Gender Inequality in Science: A Selective Review

I continue to move some of the posts I especially liked from my Telegram channel to my blog. This time, it is a post about gender inequality in science. I originally wrote it in 2024; now I have slightly updated it and added the results of two new studies published in 2025.

In general, I am rather cautious about the topic of gender inequality, because it is very often exploited and discussed excessively, sometimes with distorted facts, and it is not always clear where there is real inequality and where there is “agenda-setting.” But then I came across one paper, then another, and eventually this post emerged, inspired by several studies. I do not claim this to be a comprehensive review: I selected papers that I personally found interesting and that, while reading, seemed to naturally connect to one another.

The Beginning

It all started quite innocently, with a paper on how researchers’ personal characteristics may be related to the statistical methods they use. This was somewhat unexpected, so I decided to read it (Dunn et al., 2020). The authors examined which personality traits are associated with how researchers specify parameters in Bayesian analysis, in particular prior distributions. They found (in a sample of 900 psychology researchers) that researchers with higher levels of confidence tend to choose higher prior values. Male researchers also tend to use higher prior values more often than female researchers. This is partly explained by the fact that male researchers report higher levels of self-confidence than female researchers.

And at this point, I got drawn in 😊 I started following various references to see what is actually known about differences between male and female scientists.

Male authors use positive words more often

In a paper by Lerchenmueller and colleagues, published in 2019, some interesting results were reported. They downloaded more than 100,000 articles from clinical research (which I will loosely refer to as medical research) and more than 6 million life sciences articles published between 2002 and 2017. They then analyzed article titles and abstracts to examine whether men and women differ in how often they use “positive” words. These included words such as novel, excellent, unique, promising, favorable, and similar terms.

They divided all articles into four categories:

1. first and last author are men;

2. first author is a man, last author is a woman;

3. first author is a woman, last author is a man;

4. first and last author are women.

The importance of the first author is obvious, and the last author matters because, in papers with many authors, the last author is most often the principal investigator (project leader, grant holder, main driver, etc.).

Already at this stage, a difference becomes apparent. Among medical studies, 57% of papers had both first and last authors who were men (51% in life sciences). Only 17% of papers had both first and last authors who were women (16% in life sciences). Overall, men were first authors in 66% of medical papers and in 62% of life sciences papers. Men were last authors in 73% of cases in medical research and in 74% in life sciences.

Of course, there is room for interpretation as to why this difference exists. One could assume that men publish more papers (for various reasons). One could also assume that men are published more often because of biases among editors and reviewers, that men more often participate in higher-quality projects and studies (again, for various reasons), that men receive grants more often, that men more often lead research groups, and so on (I will mention some relevant data on this later).

But let us return to positive words and how men and women use them differently. The study showed that in papers where the first author is a man, positive words are used significantly more often in titles and abstracts. The largest difference was found for the word novel, as well as for unique, promising, and favorable.

This observation is, of course, amusing, but can it be interpreted as a marker of greater male confidence? Possibly, although this is not the only interpretation. For example, a man may come up with a title containing the word novel (here is a real example: “Antibody-drug conjugates: A promising novel therapeutic approach in lung cancer,” which includes both novel and promising). Why did he choose such a title?

• Perhaps the method is genuinely new. In that case, the gender difference may reflect the fact that men more often participate in projects that test and develop new approaches, either because they are more often invited or because they themselves take more risks.

• Or perhaps the method is not particularly new, but the title was chosen to attract attention and increase citation chances.

• Or perhaps men are indeed more confident and perceive their methods as novel and unique, whereas women are more likely to doubt themselves.

As always, things are complicated 😊

Positive words are important

But why does it matter which words are used in titles at all? Because there are studies showing that articles whose titles and abstracts contain such positive words are cited more often (e.g., Wu, Wu, & Li, 2024).

It is also worth noting that with the widespread use of large language models (LLMs) in research, things have become even more complicated. There are studies showing that abstracts and titles written with the help of LLMs use certain words significantly more often, including words that can be classified as “positive,” such as potential, crucial, comprehensive, enhancing, exhibited, and insights (Kobak et al., 2025). Incidentally, there is also research showing that LLMs themselves exhibit gender bias, but that is a separate topic.

