There have been growing concerns around sexism in science. Studies have found that women in science are often paid less, are less likely to get credit for their work and receive fewer and smaller grants than men at similar stages in their careers. This can make it harder for women to advance in their careers, resulting in less women than men taking up positions of leadership.

There are also gender imbalances between scientific disciplines, with a higher proportion of women working in some fields compared to others. Here, James et al. set out to find whether having more women working in a discipline leads to biases in how the research is evaluated.

The team examined four datasets which included information on the research evaluations and funding success of thousands of researchers across 30 different countries. The analysis suggested that scientists working in women-dominated disciplines were less likely to succeed in their grant applications. Their research was also often evaluated as being lower quality compared to researchers working in fields dominated by men. These biases applied to both men and women working in these disciplines. There were not sufficient data to analyse patterns faced by non-binary individuals.

The study by James et al. cannot pinpoint a specific cause for these outcomes. However, it suggests that funding organisations should analyse the pattern of successful applications across disciplines and consider taking steps to ensure all disciplines have similar success rates. James et al. also propose that when hiring or making promotions, scientific institutions should take care when comparing researchers across disciplines and ensure there is no built-in assumption that fields dominated by men are intrinsically better.

Many have observed that women’s research is evaluated as lower quality than men’s research (Brower and James, 2020; West et al., 2013). And apart from a few exceptions (Ceci et al., 2023; Ceci et al., 2014; Albers, 2015), most analyses show that women’s funding success rates are lower than men’s (Witteman et al., 2019; Grogan, 2019; Brower and James, 2023) both in terms of number of grants and size of grants (Grogan, 2019; Ley and Hamilton, 2008). Evaluations of research quality affect many aspects of an academic career, from hiring and promotion decisions to funding success. Other studies have explored possible explanations for this seemingly gendered pattern in evaluation of research quality, such as societal (Brower and James, 2020; Brower and James, 2023; Hunter and Leahey, 2010), institutional (James and Brower, 2022), and academic factors (El-Alayli et al., 2018; O’Meara et al., 2017), including gender bias (Shen, 2013).

We find a strong correlation between the evaluated quality of a researcher’s work and the gender balance of the researcher’s discipline. Researchers, both male and female, in female-dominated disciplines receive lower evaluations. This finding is robust to bibliometric differences across disciplines. We also find that funding success rates display a similar correlation. Applications from researchers, both male and female, in female-dominated disciplines have lower success rates.

Academic disciplines can be quite different to one another in ways such as publication rates (Olejniczak et al., 2022), citation rates (Lancho-Barrantes et al., 2010), and expected h-indices (Harzing, 2008). There are different norms for the expected number of co-authors (Puuska et al., 2014), order of authors, and attributions (Street et al., 2010). Gender balance is not constant across disciplines either. There are fewer women in STEM subjects, and more women than men use qualitative methods, rather than quantitative, in research (Shen, 2013). Recent research suggests there are interactions between individual- and discipline-level characteristics. Higher ‘expectations of brilliance’ in a discipline is related to a lower proportion of women entering Leslie et al., 2015; and research area is a strong contributor to the lower funding success rates of African American/Black scientists (Hoppe et al., 2019).

That women score lower on evaluations of research quality is not surprising. On an individual level, women publish less than men (West et al., 2013; Mairesse and Pezzoni, 2015; Symonds et al., 2006), receive fewer citations (Dion et al., 2018; Larivière et al., 2013), and have lower h-indices (Chatterjee and Werner, 2021). Women have fewer first authorships (West et al., 2013) and are less likely to receive acknowledgement (Ross et al., 2022). Women are also less likely to get credit for their innovations (Hofstra et al., 2020; Ding et al., 2006) and awards, particularly prestigious ones, are less likely to be given to women (James et al., 2019; Meho, 2021). Recent research suggests women react differently than men to a change in a journal’s prestige or impact factor (Schmal et al., 2023).

