On microaggressions


December 19, 2017

I was glad to see Karl Broman’s blog post on a troubling conference talk. In the post, Karl describes how the speaker repeatedly used the word “guys” to describe people involved in statistics, and how he and Hadley Wickham reached out to him in an effort to get him to change his ways.

If you were in the room during the talk, I’m sure the experience is still viscerally available to you, but for those who were not, I think Karl’s post falls far short of the necessary description to underscore the problem. As I said in a recent tweet, it’s hard to convey in words how icky the talk was. But, I’m going to make an attempt.

The talk was part of a session honoring Bill Cleveland. Other talks in the session were fantastic– in particular, Di Cook’s talk, where she replicated Cleveland and McGill’s findings live (!!). But this particular speaker, who has now come under scrutiny for sexual harassment, gave a talk that was broadly derisive and dismissive of anyone who is not a white man pursuing statistics. I have a record of many of the comments he made as part of email threads from the time.

In a talk that included “data science” in the title, he was dismissive of data science, making the comparison that statistics is like architecture (accompanied by a photo of the Sydney Opera House) while data science is like cookie-cutter tract housing (accompanied by a photo of a suburban development), managing to be derisive of both manual labor and the efforts of many of the people in the audience. He disparaged useR, saying that you should only go if you want to hear about “guys who made a nice plot.” The entire tenor of the talk was that work with data is less important than statistics, a completely bizarre take in a talk supposedly honoring Bill Cleveland.

Beyond that, the speaker made it clear by his references, language, and tone, that he only considers contributions by men. Every researcher he mentioned by name was male (even though he was speaking shortly after Di Cook, clearly eminent in the field). This might seem like a justification for his gendered language, but I think it speaks to a deeper issue– he simply does not see contributions by women.

The most obvious indicator of his dismissal of women was his repetitive use of the word “guys,” as Karl has mentioned. The usage become more and more grating throughout the talk, but even on his first use of the word the colleague sitting to my left said “sexist” under her breath at the same time I corrected “people” under mine. On twitter, many people chimed in to Karl’s tweet at the time saying they found the comments jarring as well. In situations where he could easily have said “people,” “statisticians,” “data scientists,” “programmers,” or many other gender-neutral terms, the speaker always used the word “guys.”

What makes this difficult to adequately describe is that in some contexts, “guys” is almost gender-neutral. However, the survey that Dave Harris linked suggests that “guys” is only gender-neutral if you say “you guys.” Phrases like “we’re going to need to hire a Python guy” or “I met a great Erlang guy” were perceived by the majority not to be gender-neutral. As Karl responded to someone saying she didn’t care about the use of “guys” as long as she was not excluded,

That’s what is so insidious about these types of microaggressions. When you try to describe them, they sound insignificant, even though for those who were present it was clear there was a pattern of dismissal. It was so blatant that


After the talk, it was clear that some action needed to be taken. Through email, Karl, Hadley, Hilary Parker (looped in for her interest in the issue, although she wasn’t present at the talk) and I discussed that while the talk offended us personally, it wasn’t a clear-cut violation of JSM’s Code of Conduct. This left making a comment to him, either publicly or privately. I knew that I (as a junior, female, data science-y person) would get no respect from him, so I encouraged Karl and Hadley to call him out. The speaker had mentioned Hadley in his talk, and since both of them are more senior, male, and ASA Fellows, I hoped their words would carry some weight. Karl, Hadley, Hilary and I went back and forth via email working on drafts.

Interestingly, we ended up softening the language in successive drafts. A first draft said people “were taken aback, to the point of being offended,” while the final version just says “were taken aback.” Karl initially mentioned that he “might also talk about the pedestrian carpenter data scientist vs the soaring architect applied statistician, but two criticisms without any positive comments seems too much.”

Within a week, the email Karl reproduced on his blog was sent, and after a couple months I followed up with Karl to see if the speaker had responded. He hadn’t, and we dropped the issue.

I regret that I didn’t feel comfortable writing publicly about this at the time, and that I didn’t push harder for follow-up when I heard he had ignored the email. We now know this particular person has a (alleged) history of sexual harassment, and perhaps publicizing this talk would have opened the issue up earlier. As Mara tweeted,

With the #metoo movement in general and the recent attention on harassment in statistics, data science, and machine learning in particular, I am reminded of the Brandeis quote that “Sunlight is said to be the best of disinfectants.” Of course, this is only true if people believe the stories that are voiced, but we seem to be at a watershed moment. I regret not shining more light on this at the time, and hope that we can be more open about violations of community norms, even if they seem “minor.” They could be indicative of a larger pattern.