In this context, it remains unclear whether the widespread adoption of LLMs will attenuate gender differences in linguistic self-presentation in science, or whether such tools will instead reproduce and possibly amplify existing patterns of bias.

Gender Gap in Psychological Research: Publication, Citation, and Pay Differences

To look at more concrete data on actual differences between men and women in science, I turned to another paper titled The Future of Women in Psychological Science. Interestingly, this paper has almost 60 co-authors, all of whom are women.

This paper presents interesting data on the gender gap in psychology. For example, it cites a study by Odic and Wojcik (2020), who examined the proportion of male and female authors in publications from 2003 to 2018 in 130 highly influential journals (more than 200,000 publications and 770,000 authors in total) and found a significant difference in favor of men. Overall, the difference is not very large (55.8% male authors), but across subfields of psychology it can be substantial. For example, in perception research, women account for only 30% of authors; in neuroscience, 36%; whereas in developmental psychology, women account for nearly 60%.

The authors reported other differences between men and women in psychology:

• the proportion of female authors decreases as journal impact factor increases: the higher the impact factor, the lower the proportion of women.

• articles with male first authors are cited 1.3 times more often than those with female first authors.

• Men also have significantly higher h-indices (by approximately 4.5 points), even after controlling for position and experience.

• Holding the same academic position, women earn less. For example, among full professors, women’s salaries amount to about 88% of those of male professors.

How does such a gap emerge?

In a paper by Lerchenmueller and Sorenson (The gender gap in early career transitions in the life sciences), the authors analyzed the career trajectories of 6,336 researchers in biology, from degree completion to principal investigator positions, using data from the U.S. National Institutes of Health. They showed that in the period after postdoctoral training, when researchers must transition to independence and begin leading their own research projects, a substantial proportion of women start to “drop out.” While the proportion of men and women during the postdoctoral period is approximately equal, by the time of reaching a professorial position, women account for only about 20%.

According to the authors, the key moment is the acquisition of an independent research grant. They examined the receipt of R01 grants from the National Institutes of Health. These grants are awarded to scientists who have demonstrated competence in a particular field and fund independent research projects. They account for nearly half of all NIH grants and are the main source of funding for most academic biomedical research groups in the United States.

According to the data used in the study, women receive such grants 20% less often than men. This, in turn, negatively affects the probability that a woman will later obtain a faculty position.

The authors also note that higher male productivity only partially explains this difference in grant success. They propose several additional explanations (beyond productivity differences):

• women spend more time on household labor than men and therefore have less time for research;

• women in research labs and centers more often engage in administrative and organizational work than men (whether voluntarily or because they are assigned these roles), again leaving less time for research;

• women have fewer role models at early career stages, because there are fewer senior women scientists, creating a vicious circle;

• grant reviewers are more often men, and they evaluate grant applications submitted by women as less valuable (possibly unconsciously).

Women Leave STEM Careers Due to Childcare Twice as Often as Men

Some of these points are supported by other studies. For example, there is evidence that women are more constrained by household responsibilities. There is also an interesting study on how childbirth and parenthood more generally affect careers (Cech & Blair-Loy, 2019). Longitudinal evidence from U.S. STEM professionals shows that parenthood is a major point of attrition from full-time STEM employment. Within 4–7 years after the birth of a first child, 43% of mothers and 23% of fathers leave full-time STEM roles, and most do not return, even after children reach school age. Among those who exited STEM for full-time work in other fields, 71% of new mothers and 38% of new fathers cited family-related reasons for their departure, compared to only about 4–5% among childless professionals. Notably, women leave after childbirth almost twice as often as men and much more frequently cite family responsibilities as the reason.

