1. Medicine

Gender differences in submission behavior exacerbate publication disparities in elite journals

  1. Chaoqun Ni
  2. Isabel Basson
  3. Giovanna Badia
  4. Nathalie Tufenkji
  5. Cassidy R Sugimoto
  6. Vincent Larivière  Is a corresponding author
  1. The Information School, University of Wisconsin-Madison, United States
  2. École de bibliothéconomie et des sciences de l’information, Université de Montréal, Canada
  3. Department of Science and Innovation-National Research Foundation Centre of Excellence in Scientometrics and Science, Technology and Innovation Policy, Stellenbosch University, South Africa
  4. Office of the Dean of Libraries, McGill University, Canada
  5. Department of Chemical Engineering, McGill University, Canada
  6. School of Public Policy, Georgia Institute of Technology, United States
  7. Observatoire des sciences et des technologies, Université du Québec à Montréal, Canada
https://doi.org/10.7554/eLife.90049.4
Share this article
Cite this article
  1. Chaoqun Ni
  2. Isabel Basson
  3. Giovanna Badia
  4. Nathalie Tufenkji
  5. Cassidy R Sugimoto
  6. Vincent Larivière
(2026)
Gender differences in submission behavior exacerbate publication disparities in elite journals
eLife 12:RP90049.
https://doi.org/10.7554/eLife.90049.4
  2. Download BibTeX
  3. Download .RIS

eLife Assessment

This convincing study, which is based on a survey of researchers, finds that women are less likely than men to submit articles to elite journals. It also finds that there is no relation between gender and reported desk rejection. The study is an important contribution to work on gender bias in the scientific literature.

https://doi.org/10.7554/eLife.90049.4.sa0

Abstract

Women are particularly underrepresented as leading authors of papers in journals of the highest impact factor, with substantial consequences for their careers. While a large body of research has focused on the outcome and the process of peer review, fewer articles have explicitly focused on gendered submission behavior and the explanations for these differences. In our study of nearly 5000 active authors, we find that women are less likely to report having submitted papers to journals of the highest impact (e.g., Science, Nature, or PNAS) and to submit fewer manuscripts, on average, than men when they do submit. Women were more likely to indicate that they did not submit their papers (in general and their subsequently most cited papers) to high-impact journals because they were advised not to. In the aggregate, no statistically significant difference was observed between men and women in how they rated the quality of their work. Nevertheless, regardless of discipline, women were more likely than men to indicate that their ‘work was not ground-breaking or sufficiently novel’ as a rationale for not submitting to one of the listed prestigious journals. Men were more likely than women to indicate that the ‘work would fit better in a more specialized journal’. We discuss the implications of these findings and interventions that can serve to mitigate the disparities caused by gendered differences in submission behavior.

Introduction

The rise of the research evaluation system has created a market of intense competition for a few hallowed venues. Scientific capital is strongly concentrated in generalist journals founded in the late 19th and early 20th centuries (Baldwin, 2015), such as Science (1880), Nature (1869), and Proceedings of the National Academy of Sciences of the United States of America (PNAS, 1914). Articles published in these journals garner greater news media attention, and authors tend to receive more career opportunities and grants (Samuel Reich, 2013). Since publishing in highly influential journals leads to wider dissemination and visibility of researchers’ work, the volume of manuscript submissions to these journals is high and acceptance low, with a large proportion of manuscripts rejected by editors without undergoing peer review. It is noted on Science’s author portal that less than 6% of originally submitted papers are accepted (Science, 2025). Nature also notes low acceptance rates on its website, from 8% to 12% depending on the year, and states that ‘many submissions are declined without being sent for review’ (Nature, 2025). A similar practice is espoused by the Proceedings of the National Academy of Sciences of the United States of America (PNAS), with most publications rejected before peer review and a final acceptance rate of 14.7% (PNAS, 2025). Prestige and hierarchy among journals are not inherently negative for scholarly communication; however, if entry into these highly visible venues varies for different sociodemographic populations of authors, it may have adverse consequences for science (Kozlowski et al., 2022).

Editors and peer reviewers evaluate both the quality of manuscripts submitted as well as whether the topic fits within the scope of the journal. Idealized gatekeeping reflects the Mertonian norms of organized skepticism and universalism, evaluating scientific work independently of the social characteristics of authors (Merton, 1968). However, various critiques of scholarly peer review have been discussed (Ware, 2008), among which is a concern of social biases, that is, ‘differential evaluation of an author’s submission as a result of her/his perceived membership in a particular social category’ (Lee et al., 2013). These biases are not necessarily conscious; yet, if no biases were present in the case of peer review, then ‘we should expect the rate with which members of less powerful social groups enjoy successful peer review outcomes to be proportionate to their representation in submission rates’ (Lee et al., 2013). Results on gender disparities and bias in scholarly publishing have arrived at mixed results (Borsuk et al., 2009; Day et al., 2020; Djupe et al., 2019; Edwards et al., 2018; Fox and Paine, 2019; Gilbert et al., 1994; Grossman, 2020; Helmer et al., 2017; Murray et al., 2018; Squazzoni et al., 2021; Walker et al., 2015). However, many of these fail to disentangle components of the publication process, focusing only on the outcome of peer-reviewed publications rather than the effect of different groups having different submission rates or desk rejections. These components have fundamentally different attributes—with one being an effect of self-selection and the other a potential indicator of bias. We emphasize here that observations of disparities are a necessary, but insufficient indicator of bias as there are several other potential explanations for disparities. We focus, therefore, on the concept of disparity.

Most studies focused on disparities in science publishing have demonstrated that women submit fewer papers than men, with little difference in desk rejections (or a disparity in favor of women) (Bendels et al., 2018; Garand and Harman, 2021; Grossman, 2020; Martinsen et al., 2022; Squazzoni et al., 2021). The lower submission rate is particularly pronounced in the most prestigious journals (Bendels et al., 2018). Interviews with women authors explained these lower submission rates with higher selectivity: they argued that they only submit when they are confident the paper would be accepted, not wanting to risk a rejection as they have limited time for conducting research and attending to lengthy reviews (Closa et al., 2020). Women were also less likely than men to indicate that they submit their manuscripts to top journals in their field as their first choice and more likely than men to select a journal based on whether a journal is most likely to accept the manuscript (Djupe et al., 2019). Other perceptions and behavioral differences have also been postulated to account for the differences in manuscript submission behaviors, including risk aversion, perfectionism (Borghans et al., 2009; Closa et al., 2020; Djupe et al., 2019), and assessment of contribution (Lincoln et al., 2012; Wennerås and Wold, 2008). However, few large-scale studies have examined this question from the perspective of the authors. In this context, our paper seeks to address whether the sociodemographic characteristics of lead authors influence submission rates to elite journals. More specifically, based on a survey of more than 4700 active authors (29.9% women), we assess differences in gender, academic rank, and discipline in the likelihood of submitting to Science, Nature, and PNAS and the self-reported reasons for not submitting to these journals.

Results

Journal submission behavior

Women were less likely than men to report that they submitted a paper to Science, Nature, or PNAS (48.7% of men and 37.0% of women, see Appendix 1—table 7). These differences are statistically significant (see Appendix 1—table 8), with men, in general, being more likely to submit to Science, Nature, or PNAS than women while controlling for academic rank (OR = 0.61, 95% CI [0.54, 0.70]). This relationship remains significant for each discipline (see Figure 1), with relatively higher percentages of men in each of the disciplines submitting to these journals (ranging from 44.2% to 53.4%), compared to women (ranging from 33.5% to 37.5%). We also investigated the mean number of manuscripts that respondents reported submitting to Science, Nature, or PNAS (see Figure 1 and Appendix 1—table 9). All disciplines combined, 552 women and 1583 men submitted to these three elite journals. The mean was calculated by investigating only respondents who submitted at least one manuscript to Science, Nature, or PNAS. The number of submissions an author could have submitted depends on the discipline (due to the difference in publication behavior between disciplines) and career status (junior, senior, non-academic). In medical sciences (coefficient = −1.46, 95% CI [−2.14, –0.79], p = 0.00) and natural sciences and engineering (coefficient = −0.80, 95% CI [−1.39, –0.22], p = 0.01) women submitted fewer manuscripts than men. No statistically significant difference was observed for the social sciences (coefficient = −1.35, 95% CI [–3.00, 0.31], p = 0.11, see Appendix 1—table 10).

