I don't really see a full fledged regression analysis here: http://econlog.econlib.org/archives/2012/01/the_college_pre.html http://econlog.econlib.org/archives

I don't really see a full fledged regression analysis here: http://econlog.econlib.org/archives/2012/01/the_college_pre.html http://econlog.econlib.org/archives/2012/01/the_college_pre.... There are several potentially correlated variables (College grad and cognitive ability?). That's not necessarily a problem but I think it would necessitate showing different models, not just one. It's also atheoretical - it's not advisable just to slap a model together and find regression coefficients with high t-stats. In economics, it's important to start with a believable hypothesis grounded in reasonable theory. This study is like a half-hearted attempt at exploratory data analysis - half-hearted because only the results of one model are shown and there are no charts or graphs to give a feel for the data. You really can't just jump to a conversation about causes unless you've done more work on the data (maybe Caplan did, but he really needs to show it to stimulate the conversation he wants to have).