# Historical comparison of gender inequality in scientific careers across countries and disciplines

> An annotated edition of this paper on Fermat's Library, where scientists and readers explain its ideas, equations and history in the margins. Annotated edition: https://fermatslibrary.com/s/historical-comparison-of-gender-inequality-in-scientific-careers-across-countries-and-disciplines

- **Authors:** Barabasi et al.
- **Original:** https://s3-us-west-1.amazonaws.com/fermatslibrary/uploads%2F1589244918701-x4q9lmp33x-0da825a8831fa3e12afe9e260d445a0c%2F4609.full.pdf
- **Pages:** 8
- **Annotations:** 4 comments in 4 threads
- **Featured:** Fermat's Library Journal Club
- **Annotated edition:** https://fermatslibrary.com/s/historical-comparison-of-gender-inequality-in-scientific-careers-across-countries-and-disciplines

## Annotations

### Annotation 1 (page 1)

Beside the passage: “Junming Huang a,b,c,1 , Alexander J. Gates a,1 , Roberta Sinatra d,e , and Albert-L ´ aszl ´ o Barab ´ asi a,f,g,h,2”

**Annotation** by Tymor Hamamsy (2020-05-12):

> Albert-Laszlo Barabasi is a physicist and network scientist who introduced the concept of scale-free networks and proposed the Barabási–Albert model to explain their widespread emergence in natural, technological and social systems. He is the director of the Center for Network Science at Northeastern University. Here is a link to the "Emergence of Scaling in Random Networks": https://science.sciencemag.org/content/286/5439/509.full
>
>
>
> Source: https://en.wikipedia.org/wiki/Albert-L%C3%A1szl%C3%B3_Barab%C3%A1si

### Annotation 2 (page 1)

Beside the passage: “Empirical evidence suggests signiﬁcant gender differences in the total productivity and impact of academic careers across science, technology, engineering, and mathematics (STEM) ﬁelds. Paradoxically, the increase in the number of women academics over the past 60 years has increased these gender differences. Yet, we …”

**Annotation** by Tymor Hamamsy (2020-05-12):

> An important finding from this paper, "This suggests the productivity and impact of gender differences are explained by different publishing career lengths and dropout rates."

### Annotation 3 (page 3)

Beside the passage: “The Increasing and Persistent Gender Gap Across all years and disciplines, women account for 27% of authors, a number that hides important trends: while in 1955 women represented only 12% of all active authors, that fraction steadily increased over the last century, reaching 35% by 2005 (Fig. 1A). Yet, these aggregate …”

**Annotation** by Tymor Hamamsy (2020-05-12):

> Always be careful when evaluating proportions without controlling for  disciplinary differences! Just take it from Simpson...
> https://en.wikipedia.org/wiki/Simpson%27s_paradox

### Annotation 4 (page 5)

Beside the passage: “Annual Productivity and Career Length As total productivity and impact over a career represent a convolution of annual productivity and publishing career length, to identify the roots of the gender gap, we must separate these two factors. Traditionally, the difﬁculty of reconstructing full publishing careers has …”

**Annotation** by Tymor Hamamsy (2020-05-12):

> This access to the full publishing career data was key to their discovery of the different career lengths of male and female researchers.

## Full text

Extracted from the PDF. [Annotation N] marks where each annotation above sits.

#### Page 1

SOCIAL SCIENCES

Historical comparison of gender inequality in scientiﬁc

careers across countries and disciplines

Junming Huang a,b,c,1 , Alexander J. Gates a,1 , Roberta Sinatra d,e , and Albert-L [Annotation 1] ´ aszl ´ o Barab ´ asi a,f,g,h,2

a Network Science Institute and Department of Physics, Northeastern University, Boston, MA 02115; b CompleX Lab, School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu 611731, China; c Paul and Marcia Wythes Center on Contemporary China, Princeton University, Princeton, NJ 08540; d Department of Computer Science, IT University of Copenhagen, 2300 Copenhagen, Denmark; e ISI Foundation, 10126 Turin, Italy; f Channing Division of Network Medicine, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA 02115; g Department of Medicine, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA 02115; and h Department of Network and Data Science, Central European University, 1051 Budapest, Hungary

Edited by Susan T. Fiske, Princeton University, Princeton, NJ, and approved January 22, 2020 (received for review August 15, 2019)

There is extensive, yet fragmented, evidence of gender differences in academia suggesting that women are underrepresented in most scientiﬁc disciplines and publish fewer articles throughout a career, and their work acquires fewer citations. Here, we offer a comprehensive picture of longitudinal gender differences in performance through a bibliometric analysis of academic publishing careers by reconstructing the complete publication history of over 1.5 million gender-identiﬁed authors whose publishing career ended between 1955 and 2010, covering 83 countries and 13 disciplines. We ﬁnd that, paradoxically, the increase of participation of women in science over the past 60 years was accompanied by an increase of gender differences in both productivity and impact. Most surprisingly, though, we uncover two gender invariants, ﬁnding that men and women publish at a comparable annual rate and have equivalent career-wise impact for the same size body of work. Finally, we demonstrate that differences in publishing career lengths and dropout rates explain a large portion of the reported career-wise differences in productivity and impact, although productivity differences still remain. This comprehensive picture of gender inequality in academia can help rephrase the conversation around the sustainability of women’s careers in academia, with important consequences for institutions and policy makers.