Figure 1
Download asset Open asset
Measures of submission rate, acceptance rate, and rejection rate with tests of statistical significance.

Logistic regression was used for whether they submitted to an elite journal by individual discipline. Linear regression (mixed-effect model) was used to understand the gender difference in number of submissions, acceptance rate, and desk rejection rate to these journals by individual discipline. We controlled for rank in all regression analysis. Error bars represent the 95% confidence interval. All odds ratio and coefficient values are based on women over men. MS: Medical Sciences; NSE: Natural Sciences and Engineering; SS: Social Sciences. ** indicates p < 0.01, * indicates p < 0.05.

Self-reported acceptance and rejection rate

We examined the difference between women and men regarding the self-reported acceptance and rejection rates for Science, Nature, and PNAS, using linear regression analysis while controlling for academic rank. The result shows no significant difference between women and men in getting their papers accepted (see Figure 1, Appendix 1—table 11, and Appendix 1—table 12). We also used a mixed-effect model to analyze the role of gender in the chances of getting manuscripts accepted, which yields similar conclusions to those above. We defined the desk rejection rate as the number of manuscripts that did not go out for peer review divided by the number of manuscripts submitted for each survey respondent. The relationship between desk rejection rate and gender was examined using linear regression and a mixed-effect model, again controlling for rank. Both models yield similar results: gender did not significantly impact the possibility of getting a desk rejection for manuscripts submitted to these journals, as illustrated in Figure 1 (as well as Appendix 1—table 13 and Appendix 1—table 14).

Submission and desk rejections of authors’ highest cited papers

To understand whether the gender differences observed above also hold for the subset of higher-impact research, we investigated respondents' submission pattern for their most cited papers (not published in Science, Nature, PNAS, but also Cell, Nature Communications, NEJM, and Science Advances; see Materials and methods for more details). Results show that a substantial proportion of the authors (48.6%) considered other journals than the one in which their paper was published, with no significant gender difference (see Appendix 1—table 17 and Appendix 1—table 18). However, only a small proportion of authors (9.1%) considered Science, Nature, or PNAS, again with no significant gender difference (see Appendix 1—table 19 and Appendix 1—table 20). As only a few authors considered publishing in these journals, it is not surprising to see that few authors submitted their most cited paper to Science, Nature, or PNAS (228 respondents), again with no significant gender gap (see Appendix 1—table 21 and Appendix 1—table 22). Similarly, no significant difference was observed in desk rejection rate for their most cited paper between men and women in any of the disciplines when submitting to Science, Nature, and PNAS (see Appendix 1—table 23 and Appendix 1—table 24). Submission behavior and the chance of getting a desk rejection for higher quality papers thus do not seem to differ significantly between men and women (Figure 2), although these conclusions are based on a very limited number of responses: 480 women and 1404 men across all disciplines.

Figure 2
Download asset Open asset
Measures of journal consideration, submission rate, and desk rejection rate for most cited papers with odds ratios (women/men).

Error bars represent the 95% confidence interval. We used logistic regression to analyze the relationship between gender and the four variables, while controlling for rank. Regression was done by each discipline separately. MS: Medical Sciences; NSE: Natural Sciences and Engineering; SS: Social Sciences; SNP: Science, Nature, PNAS.

Perception of the quality of research

Respondents were asked to rate their perception of the quality of their research (not their papers) in comparison to their peers on a five-point Likert scale ranging from Excellent (5) to Poor (1). Most respondents rated their work as good or excellent (86.7%), regardless of discipline, with very few rating their work as poor or fair (1.0%), and only a few respondents rated it as average (12.4%) (Figure 3 and Appendix 1—table 27). When considering responses regardless of discipline, no statistically significant difference was observed between men and women regarding how they rated the quality of their work (see Appendix 1—table 28). When disaggregating the respondents by discipline, it was found that compared with men, women in the natural sciences and engineering are more likely to rank their research quality lower (OR = 0.83, 95% CI [0.67, 0.99], p = 0.04). No statistically significant difference was observed in the medical (OR = 1.18, 95% CI [0.99, 1.40], p = 0.07) and social sciences (OR = 1.13, 95% CI [0.93, 1.22], p = 0.58).

Figure 3
Download asset Open asset
Odds ratio (women/men) and mean value for research quality comparison with peers.

Ordinal logistic regression was used to analyze the relationship between gender and the rating of research quality while controlling for respondents’ rank. Regression was done at the discipline level. Error bars represent the 95% confidence interval. MS: Medical Sciences; NSE: Natural Sciences and Engineering; SS: Social Sciences. * indicates p < 0.05.

Reasons for not submitting to elite journals

We also investigated the authors’ rationale for why they did not submit their manuscripts (in general and their most cited) to the selected journals by asking respondents to select from a list of potential reasons. The results of the regression analysis are represented in Figure 4. None of the respondents with a known academic rank indicated that they did not submit to the journals because they were unaware of the journals. The most common reason men and women gave was that their work would fit better in a more specialized journal.

Figure 4
Download asset Open asset
Reasons for not submitting to top journals, overall and most cited paper, with odds ratios (women/men).

Logistic regression was employed to analyze the relationship between gender and each of the reasons for not submitting. Regression was done by reason and discipline while controlling for rank. Error bars represent the 95% confidence interval. MS: Medical Sciences; NSE: Natural Sciences and Engineering; SS: Social Sciences. ** indicates p < 0.01, * indicates p < 0.05.

Two of the listed reasons are related to the quality and the novelty of the research. Women in natural sciences and engineering were more likely than men to indicate that ‘the work was not of high enough quality’ as one of the reasons why they did not submit to Science, Nature, or PNAS (OR = 1.59, 95% CI [1.22, 2.08], p = 0.001). A similar finding was also found for the most cited paper, although only significant in the medical sciences (OR = 0.75, 95% CI [0.60, 0.94], p = 0.011). Regardless of discipline, women were more likely than men to (OR = 1.26, 95% CI [1.07, 1.50]) indicate that they consider their ‘work was not ground-breaking or sufficiently novel’ for the journal in question. For their most cited papers, it was also the case for authors publishing in natural sciences and engineering (OR = 1.36, 95% CI [1.10, 1.67], p = 0.004), as well as all disciplines combined (OR = 1.19, 95% CI [1.04, 1.37], p = 0.011) (see Appendix 1—table 16 and Appendix 1—table 26).

Reasons related to the scope and audience of the journals mostly show non-significant gender differences, with one exception. In general, men were more likely than women to indicate that the ‘work would fit better in a more specialized journal’ for their general submissions (OR = 0.72, 95% CI [0.60, 0.85], p = 0.000). When disaggregated by discipline, a significant difference was observed both for the medical sciences (OR = 0.74, 95% CI [0.57, 0.95], p = 0.018) and for the natural sciences and engineering (OR = 0.72, 95% CI [0.55, 0.95], p = 0.018). However, no statistically significant difference was observed for the most cited papers. Men in the medical sciences were also more likely than women to indicate that the ‘work would reach a wider audience in another journal’ (OR = 0.61, 95% CI [0.41, 0.92], p = 0.017). No statistically significant difference was observed for the most cited papers. In contrast to the above results, women submitting in the social sciences were more likely to indicate that the ‘work fell out of the scope of the journal’ (OR = 1.84, 95% CI [1.03, 3.28], p = 0.04).

