Pages

Friday, November 30, 2012

Phrenology 2.0

Withywindle links to a very long and questionable study of disparities in Jewish and Asian elite college admissions arguing that Asians and whites are being actively excluded while underqualified Jews are getting preference. The study is done with what I suspect are high-school level statistical methods, and I've already registered my many methodological complaints in the comments at A&J. But in thinking what this article might more broadly accomplish, I've decided to be optimistic that such aggressive accusations coupled with such shoddy data will encourage the schools in question to actually release their own closely-guarded data on admissions in order to dispel these charges.

In other news, despite the supposed shrinkage of my co-ethnics' skull sizes, my own head swelled substantially this week when I passed my prospectus defense and became ABD.

16 comments:

Phoebe said...

Woohoo, ABD, congrats! Now the fun begins.

Re: counting Jews, if the French had done more of this, my dissertation would be a lot easier (or better?), but as a rule I'm against it. It's kind of futile, because who self-identifies as Jewish is going to be quite different from who does so in some official capacity (i.e. Hillel), and, in turn, as you say, from who has a Jewish last name. Like, I'm certainly Jewish, but I'm never going to tell that to anyone counting Jews, and my last name isn't giving any hints. It's also as they say in academia problematic, because a big part of anti-Semitism has long been defining people as Jews who do not self-identify as such. If David Goldberg is Methodist and so's his mom, and the only actual Jew was his great-great-great-grandfather, he's still Jewish if we're studying anti-Semitism, but maybe not if the question is representation of Jews at colleges.

And, I can't plod through that article long enough to sort out why the author thinks Jews came to be overrepresented these days despite our mediocrity. Was it just b/c of legacy admits, whose parents were the actual smart Jews? Meh.

Withywindle said...

MSI: Carl Schorske, Fin de Siecle Vienna. The procedures are substantially the same--count names--but that counts as good history because that was in another country, and besides those Jews are dead.

Miss Self-Important said...

Phoebe and Withy: I'm not sure counting Jews is the problem here so much as how Unz has gone about counting them. Assuming from surnames just seems very inaccurate in a context where many Jews don't have distinctively Jewish names because they've intermarried or changed their names. (In his appendix, Unz himself says his set accounts for on average only 1 in 12 Jews. When we're dealing with a dataset of only hundreds, that's a huge potential error.) Phoebe is one example, Withywindle is, when he considers himself a Jew, another. I have a name that is distinctly Jewish if you know Russian, and totally indistinct and unpronounceable if you don't. Jews not descended from Central Europe have a wide variety of surnames that this methodology would clearly miss.

And this is all there as a problem even before you get into questions like, who does Unz take to be a Jew in his study in the first place. Two Jewish parents only? One? A grandparent? Vague personal sense of identity? That problem is particularly salient if you propose to study a distinction between Jews and whites in America, since Jewish intermarriage most frequently occurs with whites. I don't know about Schorske and fin-de-siecle Vienna. This method would be more reliable if the Jews of Vienna were more like the Asians of present-day America - unlikely to intermarry and carrying a set of very distinct names. I don't think Unz's use of this method for identifying Asians is nearly as questionable as it is for Jews, although it may well be if he replicates the experiment in 20 years when Asian intermarriage and Americanization has accelerated. My beef is not with the idea of a census, but with the crude and unreliable way it's been carried out.

Phoebe: He claims that it's b/c admissions officers are themselves Jews or people with cultural sympathies that reflect those of affluent, urbanized lefties - which Jews in America overwhelmingly are - and so they're biased in favor of admitting such Jews despite lower academic credentials over white farmer children from Nebraska whose extra-currics are all church volunteering and FFA, and robotic-seeming Chinese hyper-achievers whose ECs are limited to cram school attendance. You should read this piece - it's up your alley in every way, even if it's a bit in the phrenology corner of that alley.

Emily Hale said...

Congrats!

Withywindle said...

