Guest post from Page Piccinini (postdoc at École Normale Supérieure, Paris)
I’ve been an R user for about 8 years (with the occasional break, including that year I took off from science and worked as a real estate agent’s personal assistant). Recently I’ve started teaching an informal R course in our department*. How I went from being self-taught to teaching a course was the prompt for this blog post. Honestly though, it feels weird to say that I’m an R programmer, I guess because I feel like it’s been ingrained in me that I’m not really an R programmer. I don’t do the most complicated things. There are so many people who know so much more than me. I’ve spent a lot of time thinking about this disconnect, the fact that people come to me to ask for help in R and I (almost) always have an answer, yet I’ll still have those moments of panic where I think I’m doing everything wrong. I’ve decided my impostor syndrome can be reduced to a couple key issues in how coding is taught and treated inside and outside of academia. First is gender discrimination. This isn’t a new idea, plenty of women have discussed gender discrimination they’ve experienced in science and the tech industry. Second is how people outside of my field (Linguistics) view my field. And third, and probably the most unfortunate of the three, is how people within my field (and potentially science at large) treat each other when it comes to coding and statistics.
Starting with gender discrimination, it’s not the main factor that stopped me from calling myself an R programmer, but it is still a very real issue that I’ve faced as a woman in science and a female R programmer. First, my boyfriend is a data scientist at a start-up in Paris, and before that he was a PhD student in a bioinformatics lab. His main programming language is also R, which has been great in a lot of ways. He does mostly predictive analytics while I do mostly inferential statistics. As a result we can give advice to each other on how to do different problems that one or the other of us is more equipped to handle. We’ll often exchange links to new exciting packages or discuss R packages we could make together. All of this is great…until we talk to anyone we know outside of my lab. As soon as I mention some R problem I’m having or that I spent all day coding I inevitably get asked “Oh, did Eric teach you R?”. I used to have a series of responses to this like “No, actually I started using R before him” or “No, we do pretty different things in R” but that usually came with a smile and the look of someone desperately wishing they could get out of this conversation. Now, these aren’t bad people. These are friends of mine who I like a lot, but when it comes to programming the default assumption is just that surely if both of us program in the same language I must have learned from him. I’ve tried to rationalize this away as a fieldist comment. I work in linguistics he works in bioinformatics, so that’s where the assumption comes from. It’s not sexism it’s just judgement about my field of study! (More on that below.) Inevitably though I decide that it can’t just be that because if our fields were reversed I doubt he would be asked “Oh, did Page teach you R?”. And then the impostor syndrome starts to kick in. “Well, maybe they have something. I mean, he does help me. I do ask him questions about code. He did introduce me to dplyr.” And then suddenly all of that back and forth discussion we’ve had, every time I’ve helped him, or told him about a new package disappears from memory. This is what I think is the worst part of the gender discrimination in science and coding, I start to believe it. Not letting those feelings feed my impostor syndrome has been (and continues to be) difficult, but I’ve found being able to talk about it and exchange stories has been a huge help.
Gender isn’t the only type of discrimination I’ve experienced though. The second issue I’ve encountered is fieldism. Every linguist has the questions that they get asked from people outside of our field “How many languages do you speak?”, “Do you correct people’s grammar a lot?” etc. Along with that comes “What do you even use R for?”. I try and explain, but once again I usually get that smile and look of panic. One of the most frustrating parts of this is in my annoyance that my knowledge is being questioned solely based on my field of study, I start judging the person asking me the question about their field of study. In my explanation I start throwing in names of statistical tests I know they have never heard of and have never used. And why? Because somehow it makes me feel better to attack someone else’s statistics knowledge to validate my own. Of course this comes from somewhere else than the person who asks the question. And this brings me to the third, and I think most important factor that has led to my impostor syndrome with R, other people in my field of study.
After five plus years in graduate school I can say confidently that the number one cause of my impostor syndrome for statistics is how other linguists (and potentially social scientists in general) talk about statistics. We are taught that only the smartest people in the field really know what they are doing, and the rest of us are just fumbling through. We see this in how classes are taught, and in conference presentations where people are more focused on questioning the presenter’s statistics than discussing the implications of the results. All of this leads a self-taught person like me to think “Sure, I spend a lot of time programming in R, but how many analyses do I really use? It’s not even that complicated of a test, right? I mean, if I understand it it can’t be that difficult, right?”. This really became apparent to me when I went to the R conference useR! in 2014. I can easily say it was the best conference I’ve ever gone to. It brought together people from both academia and industry for presentations on new work in R. I participated in a couple tutorials and learned about new packages just being released. Most importantly though I never felt like I shouldn’t be there. Tutorials were explained in an accessible way and without a condescension that I had previously experienced when discussing R with people in my field. I walked away from that conference with a new found self-confidence in my abilities and a realization that R and statistics could be taught in a completely different way than what I had experienced before.
This brings us today. Since attending useR! 2014 I still fall back into moments of self doubt, but it gave me the confidence to accept that I do know what I’m talking about and that I could teach R in the way I wish I had been taught. As a result one of my goals for my R course is to teach R in a manner so that people can realize anyone can learn how to be an R programmer.
*Stay tuned for a digital version of the tutorial series on this blog.
Page Piccinini is currently a postdoc at École Normale Supérieure in Paris. She has recently obtained a Ph.D. from University of California, San Diego, working on the phonetics of code-switching in Spanish-English bilinguals.