gender inequality | science of science | STEM | scientiﬁc careers

G

ender differences in academia, captured by disparities in the number of female and male authors, their productivity, citations, recognition, and salary, are well documented across all disciplines and countries (1–8). The epitome of gender difference is the “productivity puzzle” (9–13)—the persistent evidence that men publish more than women over the course of their career, which has inspired a plethora of possible explanations (14–16), from differences in family responsibilities (17–19), to career absences (20), resource allocation (21), the role of peer review (22), collaboration (23, 24), role stereotypes (25), academic rank (26), specialization (27), and work climate (28). The persistence of these gender differences could perpetuate the naive interpretation that the research programs of female and male scientists are not equivalent. However, such simplistic reading of the data dismisses increasing evidence that systemic barriers impede the female academic. Indeed, the deep interrelatedness of these factors has limited our ability to differentiate the causes from the consequences of the productivity puzzle, complicating the scientiﬁc community’s ability to enact effective policies to address it.

A key methodological obstacle has been the difﬁculty to reconstruct full publishing careers for scientists of both genders across the diverse academic population. Consequently, much of the available evidence on gender differences is based on case studies limited to subsets of active scientists in speciﬁc countries, disciplines, or institutions, making it difﬁcult to compare and generalize the ﬁnding to all of science. A further complication arises from the heavy-tailed nature of academia: a dispropor-

tionately small number of authors produce a large fraction of the publications and receive the majority of the citations (29), an effect that is exacerbated in small sample sizes (30). To truly understand the roots of the gender inequality, we need to survey the whole longitudinal, disciplinary, and geographical landscape, which is possible only if we capture complete publishing careers for all scientists across disciplinary and national boundaries.

Here, we reconstructed the full publishing career of 7,863,861 scientists from their publication record in the Web of Science (WoS) database between 1900 and 2016. By deploying a state-of-the-art method for gender identiﬁcation (SI Appendix, section S2.E), we identiﬁed the gender of over 3 million authors (856,889 female and 2,146,926 male) spanning 83 countries and 13 major disciplines (SI Appendix, section S2). We then focused on 1,523,002 scientists (412,808 female and 1,110,194 male) whose publishing careers ended between 1955 and 2010 (SI Appendix, sections S1 and S2.H), allowing us to systematically compare complete male and female careers. This extensive sample covers 33% of all papers published between 1955 and 2010 but due to methodological limitations, systematically lacks

Signiﬁcance

Empirical evidence suggests signiﬁcant gender differences in the total productivity and impact of academic careers across science, technology, engineering, and mathematics (STEM) ﬁelds. Paradoxically, the increase in the number of women academics over the past 60 years has increased these gender differences. Yet, we ﬁnd that men and women publish a comparable number of papers per year and have equivalent career-wise impact for the same total number of publications. This suggests the productivity and impact of gender differences are explained by different publishing career lengths [Annotation 2] and dropout rates. This comprehensive picture of gender inequality in academic publishing can help rephrase the conversation around the sustainability of women’s careers in academia, with important consequences for institutions and policy makers.

Author contributions: J.H., A.J.G., R.S., and A.-L.B. designed research; J.H. and A.J.G. performed research; J.H. and A.J.G. analyzed data; and J.H., A.J.G., R.S., and A.-L.B. wrote the paper.y Competing interest statement: A.-L.B. is founder of Nomix, Foodome, and Scipher Medicine, companies that explore the role of networks in health. All other authors declare no competing interest.y This article is a PNAS Direct Submission.y This open access article is distributed under Creative Commons Attribution License 4.0 (CC BY).y See online for related content such as Commentaries.y 1 J.H. and A.J.G. contributed equally to this work.y 2 To whom correspondence may be addressed. Email: A.Barabasi@northeastern.edu.y This article contains supporting information online at https://www.pnas.org/lookup/suppl/ doi:10.1073/pnas.1914221117/-/DCSupplemental.y First published February 18, 2020.

www.pnas.org/cgi/doi/10.1073/pnas.1914221117 PNAS | March 3, 2020 | vol. 117 | no. 9 | 4609–4616

