18/01/2012

You Heard It Here First: Rubinian Turn in Sociology

Quantitative empirical sociology is in the early stages of a Rubinian turn. Rubinian as in Rubin causal model. There was a long article by Markus Gangl in the 2010 volume of the much-read Annual Review of Sociology, which basically equated the counterfactual concept of causality with the Rubin Causal Model. Sure, Winship and Morgan already had a pretty similar paper in the 1999 edition, apparently without much immediate effect, but now you also see Rubinian analyses more and more in the top journals (heck, there's even one coming up in the leading German-language journal). I think the model and its applications do have their shortcomings, but it's certainly a huge improvement over the "let's throw all variables that may also influence the outcome onto the right-hand side" approach and its cousin, the "let's put variables from three approaches into the regression and see who wins" school of theory testing.

I expect lots of papers getting published in good journals just because they use propensity score weighting. Or, as Robert Sampson once put it, "remember LISREL?" Nonetheless, best thing that has happened to the field in my lifetime.

No comments: