Observation In recent decades, the proportion of women entering scientific careers has increased substantially, but women remain underrepresented in academic ranks. A major problem contributing to the latter is the “leaky pipeline,” whereby women drop from the academic ladder and thus deplete the potential pool of women in senior academic faculty ranks. The leaky pipeline has been attributed to many causes, including discrimination, inadequate mentoring, lack of role models, and the difficulty in balancing academic and family life (1, 2). In this article, we identify a strategy that might encourage more participation of women in science. In a search for points of leverage that might influence women’s success and retention in academic careers, we identified invitations to speak at national meetings as key professional events. Presentations at scientific conferences provide self-evident benefits—they are venues for disseminating research findings, becoming known in science, and networking with colleagues who can enable future professional opportunities. However, the career ramifications extend far beyond those associated directly with the speaking event itself. First, promotion committees typically view invited meeting presentations as evidence for external recognition when considering whether to advance a candidate to a higher academic rank. Second, speaker rosters are often used as a starting point for planning other meetings, amplifying the impact of each event. Third, invited speakers are often provided financial support for travel to the meeting, thereby either making it possible for them to attend or releasing their own travel funds for other uses. In addition to the impact on the speaker herself, speaking at meetings may influence younger women’s career choices by providing them female role models. Hence, invitations to speak at scientific meetings are a critical feature of academic science and thus are important to consider in efforts to increase the participation of women in science. Several studies have shown that women are underrepresented as speakers in scientific meetings in absolute terms and relative to attendees (2–4). A retrospective analysis of participation at meetings of the American Association of Physical Anthropologists found that symposia organized by men were significantly less likely to include women as first authors in posters, talks, and symposia, despite the fact that women constituted the numerical majority in primatology and were well recognized for their contributions to that field (5). We sought to identify factors that affect the proportion of women among invited speakers. The American Society for Microbiology (ASM) has hosted its General Meeting (GM) covering all aspects of microbiology for over a century. In addition, ASM runs the Interscience Conference on Antimicrobial Agents and Chemotherapy (ICAAC), which focuses primarily on the science of antimicrobial drugs and their clinical applications. Historically, the GM was organized by the divisions of the ASM, but beginning in 2011, a new format was instituted whereby a program committee met and designed a meeting composed of plenary sessions and symposia (6). ICAAC is also organized by a program committee. Both program committees provide overall guidance for the scientific programs and assign conveners, but there are differences in how they operate. For the GM, conveners invite speakers, moderate sessions, and shepherd the symposia. Depending on the GM session, the conveners may follow recommendations from the program committee and/or invite speakers themselves in consultation with the program committee. In contrast, the ICAAC Program Committee is composed of categories that are represented by committee members who select topics for symposia and identify speakers. ICAAC symposium conveners are selected to represent topics of discussion by the program committee members in each category, generally after the symposium is approved by the entire program committee. Thus, conveners are selected by program committee members, rather than the whole committee. Conveners are selected for their expertise and are often, but not always, speakers in the symposium that they convene. The GM and ICAAC symposium take place in the spring and fall, respectively. The size and demographics of these meetings and the ASM membership are shown in Fig. 1. FIG 1 Proportion of women speakers as a function of convener gender composition for the years 2011, 2012, and 2013 at the GM and ICAAC meeting. All comparisons were significant at P < 0.05 by Student’s t test. We examined the gender representation among speakers for plenary sessions and symposia for both the GM and ICAAC as a function of the gender composition of the session conveners. The gender of invited speakers was inferred from the first name, and in cases in which the name was ambiguous, the speaker’s sex was determined by examining a picture of them. Sessions were classified as to whether they had been convened by two men, a man and woman, or two women. Data were assembled by calculating percentages of female speakers and/or comparing the absolute ratios of male and female speakers. Analysis of 216 sessions from 3 GMs held in 2011, 2012, and 2013 revealed that 104 were convened by male-only convener teams, and 112 had at least one female in the convener team. Too few sessions were organized by female-only convener teams to analyze independently, and consequently, we focused only on convener teams that were all male versus those that included at least one female convener. In sessions convened by all men, invited female speakers comprised 22 to 27% (Fig. 1; see Table 2) and averaged 25%. In sessions in which the convener team included at least one woman, woman speakers represented 39 to 46% (Fig. 1 and Table 1) and averaged 43%. Inclusion of at least one woman among the conveners increased the proportion of female speakers by 72% compared with those convened by men alone. Irrespective of whether the session was analyzed on a yearly basis or combined, the presence of a female in the convener team resulted in significantly more women among invited speakers (Table 2). TABLE 1 Gender numbers and percentages of GM and ICAAC attendees from 2011 through 2013 Yr No. of attendees by meeting Gender % from total membership data a GM ICAAC Total Full Postdoc Student Total b Full Postdoc Student Total b Full Postdoc Student Male Female Male Female Male Female Male Female 2011 5,345 3,391 420 1,534 6,840 6,204 312 324 63 38 63 37 49 51 40 60 2012 6,102 3,948 513 1,641 6,049 5,399 325 325 62 46 63 37 45 55 38 62 2013 4,871 3,096 377 1,398 4,470 3,979 263 228 61 47 63 37 44 56 37 63 a The gender percentage of ASM members was estimated from the subset of membership data for which the gender information was known. These percentages apply to membership of the society and are not the gender percentages of meeting attendees, since that information was not available. However, the gender percentages do give an indication of likely gender percentages for the full, postdoc, and student attendees. b The total numbers each at the GM and ICAAC meeting include only the full, postdoc, and student members who attended and do