My beef is not with the idea of a census, but with the crude and unreliable way it's been carried out.

In that case, I would only be quarreling with you over trivialities.

Miss Self-Important said...

EH: Thanks!

Withy: Sure, if we allow that this article is a trivialized by its poor methodology.

Withywindle said...

Revise and resubmit?

Miss Self-Important said...

Yes! But to whom does one resubmit if one is the journal's publisher?

Flavia said...

Congrats! All downhill from here, amiright?

(And on the subject of those with swollen heads, my captcha is "asholent." Which I take to be the adjectival form.)

Miss Self-Important said...

Well, since I take something close to your view of grad school, the whole thing has, against all physical logic, been kind of downhill. Although my appreciation for my advisor has grown, so that's the only counter-force against decline.

Flavia said...

Actually, I meant "downhill" in the sense of coasting downhill (e.g. on a bike, after the tough upward climb).

But since I meant it at least partly ironically, the other sense also obtains.

Miss Self-Important said...

Oh, well in that case, no. Actually starting the dissertation has been a lot harder than defending the prospectus.

Withywindle said...

Congratulations on the ABD too, BTW.

Re being a publisher--I do think it's a good thing, all in all, to have a world where someone can be richer than Croesus, own their own mag, publish a 30K article without going through all that revise-and-resubmittery that would keep your work from getting published at all, and thus start to move the national conversation even if you haven't gotten past peer review.

Anonymous said...

I think Unz anticipates a lot of the methodological objections, even if his answers aren't totally satisfactory. (As Sam Goldman and you noted, it's at least partly not his fault, since a lot of the necessary data aren't publicly available.)

For example, the relative drop in Jewish NMS semifinalists isn't explained by differences in the share of the Jewish population that's college-age: Unz says that the difference persists even after adjusting for age structure. ("Thus, it appears that California Asians are perhaps three times as likely as Jews to do extremely well on academic tests, and this result remains unchanged if we adjust for the age distributions of the two populations.") There's more detail on that in Appendix B.

The objection that surname-counting is a less reliable method of estimating Jews than Asians is true, but it goes to the precision of the proxy rather than the accuracy unless it systematically under- or overcounts. (Random error may still be a problem where the measures have small sample sizes, as with IMO competitors, but less so with the large 25-state NMS dataset.) And even systematic bias doesn't explain the fall in Jewish numbers over time, unless the level of systematic bias has shifted in the last few decades. Finally, the sheer size of the disparities means that the potential methodological problems would have to be more severe than seems plausible in order to call the basic conclusions into question.

Miss Self-Important said...

Anonymous post, but apparently only to my inbox, so I reproduce for posterity:

I think Unz anticipates a lot of the methodological objections, even if his answers aren't totally satisfactory. (As Sam Goldman and you noted, it's at least partly not his fault, since a lot of the necessary data aren't publicly available.)

For example, the relative drop in Jewish NMS semifinalists isn't explained by differences in the share of the Jewish population that's college-age: Unz says that the difference persists even after adjusting for age structure. ("Thus, it appears that California Asians are perhaps three times as likely as Jews to do extremely well on academic tests, and this result remains unchanged if we adjust for the age distributions of the two populations.") There's more detail on that in Appendix B.

The objection that surname-counting is a less reliable method of estimating Jews than Asians is true, but it goes to the precision of the proxy rather than the accuracy unless it systematically under- or overcounts. (Random error may still be a problem where the measures have small sample sizes, as with IMO competitors, but less so with the large 25-state NMS dataset.) And even systematic bias doesn't explain the fall in Jewish numbers over time, unless the level of systematic bias has shifted in the last few decades. Finally, the sheer size of the disparities means that the potential methodological problems would have to be more severe than seems plausible in order to call the basic conclusions into question.

loafingcactus said...

Congratulations on your ABD from someone who is mostly silent on comments but thoroughly enjoys your blog!