Downloaded by guest on May 11, 2020

#### Page 2

AB

C

D

Fig. 1. Gender imbalance since 1955. (A) The number of active female (orange) and male (blue) authors over time and the total proportions of authors (Inset). (B and C) The proportion of female authors in several disciplines (B) and countries (C); for the full list, see SI Appendix, Tables S3 and S4. (D) The academic publishing career of a scientist is characterized by his or her temporal publication record. For each publication, we identify the date (gold dot) and number of citations after 10 years c 10 (gold line, lower). The aggregation by year provides the yearly productivity (light gold bars), while the aggregation over the entire career yields the total productivity (solid yellow bar, right) and total impact (solid yellow bar, right). Career length is calculated as the time between the ﬁrst and last publication, and the annual productivity (dashed gold line) represents the average yearly productivity. Authors drop out from our data when they published their last article.

authors from China, Japan, Korea, Brazil, Malaysia, and Singapore (SI Appendix, section S2). To demonstrate the robustness of our ﬁndings to database bias and author disambiguation errors, we independently replicated our results in two additional datasets: the Microsoft Academic Graph (MAG) (31) and the Digital Bibliography & Library Project (DBLP), each

using different criteria for publication inclusion and methodologies for career reconstruction (SI Appendix, sections S1 and S6). Our focus on bibliometric data limits our analysis to publishing careers and is unable to capture the career dynamics of teaching, administrative, industrial, or government related research activities. Nevertheless, our efforts constitute an extensive attempt

4610 | www.pnas.org/cgi/doi/10.1073/pnas.1914221117 Huang et al.

Downloaded by guest on May 11, 2020

#### Page 3

SOCIAL SCIENCES

to quantify gender inequality in science, technology, engineering, and mathematics (STEM) publications and citations, offering a longitudinal, career-wise perspective across national and disciplinary boundaries.

The Increasing and Persistent Gender Gap Across all years and disciplines, women account for 27% of authors, a number that hides important trends: while in 1955 women represented only 12% of all active authors, that fraction steadily increased over the last century, reaching 35% by 2005 (Fig. 1A). Yet, these aggregate numbers hide considerable disciplinary differences, as the fraction of women is as low as 15% in math, physics, and computer science and reaches 33% [Annotation 3] in psychology (Fig. 1B). We also observe signiﬁcant variations by country, ﬁnding that the proportion of female scientists can be as low as 28% in Germany and reaches parity with 50% in Russia (Fig. 1C).

The low proportion of women actively publishing in STEM captures only one aspect of gender inequality. Equally important are the persistent productivity and impact differences between the genders (Fig. 1D). We ﬁnd that while, on average, male scientists publish 13.2 papers during their career, female authors publish only 9.6, resulting in a 27% gender gap in total productivity (Fig. 2A). The difference is particularly pronounced among productive authors, as male authors in the top 20% productivity bracket publish 37% more papers than female authors (Fig. 2A). Interestingly, the gender gap disappears for median productive authors (middle 20%) and reverses for the authors in the bottom 20%. The gender gap in total productivity persists for all disciplines and almost all countries (Fig. 2 B and C). We also observe a large gender gap in total productivity for the highest-ranked afﬁliations (Fig. 2D) (determined from the 2019 Times Higher Education World University Rankings; SI Appendix, section S2.D).

AB C DE

FG H IJ

O

TS

NM

RQ

LK

P

Fig. 2. Gender gap in scientiﬁc publishing careers. The gender gap is quantiﬁed by the relative difference between the mean for male (blue) and female (orange) authors. In all cases the, relative gender differences are statistically signiﬁcant, as established by the two-sided t test, with P values < 10 −4 , unless otherwise stated (see SI Appendix, section S4.A for test statistics). (A–E) Total productivity broken down by percentile (A), discipline (B), country (C), afﬁliation rank (D), and decade (E). The gender gap in productivity has been increasing from the 1950s to the 2000s. (F–J) Total impact subdivided by percentile (F), discipline (G), country (H), afﬁliation rank (I), and decade (J). (K–O) Annual productivity is nearly identical for male and female authors when subdivided by percentile (K), discipline (L), country (M), afﬁliation rank (N), and decade (O). (P–T) Career length broken down by percentile (P), discipline (Q), country (R), afﬁliation rank (S), and decade (T).

Huang et al. PNAS | March 3, 2020 | vol. 117 | no. 9 | 4611

Downloaded by guest on May 11, 2020

#### Page 4

A

C

DE

B

Fig. 3. Controlling for career length. (A and B) The gender gap in career length strongly correlates with the gender gap in productivity across disciplines (Pearson correlation, 0.80) (A) and countries (Pearson correlation, 0.56) (B). A gender gap of 0.0% indicates gender equality, while negative gaps indicate the career length or productivity is greater for male careers, and positive gaps indicate the feature is greater for female careers. (C) In a matching experiment, equal samples are constructed by matching every female author with a male author having an identical discipline, country, and career length. (D) The average productivity provided by the matching experiment for career length compared to the population; the gender gap is reduced from 27.4% in the population to 7.8% in the matched samples. (E) The average impact provided by the matching experiment for career length compared to the original unmatched sample. Where visible, error bars denote 1 SD.