The last two reasons are much less frequent and relate to the wishes of the co-authors and the advice received. In the natural sciences and engineering, women were more likely than men to indicate that their co-authors wished to submit elsewhere (OR = 1.70, 95% CI [1.06, 2.74], p = 0.029). This was also the case for their most cited paper (OR = 1.57, 95% CI [1.02, 2.40], p = 0.039). Without disaggregating by discipline, women were more likely to indicate that they did not submit their papers (in general and their most cited papers) to Science, Nature, or PNAS because they were advised not to. When disaggregating by discipline, this difference was also observed for respondents in natural sciences and engineering for their papers in general (OR = 1.69, 95% CI [1.02, 2.81], p = 0.042) and for medical sciences respondents for their most cited papers (OR = 1.75, 95% CI [1.05, 2.93], p = 0.032).

Discussion

Women are particularly underrepresented in journals with the highest impact factor (Huang et al., 2020), with substantial consequences for their careers. While a large body of research has focused on the outcome and the process of peer review, fewer articles have explicitly focused on gendered submission behavior and the explanations for these differences. In our study of nearly 5000 active authors, we find that women are less likely to report having submitted papers and, when they have, to submit fewer manuscripts, on average, than men. This reinforces much of the literature demonstrating lower submission rates among women (Bendels et al., 2018; Grossman, 2020; Martinsen et al., 2022; Squazzoni et al., 2021). One strategy might be to optimize around homophily: women editors tend to disproportionately select women reviewers (Helmer et al., 2017), and women reviewers are more favorable to women-authored work (Murray et al., 2018). The positive reinforcement in the process will increase women reviewers and authors, which may influence submission behavior on the broader network. Another strategy might be to introduce quotas into peer review; however, this has been shown to have negative effects on women (Leibbrandt et al., 2018).

There is a demonstrated relationship between self-efficacy and publication output (Hemmings and Kay, 2009); however, our results on this were mixed. In the aggregate, no statistically significant difference was observed between men and women in how they rated the quality of their work. A difference was only observed in the natural sciences and engineering, where women ranked their research quality lower. The latter is particularly interesting, given that women tend to outperform in Engineering on several indicators (Ghiasi et al., 2015). The weak signal observed suggests that no particular intervention may be necessary to improve self-efficacy for this population; however, there may be some internalization of value premised on past experiences in peer review. Therefore, addressing larger issues of bias in evaluation (Lee et al., 2013) may serve to mitigate any gendered differences in self-efficacy.

Regardless of discipline, women were more likely than men to indicate that their ‘work was not ground-breaking or sufficiently novel’ for the listed prestigious journals. Men were more likely than women to indicate that the ‘work would fit better in a more specialized journal’. Women were more likely to indicate that they did not submit their papers (in general and their most cited papers) to Science, Nature, or PNAS because they were advised not to. These results may reinforce the notion of risk aversion and perfectionism (Closa et al., 2020), suggesting that women have a higher internal standard for what constitutes novel research. However, the more alarming result is that women seem to be deterred from submitting from people within their scholarly network. One interpretation could be that the work is truly not sufficiently novel; this would suggest that men are innately predisposed to higher novelty research designs or that they are trained in this manner. Given no evidence of the former and sufficient evidence of differential mentoring for women (Nolan et al., 2008), it would stand to reason that greater attention should be made to equity in research training. However, the alternative interpretation is that women are not receiving the simple encouragement necessary to seek out higher-impact venues. Research administrators may seek simple interventions to increase this support. Journal editors may also want to consider dedicated outreach to women-led labs.

Addressing gender disparities in science requires a multifaceted approach that considers all components of the scientific system. In that context, we cannot overemphasize the need for a research evaluation reform—including initiatives that span from the Leiden Manifesto to the San Francisco Declaration on Research Assessment. But until those reforms are actually adopted by universities and funders and actions move beyond pledges, publishing remains a cornerstone of this system and central to research evaluation. Therefore, addressing mechanisms that create disparities in journal submission—be they self-imposed or otherwise—is essential for creating a robust and responsible research ecosystem.

Materials and methods

A survey was conducted to explore the relationship between author gender and manuscript submission, rejection, and acceptance rates for high-impact, multidisciplinary journals in physical sciences, life sciences, and engineering (see S.I. appendix for the questionnaire). This study focuses on the results for Science, Nature, and PNAS. The target population consisted of active authors who published articles from 2008 to 2017 in journals indexed by WoS in medical sciences, natural sciences and engineering, and social sciences (Appendix 1—table 2). The questionnaire asked respondents to report on their manuscript submission experiences. It included a subsection focusing on their most cited paper published after 2010 not published in selected elite journals, that is, Science, Nature, PNAS (which are analyzed in the main manuscript), as well as Cell, Nature Communications, NEJM, and Science Advances (which are analyzed in the S.I. appendix). Specific questions addressed their experiences submitting their most cited manuscripts to these journals, as well as their experiences submitting to these journals in general. Stata software (Standard Edition 17) was used to conduct the analyses, which consisted of applying ordinal logistic regression, multinomial logistic regression, and multiple logistic regression with a significance level or alpha of 0.05. Although we collected data on genders other than men and women, the independent variable studied is a binary gender variable (i.e., women or men), given the low number of respondents outside these two categories. Respondents were disaggregated by academic rank (i.e., junior, senior, or non-academic, excluding all unknowns) and discipline (i.e., social sciences, medical sciences, and natural sciences and engineering). More details about the regression analysis can be found in the appendix. The sample consisted of 6002 respondents who participated in the survey, of whom 4857 finished the questionnaire. An analysis of the attrition failed to identify a common point of departure, suggesting individual variability in dropout rather than failed survey construction. The final number of respondents in this study is 4805, after the removal of 19 responses in Arts and Humanities and additional responses due to a lack of information for critical variables. Of the final number of respondents, 4740 (98.6%) have a known academic rank and are included in the regression analysis (more details available in the S.I. Appendix 1—table 6).

Appendix 1

Expansion on materials and methods

Sampling frame construction

The sampling frame was constructed using data extracted from Web of Science (WoS), as hosted by the Centre for Science and Technology Studies (CWTS) at Leiden University. Using the author disambiguation algorithm developed by Caron and van Eck (2014) (Caron, E. & Van Eck, N.J. (2014). "Large scale author name disambiguation using rule-based scoring and clustering." in Proceedings of the 19th international conference on science and technology indicators, Leiden), we extracted metadata of all publications for authors who have been corresponding authors of at least one paper between 2008 and 2017. The metadata included the email addresses of these authors; these email addresses were used to contact the participant population. A minimum cut-off point of 10 papers for the identification of prolific authors was selected. When considering whether an author is a prolific author, only research papers were considered. The total study population consists of 494,781 prolific authors, data on the gender and discipline in which they publish were known for 415,697 authors (84.0%). The gender of an author for the sampling frame was determined by the algorithm developed in Larivière et al. (2013) (Larivière, V., Ni, C., Gingras, Y., Cronin, B. & Sugimoto, C.R. (2013). "Bibliometrics: Global gender disparities in science." Nature, 504: 211–213. doi:10.1038/504211a) whereas the gender variable used in the analysis was determined by the responses of the authors. The discipline in which an author publishes was determined by the WoS journal classification.

Identifying most cited papers

Several questions in the questionnaire ask the authors about their decisions pertaining to their most cited paper. The questionnaires are customized to each respondent, showing them the title of their most cited paper when these questions are asked. The most cited paper was identified from the previous corpus of papers by the author published between 2008 and 2017, that has received the most citations at the time of data extraction and were not published in the selected elite journals. Despite the fact that citations take time to accumulate, the skewed nature of citation distributions—where most papers remain lowly cited, and a few papers get highly cited—leads to a situation where most cited papers are not necessarily the oldest papers published by respondents.

Elite journal selection

The questionnaire mainly consists of questions related to the author’s submission behavior as they pertain to elite journals. The seven journals were chosen given their general reputation and research impact. All seven journals were in the top five journals with the highest impact factors in the 2016 Journal Citation Reports categories of ‘Multidisciplinary sciences’, ‘General & internal medicine’, and ‘Biology’. Journal Citation Reports (more information at https://clarivate.libguides.com/jcr) (https://jcr.clarivate.com/jcr/) is a resource published annually by Clarivate Analytics that provides the impact factor for all journals included in WoS. The journal impact factor (Garfield, E. (2006). “The history and meaning of the journal impact factor.” JAMA, 295(1):90–93. doi:10.1001/jama.295.1.90) is a calculation that employs citations that articles in a journal received as an indication of the influence of that journal in its field.