However, our research shows a different correlation: a higher proportion of women in a discipline is correlated with lower research quality evaluations and funding success rates, for everyone in the field.

We use independent datasets spanning thousands of researchers and 30 countries and look separately at research funding success rates and holistic research quality scores. Where possible we account for researcher characteristics like age, research institution, and publishing patterns. We consistently find that researchers in male-dominated disciplines have both higher funding success rates, and higher research quality scores, than researchers in female-dominated disciplines. This applies to both men and women.

We analysed research quality and funding success separately. We used data from four sources (Performance-Based Research Fund [PBRF], Australian Research Council [ARC], Canadian Institute of Health Research [CIHR], European Institute of Gender Equality [EIGE]), spanning 30 countries (described fully in Supplementary Materials). The ARC, CIHR, and EIGE datasets cover funding success rates across 29 countries. The PBRF dataset covers research quality evaluations for all university academic staff in New Zealand. Summary data from the four datasets are reported in Table 1. The discipline category definitions varied in each dataset, e.g., EIGE data used Science as a discipline whereas PBRF broke this down into smaller groupings like physics, biology, etc. Supplementary Materials contains full information for each dataset. Due to data limitations, we consider gender as binary, data on non-binary individuals were either not available (EIGE and CIHR) or the group was too small to be included (PBRF and ARC).

Research quality evaluations: We use the Aotearoa New Zealand (ANZ) PBRF data. This dataset contains holistic research portfolio evaluations of every research-active academic in ANZ over three time spans (2000–06, 2007–12, 2013–18). The holistic research portfolio includes an individual’s self-nominated best three or four publications, in full, with descriptions of the importance and impact of the publications. It also includes non-publication aspects such as contributions to the research environment like chairing a conference organisation committee. Evaluation is done by subject area panels of experts, who read the publications and evaluate the individual’s holistic contribution. It is not an automated process and does not explicitly use measures such as h-indices, number of publications, or number of citations. The evaluation explicitly claims to focus on quality over quantity (Commission TE, 2017). Panels are likely to be more gender balanced than the fields they assess but in the more heavily skewed fields (e.g. Education, Engineering, Mathematics) they are skewed as would be expected. The PBRF is described in more detail in Supplementary Materials and summary statistics by year and discipline are given in Supplementary file 2.

With over 18,000 individual data points this dataset is a unique opportunity to explore a question such as this. To examine the robustness of any findings to bibliometric measures, we use a subset of the PBRF data from one university to compare anonymised detailed research output data to each individual’s PBRF score.

Funding success rates: We use three independent publicly available datasets: ARC application success rates from 2010 to 2019; Canadian Institute of Health Research application data from 2013 to 2016; and data covering government research grants across the EU and UK in 2019 from the EIGE. These datasets all provided aggregated data on the number of applications and successes by gender for each research discipline, i.e., not individual-level data.

The datasets are briefly described in the text, Table 1, and Supplementary file 1, and full data details are in Supplementary Materials. Each dataset is analysed separately and the analysis includes the sample size of each aggregated data point. We define a maximal model including two-way interactions between the three key variables: the gender balance of the discipline measured by the proportion of researchers who are men, $p_{men}$; researcher gender, $Gender$; and, where appropriate, $Time$. Any other available variables, e.g., researcher age, institute, or country were included as linear terms with either a fixed (Age) or random (Institute or Country) effect. We measure ‘best’ by the lowest AIC and the choice is robust by other similar measures, e.g., maximising Pearson’s r-squared or omitting non-significant variables. Table 2 shows the maximal models and the best models for each of the datasets.