4612 | www.pnas.org/cgi/doi/10.1073/pnas.1914221117 Huang et al.

Downloaded by guest on May 11, 2020

#### Page 5

SOCIAL SCIENCES

We measure the total impact during an academic career by the number of citations accrued 10 years after publication (c 10 ) by each paper published during a career (Fig. 1D), after removing self-citations and rescaling to account for citation inﬂation (32–34) (SI Appendix, section S2.F). We ﬁnd that male scientists receive 30% more citations for their publications than female scientists (Fig. 2F). Once again, the total impact difference is the largest for high-impact authors and reverses for medianand low-impact authors: male authors in the top 20% in career impact receive 36% more citations than their female counterparts. The disparity in impact persists in almost all countries and all disciplines (Fig. 1 G and H), and can be found, to a lesser extent, across all afﬁliations regardless of afﬁliation rank (Fig. 1I).

Paradoxically, the gradual increase in the fraction of women in science (5) (Fig. 1A) is accompanied by a steady increase in both the productivity and impact gender gaps (Fig. 2 E and J). The gender gap in total productivity rose from near 10% in the 1950s to a strong bias toward male productivity (35% gap) in the 2000s. The gender gap in total impact actually switches from slightly more female impact in the 1950s to a 34% gap favoring male authors in the same time frame. These observations disrupt the conventional wisdom that academia can achieve gender equality simply by increasing the number of participating female authors.

In summary, despite recent attempts to level the playing ﬁeld, men continue to outnumber women 2 to 1 in the scientiﬁc workforce and, on average, have more productive careers and accumulate more impact. These results conﬁrm, using a uniﬁed methodology spanning most of science, previous observations in speciﬁc disciplines and countries (2, 9, 11, 12, 16, 35–38) and support in a quantitative manner the perception that global gender differences in academia is a universal phenomenon persisting in every STEM discipline and in most geographic regions. Moreover, we ﬁnd that the gender gaps in productivity and impact have increased signiﬁcantly over the last 60 years. The universality of the phenomenon prompts us to ask: What characteristics of academic careers drive the observed gender-based differences in total productivity and impact?

Annual Productivity and Career Length As total productivity and impact over a career represent a convolution of annual productivity and publishing career length, to identify the roots of the gender gap, we must separate these two factors. Traditionally, the difﬁculty of reconstructing full publishing careers has limited the study of annual productivity to a small subset of authors or to career patterns observable during a ﬁxed time frame (39–46). Access to the full publishing career data allows us to decompose each author’s total productivity into his or her annual productivity and career length, deﬁned [Annotation 4] as the time span between a scientist’s ﬁrst and last publication (Fig. 1D and SI Appendix, section S3). We ﬁnd that the annual productivity differences between men and women are negligible: female authors publish, on average, 1.33 papers per year, while male authors publish, on average, 1.32 papers per year, a difference, that while statistically signiﬁcant, is considerably smaller than other gender disparities (0.9%, P value < 10 −9 ; Fig. 2K). This result is observed in all countries and disciplines (Fig. 2 L and M), and we replicated it in all three datasets (SI Appendix, section S6). The gender difference in annual productivity is small even among the most productive authors (4% for the top 20%) and is reversed for authors of median and low productivity.

The average annual productivity of scientists has slightly decreased over time; yet, there is consistently no fundamental difference between the genders (Fig. 2O). In other words, when it comes to the number of publications per year, female and male authors are largely indistinguishable, representing the ﬁrst gen-

der invariant quantity in performance metrics. As we show next, this invariant, our key result, helps us probe the possible roots of the observed gender gaps.

The comparable annual productivity of male and female scientists suggests that the large gender gap in total career productivity is determined by differences in career length. To test if this is the case, we measured the career length (Fig. 1D) of each scientist in the database, ﬁnding that, on average, male authors reach an academic age of 11.0 years before ceasing to publish, while the average terminal academic age of female authors is only 9.3 years (Fig. 2P). This gap persists when authors are grouped by either discipline, country, or afﬁliation (Fig. 2 Q, R, and S) and has been increasing over the past 60 years (Fig. 2T). Taken together, Fig. 2 K and T suggests that a signiﬁcant fraction of the variation in total productivity is rooted in variations in career lengths. This conclusion is supported by a strong correlation between the career-length gap and the careerwise productivity gap when we subdivide scientists by discipline (Fig. 3A; Pearson correlation, 0.80) and country (Fig. 3B; Pearson correlation, 0.58). In other words, this strong correlation implies that disciplines or countries with a large gender difference in the career length also have a large gender difference in total productivity, while those disciplines or countries with small gender differences in the career length also have a small gender difference in total productivity. For example, the gender gap in career length is smallest in applied physics (2.6%), as so is the gender gap in total productivity (7.8%). In contrast, in biology and chemistry, men have 19.2% longer careers on average, resulting in a total productivity gender gap that exceeds 35.1%.