Regression analysis

This study used regression analysis to analyze the data. The analysis described here is similar to Ni et al., 2021 and how it is reported in its supplementary materials. Here, the discussion is slightly adapted to reflect the variables and questions used within this study. The regression procedures include binary logistic regression, ordinal logistic regression, linear regression, and mixed-effects model. Regression analysis is usually used to explore relationships between dependent and independent variables. The most common is linear regression, in which:

Yi=β0+β1Xi+β2Zi+⋯+ei

where with one unit change in X there is β1 differences in Y after Z and other variables are controlled. However, this approach assumes that all dependent variables are normally distributed. Furthermore, it often assumes that the dependent variable Y is continuous. Our dataset, like many social science surveys, is replete with categorical variables. We want to understand whether there is a gendered difference when selecting one category over another. Therefore, we turned to logistic regression. Specific procedures and analysis methods vary by the scale of dependent variables, as well as the number of variable categories.

For questions with answers measured in ordinal scales (e.g., Likert scaling), ordinal logistic regression analysis was employed. For questions with answers (the dependent variables) presented in a categorical, unordered scale, binary logistic regression (when the dependent variable only contains two categories) was performed. The point statistic was still the odds ratio of women over men. Here again, gender is the independent variable, with the academic rank being the controlled variable in the case when a dependent variable only has two categories (e.g., ‘yes’ or ‘no’ to an answer), binary logistic regression analysis was conducted to analyze the data. An odds ratio value significantly over 1 indicates that women were more likely than men to select ‘Yes’ for this question. A separate regression analysis is performed for each discipline with rank being controlled. For analyses concerning all disciplines, we control for both academic rank and discipline.

The binary logistic regression model for each discipline can be described as follows:

Yi=Log(p1−p)=β0+β1Xi+β2Zi+⋯+ei

where Yi is the log probability of the binary dependent variable (such as whether you submitted to a journal), Xi is the binary independent variable gender (woman = 1 and man = 0), and Zi is the other control variable(s).

The mixed-effect linear model for each discipline can be described as follows:

Yi=β0+β1Xi+Zi+⋯+ei

where Yi is the dependent variable (number of submissions, acceptance rate, or rejection rate), Xi is the gender of the individual (the fixed effect), and Zi is the rank of the individual (random effect variable).

Recoded categories

The academic rank variable was created by consulting the respondents’ answers on the questions regarding academic rank (for those respondents that indicated their primary sector affiliation as Academia), and position (for those respondents from the Government, Private, or Not_for_Profit sector). Responses were recategorized into junior, senior, and non-academic, with unknowns removed from analyses, as per Appendix 1—table 1.

Appendix 1—table 1
Recategorized academic rank.
Questionnaire categoriesNew categoriesCount
Graduate studentJunior991
Postdoctoral fellow
Research Associate
Assistant Professor
Associate ProfessorSenior3012
Full Professor
Emeritus Professor
Non-academic respondentsNon-academic737
No answer and otherUnknowns65

Due to limited responses, analysis per journal was not always viable. The results pertaining to the journals were aggregated, with new categories based on the shared similarities in disciplinary foci of the journals and their prestige. The categories are as follows:

  • Science, Nature, and PNAS (S.N.P.)

  • Nature Communications and Science Advances (NC.SA.)

  • NEJM and Cell (NEJM.C.)

Similarly, the various options for research areas in the questionnaire were reclassified into three broader areas that share common research and dissemination practices, namely natural science and engineering (NSE), medical science (MS), and social sciences (SS), as shown in Appendix 1—table 2.

Appendix 1—table 2
Recategorized research areas to disciplines.
Disciplinary areaFieldSurveyedRespondentsAnalytical sample
Medical SciencesBiomedical Research63,335832678
Clinical Medicine152,05613941134
Health12,807245208
Natural Sciences and EngineeringBiology47,539670543
Chemistry36,830395321
Earth and Space36,573576467
Engineering and Technology61,746559406
Mathematics18,674275206
Physics46,694592475
Social SciencesProfessional Fields34808266
Psychology9885243207
Social Sciences435411494
Arts and HumanitiesArts2919/
Humanities47514/
UnknownUnknown4220
Total494,78160024805

Description of the analytical sample

The Indiana University Centre for Survey Research (CSR) was contracted to administer the survey. The description of the survey response is based on the technical report produced by the CSR. After data cleaning (e.g., removing duplicates), a total of 489,777 email invitations were sent, and 6002 individuals participated in the online survey. The responses consisted of 4857 completed questionnaires. After reviewing the complete questionnaires, responses of authors in Arts (8) and humanities (11) questionnaires were excluded. Lastly, we further removed additional responses due to lack of information for critical variables (e.g., gender and discipline). An analysis of the attrition failed to identify a common point of departure, suggesting individual variability in dropout rather than failed survey construction. The final analytical sample consists of 4805 respondents. Depending on the variable, the number of responses for analyzing them may vary.

The response rate of the survey was 1.2%, as classified according to The American Association for Public Opinion Research, 2015. Standard Definitions: Final Dispositions of Case Codes and Outcome Rates for Surveys. 8th edition (AAPOR), as summarized in Appendix 1—table 3.

Appendix 1—table 3
Summary of response rate of questionnaire adapted from McGill University – Gender Ethnicity Publishing Study Methods Summary technical report (Indiana University Centre for Survey Research 2019).

McGill University – Gender Ethnicity Publishing Study Methods Summary. Technical Report. Unpublished.

DispositionDefinitionCountResponse rate (RR2)
Complete (I)Respondent completed survey49441.2%
Partial (P)Respondent answered at least one question item but did not complete survey1058
Implicit Refusal (R)Respondent clicked survey link but did not answer any items1655
Nothing Returned (UH)Respondent did not respond to survey; it is unknown if any email messages were read374,341
Undeliverable (UO)Recruitment message was not received by intended recipient due to email and/or mailing returns108,163
Not EligibleSample member indicated they were not eligible to participate; also includes cases of authors who have deceased since their publication listed in the database58
Total490,219

Appendix 1—tables 4–6 provide a description of the population, respondents, and analytical sample by gender.

Appendix 1—table 4
Gender composition of population, respondents, and analysis sample.
DomainPopulation/surveyedRespondentsAnalytical sample
N%N%N%
Men286,57157.9350858.4283258.9
Women129,59226.2177429.6143529.9
Unknown78,61815.972012.053811.2
All494,781100.06002100.04805100.0
Appendix 1—table 5
Discipline composition of population, respondents, and analysis sample.
DomainPopulation/surveyedRespondentsAnalytical sample
N%N%N%
MS228,19846.1247141.2202042.0
SS17,7193.64397.33667.6
NSE248,09850.1306951.1241950.3
AH7660.2230.40.0
All disciplines494,781100.06002100.04805100.0
Appendix 1—table 6
Discipline, gender, and academic rank composition of analytical sample.
MS (N = 2020)NSE (N = 2419)SS (N = 366)
Rank/roleWomenMenWomenMenWomenMen
N%N%N%N%N%N%
Junior20024.522818.914727.636019.13118.72512.5
Senior46557.076263.328152.7121964.612474.716180.5
Non-academic13116.118915.710018.829415.6106.0136.5
Unknown202.5252.150.9130.710.610.5

Limitations

Several limitations should be noted regarding the findings of this study. First, the survey response rate is less than 2%, which may limit the generalizability of the findings. In addition, asking researchers about their behavior and intentions regarding their manuscript submissions retroactively can be complicated. There is a concern related to the capacity of researchers to remember their motivations at the time of submission; researchers with longer careers—and therefore a higher number of papers—may have submitted to various journals for various reasons. This would also make it difficult to ascertain the reasons why they do not submit to the journals, as the motivations would be different for each paper. Moreover, reasons for not submitting to these journals were not exhaustive, as noted by the various comments made by respondents. One of the notable reasons given by respondents that relate to the gender component of the study is that they did not submit to these journals due to the low percentage of women on the journal editorial boards. Based on respondents’ comments, page fees and article processing charges (APC) were also a serious consideration for some authors when deciding where to publish their work. Unfortunately, this consideration was not incorporated into the listed reasons in the questionnaire.