Scoring research quality

First, we look at the only country that scores all individual university researchers. The ANZ PBRF scores are individual-level micro-data on the results of 18,371 holistic research quality assessment scores of 13,555 individuals in ANZ for the PBRF assessments. The assessments were carried out at three separate time points. In each of 2006, 2012, and 2018, staff members at all of the eight universities in ANZ were given a research performance score from 0 to 700 based on a holistic assessment of their research portfolio over the previous 6 years. Individuals with scores over 200 were considered research active. At each assessment, between 4000 and 8000 researchers were assessed. The scores, based on assessments by local and international experts on 14 different subject area panels, were used to allocate government performance-based research funding to universities. Although many researchers were present for more than one round, we analyse each of the three time points separately to examine the link between research score and the gender balance of a researcher’s discipline.

We use a linear regression model to predict the research quality score using the available variables, in particular the gender balance of the discipline, $p_{men}$. Although the data is strictly count data requiring a Poisson model we use a simple linear model as all the predicted values are large, i.e., between 200 and 700. We start with the maximal model of Table 2 that includes our key variables of interest $Gender$ and $p_{men}$ with interactions and the confounding variables $Age$ and $Institute$ as linear terms with no interactions. We compare all candidate models up to the maximal model and choose the best model by minimising AIC. The Supplementary file 3 gives the full output from all candidate models. For all datasets all terms in the best model (PBRF3 in Supplementary file 3) were significant (${\displaystyle \mathrm{p}<0.05}$) and the best model had a ${\displaystyle \mathrm{\Delta }AIC>2}$ compared to the second best.

The best model for all three time points was

$Score\sim Age+Institute+Gender\ast p_{men}.$

The model predicts that individuals in disciplines dominated by men have higher research performance scores than individuals in female-dominated disciplines; all variables are significant (${\displaystyle \mathrm{p}<0.05}$, see Supplementary file 3) at all time points.

Even after accounting for age in this way, men have higher scores than women; but the difference between the most male-dominated disciplines and the most female-dominated is larger than the gap between men and women within any particular discipline. Figure 1 shows the mean raw score of each discipline by gender (women red +, men blue x) in each assessment period (Figure 1A) 2006, (Figure 1B) 2012, (Figure 1C) 2018. The line is the expected model output for a 50-year-old male (blue line) or female (red line) researcher at the University of Canterbury (UC) (other ages and institutions differ by a constant as they are included as linear terms with no interactions). Note that the full model predictions (lines on Figure 1) are adjusted for age so may appear different from the raw data (points on Figure 1). In particular, raw data points from discipline/gender groups with many young researchers will appear much lower than might be expected at first thought.

Figure 1

Download asset Open asset

By way of example, we consider a 50-year-old researcher in 2018 at UC (i.e. lines in Figure 1C), $Institute$ and $Age$ are linear terms in the model so a different aged researcher at another university would show a very similar pattern. We assume our researcher comes from either a male-dominated discipline ($p_{men}=75\text{\%}$) like Philosophy or Physics; a gender-balanced discipline ($p_{men}=50\text{\%}$) such as Psychology or Law; or a female-dominated discipline ($p_{men}=25\text{\%}$) like Nursing or Education. In the gender-balanced discipline (Psychology or Law) a woman’s expected score is 396 out of a possible 700. This is around 8% lower than a man’s expected score of 432/700 in the same discipline. If a man works in a male-dominated discipline (Physics or Philosophy), his mean score is now expected to be higher at 452/700. If a man works in a female-dominated discipline (Nursing or Education), his score will drop to 411/700. In comparison, a female Philosopher or Physicist can expect a score of 433/700, only slightly lower than a male researcher in the same discipline. If a woman works in Nursing or Education, her expected score is now 359/700, a drop of over 20% from the female Philosopher’s expected score and also lower than the male Nursing researcher’s score.

This effect, of researchers in female-dominated disciplines being predicted to have lower scores than researchers in male-dominated disciplines, persists for all three assessment rounds but was much larger in the 2006 round (Figure 1A). In all three rounds, the score drop associated with moving from a male- to a female-dominated discipline is larger for women than for men. Finally, the gender gap within a discipline is much larger in female-dominated disciplines than in male-dominated ones, though the gap has closed considerably between 2006 and 2018.

