DATASET ON GENDER IN THE JOURNALS

Our research focuses on ten journals that are widely accepted as leading forums for high-quality political science research in the United States, including the APSA’s own two flagship journals, the American Political Science Review (APSR) and Perspectives on Politics (POP). The other eight journals in our sample are: the American Journal of Political Science (AJPS), Comparative Politics (CP), Comparative Political Studies (CPS), International Organization (IO), Journal of Conflict Research (JCR), Journal of Politics (JOP), Political Theory (PT), and World Politics (WP). We sought to include the “top ranked” journals, though as Giles and Garand (Reference Giles and Garand2007) have noted, there is a wide range of ranking schemes, and different approaches produce somewhat different rankings. We settled on a sample of journals that show up regularly, despite differences in ranking criteria, thus reflecting some consensus in the discipline. However, we also sought to include at least one journal from each of the main recognized subfields in political science, as well as journals associated with different methodological leanings. ^{Footnote 8}

Our data collection efforts began by acquiring the meta-data on all articles published in these 10 journals from 2000 to 2015. Web-scraping techniques allowed us to gather information on nearly 8,000 articles (7,915), including approximately 6,000 research articles (5,970). The journals vary in terms of the level of information they provide about the nature of each article, but we were generally able to determine the type of article (whether a research article, book review, or symposium contribution), the names of all authors—from which we could calculate the number of authors—and often the institutional rank of each author (for example, assistant professor, full professor, etc.). In what follows, we describe the variable generation process for all types of articles in the dataset, but note that the findings we report stem from an analysis of authorship for research articles only, and not reviews or symposia.

Using an intelligent guessing technique (compared against a hand-coding method) we used authors’ first names to code author gender for all articles in the database. We also hand-coded the dominant research method employed by each research article. ^{Footnote 9} We were further able to generate women among authors (%) which is the share of women among all authors published in each journal, as well as other variables related to the gender composition for each article, which include information about whether each article was written by a man working alone, a woman working alone, an all-male team, an all-female team, or a co-ed team of authors. Because the convention in political science is generally to display author names alphabetically, we have not coded categories like “first author” or “last author” which are important in the natural sciences.

An accurate assessment of an author’s gender is critically important for the validity of our analysis. Given the large number of articles in the sample and the high levels of coauthorship, we automated the process of assigning a gender to authors. In practice, we are “imputing” gender, because we do not know about authors’ gender identity. ^{Footnote 10} This process involved four steps. First, we employed research assistants to hand-code the gender of each author for 1,311 articles. Second, using Python scripts, we used authors’ first names to query a database of names drawn from user profiles of major online social networks to guess the likely gender of all authors. ^{Footnote 11} Along with assigning genders to first names, the guessing algorithm provided a certainty measure from 0 to 1 that recorded its confidence in the guess. Third, we compared the hand-coded genders with the guessed values to evaluate the accuracy of the technique. Fourth, we fixed discrepancies and conducted web searches for any author for which the algorithm produced a certainty score of less than 1.

To assess the accuracy of the gender-guessing algorithm we compared the 32 cases where the computer-generated guess did not match our RA’s hand coding in the 1,311 articles coded by both methods (2%). The algorithm was incorrect in 26 cases (2%), almost always incorrectly guessing that a scholar was a woman (23 out of 26). In the six cases where the hand coding was incorrect, the RAs over-guessed men. Note that of all the cases where the RAs were incorrect, the algorithm had a very high certainty score of 1. In cases where the algorithm was incorrect, the average certainty score was 0.8. Along with fixing the gender assignment of all of the authors revealed by the comparison, we had a second RA re-check the assignment for all authors whose certainty score was less than 1. The RA found an additional 92 mistakes in the 5,510 articles where gender was only assigned by the algorithm (1.7%), and these were all fixed. Note that repeat incorrect guesses for the same author were very common, so we are confident that we have eliminated most of the incorrect assignments. Note, too, that because the algorithm tended to over-guess women, this would inflate the presence of women among authors and not undermine our findings.

Table 1 gives an overview of the sample, and also reports the share of women as a total of all authors in the sample period for each journal. Figure 1 provides a first look at the share of male and female authors for each journal over the full time period. The sections to follow consider a series of possible explanations for the patterns we observe here.