Given the largely indistinguishable annual productivity patterns, we next ask how much of the total productivity and the total impact gender gaps observed above (Fig. 2 A and F) could be explained by the variation in career length. For this, we perform a matching experiment designed to eliminate the gender gaps in career length. In the ﬁrst population, for each female scientist, we select a male scientist from the same discipline (Fig. 3C and SI Appendix, section S4.B). We then constructed a second matched population, as a subset of the ﬁrst, in which each female scientist is matched to a male scientist from the same discipline and with exactly the same career length. In these career length-matched samples, the gender gap in total productivity reduces from 31.0 to 7.8% (Fig. 3D). Furthermore, the gender gap in the total impact is also reduced from 38.4 to 12.0% (Fig. 3E). By matching pairs of authors based on observable confounding variables, such as their discipline, we mitigate the inﬂuence of these variables on the gender gaps. More strenuous matching criteria controlling for country and afﬁliation rank do not greatly affect these results, although they limit us to much smaller matched populations (SI Appendix, section S4.B and Fig. S1). While matching cannot rule out that gender differences are inﬂuenced by unmatched variables that are unobserved here, the signiﬁcant decrease in the productivity and impact gender gaps when we control for career length suggests that publication career length is a signiﬁcant correlate of gender differences in academia.

To address the factors governing the end of a publishing career, we calculated the dropout rate, deﬁned as the yearly fraction of authors in the population who have just published their last paper (42, 47). We ﬁnd that, on average, 9.0% of active male scientists stop publishing each year, while the yearly dropout rate for women is nearly 10.8% (Fig. 4A). In other words, each year, women scientists have a 19.5% higher risk to leave academia than male scientists, giving male authors a major cumulative advantage over time. Moreover, this observation demonstrates that the dropout gap is not limited to junior researchers but persists at similar rates throughout scientiﬁc careers.

Huang et al. PNAS | March 3, 2020 | vol. 117 | no. 9 | 4613

Downloaded by guest on May 11, 2020

#### Page 6

AB

CD

EF

Fig. 4. Author’s age-dependent dropout rate. (A) Dropout rate for male (blue) and female (orange) authors over their academic ages. (B) The cumulative survival rate for male and female authors over their academic ages. (C and D) The effect of controlling for the age-dependent dropout rate on the gender gaps in total productivity (C) and impact (D). (E) The total impact gap is eliminated in the matched sample based on total productivity. (F) The gender gap in the total number of collaborators is eliminated in the matched sample based on total productivity.

The average causal effect of this differential attrition is demonstrated through a counterfactual experiment in which we shorten the careers of male authors to simulate dropout rates matching their female counterparts at the same career stage (Fig. 4 C and D and SI Appendix, section S4.F). We ﬁnd that under similar dropout rates, the differences in total productivity and total impact reduce by roughly two-thirds, namely from 27.4 to 9.0% and from 30.5 to 12.1%, respectively. This result, combined with our previous matching experiment (Fig. 3 D and E), suggests that the difference in dropout rates is a key factor in the observed total productivity and impact differences,

accounting for about 67% of the productivity and impact gaps. Yet, the differential dropout rates do not account for the whole effect, suggesting that auxiliary disruptive effects, from perception of talent to resource allocation (15, 21), may also play a potential role.

The reduction of the gender gaps in both total productivity and total impact by similar amounts suggests that total impact, being the summation over individual articles, may be primarily dependent on productivity (15). To test this hypothesis, we conducted another matching experiment in which we selected a male author from the same discipline and with exactly

4614 | www.pnas.org/cgi/doi/10.1073/pnas.1914221117 Huang et al.

Downloaded by guest on May 11, 2020

#### Page 7

SOCIAL SCIENCES

the same number of total publications as each female author (SI Appendix, section S4.D). In these matched samples, the gender gap in the total impact is completely eliminated, dropping from 38.4% in favor of male authors to 0.8% in favor of female authors (Fig. 4E). This reveals a second genderinvariant quantity—there is no discernible difference in impact between male and female scientists for the same size body of work. This second gender invariant reinforces our main ﬁnding that it is career-length differences that drive the total productivity gap, which consequently drives the impact gender gap in academia. Interestingly, controlling for productivity similarly ﬂips the gender gap in the total number of collaborators throughout a career, from 13.3% in favor of male authors to 16% in favor of female authors (Fig. 4F and SI Appendix, section S4.E).