Accounting for bibliometric differences between individuals

Here, we explore whether gendered patterns in publication explain the variation in research scores across disciplines. We show that publications are important but they do not negate our previous result that researchers in female-dominated disciplines receive lower research scores.

We analysed a subset of data available from UC, one of the eight universities contained in the PBRF datasets. We included research-active staff (i.e. PBRF score >200) with at least one publication between 2013 and 2018, working at UC in 2018. Using date of birth, gender, ethnicity, and discipline we were able to link the university’s internal research publications dataset to individual PBRF scores for over 80% of individuals ($N=384$, 141 female, 243 male). Data records that could not be linked were excluded.

These publication records allowed us to calculate a series of research bibliometrics, e.g., number of outputs, number of citations, expected field weighted citation index of publications, etc., for each researcher. We then used these bibliometrics to predict each individual’s expected research score (see Supplementary Materials for a more detailed description of the data and the analysis). The bibliometrics were tested separately in an individual regression model

$Score\sim Age+Gender+log\left(Bibliometric\right).$

The bibliometric that was the best predictor of research score (with minimum AIC and maximum r-squared) was the total number of outputs $N_{outputs}$ (see Supplementary file 4, model 2). This output gave a model correlation of $r^2=0.325$ and the coefficient of $N_{outputs}$ was highly significant (${\displaystyle \mathrm{p}<0.001}$). In this model $Age$ was also significant (${\displaystyle \mathrm{p}<0.01}$) but $Gender$ was not (${\displaystyle \mathrm{p}>0.05}$). The second-best bibliometric predictor by AIC, number of outputs weighted by number of authors, was also a measure of research quantity. It had correlation $r^2=0.280$. By comparison the other bibliometrics, all measures more related to research quality, had ${\displaystyle r^2<0.09}$. We also tested a two bibliometric model by combining the best predictor, $N_{outputs}$, with each of the other bibliometrics separately. In all cases the change in AIC due to the addition of the second bibliometric was very small (${\displaystyle \mathrm{\Delta }AIC<2}$) and the secondary bibliometric was not significant (${\displaystyle \mathrm{p}>0.05}$) (see Supplementary file 4).

The bibliometric analysis shows several things. First, Figure 2A shows the best available predictor of research score was number of research outputs; and bibliometrics related to quality (e.g. expected field weighted citation index) had almost no effect. Alone, this result is surprising as documentation on PBRF clearly states that the evaluations are based on quality not quantity of research (Commission TE, 2017). Similarly, $Gender$ is not a predictor of research quality, a man and a woman of the same age with the same number of research outputs have the same expected score. Figure 2A shows the raw scores of all individuals in the single university sample and the predicted relationship between number of outputs and score for a 50-year-old of either gender. On average, men have more outputs than women (Figure 2A, vertical error bars) and get higher scores (horizontal error bars).

Figure 2

Download asset Open asset

Having seen that score is strongly correlated to the number of research outputs (Figure 2A), we then test the relationship between the number of outputs and the proportion of men in the discipline using the same sample of 384 individuals at UC (Figure 2B). We start with the maximal model to allow the relationship to change for each gender

$log\left(N_{outputs}\right)\sim p_{men}\ast Gender.$

The gender balance of the discipline is highly significant (coefficient $p_{men}$: ${\displaystyle p<0.001}$) but again $Gender$ is not significant either alone or with an interaction (${\displaystyle p>0.05}$). So the most parsimonious model is

$log\left(N_{outputs}\right)\sim p_{men}$

meaning a man and woman in the same discipline have, on average, the same number of outputs and researchers in male-dominated disciplines have, on average, more publications. Taken together, it is tempting to use these correlations as an explanation for the lower scores for researchers in female-dominated disciplines. In other words, both men and women in female-dominated disciplines publish less, and fewer publications lead to a lower score. This explanation fails the empirical test because, as seen in Figure 1, on average a man receives a higher score than a woman in the same discipline. This is despite, on average, having the same number of outputs (Figure 2B).