Table 1

Database of Gender in the Journals Overview

Figure 1

Women and Men as a Share of All Authors of Research Articles in 10 Political Science Journals, 2000–2015

Note: Over the entire time period, the share of women among all authors was highest for Perspectives on Politics and Political Theory (nearly 34%) and lowest for the AJPS (18%) and the JOP (23%).

Is it the Pool?

As a first step, we sought to develop an overview of the share of women in the profession. This is, of course, a moving target because, while some attrition is to be expected among both men and women, research has shown that the career pipeline in academia is especially leaky for women (e.g., APSA 2005). We establish three benchmarks that can serve to orient the discussion—the share of female PhDs in political science, the share of female members belonging to the APSA, and the share of women among tenure ladder faculty in the 20 largest PhD-granting institutions.

Earning a PhD is the crucial first step toward a career as a political scientist. Thus, we begin with the female share of all political science PhDs, which over the past decade has been steady and rather high, at around 40% (NSF survey of earned doctorates). As noted earlier, an alternative measure—also based on NSF data—that combines earned doctorates in political science and public administration records an even higher share (47.1% in 2008). However, we use the more conservative figure in what follows.

We do not have reliable data on the share of women who, after having earned a PhD, then go on to a career in political science in the academy. The best surrogate measure we have available is the share of all members of the APSA who are women. The aggregate figure (31% in 2014) is lower than the share of women who earned a doctorate in that year. However, the corresponding APSA membership figures for women earlier on in their careers (at the lower ranks of the tenure track) are significantly higher, and essentially the same as the PhD share for assistant professors (nearly 40%), dropping somewhat for associate professors (33%). Figure 2 provides a breakdown of APSA membership by gender and rank.

Figure 2

APSA Membership by Gender and Academic Rank

Source: APSA

We know from previous research (e.g., APSA 2005, 4) that women are relatively more likely than men to wind up in permanent lectureships or at teaching institutions and liberal arts colleges. Hence, a third benchmark is the share of women in the discipline’s largest PhD-granting departments. ^{Footnote 12} Here, we find that overall, 27% of ladder faculty are female (as of 2015). However, once again, the share of women rises amongst the younger cohorts: women make up 38% of untenured ladder faculty at such institutions.

We note that even though the share of women (both within the APSA and within the largest PhD-granting departments) is higher at these lower ranks—which, of course, is precisely the group whose advancement depends on demonstrating scholarly productivity—in all that follows, we will refer to the more conservative (overall) share of women for each of the three benchmarks.

Are female scholars published in these journals in approximately the same proportion as their presence in the discipline? In PhD-granting institutions? In the top-ranked departments? The answer to almost all of these questions is “no” for most of the journals in our sample, as figure 3

Figure 3

Women as Share of All Authors in 10 Political Science Journals, 2000–2015

Note: Line A represents the share of women in the ladder faculty at the largest 20 PhD-granting departments in the discipline (27%). Line B represents the share of women among all APSA members (31%). Line C represents the share of women among all newly minted PhDs as reported in the NSF’s survey of earned doctorates.

illustrates.

How well represented are these women in the pages of the top journals? Are female scholars published in these journals in approximately the same proportion as their presence in the discipline? In PhD-granting institutions? In the top-ranked departments? ^{Footnote 13} The answer to almost all of these questions is “no” for most of the journals in our sample, as figure 3 illustrates.

Figure 3 plots the share of all authors contributing to research articles (as opposed to review essays) who are women. Journals higher up on the plot have a higher proportion of women among authors—for example 33.5% of authors in Perspectives on Politics are women—while journals lower down have a lower share of women among their authors; only 18% of authors in the AJPS over this period are women (similar to the APSR, with 23.4%). The figure also includes three lines—line A depicts the proportion of women among tenure-ladder faculty at the largest 20 PhD-granting departments in 2015 (27%); line B shows the share of women among all APSA members in 2015 (31%); line C charts the share of women among newly conferred doctorates (40%). What is striking is that, compared against these benchmarks, only four journals—Perspectives on Politics, Political Theory, Comparative Politics, and Comparative Political Studies—are publishing women at rates consistent with women’s presence in the discipline as a whole.

This mix of journals suggests a hypothesis that we are not able to fully test with the data available to us. It could be that publication rates across these journals reflect differences in the composition of scholars working within specific subfields. For example, we know that women have traditionally been underrepresented among scholars in American politics and international relations, and better represented among students of comparative politics (Shames and Wise Reference Shames and Wise2017, figure 2). This could explain the fact that those journals in our sample that explicitly focus on comparative politics (CP, CPS) publish, on average, more female authors than those with a significant international relations emphasis (IO, JCR, WP).

