Column: Student Angle
Danielle Quinn | Memorial University, 230 Elizabeth Ave, St. John’s, Newfoundland, Canada, A1C5S7. Email: [email protected]
In 2017, Penaluna et al. presented nine proposed action areas that could be used to enhance diversity and inclusion in the American Fisheries Society (AFS). Some wonderful examples of the Society actively working toward these goals, particularly through the efforts of the Equal Opportunities Section, include an entire symposium on diversity and inclusion at both the 2017 and 2018 Annual Meetings. One of the action areas proposed was to “undertake self‐reflection and evaluation,” which includes consciously evaluating who is being heard at events, on panel discussions, and in the scientific literature. Earlier in 2018, I was introduced to Cathy Deng’s online tool, “Who’s Talking?” (http://arementalkingtoomuch.com/), which helps to identify who is dominating a conversation by allowing the user to record the proportion of time in which a “Dude” or “Not a Dude” is speaking. I started thinking about scenarios where I could use this tool to quantitatively demonstrate some of the underlying problems that I see firsthand in conversations around science, research, and fisheries. I decided that I would apply this method at the AFS 148th Annual Meeting in Atlantic City, New Jersey, to see if any useful or interesting information could be gathered about who is dominating conversations at scientific conferences from a gender perspective.
The conference took place August 20–24 and included scientific talks spanning a broad range of fisheries topics, each 15‐min presentation was followed by a 5‐min question periods. At 26 talks, I recorded the number of males and females in the audience, the number and duration of questions asked during the question period, the gender of the person asking each question, and the duration of the question. The data and code are available online at https://github.com/DanielleQuinn/blog-posts/tree/gh-pages/afs148. As recognized by a similarly structured study by Hinsley et al. (2017), “…these methods treat gender as a binary dichotomy, which is not the case. In addition, the visual identification of gender assumes that people identify as the gender they appear. This will not always be true.” I apologize for the exclusion and acknowledge these limitations of the data. I hope that others will share ideas and suggestions about how to make subsequent observational studies more inclusive.
A total of 79 questions were asked by the 864 audience members attending the 26 talks, with a total duration of 934 s. Of these, questions from women accounted for 208 s (Figure 1). Women represented 39.7% of the overall audience and an average of 39.1% of the audience at each talk, which is encouraging given that women only represented 25.0% of AFS members as of 2015 (Penaluna et al. 2017). However, a disproportionately low percentage of questions (21.5%) were asked by women, and men asked significantly more (2.38 ± 1.44) questions per talk than women (0.65 ± 1.16) (Figure 2; Wilcoxon test, P < 0.010). This translates to men asking 3.60 questions for every 1.00 question asked by women, which far outweighs the relative proportions of men and women in the audience. Even as I recorded the data into my notebook, I was shocked at the number of times I found myself writing 0.00 in the column called “# of questions – women.” It turns out that at 18 talks (that is nearly 70.0% of the talks where data were recorded), no questions were asked by women! In these instances, men asked 1.0 to 5.0 questions (mean = 2.7). In comparison, men asked no questions at only two talks (7.7%), and in these instances, women asked one and two questions, respectively. Although men asked more questions than women, the duration of questions did not differ significantly between the two groups, with the average question lasting 12.2 s for women, and 11.7 s for men (Wilcoxon test, P = 0.616).
Interestingly, in seven talks, women asked proportionally as many or more questions than men. I struggled to identify anything that might explain what made these talks different than the others. I thought that the gender of the speaker may play a role, but neither audience composition nor the proportion of questions asked by women was found to significantly differ based on those data (t‐test, both P > 0.252). However, it turns out that the gender of the person asking the first question was a strong predictor of whether or not women asked subsequent questions. A woman asked the first question at five talks and in every single case women ended up asking proportionally as many or more questions than men (Figure 3). This was only true in 2 of the 21 talks where a man asked the first question. Revisiting the idea that the gender of the speaker may somehow influence these trends, I noticed that in 60% of cases where a woman asked the first question, the speaker was also a woman. So, while having a woman present the talk may not necessarily lead to more questions from women in the audience, it may be slightly more likely that a woman asks the first question, which, in turn, may encourage more women to ask subsequent questions.
Since including these results in a blog post in late August 2018 (available: https://daniellequinn.github.io/blog-posts/afs148/blogtext.html), I have had some great follow‐up discussions that were sparked by responses from a wide variety of people, including several who attended the conference. One thing that strikes me is how often people’s responses sound something like, “I’ve heard it was an issue sometimes, but I didn’t really think it was this bad!” It is certainly something that is easy to miss unless you are paying very close attention, and I think that as scientists, it is often in our nature to want to see things from a quantitative perspective. I want to emphasize that this was far from a perfectly controlled scientific study. I collected data while attending talks on topics that I was specifically interested in, including data science, sharks, and community outreach. It is possible that different trends may have been revealed in audiences attending talks focused on other domains (e.g., genomics or physiology). If this study were to be followed up, a randomized design would be helpful in formalizing the data collection, and additional information such as age group or academic role could be included. Despite these limitations, I hope that my observations can help fuel the conversation and provide quantitative insight into a problematic pattern.
In my original blog post, I was careful to avoid suggesting that men ask fewer questions and instead encouraged women to ask more questions. However, I recognize that asking members of a marginalized group to take on additional tasks and responsibilities is unfair and often ineffective for a wide variety of reasons. Here, I simply ask that any readers who are surprised, disappointed, or otherwise moved by these results to do their best to be an ally by using their own privilege (in whatever context that might be) to take actions that will help make a difference. If you are an advisor, help all your students to become equally comfortable asking questions at scientific talks. If you are moderating a session, try to select a woman to ask the first question whenever possible. If you’re aware of someone from an underrepresented group who wants to speak, support them and help their voice be heard. I hope that readers who will be attending AFS 149 in Reno, Nevada, will find themselves more aware of the gender discrepancy in conversation and consider how their own actions may help counter it. No single action is going to change the landscape of fisheries science, but as a member of the AFS community, I’m excited about our potential to move things in the right direction.