Finally, we test the relationship of our previous analysis relating $Score$ to $Gender$ and gender balance, $p_{men}$, but now we add the best bibliometric predictor $N_{outputs}$

$Score\sim Age+Gender\ast p_{men}+log\left(N_{outputs}\right).$

After using this model to account for the difference in publication norms across disciplines with different gender balance, the proportion of men in the discipline is still a significant predictor of score coefficient and interaction ${\displaystyle \mathrm{p}<0.05}$, see Supplementary file 4, but now the relationship is more intricate.

In a female-dominated ($p_{men}=25\text{\%}$) discipline, like Nursing or Education, the expected number of outputs is low for both genders (Figure 2B) and scores are correspondingly low. But given that a man and a woman have the same expected number of outputs for their discipline (as predicted by Figure 2B), the man’s expected score of 412/700 is higher than the woman’s of 361/700 (Figure 2C, left bars). When we consider a male-dominated (${\displaystyle p_{men}=75\mathrm{\%}}$) discipline, like Physics or Philosophy, given gender-equal (but now likely to be high) publication rates, overall scores are much higher, a man’s score is predicted to be 467/700 and a woman’s is higher still at 505/700 (Figure 2C, right bars). In short, even after accounting for variations in publication rates, researchers in female-dominated disciplines still have lower expected research scores.

In this detailed look at PBRF scores compared to bibliometric measures, the gender minority has an advantage over the majority at both ends of the gender balance spectrum. If those in the gender minority of a discipline have the same number of publications as their majority colleagues (as our findings suggest they usually do), the gender minority will, on average, receive a higher research evaluation score.

Here, we have presented a more detailed analysis which accounts for publication differences. Whilst differences between the gender minority and majority in a discipline continue to exist, the key point remains: when women are in the majority, the scores are lower for everyone (Figure 2C).

Far more men work in male-dominated disciplines (PBRF 2018: 74% of men in disciplines with ${\displaystyle p_{men}>50\mathrm{\%}}$) than women (PBRF 2018: 47% of women in disciplines with ${\displaystyle p_{men}>50\mathrm{\%}}$). Hence, men will benefit overall from a pattern of evaluating research in male-dominated disciplines as higher quality than research in female-dominated disciplines.

We present two similar, yet separate, findings: (1) gender balance in a discipline correlates with research quality scores; and (2) gender balance in a discipline correlates with research funding success rates. These findings do not identify a causal relationship in which gender balance causes low research scores or funding success or vice versa. But we can highlight and explore some possible explanations and theories that might link gender balance and research score/funding success. Further, even without establishing causality, these paired findings offer another piece of the academic gender puzzle.

All publicly available data analysed during this study are included in the manuscript and supporting files. The NZ PBRF data is not publicly available but can be requested from the NZ TEC via the Official Information Act.

1. 1. Albers CJ

   (2015) Dutch research funding, gender bias, and Simpson’s paradox

   PNAS 112:E6828–E6829.

   https://doi.org/10.1073/pnas.1518936112

   • PubMed
   • Google Scholar
2. 1. Baker M
   2. Fortin NM

   (2001) Occupational gender composition and wages in Canada, 1987–1988

   Canadian Journal of Economics/Revue Canadienne d’économique 34:345–376.

   https://doi.org/10.1111/0008-4085.00078

   • Google Scholar
3. 1. Brower A
   2. James A

   (2020) Research performance and age explain less than half of the gender pay gap in New Zealand universities

   PLOS ONE 15:e0226392.

   https://doi.org/10.1371/journal.pone.0226392

   • PubMed
   • Google Scholar
4. 1. Brower A
   2. James A

   (2023) Sticky floors, double-binds, and double whammies: adjusting for research performance reveals universities’ gender pay gap is not disappearing

   SAGE Open 13:2323.

   https://doi.org/10.1177/21582440231192323

   • Google Scholar
5. 1. Carli G
   2. Tagliaventi MR
   3. Cutolo D

   (2019) One size does not fit all: the influence of individual and contextual factors on research excellence in academia