However, gendered sorting into subfields would not explain is the pattern we observe for the four “generalist” journals in our sample (AJPS, APSR, JOP and POP). These four journals—official journals either of the national association or one of its regional affiliates—are all “generalist” outlets, in that their websites indicate that they are open to submissions across all subfields. ^{Footnote 14} Yet, as figure 3 shows, women are underrepresented, against all three benchmarks, in three of those four “generalist” journals.

Returning to the full sample, we can use the APSA’s figures on the share of women at different academic ranks (e.g., graduate students, post-doctorates, assistant, associate, and full professors) to examine the gender publication gap at each rank. In figure 4, equal representation would imply clustering along the 45-degree line. Instead, we see a significant gender gap for female authors at nearly every professional rank. For example, although about 23% of full professors are women, only 11% of the full professors who write for these top journals are women, meaning that there is a gender gap of 12 points at the full professor level. Thirty-nine percent of assistant professors are women, but women make up only 27% of assistant professor authors in the journals, again a gap of 12 points. In terms of the academic career ladder, the gap for female assistant professors is especially consequential, as this is the group most needing publications to advance to tenure.

Figure 4

Representation of Women in the Journals by Academic Rank

Note: The share of women at a given ladder level is on the x-axis, while their representation among authors at that rank in the top journals is along the y-axis. Equal representation would mean that the points fall along the 45-degree line.

Summing up the results with respect to our first question (women as a share of the pool of all authors), we find that publication rates at all of the journals in our sample are well below the share of female PhDs. More significantly, with the notable exceptions of Perspectives on Politics, Political Theory, Comparative Politics and Comparative Political Studies, they are also below the share of women in the APSA. Many of these journals are, in fact, also publishing a smaller share of female authors than the share of women on the faculty of the 20 largest PhD-granting departments. Strikingly large gender gaps exist at almost every academic rank. Relative to their share among full professors, for example, women publish in the top journals at far lower rates. Most consequentially of all, female assistant professors seem to be quite disadvantaged relative to their male peers when it comes to placing their work in the top journals.

Coauthorship

Additional factors that could play a role in the gender gap in publication patterns include changing norms and practices with respect to coauthorship. A 2005 APSA report on “Advancement of Women in Academic Political Science in the United States” emphasized the importance of inclusion in collaborative research and publication networks (2005, 2, 12 and passim). We know from previous research (e.g., Fisher et al. Reference Fisher, Cobane, Vander Ven and Cullen1998) that coauthorship and scholarship by research teams has been growing in the discipline. Yet, we find evidence that women are not benefiting equally in the resulting multi-authored publications in the field’s top journals (see also Evans and Moulder Reference Evans and Moulder2011).

Previous research on coauthorship by Fisher et al. (Reference Fisher, Cobane, Vander Ven and Cullen1998) focused on three journals (APSR, AJPS, and JOP) over the period from 1950 to 1996. The authors document an increase in multi-authored articles over this period, and show that by the mid-1990s almost half the articles that appeared in these three journals were co- or multi-authored. The context in which Fisher et al.’s study took place was one in which the discipline was just beginning to grapple with the issue of how to evaluate multi-authored work, and so the core question motivating their research was whether articles authored by women were more or less likely to be collaborative across the sexes than articles authored by men. The answer to this was a resounding yes: women were nearly four times more likely to publish articles coauthored with men than the other way around (1998, 852). For Fisher et al., the worry was that if multi-authored work were underrated, women would not benefit much from their apparently greater proclivity to collaborate with others, raising the “issue of how multi-authored articles should be evaluated” (1998, 854). We return to the issue of how women’s contribution to collaborative research is evaluated below.

Our core concern at this point, however, is a somewhat different one, namely whether women have the same opportunities as men to engage in collaborative research and publication as reflected in overall publication patterns. Viewed in this light, Fisher et al.’s data reveal a different pattern. Their figures suggest that the decline in single-authored work from the 1950s to the 1990s was largely made up by an increase in all-male collaborations. Articles published by exclusively male multi-author groups rose as a percentage of all articles published across all three journals to 39.2%, while cross-gender collaborations rose but only to 13.5%.

