Over the last decade, significant attention has been drawn to the gender ratio of speakers at conferences, with ongoing efforts for meetings to better reflect the gender representation in the field. We find that women are significantly under-represented, however, among the astronomers asking questions after talks.
Conferences act as a microcosm for our field, with the same biases and barriers to inclusion that affect the careers of astronomers played out in person instead of in abstract. The networking that occurs at conferences is essential to career development, and the in-person interactions between researchers have the potential to shape entire subfields. Recently there have been strong gains in the fraction of women speaking at conferences, as seen in the data collected from 2008 to 2013 by the American Astronomical Society Committee on the Status of Women in Astronomy (https://cswa.aas.org/percent.html). However, less is known about the nature and equity of the actual interactions scientists have at these conferences, such as in the question and answer (Q and A) sessions following each conference talk. This prompted us to conduct a volunteer-driven study of gender ratios in conference talks. Now in its fourth year, this ongoing survey has revealed many interesting biases in conference talks and questions.
Gathering data on gendered questions
During the 223rd meeting of the American Astronomical Society (AAS) in January 2014, we first began to gather data on the gender of people asking questions after talks. The project was announced via Twitter, with a short explanation of the basic goals as well as a link to a simple online form optimized for both computer and phone browsers. This form asked voluntary and anonymous participants to identify the session, the perceived gender of the speaker, and the perceived gender of each person asking questions. The results from AAS 223, examined during the AAS Hack Day event (http://www.astrobetter.com/wiki/AAS_HackDay_2014), confirmed the hypothesis that men are over-represented in conference questions compared to speakers or overall conference attendees1.
To date, a version of the gender question survey has been deployed at more than 12 conferences, including five AAS meetings. So far data from the initial AAS 2231, the 2014 UK National Astronomy Meeting2 and Cool Stars 18 and 193 have been analysed in detail. The most complete set of data so far is a compilation of responses from the four successive AAS winter meetings in 2014–2017 (223, 225, 227 and 229), each with responses from at least 150 talks. In total the dataset samples nearly 1,000 talks and over 2,500 questions.
The overall results, shown in Fig. 1, are consistent between all four AAS meetings. The gender ratio of speakers (36% women and 64% men) typically mirrors the ratio of attendees (35% women and 65% men at AAS 2231), indicating the speakers are drawn representatively from the astronomers in attendance. There are more female attendees, on average, than the overall AAS membership (73% male and 25% female), as they may be drawn from a more junior portion of the field (AAS members born after 1980 are 60% male and 40% female4). The gender ratio of people asking questions did not mirror the ratio of attendees, with men asking 75% of questions while women asked 25%. If we assume the data gathered is a representative sampling of the entire conference, each male AAS attendee asks an average of 0.93 questions per meeting, while each female attendee asks 0.57 (based on data from the AAS 223 attendees).
Toward a more inclusive survey
The largest drawback of the gender question dataset is the binary treatment of gender, with men and women as two distinct groups and where gender is the proximate factor in asking a question. This reinforces a false gender binary, and excludes or inaccurately represents scientists who are non-binary and/or gender-fluid, genderqueer, or agender. The main hurdle in properly including all genders is in self-identification; the survey relies on volunteers gendering people from an observer's standpoint, and may be particularly prone to misgendering.
Gender is also not the sole determining factor in who asks questions at conferences. As discussed above, seniority is a strong factor in who speaks and how frequently6,7, and the relatively higher fraction of men at higher levels of seniority can skew the overall demographics, as observed in our data. Other likely important factors that are not considered in our data are race, ethnicity, culture, national origin, gender self-identity, sexual orientation, disability and the manner people who hold one or many of these identities interact with scientific culture8. A more comprehensive survey that includes self-identification of conference attendees should be adopted, but as the survey increases in complexity the involvement of trained sociologists or ethnographers, and dedicated data gathering personnel are desperately needed. We aim to not only know who is asking questions at astronomy meetings, but also how to include everyone in the conversation.
Looking to the future
Previous Hack Day studies of the AAS survey data suggested that the gender of the first questioner could strongly influence the subsequent questions1. This indicated that early participation of women in the Q and A sessions could encourage other women to ask subsequent questions. However, our analysis of the entire four-year AAS dataset found that while this trend is present, the effect is weak and may only have impact in longer Q and A sessions.
These types of subtle correlations in our data indicate that an ongoing survey has great potential to change how we conduct conference talks and Q and A sessions. As a growing team of Hack Day participants, both young students and seasoned researchers alike, help analyse the data from the survey, we hope to ask new and more challenging questions. This may reveal particular sessions or subfields that out-perform others in gender parity during Q and A sessions; ultimately, this may be of use to conference organizers working to design inclusive sessions.
Informally, we have heard feedback from some women that their knowledge of this project has encouraged them to actively participate in Q and A sessions. While this effect biases our sample (towards a higher fraction of questions from women), we consider it a positive outcome, as our ultimate goal is to work towards more equitable and inclusive conference interactions. As such we will continue to call on the community to help gather gender survey data from future AAS meetings, and to initiate the survey at smaller subject-focused and regional meetings. Ideally, conference organizers and professional societies will adopt ours or similar methods, perhaps incorporating self-identification of identities beyond male or female, to both improve conferences in astronomy and work to become an example of inclusion for conferences across many fields.
This work is supported by a National Science Foundation Astronomy and Astrophysics Postdoctoral Fellowship under award AST-1501418.