   Studies in Higher Education 44:1912–1930.

   https://doi.org/10.1080/03075079.2018.1466873

   • Google Scholar
6. 1. Ceci SJ
   2. Ginther DK
   3. Kahn S
   4. Williams WM

   (2014) Women in academic science: A changing landscape

   Psychological Science in the Public Interest 15:75–141.

   https://doi.org/10.1177/1529100614541236

   • PubMed
   • Google Scholar
7. 1. Ceci SJ
   2. Kahn S
   3. Williams WM

   (2023) Exploring gender bias in six key domains of academic science: An adversarial collaboration

   Psychological Science in the Public Interest 24:15–73.

   https://doi.org/10.1177/15291006231163179

   • PubMed
   • Google Scholar
8. 1. Chatterjee P
   2. Werner RM

   (2021) Gender disparity in citations in high-impact journal articles

   JAMA Network Open 4:e2114509.

   https://doi.org/10.1001/jamanetworkopen.2021.14509

   • PubMed
   • Google Scholar
9. Book

   1. Commission TE

   (2017)

   Performance-based research fund: a guide for staff members participating in the 2018 quality evaluation

   Tertiary Education Commission.

   • Google Scholar
10. 1. Ding WW
    2. Murray F
    3. Stuart TE

    (2006) Gender differences in patenting in the academic life sciences

    Science 313:665–667.

    https://doi.org/10.1126/science.1124832

    • PubMed
    • Google Scholar
11. 1. Dion ML
    2. Sumner JL
    3. Mitchell SM

    (2018) Gendered citation patterns across political science and social science methodology fields

    Political Analysis 26:312–327.

    https://doi.org/10.1017/pan.2018.12

    • Google Scholar
12. 1. El-Alayli A
    2. Hansen-Brown AA
    3. Ceynar M

    (2018) Dancing backwards in high heels: female professors experience more work demands and special favor requests, particularly from academically entitled students

    Sex Roles 79:136–150.

    https://doi.org/10.1007/s11199-017-0872-6

    • Google Scholar
13. 1. England P
    2. Allison P
    3. Wu Y

    (2007) Does bad pay cause occupations to feminize, does feminization reduce pay, and how can we tell with longitudinal data?

    Social Science Research 36:1237–1256.

    https://doi.org/10.1016/j.ssresearch.2006.08.003

    • Google Scholar
14. 1. Grogan KE

    (2019) How the entire scientific community can confront gender bias in the workplace

    Nature Ecology & Evolution 3:3–6.

    https://doi.org/10.1038/s41559-018-0747-4

    • Google Scholar
15. Website

    1. Harzing AW

    (2008) Reflections on the h-index

    Accessed April 23, 2008.

    http://wwwharzingcom/pop_hindexhtm
16. 1. Hofstra B
    2. Kulkarni VV
    3. Munoz-Najar Galvez S
    4. He B
    5. Jurafsky D
    6. McFarland DA

    (2020) The diversity-innovation paradox in science

    PNAS 117:9284–9291.

    https://doi.org/10.1073/pnas.1915378117

    • PubMed
    • Google Scholar
17. 1. Hoppe TA
    2. Litovitz A
    3. Willis KA
    4. Meseroll RA
    5. Perkins MJ
    6. Hutchins BI
    7. Davis AF
    8. Lauer MS
    9. Valantine HA
    10. Anderson JM
    11. Santangelo GM

    (2019) Topic choice contributes to the lower rate of NIH awards to African-American/black scientists

    Science Advances 5:eaaw7238.

    https://doi.org/10.1126/sciadv.aaw7238

    • PubMed
    • Google Scholar
18. 1. Hunter LA
    2. Leahey E

    (2010) Parenting and research productivity: New evidence and methods

    Social Studies of Science 40:433–451.

