Friday, February 15, 2013

Paging Withywindle

Andrew Gelman also finds some problems w/ Ron Unz's statistics, including some of the problems I indicated about counting Jews of intermarried parents with non-obvious names, and the drop in the percentage of the Jewish population that is not Ultra-Orthodox and so competing in elite high school achievement contests like NMS and Intel and applying for elite college spots. He seems to be coming from the same position as me - not primarily concerned with the political correctness of making Unz's claims, but that the claims be based on methods that take account of demographic realities.


Withywindle said...

Yah, I read that--starting from Unz's response to it, working backward. I thought of mentioning it on my blog, but thought I might be boring you (and everyone else) if I kept on about it myself. I thought he was actually missing your point about how to count .5 and .25 Jews, and so on. I didn't delve too much into the weeds, since I am innumerate anyway. My reactions were: 1) both he and Unz were unpleasantly snide to one another, and their made me mistrust both of them; 2) it wouldn't have hurt him to say something like, "You are right that everybody else has used Hillel numbers when it suits them, it is a double-standard that everyone else got a pass and you don't, anyone else who uses those numbers in the future should be slapped down briskly--nonetheless, since the numbers don't seem to be accurate, we should adjust the estimates like so"; 3) he should also acknowledge that everybody in popular punditry combines different statistical sources blithely, and Unz is not really doing anything out of the ordinary for punditry.

So, yeah, he's making me doubt Unz's thesis a bit more, but he's also making me thinking that you could probably apply equally withering critiques to virtually every Popularly Known Statistic--especially to the ones Liberals Love.

Miss Self-Important said...

He does account for the half-Jewish phenomenon in the section where he quotes someone doing a re-count of Jews in the small sample pools of the math and science competitions "based on personal knowledge of these students", which I assume means the person is not going by last names but by information about family background. That person's re-count takes each half-Jew to count as one Jew and one white, I think, although I'm not quite sure what he means by "with 1/2 Jews being counted 50-50 between Jewish and non-Jewish white."

Everyone's an asshole on the internet, so I ignored that part, since I have no prior evaluation of either Gelman or Unz to go on. But as for the possibilities of using statistics to destroy other statistics - completely do-able if you want to slog through the two-year sequence of courses on statistical methods in the social sciences. Few people do on the whole, with conservatives among them probably proportional to the minority of conservatives in the social sciences as a whole. This Unz stuff, however, looks more like the kind of statistics you can manage with a high school math education.

Withywindle said...


Nothing particularly intelligent to add right now; I just wanted you to know I'd read your comment.

I also want you to know that it has been cold for the last month and I envy you bitterly, Miss "I live in San Diego and yes it is pleasantly warm year-round" Self Important.

Miss Self-Important said...

It's only relatively warmer, and one's expectations always climb higher than reality on that front. You can move here, and give me your place in NY so I can return to civilization.