Evaluation Bias and the Use of Scientific Ideas

There is also another study confirming biases in evaluation, this time focusing on the use of scientific ideas in technological development. In this work, the authors examined how often scientific articles are cited in patents as a measure of technological impact (Bikard, Fernandez-Mateo, & Mogra, 2025). The analysis drew on data from tens of millions of scientific publications published between 1980 and 2020, as well as information on which of these papers were subsequently cited in patent documentation. The main result showed that articles with female lead authors are cited in patents significantly less often than articles authored by men, even when scientific quality and topic are comparable. This effect was stable over time, observed across scientific fields, and persisted after controlling for numerous characteristics, including journal features, number of co-authors, academic reputation, type of research, and scientific networks. To further reduce the influence of unobserved quality differences, the authors used a “paper twins” design, comparing papers describing independent but nearly simultaneous discoveries in the same field. Even within these matched pairs, women’s papers were less likely to become the basis for technological developments.

What interested me most, however, were the results of the experiment conducted by the authors. They asked 400 researchers with PhDs to evaluate the value of a scientific idea presented in an abstract, randomly assigning a male or female author name. The results showed that the “male” versions of abstracts were read longer, rated as more important, and perceived as more promising for future use. Taken together, these results point to a clear bias: even when content is identical, women’s scientific ideas receive less attention and are evaluated as less valuable and less promising.

Personal thoughts

I will conclude my post here. Clearly, this topic is inexhaustible. Clearly, the gap exists, and it may be experienced by women as unfair and difficult to overcome. But if we compare the current situation with what existed 100 years ago when, for example, Titchener did not allow women to attend his psychology seminars - the changes are obvious. So perhaps the situation will continue to change. There is no need to despair.

As a woman researcher, I have ambivalent feelings. On the one hand, I would like to be evaluated solely based on what I do, without any adjustment for the fact that I am a woman, because that feels demeaning. At the moment, I feel substantial inequality in the research environment, but mainly because everything is designed for researchers with institutional affiliation. For independent researchers, things are much more difficult (for example, I cannot even verify my Google Scholar profile because they do not accept Gmail addresses).

On the other hand, if inequality exists and is sustained even at an unconscious level, then perhaps granting some

form of “advantage”—for example, in hiring women—simply repairs an organizational system that is fundamentally flawed. Similarly, if grant reviews were fully blind, or at least if names were removed from CVs when evaluating a researcher’s ability to carry out a grant, some of these biases might be reduced.

Reference:

Bikard, M., Fernandez-Mateo, I., & Mogra, R. (2025). Standing on the shoulders of (male) giants: Gender inequality and the technological impact of scientific ideas. Administrative Science Quarterly, 00018392251331957.

Cech, E. A., & Blair-Loy, M. (2019). The changing career trajectories of new parents in STEM. Proceedings of the National Academy of Sciences, 116(10), 4182-4187.

Dmitry Kobak et al. Delving into LLM-assisted writing in biomedical publications through excess vocabulary.Sci. Adv.11,eadt3813(2025).DOI:10.1126/sciadv.adt3813

Dunn, E. W., Chen, L., Proulx, J. D. E., Ehrlinger, J., & Savalei, V. (2020). Can Researchers’ Personal Characteristics Shape Their Statistical Inferences? Personality and Social Psychology Bulletin, 47(6), 969-984. https://doi.org/10.1177/0146167220950522 (Original work published 2021)

Gruber, J., Mendle, J., Lindquist, K. A., Schmader, T., Clark, L. A., Bliss-Moreau, E., ... & Williams, L. A. (2021). The future of women in psychological science. Perspectives on Psychological Science, 16(3), 483-516.

Lerchenmueller, M. J., & Sorenson, O. (2018). The gender gap in early career transitions in the life sciences. Research Policy, 47(6), 1007-1017.

Lerchenmueller, M. J., Sorenson, O., & Jena, A. B. (2019). Gender differences in how scientists present the importance of their research: observational study. bmj, 367.

Wu, D., Wu, H. & Li, J. Citation advantage of positive words: predictability, temporal evolution, and universality in varied quality journals. Scientometrics 129, 4275–4293 (2024). https://doi.org/10.1007/s11192-024-05074-4