    https://doi.org/10.1177/0306312709358472

    • Google Scholar
19. 1. James A
    2. Chisnall R
    3. Plank MJ

    (2019) Gender and societies: a grassroots approach to women in science

    Royal Society Open Science 6:190633.

    https://doi.org/10.1098/rsos.190633

    • PubMed
    • Google Scholar
20. 1. James A
    2. Brower A

    (2022) Levers of change: using mathematical models to compare gender equity interventions in universities

    Royal Society Open Science 9:220785.

    https://doi.org/10.1098/rsos.220785

    • PubMed
    • Google Scholar
21. 1. Key EM
    2. Sumner JL

    (2019) You research like a girl: gendered research agendas and their implications

    PS 52:663–668.

    https://doi.org/10.1017/S1049096519000945

    • Google Scholar
22. 1. Kieokaew R
    2. IDEEA collaboration

    (2023) Unconscious bias in academia: its effects in geoscience and space science research communities and call for actions

    Display 01:4670.

    https://doi.org/10.5194/egusphere-egu23-14670

    • Google Scholar
23. 1. Kim L
    2. Smith DS
    3. Hofstra B
    4. McFarland DA

    (2022) Gendered knowledge in fields and academic careers

    Research Policy 51:104411.

    https://doi.org/10.1016/j.respol.2021.104411

    • Google Scholar
24. 1. Lancho-Barrantes BS
    2. Guerrero-Bote VP
    3. Moya-Anegón F

    (2010) What lies behind the averages and significance of citation indicators in different disciplines?

    Journal of Information Science 36:371–382.

    https://doi.org/10.1177/0165551510366077

    • Google Scholar
25. 1. Larivière V
    2. Ni C
    3. Gingras Y
    4. Cronin B
    5. Sugimoto CR

    (2013) Bibliometrics: global gender disparities in science

    Nature 504:211–213.

    https://doi.org/10.1038/504211a

    • PubMed
    • Google Scholar
26. 1. Leslie SJ
    2. Cimpian A
    3. Meyer M
    4. Freeland E

    (2015) Expectations of brilliance underlie gender distributions across academic disciplines

    Science 347:262–265.

    https://doi.org/10.1126/science.1261375

    • PubMed
    • Google Scholar
27. 1. Ley TJ
    2. Hamilton BH

    (2008) Sociology. The gender gap in NIH grant applications

    Science 322:1472–1474.

    https://doi.org/10.1126/science.1165878

    • PubMed
    • Google Scholar
28. 1. Mairesse J
    2. Pezzoni M

    (2015) Does gender affect scientific productivity? A critical review of the empirical evidence and A panel data econometric analysis for French physicists

    Revue Économique 66:65–113.

    https://doi.org/10.3917/reco.661.0065

    • Google Scholar
29. 1. Meho LI

    (2021) The gender gap in highly prestigious international research awards, 2001–2020

    Quantitative Science Studies 2:976–989.

    https://doi.org/10.1162/qss_a_00148

    • Google Scholar
30. 1. Olejniczak AJ
    2. Savage WE
    3. Wheeler R

    (2022) The rhythms of scholarly publication: Suggestions to enhance bibliometric comparisons across disciplines

    Frontiers in Research Metrics and Analytics 7:812312.

    https://doi.org/10.3389/frma.2022.812312

    • PubMed
    • Google Scholar
31. 1. O’Meara K
    2. Kuvaeva A
    3. Nyunt G
    4. Waugaman C
    5. Jackson R

    (2017) Asked more often: Gender differences in faculty workload in research universities and the work interactions that shape them

    American Educational Research Journal 54:1154–1186.

    https://doi.org/10.3102/0002831217716767

    • Google Scholar
32. Conference

    1. Poppenhaeger K

    (2019) Unconscious gender bias in academia: From PhD students to professors

    AIP Conference Proceedings.

    https://doi.org/10.1063/1.5110149

    • Google Scholar
33. 1. Pritlove C
    2. Juando-Prats C
    3. Ala-leppilampi K
    4. Parsons JA