Our data allow us to extend Fisher et al.’s observations to 2015, and for a wider range of journals. As figure 5 shows, we observe a continued increase in coauthorship in most of the journals. For the AJPS, the share of coauthored articles grew by 15 percentage points, to 66.1%, by 2015; for the APSR it grew almost 9 percentage points, to 54.2%; for the JOP it grew by 12.6 percentage points, to 63.9%.

Figure 5

Average Number of Authors, Per Article, by Journal

Note: The appendix provides some data on long-term trends from the 1950s (from Fisher et al. Reference Fisher, Cobane, Vander Ven and Cullen1998).

Digging a little deeper, we investigate gender in the pattern of coauthorship for our full sample of journals. Figure 6 displays the results. The most common form of publication across all the journals is a single male author (41.1%), while the second most common form of publication is an all-male “team” of more than one author (24%). Women working alone represent 17.1% of publications, while all-female teams represent just 2.4% of all journal articles. Finally, cross-gender collaborations account for 15.4% of publications. ^{Footnote 15}

Figure 6

Gendered Patterns of Authorship for All Journals

Note: The columns report the share of all articles in our sample that reflect particular authorship patterns: the first column is of men publishing alone, followed by all male teams, women alone, cross-sex collaboration, and finally, all female teams.

The patterns of coauthorship have changed over time, but not unequivocally in a direction that is beneficial for women. Figure 7 shows that solo male authorship has declined over the last 15 years, but is still the most common form of publication in these journals. In 2000, solo male articles or all-male teams accounted for 71.4% of all articles, while in 2014 this combination accounted for 56.9%. In the meantime, cross-gender collaboration rose from 13.2% of all articles in 2000 to 24.5% in 2014. But female solo authorship has been relatively flat over this period, well under 20%, and all-women teams remain rare, at 2% in 2000 and 3.4% in 2014. Thus, the main growth of female authorship in the top journals is through their collaboration with men.

In the meantime, cross-gender collaboration rose from 13.2% of all articles in 2000 to 24.5% in 2014. But female solo authorship has been relatively flat over this period, well under 20%, and all-women teams remain rare, at 2% in 2000 and 3.4% in 2014.

Figure 7

Gendered Patterns of Authorship Across Journals, over Time

There are certainly differences across journals in the importance of coauthorship and the prevalence of collaboration. Figure 8 shows that the AJPS is slightly more likely to have articles authored by all-male teams than by men working alone, and is very unlikely to publish an article authored by only one woman (6.8%). Other journals, like JCR and JOP, exhibit a similar pattern, publishing more work by all-male teams than any other type, and very few articles (<10%) by women working alone.

Figure 8

Coauthorship Patterns Within Journals

Table 2 presents more data on patterns of collaboration for each of the journals, providing an even more granular picture and documenting significant variation across the journals for the entire time period of the study. For example, over 70% of all articles published in the AJPS and the APSR are authored exclusively by men (column 3), working either alone or in collaboration with other men (column 5). ^{Footnote 16}

Table 2

Gender and Collaboration, All Years

Note: Column 1 shows the percentage of women among all authors in each journal, while columns 2-13 take an article as the unit of analysis. Columns 2 and 3 show the percentage of all articles for each journal that emerge from a woman working alone (2) or a man working alone (3). Columns 4-6 consider the profile of authorship within multi-authored publications. Columns 2-6 should therefore sum to 100, save rounding errors. Columns 7-9 consider the gendered origins of all publications that have at least one female author (male author for columns 10-13). Column 7 finds the share of publications with at least one female author that is a collaborative project (similarly for men in column 10). The equation used to generate column 7 = (all women + cross sex) / (all women + cross sex + solo female) = (col 4 + col 6 )/(col 4 + col 6 + col 2 ). Column 8 considers the share of all publications with a female author that emerge from all female teams, and column 9 presents the share of publications with at least one female author that emerge from cross- sex collaboration. Similarly for columns 11 and 13 for men.

The underrepresentation of women in multi-authored work appears not to be function of their unwillingness to engage in collaboration. In table 2, column 7 shows that on average, 51% of publications with at least one female author emerge from collaborative work, while column 10 shows that 47.4% of publications with at least one male author are team projects. But while most of the work with at least one male author emerges from all-male teams (column 12, 28.8%), only 6.6% of the publications with at least one female author are all-female teams. Focusing on cross-sex collaborations specifically, we see that women publishing in these journals are much more likely than men to engage in such endeavors. Column 9 of table 2 shows that 44.8% of articles with at least one female author emerge from cross-sex collaboration, while only 18.6% of articles with one male author emerge from cross-gender teams (column 13). This finding mirrors what Fisher et al. found in 1998.

