I have long wanted to write a long essay on causal inference for humanists. The idea behind this essay would be to lay out the fundamentals of causal inference in the social sciences for an audience of people trained in an entirely different way of doing academic research.
This would be different than the more conversational introductions to causal inference and statistics that are very popular these days, like Mostly Harmless Econometrics, or Murray (2006) on instrumental variables, or my old favorite “Maximum Likelihood for the Masses” by Don Green. Those books and articles are nevertheless geared towards social scientists who want to use this methods. The essay I have in mind would be far less technical, much more illustrative, and most important, precise and exacting about how to be a proper critic of this type of work. I want, in other words, to bring contemporary methodological debates on causal inference to a new audience. Not within the harder social sciences, or even to the perestroikans, but outside of that community altogether.
Why do this? Because there is such a yawning gap between what most historians, anthropologists, critical area studies people, etc think that social science methodology is and what most of the contemporary methodological debate is actually about. I think it would be useful, even fun, to start these conversations, at least to the extent that others want to join in. At the very least, it would be great to have on hand an essay like this to share to my colleagues who work in a more traditional area studies vein but who are still preoccupied by early 1990s debates about quantification and rational choice theory.
I bet that lots of you reading this blog would enjoy reading a long essay of that sort. So why don’t I write it? Because I can’t think of any way to publish it. There’s no way that the American Historical Review or the Journal of Asian Studies or Comparative Studies in Society and History, three outlets that might have some reach in the communities I wish to converse with, would ever publish an essay of this sort. There’s also no way that any serious political science journal would publish it (they shouldn’t—I don’t have any novel contribution in mind, and my audience isn’t social scientists anyway). If I wrote an essay like this, it would be a permanent working paper. That means I can blog about it, put it online for any to read, share it on SSRN, but that won’t give me the reach I want. And I can’t figure out my incentive to do things that neither get broad attention nor get a nice CV line.
This, then, brings me to the title of this post. I would write an unpublishable essay of this sort if there were a serious working paper series in which it could appear. The model I have in mind is the working paper series from the National Bureau of Economic Research. These are not peer-reviewed, but they are widely shared and discussed in a way that gets the ideas out even if they never become formal journal articles. Having a paper that is a permanent NBER working paper isn’t ideal, but it’s also not that bad. There are of course problems with the NBER itself, like its clubbiness and its relative deemphasis of topics like development, but my point about the value of the working paper series still stands.
Why doesn’t political science have anything like this? Don’t know. But I sure wish there were such as a thing.
Tom Pepinsky 