Thread by @RochelleTerman

Thread: Recent events have caused me to reflect on the way I teach computational social science, and how this relates to my experiences with the data methods / data science community in political science. 1/n

Some background: For the past 6-7 years, I've taught a variety of courses and workshops on computational social science (CSS): R, python, git, webscraping, web development, machine learning, text-as-data, etc. I've taught over 500 people, mostly graduate students.

My specialty is "Introductions" -- i.e., teaching folks with no background knowledge. My pedagogical goal is to give students the tools + background they need to pursue these topics in advanced courses, and on their own. 3/n

I make this goal explicit in the beginning of every class: I am not a methodologist. I do not train students in advanced methods or computer science. I offer the skills required to be successful in more advanced methods programs. 4/n

The vast majority of my experience has been super positive. Most of my students are women + URM, and it is especially motivating to see these students become interested in these topics, succeed in the course, and pursue more advanced training. 5/n

However, I have also faced criticism for the way I teach by (some) colleagues -- all advanced methodologists. (Why are advanced methodologists in my intro courses? More on that later.) 6/n

These criticisms center on a common theme: My courses are too remedial, too accessible, not technical enough, not mathy, not rigorous. The critics rarely define what the appropriate level of "rigor" is, or provide a pedagogical justification for their preferred mode of rigor.

For example: When teaching R, how far to you go? Should you cover object-oriented programming, compiling in C, assembly, electric circuits and motherboards? 8/n

In any case -- I am told that my level of technical exposition -- however ill-defined -- is insufficient. 9/n

For example, in 2015 I created PS239T: Intro to computing for social scientists. The idea was to teach social scientists the computing they need to know, without the CS material that was irrelevant to them. Now there are dozens of such courses, but at the time it was novel. 10/n

At the time, a faculty member in my dept expressed hostility to the idea, demanding that grad students take CS courses if they wanted to learn how to program. He called my class "remedial." 11/n

That course, I am proud to say, is now in it's 5th iteration, and is one of the most popular grad courses in Poli Sci, while drawing heavy enrollment by other departments. 12/n

Another example: I taught Intro to ML at stanford, geared towards undergrads with no STEM background. An adv grad student in engineering insisted he take the course, even when I advised him that another ML course might better suit his needs. 13/n

He consistently disrupted the class with highly-technical comments that were irrelevant to the subject matter. I imagine the did this mainly to signal his technical superiority and undermine my authority. His final evaluation degraded the class as amateurish. 14/n

Most recently, I taught intro to computational text analysis to early grad students. A senior faculty member sat in, and interrupted my lecture, calling it "useless". Why? In a 1-hour unit on topic modeling, I skipped the statistical properties + optimization of LDA.15/n

All of this raises a series of questions for me. 1) why do advanced methodologists feel compelled to sit in on my INTRO courses if they don't find it helpful for their needs? (I make my approach known up front + no one is holding a gun to their head!) 16/n

2) What is the end goal of methods training? Which people deserve to be taught? What is the intention behind teaching "technical details", and how does it actually function with our students? 17/n

I suspect the answer to both questions has to do with the endless gate-keeping that ensure the exclusivity of a club called "methods," and the status associated with membership into that elite circle. 18/n

Ironically, such gatekeeping does nothing to improve the methodological sophistication of the discipline, and actively undermines that goal. 19/n

For example, which course constitutes "better training": 1) a highly technical course taught by a brilliant methodologist but awful teacher, so poorly and unskillfully taught that every student (except 1-2) checks out and gives up? or 2) a less technical, well taught class? 20/n

I suspect that many of my critics would inherently find the more-technical class to constitute better training, even though the average skill level of students faired worse than the second course. 21/n

Thus there's a conflation between accessibility and poor training; and between rigor and "weeding out". But I suppose that's a feature, not a bug, of (a portion of) the methods community in Political Science. Much like a LV handbag, greater popularity means less prestige. 22/n

I'll end this thread with a question I posed earlier: What is the purpose, the goal, the point of teaching methods? Is it to increase the methodological sophistication of the discipline, or to maintain a hierarchy? As a community, we need to wrestle with that question. /END

Addendum: It has occurred to me some readers might think I'm talking about @UChicagoPoliSci and I just want to make clear -- none of this happened at UChicago, and the department has been nothing but positive in my experience so far!