performances of one-stage and two-stage meta-analysis in matched case-control studies.  Pharmacoepidemiol Drug Saf. 2017;26(10):1213-1219. doi:10.1002/pds.4280PubMedGoogle ScholarCrossref
    36.
     Data set for investigation of gender disparities in invited commentary authorship for medical journals. https://github.com/emgthomas/gender_and_invited_commentaries. Accessed September 9, 2019.
    37.
    Westring  A, McDonald  JM, Carr  P, Grisso  JA.  An integrated framework for gender equity in academic medicine.  Acad Med. 2016;91(8):1041-1044. doi:10.1097/ACM.0000000000001275PubMedGoogle ScholarCrossref
    38.
    Ellinas  EH, Kaljo  K, Patitucci  TN, Novalija  J, Byars-Winston  A, Fouad  NA.  No room to “lean in”: a qualitative study on gendered barriers to promotion and leadership.  J Womens Health (Larchmt). 2019;28(3):393-402. doi:10.1089/jwh.2018.7252PubMedGoogle ScholarCrossref
    39.
    Babcock  L, Recalde  MP, Vesterlund  L, Weingart  L.  Gender differences in accepting and receiving requests for tasks with low promotability.  Am Econ Rev. 2017;107(3):714-747. doi:10.1257/aer.20141734Google ScholarCrossref
    40.
    Guarino  CM, Borden  VM.  Faculty service loads and gender: are women taking care of the academic family?  Res High Educ. 2017;58(6):672-694. doi:10.1007/s11162-017-9454-2Google ScholarCrossref
    41.
    Gupta  N, Kemelgor  C, Fuchs  S, Etzkowitz  H.  The “Triple Burden”: a cross-cultural analysis of the consequences of discrimination for women in science.  Curr Sci. 2004;89(8):1382-1386.Google Scholar
    42.
    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(41):16474-16479. doi:10.1073/pnas.1211286109PubMedGoogle ScholarCrossref
    43.
    Quadlin  N.  The mark of a woman’s record: gender and academic performance in hiring.  Am Sociol Rev. 2018;83(2):331-360. doi:10.1177/0003122418762291Google ScholarCrossref
    44.
    Knobloch-Westerwick  S, Glynn  CJ, Huge  M.  The Matilda effect in science communication: an experiment on gender bias in publication quality perceptions and collaboration interest.  Sci Commun. 2013;35(5):603-625. doi:10.1177/1075547012472684Google ScholarCrossref
    45.
    Krawczyk  M, Smyk  M.  Author’s gender affects rating of academic articles: evidence from an incentivized, deception-free laboratory experiment.  Eur Econ Rev. 2016;90:326-335. doi:10.1016/j.euroecorev.2016.02.017Google ScholarCrossref
    46.
    Hengel  E.  Two Essays on Bankruptcy and One Essay on Gender Differences in Academic Publishing [doctoral thesis]. Cambridge, UK: University of Cambridge; 2015.
    47.
    Merton  RK.  The Matthew effect in science: the reward and communication systems of science are considered.  Science. 1968;159(3810):56-63. doi:10.1126/science.159.3810.56Google ScholarCrossref
    48.
    Bol  T, de Vaan  M, van de Rijt  A.  The Matthew effect in science funding.  Proc Natl Acad Sci U S A. 2018;115(19):4887-4890. doi:10.1073/pnas.1719557115PubMedGoogle ScholarCrossref
    49.
    Rossiter  MW.  The Matthew Matilda effect in science.  Soc Stud Sci. 1993;23(2):325-341. doi:10.1177/030631293023002004Google ScholarCrossref
    50.
    Lerchenmueller  MJ, Sorenson  O.  The gender gap in early career transitions in the life sciences.  Res Policy. 2018;47(6):1007-1017. doi:10.1016/j.respol.2018.02.009Google ScholarCrossref
    51.
    Santamaría  L, Mihaljević  H.  Comparison and benchmark of name-to-gender inference services.  PeerJ Comput Sci. 2018;4:e156. doi:10.7717/peerj-cs.156Google Scholar
    52.
    Zhang  C, Bu  Y, Ding  Y, Xu  J.  Understanding scientific collaboration: homophily, transitivity, and preferential attachment.  J Assoc Inf Sci Technol. 2018;69(1):72-86. doi:10.1002/asi.23916Google ScholarCrossref
    53.
    Bates  C, Gordon  L, Travis  E,  et al.  Striving for gender equity in academic medicine careers: a call to action.  Acad Med. 2016;91(8):1050-1052. doi:10.1097/ACM.0000000000001283PubMedGoogle ScholarCrossref
    54.
    Zou  J, Schiebinger  L.  AI can be sexist and racist: it’s time to make it fair.  Nature. 2018;559(7714):324-326. doi:10.1038/d41586-018-05707-8PubMedGoogle ScholarCrossref
    ×