1. The first problem isn't a problem. It's not what they were looking at. I didn't understand the results of this study as reported to mean anything that what the study said.
I thought that was crystal clear. Of course, the study isn't a different study, but I've never faulted researchers for not doing the research I wanted them to do, but instead pursued their own research agenda. The claim is that in an encounter the risks vary for non-lethal force but not for lethal force. If you don't read the whole sentence, you don't get the whole meaning.
2. The researchers deal with this in the paper. They know it's a limited dataset, and they even point out problems within the data set and use a national survey to compare their findings with--sometimes finding similarities, but not always. It's only a problem if you don't actually bother to read even the summaries. Again, I knew they had a limited data set, and the summary article I read in the MSM pointed out comparisons with the survey. Civilian-reported differences typically cited greater use of non-lethal force, unless you're white--then the civilian-reported incidents were far lower than what the police reported.
Some of the details--such as "precincts matter"--is worth trumpeting from the rooftops, though. Instead, most activists smooth over the fact that even within a city like Houston you get a wide range of outcomes, and instead try to say that "police" are this or that. It's worth pointing out that if they'd used a national dataset, they would not have gotten that information. Perhaps variation by city might have popped out, but perhaps only by quadrant (which is how the FBI breaks things down when they do geography).
3. NBER studies aren't peer reviewed. That's not a surprise. However, they're often widely cited in the press because it's a rather prestigious organization and they often have interesting things to say. Their determination of the start and conclusion of recessions is a scholarly conclusion, but that's also not peer reviewed. I don't know how many times I've read something from the NYT or WaPo, often favorably viewed on DU, where the paper was on the NBER site. (IIRC, there's also a sign-in wall with more goodies behind it, but I don't have any credentials that I think would get me access. Pity, that.)
"Working papers" don't all necessarily fall into the "shoddy or incomplete." Sometimes they're chunks of research that will wind up in a book (books, by the way, are very often not peer reviewed). Sometimes they make no big theoretical point or the people doing the research don't want to take the time to polish it for print. Or even it's a way of getting things into print quickly while the peer-review process works in the background.
My wife did that with a number of things. Working papers or conference proceedings, then publish in good form. Typically the "good form" results varied at best slightly from the working papers. In some cases the argument was strengthened, or two papers merged if thematically similar enough. What I've published that underwent peer review was little changed. At the same time, mss. and working papers are often widespread in some fields, simply because it's not worth the hassle to dot every i and cross every t.
By the way, after "considering" the idea of genetically lower intelligence, what was his finding?
= new reply since forum marked as read