There are also several factors pertaining to the nature of most cited papers that complicate the interpretation of the results. For example, some respondents stated that their most cited paper was an invited paper of some kind, thus the choice of journal was already made for them and, therefore, the survey questions were irrelevant. In addition, even though their most cited paper was used as a proxy of a high-quality paper, the paper in question was not necessarily a highly cited paper, or necessarily significantly more cited than their other papers, or a paper of high quality (e.g., all their other papers could have been cited once and their most cited paper twice). That being said, the papers analyzed had, on average, a research impact 2.3 times higher than the world average of their field, which shows that they can be categorized as high impact. On the whole, this was the best proxy available for assignment of high-quality papers to researchers, and the effect is reduced by only investigating the subset of prolific researchers.

Some limitations relate to the passage of time between manuscript submission and data collection. Academic rank refers to the academic rank of the author at the time of the survey and not at the time of manuscript submission. This could result in a scenario where a respondent’s highest cited paper could have been submitted in 2008 while they were a junior academic but in 2019, at the time of the survey, they have since changed academic rank to senior academic. Therefore, this should be interpreted as the maximum rank reached by the respondent. Review policies of journals are also a potential limitation as data pertaining to the review policies at the time of manuscript submission were not collected. If a journal had a double-blind review policy at the time of submission, it might reduce potential gender bias and be a confounding effect if journals of different peer review types were grouped together. It is likely that the review policy of the journals was single-blind at the time of submission of the papers as it is the review policy most used in the fields of life sciences, physical sciences, and engineering (Ware, M. (2008). "Peer review in scholarly journals: Perspective of the scholarly community — an international study." Information Services & Use, 28(2). doi:10.5555/1454388.1454399) and thus unlikely to have changed. While this could be a factor when we investigated acceptance rate, it is not a consideration when investigating desk rejections as none of the journals include an editor-blinded review policy.

Regression and additional descriptive tables

The questionnaire was divided into four parts, and the following sections are ordered by the sections of the questionnaire. The sections are as follows:

  1. Part 1: Consent for Research Study Participation (information regarding questionnaire – no questions)

  2. Part 2: Your manuscript submissions (i.e., questions pertaining to submissions in general)

  3. Part 3: Your journal article (i.e., questions pertaining to most cited paper)

  4. Part 4: Demographics (as well as a question pertaining to quality of research in general and open comments)

Part 2: Submissions in general

Journal submission behavior

Appendix 1—table 7
Question: ‘For each of the journals listed in the table below, please indicate the appropriate number of manuscripts submitted’—percentage by discipline, journal, and gender ever submitted.
S.N.P.WomenMen
NoYesNo response%YesNoYesNo response%Yes
MS504303937.5%5556361353.4%
NSE3281941137.2%10008602646.2%
SS10955233.5%11087344.2%
All disciplines9415522237.0%166515834248.7%
NC.SA.WomenMen
NoYesNo response%YesNoYesNo response%Yes
MS670137917.0%9022831923.9%
NSE429911317.5%14763753520.3%
SS158533.1%1811457.2%
All disciplines12572332515.6%25596725920.8%
NEJM.CWomenMen
NoYesNo response%YesNoYesNo response%Yes
MS603204925.3%7953901932.9%
NSE50512162.3%181634361.8%
SS1521136.7%1821356.7%
All disciplines12602272815.3%27934376013.5%
Appendix 1—table 8
Question: ‘For each of the journals listed in the table below, please indicate the appropriate number of manuscripts submitted’—odds ratio (women to men) values for the probability of submitting.
JournalDisciplineOdds ratioStd. Err.zp > |z|95% CI lower95% CI upperobs
S.N.P.MS0.540.05–6.450.0000.450.651937
NSE0.700.07–3.400.0010.580.862356
SS0.620.14–2.130.0330.400.96355
All disciplines0.610.04–7.320.0000.540.704648
NC.SA.MS0.650.08–3.680.0000.520.821931
NSE0.810.11–1.600.1100.631.052346
SS0.350.19–1.910.0560.121.02352
All disciplines0.700.06–4.060.0000.590.834629
NEJM.C.MS0.710.07–3.350.0010.580.871931
NSE1.220.420.590.5580.632.392342
SS1.080.460.180.8590.472.50352
All disciplines0.750.07–2.950.0030.620.914625
Appendix 1—table 9
Question: ‘For each of the journals listed in the table below, please indicate the appropriate number of manuscripts submitted’—mean number of manuscript submissions.
Average number of submissions to S.N.P.WomenMen
ObsMeanStd. dev.MinMaxObsMeanStd. dev.MinMax
MS3033.603.511346365.125.41135
NSE1943.042.381138603.953.95130
SS552.803.36119874.225.49130
All disciplines5523.323.1513415834.434.71135
Average number of submissions to NC.SA.WomenMen
ObsMeanStd. dev.MinMaxObsMeanStd. dev.MinMax
MS1371.741.341102811.901.2418
NSE891.651.08163741.861.57117
SS51.000.0011141.711.1415
All disciplines2311.691.23110141.711.1415
Average number of submissions to NEJM.C.WomenMen
ObsMeanStd. dev.MinMaxObsMeanStd. dev.MinMax
MS2031.891.341103892.252.00116
NSE121.671.4416341.761.4818
SS111.640.9214131.150.3812
All disciplines2261.861.321104362.181.95116
Appendix 1—table 10
Question: ‘For each of the journals listed in the table below, please indicate the appropriate number of manuscripts submitted’—mixed linear regression.
JournalDisciplineCoefficientStd. dev.tp > |t|95% CI lower95% CI upperobs
S.N.P.MS–1.460.34–4.260.000–2.14–0.79921
NSE–0.800.30–2.690.007–1.39–0.221042
SS–1.350.84–1.610.110–3.000.31140
All disciplines–1.190.22–5.410.000–1.62–0.762103
NC.SA.MS–0.190.13–1.410.161–0.450.08411
NSE–0.190.18–1.090.276–0.540.16459
SS–0.710.54–1.320.207–1.860.4419
All disciplines–0.200.11–1.890.059–0.420.01889
NEJM.C.MS–0.360.15–2.390.017–0.65–0.06575
NSE–0.080.50–0.170.867–1.100.9346
SS0.590.321.840.081–0.081.2624
All disciplines–0.310.14–2.230.026–0.58–0.04645

Acceptance rate

Appendix 1—table 11
Question: ‘For each of the journals listed in the table below, please indicate the appropriate number of manuscripts accepted’—mean acceptance rate.
Acceptance rate of S.N.P.WomenMen
ObsMeanStd. dev.MinMaxObsMeanStd. dev.MinMax
MS2620.300.35015650.300.3401
NSE1770.250.35017980.240.3401
SS530.280.3801800.240.3601
All disciplines4920.280.350114430.260.3401
Acceptance rate of NC.SA.WomenMen
ObsMeanStd. dev.MinMaxObsMeanStd. dev.MinMax
MS1370.270.39012810.320.3901
NSE890.300.41013740.360.4201
SS50.200.4501140.240.4201
All disciplines2310.280.40016690.340.4101
Acceptance rate of NEJM.C.WomenMen
ObsMeanStd. dev.MinMaxObsMeanStd. dev.MinMax
MS2020.270.38013880.260.3601
NSE120.170.3301340.180.3401
SS110.390.4701130.040.1400.5
All disciplines2250.270.38014350.250.3601
Appendix 1—table 12
Question: ‘For each of the journals listed in the table below, please indicate the appropriate number of manuscripts accepted’—linear regression.
JournalDisciplineCoefficientStd. dev.tp > |t|95% CI lower95% CI upperobs
S.N.P.MS0.000.030.010.989–0.050.05810
NSE0.010.030.480.628–0.040.07964
SS0.050.070.780.435–0.080.18132
All disciplines0.010.020.570.566–0.030.051906
NC.SA.MS–0.050.04–1.150.252–0.130.03411
NSE–0.070.05–1.420.156–0.170.03459
SS–0.040.22–0.160.877–0.510.4419
All disciplines0.060.03–1.780.075–0.120.01889
NEJM.C.MS0.020.030.650.517–0.040.08573
NSE–0.020.11–0.150.885–0.250.2146
SS0.430.152.790.0110.110.7424
All disciplines0.030.031.070.287–0.030.09643