Summary and Discussion The reconstruction of full publishing careers of scientists allowed us to conﬁrm the differences in total productivity and impact between female and male scientists across disciplines and countries since 1955. We showed that the gradual increase in the fraction of women in STEM was accompanied by an increase in the gender disparities in productivity and impact. It is particularly troubling that the gender gap is the most pronounced among the highly productive authors—those who train the new generations of scientists and serve as role models for them. Yet, we also found two gender invariants, revealing that active female and male scientists have largely indistinguishable yearly performance and receive a comparable number of citations for the same size body of work. These gender-invariant quantities allowed us to show that a large portion of the observed gender gaps are rooted in gender-speciﬁc dropout rates and the subsequent gender gaps in publishing career length and total productivity. This ﬁnding suggests that we must rephrase the conversation about gender inequality around the sustainability of woman’s careers in academia, with important administrative and policy implications (16, 37, 48–53).

It is often argued that in order to reduce the gender gap, the scientiﬁc community must make efforts to nurture junior female researchers. We ﬁnd, however, that the academic system is losing women at a higher rate at every stage of their careers, suggesting that focusing on junior scientists alone may not be sufﬁcient to reduce the observed career-wise gender imbalance. The cumulative impact of this career-wide effect dramatically increases the gender disparity for senior mentors in academia, perpetuating the cycle of lower retention and advancement of female faculty (10, 53–55).

Our focus on closed careers limited our study to careers that ended by 2010, eliminating currently active careers. Therefore, further work is needed to detect the impact of recent efforts by many institutions and funding agencies to support the par-

ticipation of women and minorities (41, 56). Our analysis of all careers and the factors that dominate the gender gap could offer a baseline for such experimental studies in the future. Due to the reliability of gender disambiguation, we were also unable to assess author gender for China, Japan, Korea, Brazil, Malaysia, and Singapore, whose inclusion would provide a more comprehensive global perspective of gender differences in science. Since scientists from these countries signiﬁcantly increased their contributions to the global scientiﬁc discourse, there is a pressing need for future work to develop more accurate gender identiﬁcation methodologies. Despite these limitations, our work suggests the importance of temporal controls for studying academic careers and, in particular, gender inequality in academia.

It is important to emphasize that the end of a publishing career does not always imply an end of an academic career; authors who stopped publishing often retain teaching or administrative duties or conduct productive research in industry or governmental positions, with less pressure to communicate their ﬁndings through research publications. Scientiﬁc publications represent only one of the possible academic outputs; in some academic disciplines, books and patents are equally important, and all three of our data sources (WoS, MAG, and DBLP) tend to overrepresent STEM and English language publications (57), thereby possibly biasing our analysis. Furthermore, our bibliometric approach can draw deep insight into the large-scale statistical patterns reﬂecting gender differences, and yet we cannot observe and test potential variation in the organizational context and resources available to individual researchers (13, 58). However, our results do suggest important consequences for the organizational structures within academic departments. Namely, we ﬁnd that a key component of the gender gaps in productivity and impact may not be rooted in genderspeciﬁc processes through which academics conduct research and contribute publications but by the gender-speciﬁc sustainability of that effort over the course of an entire academic career.

Data and Code Availability. The DBLP and MAG are publicly available from their source websites (SI Appendix). Other related and relevant data and code are available from the corresponding author upon request.

ACKNOWLEDGMENTS. We thank Alice Grishchenko for help with the visualizations. We also thank the wonderful research community at the Center for Complex Network Research, and in particular Yasamin Khorramzadeh, for helpful discussions, and Kathrin Zippel at Northeastern University for valuable suggestions. A.J.G. and A.-L.B. were supported in part by Templeton Foundation Contract 61066 and Air Force Ofﬁce of Scientiﬁc Research Award FA9550-19-1-0354. J.H. and A.-L.B. were supported in part by Defense Advanced Research Projects Agency Contract DARPA-BAA-15-39. R.S. acknowledges support from Air Force Ofﬁce of Scientiﬁc Research Grants FA9550-15-1-0077 and FA9550-15-1-0364.

1. T. J. Ley, B. H. Hamilton, The gender gap in NIH grant applications. Science 322, 1472–

1474 (2008). 2. V. Lariviere, C. Ni, Y. Gingras, B. Cronin, C. R. Sugimoto, Global gender disparities in

science. Nature 504, 211–213 (2013). 3. H. Shen, Inequality quantiﬁed: Mind the gender gap. Nature News 495, 22–24

(2013). 4. A. E. Lincoln, S. Pincus, J. B. Koster, P. S. Leboy, The Matilda effect in science:

Awards and prizes in the US, 1990s and 2000s. Soc. Stud. Sci. 42, 307–320 (2012). 5. L. Holman, D. Stuart-Fox, C. E. Hauser, The gender gap in science: How long until

women are equally represented? PLoS Biol. 16, e2004956 (2018). 6. K. Zippel, Women in Global Science: Advancing Academic Careers Through

International Collaboration (Stanford University Press, Stanford, CA, 2017). 7. “Gender in the global research landscape” (Tech. Rep., Elsevier, 2017; https://www.

elsevier.com/research-intelligence/campaigns/gender-17). 8. National Academy of Sciences, National Academy of Engineering, and Institute

of Medicine, Beyond Bias and Barriers: Fulﬁlling the Potential of Women in Academic Science and Engineering (The National Academies Press, Washington, DC, 2007).