One possible explanation for these patterns is that women are less likely to be invited to collaborate in these collective research and publication projects, which, if true, would then exacerbate the gender gap in publication. Our data do not allow us to examine the formation of research teams. What we can surely say, however, is that as long as coauthorship continues to rise in the top political science journals, the gender disparities in these research teams put women who come up for tenure or promotion at a distinct disadvantage.

At the same time, and indeed as Fisher et al. noted back in 1998, for collaborative efforts to work for women, the evaluation of such work needs to be gender-neutral. In particular, Fisher et al. noted the possibility that women’s greater proclivity to collaborate would not redound to their advantage “if women’s contributions are implicitly attributed to male coauthors” (1998, 854). A recent study in economics has shown that the contributions of women to multi-authored work in that field are indeed often discounted. Specifically, the study shows that women have a lower probability of getting tenure than men if most of their articles are coauthored (Sarsons Reference Sarsons2015). In short, the trend toward coauthorship will benefit female scholars only if women are included in these research teams, and if they then also get credit for their apparently greater tendency to collaborate across gender lines.

Summing up our results with respect to authorship patterns, we find that the most common form of authorship over the past 15 years has been—and continues to be—men working alone. This is true even though the average number of authors per article has risen across many of the journals in our sample. For collaborative research, despite an increase in cross-gender collaboration, all-male teams remain predominant, and all-female teams continue to be extremely rare. The number of articles published by female authors working alone has been mostly constant, which means that the (relatively small) overall increase in the share of female authors comes from coauthorship with men. We have also identified some of the dilemmas posed by coauthorship. However, for now what is certainly true—and highly consequential—is that there are multiplier effects for coauthored work in recruitment, tenure, and promotion processes. Every author on a coauthored piece can list the publication that his or her team has produced, and so patterns of inclusion/exclusion—for whatever reason—from these research teams can significantly affect one’s perceived productivity. At the same time, however, if the contribution of female authors is discounted (as appears to be the case in Sarsons’s research), then when it comes to coauthorship, women may be damned if they don’t, but also damned if they do.

METHODOLOGICAL APPROACH

Finally, we consider the possibility that the gendered publication patterns that emerge from our data reflect a disconnect between the methodological leanings of different journals and the types of work that female scholars are more likely than their male colleagues to produce and to publish. The “Perestroika” movement of a few decades ago was motivated by the sense that many of the discipline’s top journals were privileging particular kinds of work based on specific (mostly quantitative) methods. ^{Footnote 17} The question here is somewhat different, namely whether the methodological leanings of particular journals provide de facto barriers to women having their work published.

Several previous studies point to this possibility. For example, based on an analysis of articles published in eight prominent political science journals between 1995 and 2004, Breuning and Sanders (Reference Breuning and Sanders2007) find that female authors are more prevalent in journals publishing a larger share of case-study work. Conversely, they observe that “journals with higher proportions of statistical analyses…are less likely to publish work by women” (2007, 349). ^{Footnote 18} More recently, Evans and Moulder examine publication patterns in a smaller set of journals (APSR, AJPS, JOP and PS) and find a similar pattern. Their data show that “articles written by female lead authors were significantly more likely to be qualitative…than were articles written by male lead authors” (2011, 796). ^{Footnote 19}

To examine the relationship between gender and method, we coded our sample using categories that broadly conform to the categories that the APSR uses in its reports to the Association. ^{Footnote 20} Specifically, we examine whether an article’s primary evidentiary base draws from one of the following traditions: qualitative investigation, formal theory, traditional statistical analyses, experimental methods, conceptual exploration, or political theory.

What we find resonates with the results in Breuning and Sanders (Reference Breuning and Sanders2007). Figure 9 classifies articles for all journals except Political Theory by whether the primary methodology was conceptual, experimental, formal theory, political theory, qualitative, or statistical. Examining the publication patterns across diverse methods, we find that women are underrepresented in work across all methods except qualitative work and political theory. The gap is most severe in work that relies primarily on formal theory, followed by conceptual and experimental work. By contrast, among articles categorized as employing primarily “qualitative” methods (case studies, small-n), the share of women exceeds their membership share in the APSA.