    (2019) The good, the bad, and the ugly of implicit bias

    The Lancet 393:502–504.

    https://doi.org/10.1016/S0140-6736(18)32267-0

    • Google Scholar
34. 1. Puuska HM
    2. Muhonen R
    3. Leino Y

    (2014) International and domestic co-publishing and their citation impact in different disciplines

    Scientometrics 98:823–839.

    https://doi.org/10.1007/s11192-013-1181-7

    • Google Scholar
35. 1. Ross MB
    2. Glennon BM
    3. Murciano-Goroff R
    4. Berkes EG
    5. Weinberg BA
    6. Lane JI

    (2022) Women are credited lessin science than men

    Nature 608:135–145.

    https://doi.org/10.1038/s41586-022-04966-w

    • PubMed
    • Google Scholar
36. 1. Schmal WB
    2. Haucap J
    3. Knoke L

    (2023) The role of gender and coauthors in academic publication behavior

    SSRN Electronic Journal 52:241.

    https://doi.org/10.2139/ssrn.4410241

    • Google Scholar
37. 1. Shen H

    (2013) Mind the gender gap

    Nature 495:022a.

    https://doi.org/10.1038/495022a

    • Google Scholar
38. 1. Skov T

    (2020) Unconscious gender bias in academia: scarcity of empirical evidence

    Societies 10:31.

    https://doi.org/10.3390/soc10020031

    • Google Scholar
39. 1. Street JM
    2. Rogers WA
    3. Israel M
    4. Braunack-Mayer AJ

    (2010) Credit where credit is due? Regulation, research integrity and the attribution of authorship in the health sciences

    Social Science & Medicine 70:1458–1465.

    https://doi.org/10.1016/j.socscimed.2010.01.013

    • Google Scholar
40. 1. Symonds MRE
    2. Gemmell NJ
    3. Braisher TL
    4. Gorringe KL
    5. Elgar MA

    (2006) Gender differences in publication output: towards an unbiased metric of research performance

    PLOS ONE 1:e127.

    https://doi.org/10.1371/journal.pone.0000127

    • PubMed
    • Google Scholar
41. 1. West JD
    2. Jacquet J
    3. King MM
    4. Correll SJ
    5. Bergstrom CT

    (2013) The role of gender in scholarly authorship

    PLOS ONE 8:e66212.

    https://doi.org/10.1371/journal.pone.0066212

    • PubMed
    • Google Scholar
42. 1. Witteman HO
    2. Hendricks M
    3. Straus S
    4. Tannenbaum C

    (2019) Are gender gaps due to evaluations of the applicant or the science? A natural experiment at a national funding agency

    The Lancet 393:531–540.

    https://doi.org/10.1016/S0140-6736(18)32611-4

    • Google Scholar

Author details

1. Alex James

   School of Mathematics and Statistics, University of Canterbury, Christchurch, New Zealand

   Contribution

   Conceptualization, Data curation, Formal analysis, Investigation, Visualization, Methodology, Writing - original draft, Project administration, Writing - review and editing

   For correspondence

   alex.james@canterbury.ac.nz

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-1543-7139

2. Franca Buelow

   1. Bioprotection Centre of Research Excellence, University of Lincoln, Lincoln, Lincoln, New Zealand
   2. School of Earth and Environment, University of Canterbury, Christchurch, New Zealand

   Contribution

   Methodology, Writing - original draft, Writing - review and editing

   Competing interests

   No competing interests declared

3. Liam Gibson

   School of Mathematics and Statistics, University of Canterbury, Christchurch, New Zealand

   Contribution

   Investigation, Methodology

   Competing interests

   No competing interests declared

4. Ann Brower

   School of Earth and Environment, University of Canterbury, Christchurch, New Zealand

   Contribution

   Conceptualization, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing

   Competing interests

   No competing interests declared

• 1,415

  views

• 65

  downloads

• 2

  citations

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