Nascent Transcript writer Brooke Wolford interviewed Natalie Telis, a 5th year PhD student in Biomedical Informatics at Stanford University.
Asking questions about question-asking behavior
Diversity was a recurring theme at the American Society of Human Genetics’ 2017 Annual Meeting in Orlando, Florida—from the Presidential Address by Nancy Cox, PhD, to the final talk in the opening plenary session by Natalie Telis, a 5th year PhD student in Biomedical Informatics at Stanford University. Natalie’s presentation, “Scalable computational quantification of gender representation and behavior at ASHG,” described how she and colleagues quantitatively measure diversity in science using questions asked at conferences as a model. Her research showed that after adjusting for the proportion of women in the room, women were asking fewer questions than their male colleagues at the last three ASHG meetings.
Originally from Brooklyn, New York, Natalie grew up in California and completed her undergraduate work at the University of California, Davis, where she double majored in mathematics and biology. Her main work currently focuses on polygenic adaptation across complex traits and understanding the regulatory context-specific selection on Neanderthal introgression. However, she has also been collecting data on question asking behaviors in scientific settings since 2012. Natalie and her colleagues mined abstract data to identify the gender of conference attendees and speakers. They also tabulated the gender of audience members who asked questions after talks by watching recordings of invited sessions and attending sessions in person. In addition, they launched a platform to crowdsource data collection across all 2017 meeting sessions. The website also allows users to interact with existing data to ask their own questions about gender representation and behavior at ASHG.
Nascent Transcript contributor Brooke Wolford (@bnwolford) met with Natalie during the conference to discuss her work and her unique perspective. The conversation has been revised for clarity and brevity.
What motivated you to step back from genetics and look at this issue of women in scientific spaces quantitatively?
We care about the issue of gender discrepancy so much in our community, but we don’t spend a lot of time quantifying and having a data-driven discussion about it. The project started when I realized I was once the only woman who had asked a question at a meeting. I thought that was weird, but in order to know if it really was weird, I would need to know who was in the room. I wondered if there was some way to cross-apply the statistical modelling and computational skills I use to study evolution to study something completely different.
How did the American Society of Human Genetics help you to get the data?
All analyses from the abstract I submitted were done independently. After I received the plenary talk, ASHG reached out to help validate the computational inference. The core assumption of our work is that gender can be inferred from first names in a statistically accurate way. We use the gold standard methods within the field, but they are imperfect. In 2016, ASHG started collecting gender information, but not everyone reports it. If reported gender is biased, it will be biased in a different way than inferred gender. But we see overlap in the proportions estimated by both methods, so we are pretty confident about assigning gender to names. ASHG was instrumental in helping to validate our method.
What is the coolest question you have seen asked at https://telis.blog so far?
I am really excited about my personal vision of science where, instead of publishing static pieces that we passively take in, people instead create models where readers can engage with the results. For this project, I have no reason to believe that I have the best executive say on determining what is a cool question to ask of my data. So, making interactive data-sharing portals was a core goal of mine in trying to execute and believe in that scientific vision.
And indeed, people have thought of things that I totally didn’t think about. One scientist posted a figure of the proportions of males versus females who adopt either the British or the American spelling of the word analyze, which would never have occurred to me. The two usages of this word have very different gender proportions and it’s an interesting way to accidentally estimate the fraction of people represented from each particular country. The British use of this word is much more male than the American use.
Because your plenary was during the opening session, it serves as an intervention of sorts by sparking the conversation about gender-specific behavior at ASHG. How do you see women’s question asking behavior changing this year compared to your results from previous ASHG meetings?
Two full days into the meeting, we have over 600 audience member questions collected via our crowdsourcing platform. We can take some of what we observed in past invited sessions, for instance, and use that to approximate what we would expect now. The change in the number of women asking questions at this meeting compared to previous ASHG meetings seems to vary across different fields. A lot of the areas that I am closest to as a researcher, like evolutionary biology, statistical genetics, and bioinformatics, show much larger differences than other areas. I’m curious if there is an amount of, I don’t know, predisposition or awareness about this project in these fields.
There are a bunch of reasons that could explain this trend and I’m still trying to think about how to analyze it best, but I think it’s really interesting. A big reason I was excited about being able to collect this data after my talk is because we don’t test interventions. If it’s possible to create more inclusivity and a greater feeling of comfort or confidence by doing something, then I might as well test that out. And it will be really interesting to find out if that’s true across the board at the meeting, or whether it is more nuanced or discipline-specific.
I was really intrigued to see that even in a female-dominated field such as genetic counseling, women still asked fewer questions…
Right. That is another one of those hypotheses that I have always wanted to test. People always say that the issue is representation (i.e. that women are underrepresented in most scientific disciplines), and if you manage to raise female to male ratios to 50/50 or above, then all of a sudden you would see the reverse behavior, with women dominating the question asking. People essentially hypothesize that it’s a minority effect and not a sex-specific effect. Based on the effect sizes we measured, we think that in order to get a session with 50% of questions asked by females, you’d have to have a room that is 85% female. If this is culturally or intrinsically motivated, that shouldn’t be surprising. I don’t change when I walk into a new room, and we might not expect other people to either. But it definitely points to the fact that at least along this simplified axis, just changing demographics isn’t a sufficient intervention.
I am personally curious how imposter syndrome plays into question asking behavior.
If I could word cloud the things people have told me about this subject, imposter and nervous would be the two most enriched words. What is so interesting about questions is that they are binary and simple, but are tied to underlying and complex things. Trying to study imposter syndrome in a really direct way would probably be ineffective and also underpowered. But trying to study it in this indirect way should work, despite being confounded by factors such as general confidence or self-esteem outside of imposter syndrome. Sure, questions are not a perfect measure, but they’re a powerful measure in much the same way that we acknowledge in genomics that it may be better to have more data even if it’s of lower quality. Obviously, it’s not the perfect psychological survey of all attendees at this meeting.
So, what is your response to those who might ask if men should just stop asking questions?
In all the years of doing this project, many women have come up to me and told me that they asked a question, sometimes for the first time in their lives. But no man has ever come up to me and told me, “you silenced me.” I am really happy to continue giving people a voice and think it’s a false narrative that this would require that other people’s voices be taken away. It’s really easy to be threatened, maybe because science is so zero sum, by the thought that people want to take away your opportunity to shine in front of somebody who is eminent or to learn more about a topic you’re really excited about. These are all things that we all, as scientists, should have access to, and it would be a great disservice on my part to take that from a part of the scientific community, even with the greater goal of giving it to another part.
I remember your first tweets on data analysis of question asking behaviors at The Biology of Genomes meeting in 2015. How do you see Twitter as a way to democratize science and network in the scientific community?
I feel like one of the reasons that I originally used Twitter to share this at Biology of Genomes is because Twitter is so democratic. If people approve of your content, they can share it, and you have the opportunity to engage in discourse without ever approaching those people in person. At least for me, taking advantage of Twitter as a scientific participant, not just posting photos of my food or whatever, but specifically using it as a vehicle to communicate with my world of science, has been incredibly valuable. I’ve gotten postdoc and job interviews over Twitter probably because of it. I think it has been an incredibly equalizing and broadly connecting force. I connect with people at universities I’ve never visited who I’ve met at ASHG maybe once, but who continue to be part of the conversation about genetics, and then often about this part of my work as well. It’s worth saying that using Twitter really just involves participating in a way that is good for you, not stressing yourself out about posting N times a day, and in that context it can be a really useful tool.