Back in 2015 we had a discussion with economist Douglas Campbell on evolutionary psychology in economics.

Then in 2016 Campbell sent me this:

So, Spolaore and Wacziarg have a new paper out, with a brand new measure of genetic distance which they claim is robust, although their paper again repeats some of their earlier mistakes. This time they do control for latitude (I wonder where they got this idea?), but they again omit a control for Sub-Saharan Africa (they exclude it instead). They also do some cheerleading for Ashraf-Galor, one of your favorites. Any interest in a follow-up post on this?

Secondly, myself, Shiping Tang, and Jack Paine are contemplating doing a Meta-Robustness paper on the entire “Macro-Genoeconomics” literature, especially as it continues to grow (Ashraf and Galor also have a new follow-up slated for the Journal of Economic Literature). The idea would be to check the robustness of all of these studies at once. There appear to be about 14 papers in leading economics journals which use genetic distance or diversity, and all of them look as though they would fall apart on inspection. (I’m sure Spolaore and Wacziarg’s new masterpiece will also be published by the time we have a draft, so it too could be added.) Blunt continent/regional dummies would likely do the damage for most of these. We’re quite curious what the “Edlin Factor” on these scholarly works published in revered economics journals will be after testing robustness. We’d also like to implement something like what the Camerer study did, and set up a betting market and do opinion polls on which studies will prove robust.

I never got around to blogging the above story, but in the meantime Campbell updated me in 2020:

So, with coauthor Karsten Mau, I went and wrote another comment paper. We published this one in the Review of Economic Studies. We caught some guys at Stanford & MIT (two of the top 12 cited economists the past 10 years) doing some Brian Wansink-level p-hacking. It might be a teachable moment for your blog readers. These guys run a difference-in-difference, but don’t plot any (relevant) data. They have a hard-to-believe thesis that happens to be a pleasing story to the median economist (trade increases growth). They make some coding errors, and use a one plus log variable transformation.

Two of the top 12 cited economists! That doesn’t sound good. But I guess one path to citations is to make big claims, and one path to big claims is to go beyond your data.

Then in 2021 Campbell sent one more update:

Both my paper, and a response by the original authors, has now been published in the Review of Economic Studies. Here is their response.

I [Campbell] think I’m not going to write a full response, but I did make a quick response on Twitter.

In my view, their response has a clear multi-collinearity problem. In addition, their new favored “correct” regression differs in four different ways from their stated regression in their first draft.