- Mapping the results from Iowa at Offensive politics (via Revolution Analytics). Some other interesting resources about maps in R: “R” you ready?, Baptiste Coulmont.
- Matt Asher runs a simulation to look at the difference between the first results reported on TV and the final count.
- Andrew Gelman has a series of posts on possible irregularities in the Russian elections. In fact, using statistical techniques to look for signs of fraud in election data seems to be a growing field, see e.g. Walter Mebane and The Forensics of Election Fraud.
Saturday, January 28, 2012
Election season and statistics
The US primary season is now in full swing and it's as good an occasion as any to play with some data. Some interesting stuff from a statistical angle:
Thursday, January 26, 2012
Research as usual
Andrew Gelman just posted a follow-up on one of the big psychological research “scandals” of 2011: Daryl Bem's “Feeling the future” paper. Bem relates a series of experiments showing evidence of paranormal “precognition” (for example predicting on which side of the screen erotic pictures are going to appear or remembering some words better before learning them). Given the nature of the paper, the editors of the journal that published it decided to explain their decision and invite a methodological critique.
The study has been heavily commented on the web. Since then, another methodological critique and failures to replicate Bem's results have appeared (but it turns out to be more difficult to publish them in the JPSP than a paper arguing for psychic powers). One of the main point on which all commenters agree is that the original studies are actually quite banal methodologically speaking, as far as social psychology experiments go. Tal Yarkoni details all the little flaws that make it possible to find such spurious results but none of them seem very big and all are pretty common in the psychological literature.
All of this reminded me of another little scandal that unfolded last year: Satoshi Kanazawa blog post proclaiming that “black women are less attractive” (more on the content and methodological flaw in the analysis). It was not the first time that Kanazawa posted stupid and offensive stuff on his blog but this time, it started a storm of controversy, complete with calls to sack him, an official investigation, letters of support and Psychology Today finally caving in to the pressure and removing the text. True, it was not a peer-reviewed article but the sad thing is that his usual output is not much better and still his supporters are perfectly right when they stress that he has published many articles that were judged sound by reviewers.
The two scandals were quite different but in both cases boil down to offensive and ludicrous findings that still meet the current methodological standards within psychology. This should perhaps tell us something about those standards…
PS: A new post from Andrew Gelman with some thought on how to improve the situation just came in as I was writing this entry.
The study has been heavily commented on the web. Since then, another methodological critique and failures to replicate Bem's results have appeared (but it turns out to be more difficult to publish them in the JPSP than a paper arguing for psychic powers). One of the main point on which all commenters agree is that the original studies are actually quite banal methodologically speaking, as far as social psychology experiments go. Tal Yarkoni details all the little flaws that make it possible to find such spurious results but none of them seem very big and all are pretty common in the psychological literature.
All of this reminded me of another little scandal that unfolded last year: Satoshi Kanazawa blog post proclaiming that “black women are less attractive” (more on the content and methodological flaw in the analysis). It was not the first time that Kanazawa posted stupid and offensive stuff on his blog but this time, it started a storm of controversy, complete with calls to sack him, an official investigation, letters of support and Psychology Today finally caving in to the pressure and removing the text. True, it was not a peer-reviewed article but the sad thing is that his usual output is not much better and still his supporters are perfectly right when they stress that he has published many articles that were judged sound by reviewers.
The two scandals were quite different but in both cases boil down to offensive and ludicrous findings that still meet the current methodological standards within psychology. This should perhaps tell us something about those standards…
PS: A new post from Andrew Gelman with some thought on how to improve the situation just came in as I was writing this entry.
Friday, January 13, 2012
Our programme will resume shortly
It's now over two years since the last time I have posted anything on this blog. Not that I was intensively blogging before, but I had a pretty good excuse. Now that my PhD is out (it should be available soon on the TU Delft repository), I intend to return to blogging. Let's see how it goes…
Subscribe to:
Posts (Atom)