From the department of "Well Yes"

From Stephen Levitt's latest paper on the Ku Klux Klan:

"Perhaps the most limiting feature of our data is that we were unable to obtain any records on Klan members or activities in the Deep South.

Indeed. Perhaps more on this later, particularly if a non-gated version turns up. As it is, I've looked at it and it epitomises all that's good and bad about Levitt's research. The "unique dataset" that they've put together (basically a fuzzy-match of Klan membership lists to census data) is quite cool. The uses they've put it to are crazy - in particular, it looks to me as if a lot of the regressions they've run on Klan membership versus lynchings, Republican vote share, net migration etc, are structurally very similar to the "more guns less crime" model, and reproduce all of that model's problems with endogeneity, reverse causation etc.

But the big two problems they have are 1) the dataset is dominated by Pennsylvania and Ohio, and it clearly doesn't pass the laugh test to be doing a "rigorous" assessment of the effectiveness and aggressiveness of the Ku Klux Klan as a terrorist organisation based on what happened in non-Confederate states, and 2) they just totally ignore the literature. There is no mention, for example, of any secret societies other than the KKK (they claim that the Know-Nothings were the only precedent, which is just not true). This is quite an ommission in the context of Indiana - recall that the Supreme Tribe of Ben-Hur were big in Indiana too. And they act like they have made a massive discovery that the Klan acted like a pyramid-selling scheme generating money for those at the top, whereas this is like an utter commonplace of the history of secret societies in the USA - the business was so institutionalised that there were plenty of professional insurance lodge salesmen hanging about (and indeed the revival Klan was partly founded by one).

So, dodgy use of interesting data in order to partially reinvent the wheel, hyped to the sky as definitive and rigorous. Freakofreakinomics.

Update: Aha, here ye go, via the Marginal Revolution comments. Note that the stereotype of KKK members as ill-educated hicks is really a hell of a straw man; the demographics of the Klan (that they skewed lower-middle class, like the Poujadists, Blackshirts and basically every other fascist movement ever) have been known for a long time by all serious historians, who have dug through the registers and names, although not on this sort of systematic basis. To be honest, the pure and simple cost of the KKK as a hobby more or less ensured that it was mainly a well-off man's pleasure.

Update Update: (via Cian in the comments) - look at this. There surely is no excuse for that? At some point, the constant drip drip of this rubbish surely has to have some sort of effect on Levitt's academic credibility? Either he thinks that the NYT article linked is good enough research to be published under his name, which would be really worrying, or he knows it's crap and is publishing it anyway for the cash, which is also not hugely edifying.