Jan Hoffler sends along this information:
YSI workshop with Richard Ball, Johannes Pfeifer, Edward Miguel, Jan H. Höffler, and Thomas Herndon
January 6 – 7, San Francisco, Mozilla Science Lab
The workshop will take place right after the Annual Meeting of the American Social Sciences Associations, which includes the Annual Meeting of the American Economic Association (AEA) . . .
The workshop will consist of mini-courses covering research transparency in empirical research and macro models that are neglected in the conventional economics curriculum. For young scholars it can be very useful to orient themselves by looking at how established researchers do their studies. By now there is a lot of material available but then it is often frustrating when one wants to take a look at how their analyses were done just to see that it cannot so easily be redone. This workshops intends to help young scholars to find out how to replicate others’ studies and how to archive their own research for future use and for others.
The workshop also will feature student presentation sessions, which will give Ph.D. candidates the opportunity to present and discuss their research in a collaborative environment. Applicants shall enter the title and abstract in the registration form (deadline is November 1) and submit the complete version to the Institute no later than December 1, 2015. Moreover, during the joined lunch and dinner there will be ample time for social interaction with students and teachers.
P.S. The name Edward Miguel rings a bell . . . oh yeah, here he is. His website remains impressive but it no longer says that he’s from New Jersey. I wonder what happened with that.
P.P.S. A commenter reminds us that Miguel is also involved in the Worm Wars which we discussed recently in this space.
I think they revoked his New Jersey citizenship because they discovered he eats Montreal bagels. #TrueFact #SickSadWorld
With Michael Kremer, Ed Miguel is one of the authors of the original deworming paper that was the subject of a replication controversy a few months ago.
I would be interested to hear your take on Miguel’s latest coup:
http://web.stanford.edu/~mburke/climate/BurkeHsiangMiguel2015.pdf
Based on data from the last 40 years or so they predict things like 96% GDP loss in 2100 in Saudi Arabia, or 419% GDP gain in Russia:
http://web.stanford.edu/~mburke/climate/map.php
Isn’t that a bit of a stretch??
Send in the Edlin factor!
But the problem here isn’t likely to be the statistical significance filter, right? It is more likely to be out-of-sample prediction: as in, extrapolating the effects of fairly large weather changes from fairly small ones, yeah? I’m not saying it isn’t over-estimated, I’m just saying that the over-estimate would likely stem from functional form considerations and not from the statistical significance filter and/or various types of p-hacking (err… large effect size hacking), which I thought was the motivation for the Edlin factor.
I think your critique here is more like your critique of the pollution RD paper than the Jamaican babies paper. So I think it requires a differently-named reduction factor. Or is the Edlin factor more general than I’m giving it credit for? If that is the case, then perhaps we need several classes of Edlin factors associated with different kinds of statistical problems…
Also – no comment on the beautiful graphics? I once again plug Sol Hsiang as one of the best people in the business at visual presentation of results and uncertainty. But maybe you don’t like them?
Jrc:
Good point, I haven’t really thought too carefully about how the different biases interact! And, yes, I really like Hsiang’s graphs.