Figure 9

Share of Women among All Authors in Articles by Predominant Methodology

Note: The figure reports findings for all journals except Political Theory. Line A represents the share of women in the ladder faculty at the top 20 PhD-granting departments in the discipline (27%). Line B represents the share of women among all APSA members (31%). Line C represents the share of women among all newly minted PhDs, as reported in the NSF’s survey of earned doctorates (40%).

Another way to assess the extent to which methodological affinities play a role in the gender differences we observe is to examine the prevalence of different types of work across each of the 10 journals, and to compare this with share of female authors. Figure 10 reveals striking differences across journals in their propensity to publish particular types of scholarship: whereas AJPS and JOP publish almost no qualitative articles, WP, CP and POP publish a larger share of qualitative research.

Figure 10

Statistical and Qualitative Methods as a Share of All Articles, 2000–2014

Note: Political Theory excluded.

Linking these findings to gendered publication patterns, our findings are again similar to Breuning and Sanders (Reference Breuning and Sanders2007). We, too, observe that the journals that publish the largest share of qualitative work map rather well onto the journals that publish higher overall shares of female authors, documented in figure 1 above. Figure 11 shows that there is a positive correlation between the share of authors that are women and the share of all articles that are qualitative, and a negative correlation between female authors and statistical articles.

Figure 11

Women among All Authors, as a Function of Qualitative Articles (left) or Statistical Articles (right), 2000–2015

Note: The lines in each panel represent bivariate linear regression lines, meant to display the correlation between the x- and y-variables. The journal Political Theory is excluded.

If a larger share of female political scientists engages in qualitative research, this would help explain why women are especially underrepresented in those journals that tilt heavily toward publishing quantitative articles. But do female political scientists in fact gravitate more toward qualitative research than their male counterparts? Although we cannot answer this question definitively, we can tap into APSA data to gain some initial leverage on it. One possible indicator of whether female political scientists are more prominent among qualitative researchers is membership in the APSA section devoted to qualitative and multi-method research. We obtained the section’s most recent membership roster and hand-coded each of the members’ gender. ^{Footnote 21} The share of women in this section (42%) does indeed exceed the share of women in the Association as a whole (31%). A second indication is the share of women among members of APSA’s political methodology section, which is strongly associated with quantitative methods. Based on figures provided to us by its chair, women account for 18.4% of the section’s membership, with men accounting for 81.6% of its members. ^{Footnote 22} Thus, comparing the membership profiles of these two sections against overall APSA membership share does suggest that female scholars, on average, may gravitate somewhat more toward qualitative and mixed method research than their male counterparts. ^{Footnote 23}

Summing up the results of our exploration of possible connections between method, gender, and publication patterns across journals, our findings echo those of previous studies. Specifically, our data show that women are overrepresented among the authors of published qualitative studies, and they are underrepresented in the pages of those journals that publish very little such research. We have also presented some evidence that female political scientists may tend to pursue qualitative and multi-method research more than their male colleagues. Overall, our publications data show that very few of the top journals publish a significant share of qualitative research, but those that do tend, on average, to have more women among their author pool. Within our sample, the two journals that publish the most qualitative work (Perspectives on Politics and Comparative Politics) also publish the work of female authors in higher proportions. Conversely, the two journals that publish the fewest qualitative articles (AJPS and JOP) also publish the fewest female authors.

DISCUSSION

We have presented a series of rather sobering results for each of the three factors we considered: on the share of women in the discipline, on coauthorship, and on methods. First, our study has documented a striking and persistent gender gap in the share of female authors in 10 prominent political science journals, a gap that is especially pronounced in three of the discipline’s top “generalist” journals (AJPS, JOP, and APSR). Broadly speaking there are two quite different possible explanations for why women are so underrepresented in the pages of these journals: one is that women’s work is getting rejected at higher rates than men; the other is that that they are not submitting their work to these top journals in the first place.