9. J. R. Cole, H. Zuckerman, The productivity puzzle: Persistence and changes in pat-

terns of publication of men and women scientists. Adv. Motivation Achiev. 2, 217–256 (1984). 10. J. S. Long, The origins of sex differences in science. Soc. Forces 68, 1297–1316 (1990). 11. J. S. Long, Measures of sex differences in scientiﬁc productivity. Soc. Forces 71, 159–

178 (1992). 12. Y. Xie, K. A. Shauman, Sex differences in research productivity: New evidence about

an old puzzle. Am. Socio. Rev. 63, 847–870 (1998). 13. M. F. Fox, K. Whittington, M. Linkova, “Gender,(in) equity, and the scientiﬁc work-

force” in Handbook of Science and Technology Studies (MIT Press, Cambridge, MA, 2017). 14. G. Abramo, C. A. D’Angelo, A. Caprasecca, Gender differences in research productiv-

ity: A bibliometric analysis of the Italian academic system. Scientometrics 79, 517–539 (2009). 15. V. Lariviere, E. Vignola-Gagne, C. Villeneuve, P. Gelinas, Y. Gingras, Sex differences

in research funding, productivity and impact: An analysis of Qu ´ ebec university professors. Scientometrics 87, 483–498 (2011). 16. Y. Xie, K. A. Shauman, Women in Science: Career Processes and Outcomes (Harvard

University Press, 2003).

Huang et al. PNAS | March 3, 2020 | vol. 117 | no. 9 | 4615

Downloaded by guest on May 11, 2020

#### Page 8

17. P. L. Carr, Relation of family responsibilities and gender to the productivity and career

satisfaction of medical faculty. Ann. Intern. Med. 129, 532–538 (1998). 18. S. Stack, Gender, children and research productivity. Res. High. Educ. 45, 891–920

(2004). 19. M. F. Fox, Gender, family characteristics, and publication productivity among

scientists. Soc. Stud. Sci. 35, 131–150 (2005). 20. E. Z. Cameron, A. M. White, M. E. Gray, Solving the productivity and impact puzzle:

Do men outperform women, or are metrics biased? BioScience 66, 245–252 (2016). 21. J. Duch et al., The possible role of resource requirements and academic career-choice

risk on gender differences in publication rate and impact. PLoS One 7, e51332 (2012). 22. R. M. Borsuk et al., To name or not to name: The effect of changing author gender

on peer review. BioScience 59, 985–989 (2009). 23. M. Jadidi, F. Karimi, H. Lietz, C. Wagner, Gender disparities in science? Dropout, pro-

ductivity, collaborations and success of male and female computer scientists. Adv. Complex Syst. 21, 1750011 (2017). 24. K. M. Uhly, L. M. Visser, K. S. Zippel, Gendered patterns in international research

collaborations in academia. Stud. High Educ. 42, 760–782 (2015). 25. A. H. Eagly, C. Nater, D. I. Miller, M. Kaufmann, S. Sczesny, Gender stereotypes have

changed: A cross-temporal meta-analysis of U.S. public opinion polls from 1946 to 2018. Am. Psychol., 10.1037/amp0000494 (2019). 26. P. van den Besselaar, U. Sandstr ¨ om, Vicious circles of gender bias, lower positions, and lower performance: Gender differences in scholarly productivity and impact. PLoS One 12, e0183301 (2017). 27. E. Leahey, Gender differences in productivity: Research specialization as a missing

link. Gend. Soc. 20, 754–780 (2006). 28. P. Bronstein, L. Farnsworth, Gender differences in faculty experiences of inter-

personal climate and processes for advancement. Res. High. Educ. 39, 557–585 (1998). 29. S. Fortunato et al., Science of science. Science 359, eaao0185 (2018). 30. A. Clauset, C. R. Shalizi, M. E. J. Newman, Power-law distributions in empirical data.

SIAM Rev. 51, 661–703 (2009). 31. A. Sinha et al., “An overview of Microsoft Academic Service (MAS) and Applica-

tions” in Proceedings of the 24th International Conference on World Wide Web (International World Wide Web Conference Committee, 2015), pp. 243–246. 32. R. Sinatra, D. Wang, P. Deville, C. Song, A.-L. Barab ` asi, Quantifying the evolution of individual scientiﬁc impact. Science 354, aaf5239 (2016). 33. D. Wang, C. Song, A.-L. Barab ´ asi, Quantifying long-term scientiﬁc impact. Science 342, 127–132 (2013). 34. F. Radicchi, S. Fortunato, C. Castellano, Universality of citation distributions: Toward

an objective measure of scientiﬁc impact. Proc. Natl. Acad. Sci. U.S.A. 105, 17268– 17272 (2008). 35. I. E. Broder, Professional achievements and gender differences among academic

economists. Econ. Inq. 31, 116–127 (1993). 36. G. Abramo, C. A. D’Angelo, A. Caprasecca, The contribution of star scientists to overall

sex differences in research productivity. Scientometrics 81, 81–137 (2009). 37. D. Maliniak, R. Powers, B. F. Walter, The gender citation gap in international relations.