not include emeritus members, one-day attendees, exhibitors, and those who attended only workshops. Not all attendees are necessarily ASM members. TABLE 2 Effect of including female conveners on the gender composition of scientific sessions Meeting and yr No. of sessions (F/M ratio) a P value b Total No female conveners 1 or more female conveners GM 2011 65 28 (28/88) 37 (57/90) 0.0121 2012 76 35 (32/109) 41 (77/99) 0.0001 2013 75 41 (44/117) 34 (69/82) 0.0009 Total 216 104 (105/314) 112 (203/271) <0.0001 ICAAC 2011 84 55 (50/168) 29 (42/68) 0.0045 2012 85 47 (47/134) 38 (56/95) 0.0324 2013 75 47 (28/142) 32 (47/75) <0.0001 Total 244 149 (125/444) 99 (145/238) <0.0001 a F/M denotes the total number of female (F) and male (M) invited speakers. b P value of the comparison of F/M ratios calculated by chi-square test (two tailed) with the Yates correction. We also examined the impact of the conveners’ gender on the likelihood of a session containing only male speakers. This analysis is important, because all-male rosters send a subtle message, especially to junior scientists, that a field has few women. At the GM over 3 years, 23 to 36% (average 30%) of the sessions were comprised of all-male speakers if the convener team contained only men (Fig. 2 and Table 3), whereas 8 to 10% (average of 8.9%) of the sessions had all-male rosters among those organized by teams that included at least one woman (Fig. 2 and Table 2). The presence of a woman on the convening team was associated with reduced likelihood of an all-male session by about 70%. FIG 2 Proportion of sessions containing only a male speaker as a function of the composition of the convening team for 2011, 2012, and 2013 at the GM and ICAAC meeting. TABLE 3 Effect of including female conveners on the frequency of all-male sessions Meeting and yr No. of sessions P value a No female conveners 1 or more female conveners All male Total All male Total GM 2011 10 28 3 37 0.0105 2012 11 35 4 41 0.0226 2013 9 41 3 34 0.205 Total 30 104 10 112 0.0002 ICAAC 2011 15 55 6 29 0.602 2012 16 47 9 38 0.345 2013 23 47 4 32 0.0008 Total 54 149 19 99 0.0042 a Calculated by chi-square test with the Yates correction or by Fisher’s exact test. Analysis of 244 sessions from three ICAAC meetings held in 2011, 2012, and 2013 revealed that 145 were convened by male-only convener teams, and 99 had at least one female in the convener team (Table 2). Paralleling the analysis of the GM, inclusion of at least one woman in the convening team was associated with 74% more female-invited speakers than symposia organized by all-male teams. Inclusion of at least one woman in the convening team significantly reduced the number of sessions with all-male rosters (Fig. 2 and Table 3). In 2013, almost half of the sessions organized by all-male convener teams included only men in the speaker roster. Hence, the findings from the analysis of GM and ICAAC session gender composition data closely mirrored each other. Our results show an association between the presence of a woman in the session convening team and significantly greater female participation in the roster of invited speakers. We considered the obvious explanation that the increase in female speakers in sessions with one or more women conveners was a result of the convener also speaking. Clearly, having a woman convener who then chooses to speak at her symposium or selecting a woman from the speaker roster to serve as convener could account for some of the association observed. Consequently, we also analyzed the prevalence of conveners among the speaker rostrum for all symposia for both sexes. We found that on average for the 3 years studied, 24% of the male conveners and 25% of the female conveners spoke at their sessions at the GM, and 26% and 22% of the male and female conveners, respectively, spoke at their sessions at the ICAAC symposium. Hence, male and female conveners are making comparable contributions to the overall gender ratios of speakers, and the percentage of female conveners who spoke at their sessions was smaller than the increase in female participation associated with inclusion of female conveners, suggesting that female conveners increased the participation of women by other means. Furthermore, the presence of a woman on the convening team was associated with a marked reduction in the percentage of rosters comprised of all-male invited speakers. The findings on gender proportion for the GM and ASM closely paralleled each other despite the differences between the contents of the meetings; the GM is focused primarily on basic microbiology, whereas the ICAAC symposium focuses primarily on antimicrobial agents and their clinical use. We note that the data from both meetings were remarkably similar despite differences in the operating procedures of the two program committees and differences in both the roles of the conveners and how they were assigned. Furthermore, our results are similar to and consistent with analysis of invited participation in primatology meetings (5), suggesting that the tendency may extend to other fields of science. Low representation of women among speakers is consistent with the pervasive influence of implicit bias, which is the unintended, typically unconscious, bias toward certain groups that influences evaluation of people and their work (7). However, judgments made by men and women, including academic scientists, are equally influenced by implicit biases (7). If implicit bias were the only influence, we would predict that the gender of the conveners would have no effect on the composition of the speaker roster. We propose that other factors are at play. Perhaps women as conveners more often consider gender and make conscious efforts to find female speakers. Or maybe women are more likely to accept invitations from women or men are less likely to accept invitations from women, leading to an imbalance in the final rosters. Many other hypotheses could explain the results we report here. Because we do not know the mechanism underlying the association, we caution that the data presented here produce only a striking and highly significant correlation, and causation should not be inferred from these findings. At the very least, our findings suggest the need for additional studies to understand how gender choices are made in selecting speakers. Whatever the mechanism driving the results, practical actions are suggested by the data. The results suggest that an experiment in which at least one woman is included in every team of conveners might increase the proportional representation of women among the speakers at ASM meetings. An alternative might be to explicitly charge conveners with finding speakers who reflect the diversity of microbiologists. These strategies are worth testing. In the process, we might find that our meetings draw on a fuller arc of talent in microbiology and are enriched by increased gender balance. This study suggests a simple mechanism for increasing women’s participation in a critical part of a scientific life. Further research should determine whether discriminatory behaviors contribute to the outcomes and whether the outcomes contribute to the loss of women from academic science.