We cannot adjudicate between these two possibilities with the (public) data available to us, but previous research provides some valuable hints. For example, Østby et al. (Reference Østby, Strand, Nordås and Gleditsch2013) were able to examine submission data for the Journal of Peace Research from 1983 to 2008. They did not find evidence that papers with female authors were rejected at higher rates than men under either a single- or a double-blind editorial process. If anything, female authors may have had a slight advantage (p.9). Similarly, Breuning and Sanders (Reference Breuning and Sanders2007) acquired some limited submissions data from CPS, International Studies Quarterly, and WP. They, too, find that women submit at lower rates than men, but fare comparatively well in the review process, appearing in the journals at somewhat higher rates than their presence among all submitting authors (350). ^{Footnote 24} Overall, though, both of these studies find that women submit at lower rates than men. We think that differential rates of submission by women may well be a problem beyond the limited journals and time periods these authors were able to examine, and we are partnering with several journal editors to assemble the data needed to explore this issue. If submission rates are indeed a part of the problem, then journal editors will, with justification, respond, “We cannot accept what we don’t get to review.”

Our own view is that if women’s self-selection bias is contributing to the patterns we have documented, this is the beginning—and not the end—of the story. Such a finding would demand that we then look for the deeper structural causes, asking further: why are women not submitting their work to particular journals? Our own discipline has been at the forefront of crucial insights into the “second” face of power—that is, the idea that lack of contestation often reflects not the absence of power, but instead anticipated reactions and limited prospects for success (Gaventa Reference Gaventa1982; Pierson Reference Pierson, Mahoney and Thelen2015). A similar “second face of bias” may be at work in the patterns we observe—as female scholars rationally decline to send their work to journals that are not seen as hospitable—whether for reasons of method or more informal exclusionary signals. Over time, publication patterns can become self-enforcing, as authors begin to see it as a waste of time to submit to venues whose past publications do not include the kind of work they do. The fact that we observe, and have documented, significant variation across journals in publication rates for women suggests that we should be skeptical of explanations of the problem that attribute it to its victims.

Second, with respect to coauthorship, we observe a trend toward multi-authored publications across many of the journals in our sample. This phenomenon may be the product of “teams”—that is, emerging from research clusters or labs led by senior scholars, but often providing opportunities for more junior scholars to engage in collaborative research. To the extent that women are equally well represented in these research clusters, this can be valuable. However, our data suggest that most coauthored work in the top journals comes from teams made up of only men. What we cannot say, with the data available to us, is whether this imbalance is itself perhaps linked to other gender imbalances, for example, imbalances in which particular scholars—male or female—are landing the large research grants that are often necessary to launch these collaborative research efforts, or whether it is due to gender differences in methodological training in graduate school as senior team members perhaps look for graduate students or junior colleagues with an interest in the technical side of research, or some other factor altogether.

Beyond this, we have also suggested that coauthorship may present some dilemmas for women that are not shared by their male colleagues. More research is needed to determine whether the pattern of gender bias in the evaluation of the relative contribution of different scholars in multi-authored work—recently documented in the field of economics—is an issue in political science as well. As Sarsons has shown, in tenure decisions, female economists are penalized more than male economists for coauthoring with members of the other sex. While this could lead women to select out of publishing with men, our data shows that cross-sex collaboration produces almost half of the publications with at least one female author in these journals, which suggests that women themselves are probably not opting out of collaborations with men.

A third big takeaway from our research is that, in the prestigious journals, women are not publishing quantitative research at the same rate as men, and the top journals are publishing mostly quantitative research. Again, this raises important questions of interpretation: is it that female political scientists do do statistical research, but the top journals just do not publish it? Or—similar to Young’s conjecture from 1995—are they perhaps targeting less prestigious outlets for their quantitative work? Or is there a “selection” issue in which women, on average, are drawn more toward methods or modes of scholarship that are underrepresented in these journals? The APSA membership data we have presented, though not dispositive, suggests that there may be a gendered division of methodological commitments in the discipline. If this is the case, we should not expect women to show up in journals that publish primarily quantitative work, although this does raise the possibility of de facto, unintentional barriers to gender equity, and thus the question of whether such journals might look for ways to achieve more of a balance to reflect the full range of work produced by all types of scholars.

Finally, the observation that the three journals that publish the lowest share of female authors—AJPS, JOP, and APSR—are all not only “generalist” outlets (open to all subfields and methods), but also the official flagship journals of (respectively) the Western Political Science Association, the Southern Political Science Association, and the American Political Science Association raises broader issues about the larger ecology of journal publication in political science, and about how “quality” in scholarship is defined and evaluated in our discipline. Of course, these are partly normative questions that cannot be answered with data. However, given the vital role of publication in the advancement of careers in our discipline, the patterns we have documented underscore that these questions are well worthy of future research and discussion.