Desk rejections

Appendix 1—table 13
Question: ‘For each of the journals listed in the table below, please indicate the appropriate number of manuscripts rejected’—mean desk rejection rate.
Desk rejection rate of S.N.P.WomenMen
ObsMeanStd. dev.MinMaxObsMeanStd. dev.MinMax
MS2620.760.34015650.790.2801
NSE1720.810.31017640.790.2901
SS460.850.2601750.870.2401
All disciplines4800.790.320114040.800.2801
Desk rejection rate of NC.SA.WomenMen
ObsMeanStd. dev.MinMaxObsMeanStd. dev.MinMax
MS1050.800.34012100.770.3501
NSE640.720.38012660.740.3701
SS40.750.5001110.860.2501
All disciplines1730.770.36014870.760.3601
Desk rejection rate of NEJM.C.WomenMen
ObsMeanStd. dev.MinMaxObsMeanStd. dev.MinMax
MS1610.680.40013200.670.4001
NSE90.720.4401280.710.4301
SS60.830.260.51130.880.3001
All disciplines1760.680.39013610.680.4001
Appendix 1—table 14
Question: ‘For each of the journals listed in the table below, please indicate the appropriate number of manuscripts rejected’—linear regression.
JournalDisciplineCoefficientStd. dev.tp > |t|95% CI lower95% CI upperobs
S.N.P.MS–0.030.02–1.330.185–0.070.01809
NSE0.030.031.010.314–0.020.07928
SS0.000.05–0.090.928–0.090.09119
All disciplines0.000.02–0.310.757–0.040.031856
NC.SA.MS0.040.040.960.336–0.040.12311
NSE–0.030.05–0.610.542–0.140.07326
SS–0.140.20–0.710.494–0.590.3015
All disciplines0.010.030.170.867–0.060.07652
NEJM.C.MS0.020.040.370.708–0.060.09467
NSE0.030.170.180.862–0.320.3837
SS–0.090.17–0.530.607–0.450.2719
All disciplines0.010.040.330.741–0.060.09523

Reasons for not considering submitting to top journals

Appendix 1—table 15
Question: ‘Please indicate the reason(s) why you did not consider submitting a manuscript to journal name taken from response to the previous question’—number of respondents and percentage selecting response by discipline and gender.
Reasons for not submitting to S.N.P.MSNSESS
WomenMenWomenWomenMenWomen
Total%YesTotal%YesTotal%YesTotal%YesTotal%YesTotal%Yes
Work was not of high enough quality50136%55341%32837%99827%10933%11030%
Work fell outside the scope of the journal50161%55355%32846%99849%10972%11059%
Work was not groundbreaking or sufficiently novel50152%55350%32853%99844%10953%11040%
Work would fit better in a more specialized journal50156%55363%32864%99870%10958%11069%
Work would reach a wider audience in another journal5019%55313%32816%99817%1097%1108%
Co-authors wished to submit the manuscript elsewhere5017%5537%3289%9985%1096%1105%
I was advised against submitting to this journal5016%5534%3288%9985%1096%1105%
Reasons for not submitting to NC.SA.MSNSESS
WomenMenWomenWomenMenWomen
n%Yesn%Yesn%Yesn%Yesn%Yesn%Yes
Work was not of high enough quality66722%89423%42721%146715%15712%18017%
Work fell outside the scope of the journal66746%89437%42733%146734%15761%18053%
Work was not groundbreaking or sufficiently novel66732%89429%42737%146728%15729%18023%
Work would fit better in a more specialized journal66742%89447%42753%146752%15739%18051%
Work would reach a wider audience in another journal6679%89414%42713%146717%15710%18011%
Co-authors wished to submit the manuscript elsewhere6679%8949%42710%14678%1576%1804%
I was advised against submitting to this journal6674%8944%4273%14673%1573%1802%
Reasons for not submitting to NEJM.C.MSNSESS
WomenMenWomenWomenMenWomen
n%Yesn%Yesn%Yesn%Yesn%Yesn%Yes
Work was not of high enough quality60234%79131%5036%18014%15115%1808%
Work fell outside the scope of the journal60283%79180%50384%180186%15193%18090%
Work was not groundbreaking or sufficiently novel60238%79133%5038%18015%15121%18012%
Work would fit better in a more specialized journal60237%79135%50313%180113%15124%18027%
Work would reach a wider audience in another journal6025%7917%5033%18014%1515%1806%
Co-authors wished to submit the manuscript elsewhere6025%7916%5032%18012%1515%1803%
I was advised against submitting to this journal6024%7914%5030%18010%1512%1801%
Appendix 1—table 16
Question: ‘Please indicate the reason(s) why you did not consider submitting a manuscript to journal name taken from response to the previous question’—odds ratio (women to men) values.
Reasons why did not consider submitting papers to top journalsS.N.P.NC.SA.NEJM.C.
MSNSESSAll displ.MSNSESSAll displ.MSNSESSAll displ.
Work was not of high enough quality0.831.59†1.041.120.911.51†0.601.081.101.511.691.21
Work fell outside the scope of the journal1.270.941.84*1.141.42†1.011.411.24†1.200.921.411.08
Work was not groundbreaking or sufficiently novel1.24*1.47†1.69*1.26†1.131.41†1.271.26†1.181.60*1.711.31†
Work would fit better in a more specialized journal0.74*0.72*0.580.72*0.821.020.60*0.871.061.010.811.01
Work would reach a wider audience in another journal0.61*0.930.920.780.58†0.740.900.68†0.710.810.800.76
Co-authors wished to submit the manuscript elsewhere1.021.70*1.311.321.001.171.341.090.881.091.340.97
I was advised against submitting to this journal1.301.69*1.371.48*0.870.771.730.871.241.111.521.25
I was unaware of this journal‡No one reported they were unaware of the top journals
  1. *

    indicates p < 0.05.

  2. †

    indicates p < 001.

  3. ‡

    Insufficient sample.