Int. Organ. 67, 889–922 (2013). 38. J. D. West, J. Jacquet, M. M. King, S. J. Correll, C. T. Bergstrom, The role of gender in

scholarly authorship. PLoS One 8, e66212 (2013).

39. K. Rorstad, D. W. Aksnes, Publication rate expressed by age, gender and academic

position – a large-scale analysis of Norwegian academic staff. J. Informetrics 9, 317– 333 (2015). 40. M. R. E. Symonds, N. J. Gemmell, T. L. Braisher, K. L. Gorringe, M. A. Elgar, Gender dif-

ferences in publication output: Towards an unbiased metric of research performance. PLoS One 1, e127 (2006). 41. P. van Arensbergen, I. van der Weijden, P. van den Besselaar, Gender differences in

scientiﬁc productivity: A persisting phenomenon? Scientometrics 93, 857–868 (2012). 42. D. Kaminski, C. Geisler, Survival analysis of faculty retention in science and

engineering by gender. Science 335, 864–866 (2012). 43. J. M. Box-Steffensmeier et al., Survival analysis of faculty retention and promotion in

the social sciences by gender. PLoS One 10, e0143093 (2015). 44. S. F. Way, D. B. Larremore, A. Clauset, “Gender, productivity, and prestige in computer

science faculty hiring networks” in Proceedings of the 25th International Conference on World Wide Web (International World Wide Web Conferences Steering Committee, 2016), pp. 1169–1179. 45. S. F. Way, A. C. Morgan, A. Clauset, D. B. Larremore, The misleading narrative of the

canonical faculty productivity trajectory. Proc. Natl. Acad. Sci. U.S.A. 114, E9216–E9223 (2017). 46. L. A. Hechtman et al., NIH funding longevity by gender. Proc. Natl. Acad. Sci. U.S.A.

115, 7943–7948 (2018). 47. S. Milojevi ´ c, F. Radicchi, J. P. Walsh, Changing demographics of scientiﬁc careers: The rise of the temporary workforce. Proc. Natl. Acad. Sci. U.S.A. 115, 12616–12623 (2018). 48. H. Etzkowitz, C. Kemelgor, B. Uzzi, Athena Unbound: The Advancement of Women

in Science and Technology (Cambridge University Press, 2000). 49. S. J. Ceci, W. M. Williams, Understanding current causes of women’s underrepresen-

tation in science. Proc. Natl. Acad. Sci. U.S.A. 108, 3157–3162 (2011). 50. J. M. Sheltzer, J. C. Smith, Elite male faculty in the life sciences employ fewer women.

Proc. Natl. Acad. Sci. U.S.A. 111, 10107–10112 (2014). 51. W. M. Williams, S. J. Ceci, National hiring experiments reveal 2:1 faculty prefer-

ence for women on stem tenure track. Proc. Natl. Acad. Sci. U.S.A. 112, 5360–5365 (2015). 52. M. W. Nielsen et al., Opinion: Gender diversity leads to better science. Proc. Natl.

Acad. Sci. U.S.A. 114, 1740–1742 (2017). 53. E. A. Cech, M. Blair-Loy, The changing career trajectories of new parents in stem. Proc.

Natl. Acad. Sci. U.S.A. 116, 4182–4187 (2019). 54. National Research Council, Gender Differences at Critical Transitions in the Careers

of Science, Engineering, and Mathematics Faculty (The National Academies Press, 2010). 55. L. R. Martinez, K. R. O’Brien, M. R. Hebl, Fleeing the ivory tower: Gender differ-

ences in the turnover experiences of women faculty. J. Wom. Health 26, 580–586 (2017). 56. A. J. Stewart, V. Valian, An Inclusive Academy: Achieving Diversity and Excellence (MIT

Press, 2018). 57. P. Mongeon, A. Paul-Hus, The journal coverage of Web of Science and Scopus: A

comparative analysis. Scientometrics 106, 213–228 (2016). 58. C. Wennerc, A. Wold, Nepotism and sexism in peer-review. Nature 387, 341–343

(1997).

4616 | www.pnas.org/cgi/doi/10.1073/pnas.1914221117 Huang et al.

Downloaded by guest on May 11, 2020

---

Source: [Fermat's Library](https://fermatslibrary.com/) — https://fermatslibrary.com/s/historical-comparison-of-gender-inequality-in-scientific-careers-across-countries-and-disciplines. More annotated papers: https://fermatslibrary.com/llms.txt