Part 3: Most cited

Journal submission behavior

Appendix 1—table 17
Question: ‘For your published manuscript <Title of the respondent’s most cited paper>, did you ever consider submitting it to a journal other than ?’—percentage by discipline, and gender.
DisciplineWomenMen
NoYesTotal%YesNoYesTotal%Yes
MS37043280253.9%559616117552.4%
NSE27324852147.6%1038814185244.0%
SS808416451.2%1019219347.7%
All disciplines723764148751.4%16981522322047.3%
Appendix 1—table 18
Question: ‘For your published manuscript <Title of the respondent’s most cited paper>, did you ever consider submitting it to a journal other than ?’—odds ratio (women to men) values.
Other journalsF/M odds ratioStd. errzp > |z|95% CI lower95% CI upperobs
MS1.070.1000.670.500.891.281914
NSE1.120.1131.140.260.921.372351
SS1.160.2500.680.500.761.77351
All disciplines1.100.0721.400.160.961.254616
Appendix 1—table 19
Question: ‘Before publication of your manuscript,<Title of the respondent’s most cited paper>, did you ever consider submitting it to the journals listed in the table below?’—percentage by discipline, journal, and gender.
Consider submitting to S.N.P.WomenMen
NoYesTotal%YesNoYesTotal%Yes
MS721767979.5%1029143117212.2%
NSE478405187.7%170214818508.0%
SS15761633.7%185131986.6%
All disciplines135612214788.3%291630432209.4%
Consider submitting to NC.SA.WomenMen
NoYesTotal%YesNoYesTotal%Yes
MS743327754.1%10824611284.1%
NSE486175033.4%17474717942.6%
SS16111620.6%19401940.0%
All disciplines13905014403.5%30239331163.0%
Consider submitting to NEJM.C.WomenMen
NoYesTotal%YesNoYesTotal%Yes
MS749427915.3%10787311516.3%
NSE50025020.4%1788317910.2%
SS16301630.0%19221941.0%
All disciplines14124414563.0%30587831362.5%
Appendix 1—table 20
Question: ‘Before publication of your manuscript,<Title of the respondent’s most cited paper>, did you ever consider submitting it to the journals listed in the table below?’—odds ratio (women to men) values.
JournalW/M odds ratioStd. errzp > |z|95% CI lower95% CI upperobs
S.N.P.MS0.790.120–1.530.1260.591.071907
NSE0.960.181–0.210.830.661.392342
SS0.540.275–1.210.2270.201.47355
All disciplines0.830.096–1.590.1120.661.044604
NC.SA.MS1.010.2400.050.9610.641.611844
NSE1.050.3210.160.8730.581.912272
SSInsufficient sample
All disciplines1.050.1960.250.8010.731.514466
NEJM.C.MS0.840.170–0.870.3860.561.251882
NSE2.442.2460.970.3330.4014.842268
SSInsufficient sample
All disciplines0.850.166–0.850.3930.571.244501
Appendix 1—table 21
Question: ‘Before publication of your manuscript,<Title of the respondent’s most cited paper>, did you ever submit it to the journals listed in the table below?’—percentage by discipline, journal, and gender.
Submitted to S.N.P.WomenMen
NoYesTotal%YesNoYesTotal%Yes
MS37387550.7%588113958.3%
NSE22173943.6%648014455.6%
SS14580.0%581361.5%
All disciplines605911949.6%12716929657.1%
Submitted to NC.SA.WomenMen
NoYesTotal%YesNoYesTotal%Yes
MS19143342.4%25275251.9%
NSE1372035.0%35215637.5%
SS1010.0%0000
All disciplines33215438.9%604810844.4%
Submitted to NEJM.C.WomenMen
NoYesTotal%YesNoYesTotal%Yes
MS29134231.0%40317143.7%
NSE2020.0%2020.0%
SS00002020.0%
All disciplines31134429.5%44317541.3%
Appendix 1—table 22
Question: ‘Before publication of your manuscript, <Title of the respondent’s most cited paper>, did you ever submit it to the journals listed in the table below?’—odds ratio (women to men) values for the probability of submitting.
JournalDisciplineW/M odds ratioStd. errzp > |z|95% CI lower95% CI upperobs
S.N.P.MS0.750.22–0.990.3210.431.32210
NSE0.650.24–1.190.2340.311.33180
SS2.503.140.730.4650.2129.2518
All disciplines0.740.17–1.350.1770.481.15408
NC.SA.MS0.690.32–0.820.4140.281.7084
NSE1.060.610.100.9180.353.2573
SSNo one submitted to NC.SA.
All disciplines0.800.29–0.610.5400.401.62157
NEJM.C.MS0.550.23–1.420.1570.241.26111
NSENo one submitted to NEJM.C.
SSNo one submitted to NEJM.C.
All disciplines0.570.24–1.350.1780.251.29117

Desk rejections

Appendix 1—table 23
Question: ‘Was your manuscript <Title of the respondent’s most cited paper> sent out for peer review before being rejected by the journals listed in the table below?’—percentage by discipline, journal, and gender.
Sent out for peer review S.N.P.WomenMen
NoYesTotal%DeskRejNoYesTotal%DeskRej
MS26103672.2%62198176.5%
NSE1431782.4%58228072.5%
SS31475.0%808100.0%
All disciplines43145775.4%1284116975.7%
Sent out for peer review NC.SA.WomenMen
NoYesTotal%DeskRejNoYesTotal%DeskRej
MS1221485.7%2172875.0%
NSE71887.5%1742181.0%
SSNo responses
All disciplines1932286.4%31114977.6%
Sent out for peer review NEJM.C.WomenMen
NoYesTotal%DeskRejNoYesTotal%DeskRej
MS581338.5%2753284.4%
NSENo responses
SS
Appendix 1—table 24
Question: ‘Was your manuscript <Title of the respondent’s most cited paper> sent out for peer review before being rejected by the journals listed in the table below?’—odds ratio (women to men) values.
JournalDisciplineW/M odds ratioStd. errzp > |z|95% CI lower95% CI upperobs
S.N.P.MS0.480.660.740.460.563.52114
NSE1.410.33–1.050.290.121.8895
SSInsufficient sample
All disciplines1.130.410.320.750.552.31221
NC.SA.MS0.430.38–0.950.340.072.4636
NSE0.700.90–0.280.780.068.8227
SSNo responses
All disciplines0.500.37–0.950.350.122.1063
NEJM.C.MS13.0710.963.070.002.5367.6044
NSENo responses
SS

Reasons for not considering submitting to top journals

Appendix 1—table 25
Question: ‘Please indicate the reason(s) why you did not consider submitting your manuscript <Title of the respondent’s most cited paper> to the journals listed below.’—number of respondents and percentage selecting response by discipline and gender.
Reasons for not submitting to S.N.P.MSNSESS
WomenMenWomenMenWomenMen
n%Yesn%Yesn%Yesn%Yesn%Yesn%Yes
Work was not of high enough quality71925%102730%47521%169618%15627%18518%
Work fell outside the scope of the journal71945%102744%47535%169640%15662%18560%
Work was not groundbreaking or sufficiently novel71945%102744%47548%169639%15639%18530%
Work would fit better in a more specialized journal71953%102753%47560%169663%15646%18557%
Work would reach a wider audience in another journal7198%10279%47512%169714%1568%18511%
Co-authors wished to submit the manuscript elsewhere7197%10276%4758%16974%1566%1853%
I was advised against submitting to this journal7195%10273%4753%16962%1562%1852%
Reasons for not submitting to NC.SA.MSNSESS
WomenMenWomenWomenMenWomen
n%Yesn%Yesn%Yesn%Yesn%Yesn%Yes
Work was not of high enough quality73219%107622%48114%173113%15914%18912%
Work fell outside the scope of the journal73241%107638%48131%173133%15963%18958%
Work was not groundbreaking or sufficiently novel73234%107632%48138%173130%15929%18921%
Work would fit better in a more specialized journal73246%107645%48152%173155%15932%18943%
Work would reach a wider audience in another journal7328%107611%48110%173112%1595%1897%
Co-authors wished to submit the manuscript elsewhere7326%10766%4816%17325%1598%1892%
I was advised against submitting to this journal7322%10762%4812%17312%1591%1891%
Reasons for not submitting to NEJM.C.MSNSESS
WomenMenWomenWomenMenWomen
n%Yesn%Yesn%Yesn%Yesn%Yesn%Yes
Work was not of high enough quality73825%107328%4955%17715%16115%1877%
Work fell outside the scope of the journal73872%107375%49585%177286%16191%18786%
Work was not groundbreaking or sufficiently novel73837%107336%4959%17717%16118%18711%
Work would fit better in a more specialized journal73844%107342%49516%177118%16125%18730%
Work would reach a wider audience in another journal7386%10736%4954%17714%1614%1877%
Co-authors wished to submit the manuscript elsewhere7385%10734%4952%17711%1615%1872%
I was advised against submitting to this journal7382%10732%4950%17710%1611%1870%
Appendix 1—table 26
Question: ‘Please indicate the reason(s) why you did not consider submitting your manuscript <Title of the respondent’s most cited paper> to the journals listed below.’—odds ratio (women to men) values.
Reasons why did not consider submitting the most cited papers to top journalsS.N.P.NC.SA.NEJM.C.
MSNSESSAllMSNSESSAllMSNSESSAll
Work was not of high enough quality0.75*1.171.630.960.861.001.310.930.830.942.36*0.91
Work fell outside the scope of the journal1.050.821.150.951.090.931.301.040.860.941.520.92
Work was not groundbreaking or sufficiently novel1.021.39 ‡1.481.19*1.051.28*1.811.141.051.281.811.14
Work would fit better in a more specialized journal0.990.880.660.901.000.900.640.921.080.920.791.00
Work would reach a wider audience in another journal0.960.870.710.890.720.870.760.790.980.850.550.88
Co-authors wished to submit the manuscript elsewhere1.131.57*2.261.35*0.991.064.53*1.121.141.412.071.26
I was advised against submitting to this journal1.75*1.510.971.59*0.961.062.271.031.473.52+1.67
I was unaware of this journal†No one reported they were unaware of the top journals
  1. *

    indicates p < 0.05.

  2. †

    Insufficient sample.

  3. ‡

    indicates p < 0.01.

Part 4: Perception of the quality of research

Appendix 1—table 27
Question: ‘Compared to my peers, I feel that the quality of my research is’—average rank and count by discipline and gender.
Women
DisciplineAverage quality rankStd. dev#Good and Excellent#Average#Fair and PoorTotal
MS4.20.69574748656
NSE4.00.67352608420
SS4.10.62135180153
All disciplines4.10.681061152161229
Men
DisciplineAverage quality rankStd. dev#Good and Excellent#Average#Fair and PoorTotal
MS4.10.78411267974
NSE4.20.6136318591557
SS4.10.8151256182
All disciplines4.10.72355336222713

  1. Note: Excellent (5)–Poor (1).

Appendix 1—table 28
Question: ‘Compared to my peers, I feel that the quality of my research is’—ordinal logistic (women to men), controlled for rank.
DisciplinesOdds ratioStd. errzp > |z|95% CI lower95% CI upperobs
MS1.180.132.400.070.991.401630
NSE0.830.09–1.640.040.670.991977
SS1.130.240.560.580.741.71335
All disciplines1.060.070.870.380.931.223,942

Data availability

All data needed to evaluate the conclusions in the paper are present in the paper and the Supporting information. Aggregated, de-identified data by gender, discipline, and rank for analyses are available on GitHub (https://github.com/MetascienceLab/genderPublishing/ copy archived at Metascience Research Lab, 2024). Due to Institutional Review Board restrictions and the terms of participant consent, the survey data used in this study cannot be shared publicly. However, to support transparency and reproducibility, all analysis code and a synthetic dataset- constructed with randomly generated values but matching the original data structure- are available on Zenodo: https://doi.org/10.5281/zenodo.16327580.

The following data sets were generated
    1. Ni C
    2. Basson I
    3. Badia G
    4. Tufenkji N
    5. Sugimoto CR
    6. Larivière V
    (2025) Zenodo
    Data for Gender differences in submission behavior exacerbate publication disparities in elite journals.
    https://doi.org/10.5281/zenodo.16327580

References

  1. Book
    1. Baldwin M
    (2015)
    Making “Nature”: The History of a Scientific Journal
    University of Chicago Press.
    1. Gilbert JR
    2. Williams ES
    3. Lundberg GD
    (1994)
    Is there gender bias in JAMA’s peer review process?
    JAMA 272:139–142.
    1. Hemmings B
    2. Kay R
    (2009)
    Lecturer self efficacy: Its related dimensions and the influence of gender and qualifications
    Issues in Educational Research 19:243–254.
    1. Lee CJ
    2. Sugimoto CR
    3. Zhang G
    4. Cronin B
    (2013) Bias in peer review
    Journal of the American Society for Information Science and Technology 64:2–17.
    https://doi.org/10.1002/asi.22784
  34. Book
    1. Wennerås C
    2. Wold A
    (2008)
    Nepotism and sexism in peer-review
    In: Wennerås C, Wold A, editors. Women, Science, and Technology. Routledge. pp. 1–7.

Article and author information

Author details

  1. Chaoqun Ni

    The Information School, University of Wisconsin-Madison, Madison, United States
    Contribution
    Data curation, Software, Formal analysis, Validation, Investigation, Visualization, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4130-7602

  2. Isabel Basson

    1. École de bibliothéconomie et des sciences de l’information, Université de Montréal, Montréal, Canada
    2. Department of Science and Innovation-National Research Foundation Centre of Excellence in Scientometrics and Science, Technology and Innovation Policy, Stellenbosch University, Stellenbosch, South Africa
    Contribution
    Investigation, Writing – original draft, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-6150-3208

  3. Giovanna Badia

    Office of the Dean of Libraries, McGill University, Montréal, Canada
    Contribution
    Conceptualization, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-1544-2728

  4. Nathalie Tufenkji

    Department of Chemical Engineering, McGill University, Montréal, Canada
    Contribution
    Conceptualization, Methodology, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1546-3441

  5. Cassidy R Sugimoto

    1. Department of Science and Innovation-National Research Foundation Centre of Excellence in Scientometrics and Science, Technology and Innovation Policy, Stellenbosch University, Stellenbosch, South Africa
    2. School of Public Policy, Georgia Institute of Technology, Atlanta, United States
    Contribution
    Conceptualization, Methodology, Writing – original draft, Writing – review and editing
    Competing interests
    No competing interests declared

  6. Vincent Larivière

    1. École de bibliothéconomie et des sciences de l’information, Université de Montréal, Montréal, Canada
    2. Department of Science and Innovation-National Research Foundation Centre of Excellence in Scientometrics and Science, Technology and Innovation Policy, Stellenbosch University, Stellenbosch, South Africa
    3. Observatoire des sciences et des technologies, Université du Québec à Montréal, Montréal, Canada
    Contribution
    Conceptualization, Resources, Supervision, Funding acquisition, Investigation, Methodology, Writing – original draft, Project administration, Writing – review and editing
    For correspondence
    vincent.lariviere@umontreal.ca
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-2733-0689

Funding

Canada Research Chairs

  • Vincent Larivière

The funders had no role in study design, data collection, and interpretation, or the decision to submit the work for publication.

Acknowledgements

Vincent Larivière acknowledges funding from the Canada Research Chairs program. The authors would like to thank the survey respondents and the Indiana University Center for Survey Research for their critical contribution to the study. This analysis was approved by McGill University’s Research Ethics Board II, R.E.B. File #: 501-0518 Factors affecting scientific manuscript submission and acceptance rates.

Ethics

This analysis was approved by McGill University's Research Ethics Board II, R.E.B. File #: 501-0518 Factors affecting scientific manuscript submission and acceptance rates.

Version history

  1. Sent for peer review:
  2. Preprint posted:
  3. Reviewed Preprint version 1:
  4. Reviewed Preprint version 2:
  5. Reviewed Preprint version 3:
  6. Version of Record published:

Cite all versions

You can cite all versions using the DOI https://doi.org/10.7554/eLife.90049. This DOI represents all versions, and will always resolve to the latest one.

Copyright

© 2023, Ni et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 2,215
    views
  • 76
    downloads
  • 7
    citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.

Citations by DOI

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Chaoqun Ni
  2. Isabel Basson
  3. Giovanna Badia
  4. Nathalie Tufenkji
  5. Cassidy R Sugimoto
  6. Vincent Larivière
(2026)
Gender differences in submission behavior exacerbate publication disparities in elite journals
eLife 12:RP90049.
https://doi.org/10.7554/eLife.90049.4

Share this article

https://doi.org/10.7554/eLife.90049

Categories and tags

Further reading

  1. Meta-Research: A Collection of Articles

    Edited by Peter A Rodgers
    Collection

    The study of science itself is a growing field of research.

  2. Listen to Vincent Lariviere discuss why female scientists are less represented than male scientists in elite journals.

    Podcast

    Female authors are more hesitant to submit to top-tier journals