Meritocracy: The Yale Debate and Surname Analysis
I just returned from attending a couple of events at Yale University, all in connection with the controversial issues raised by my Meritocracy article.
On Tuesday, I participated in a large public debate organized by the Yale Political Union on the somewhat related question of whether Affirmative Action on college admissions should be ended. The audience was narrowly divided on the question, and often loud and boisterous in expressing their sentiments, thus resulting in a lively evening, as reported in the Yale Daily News.
One of the most interesting moments came when a girl from the audience declared she was a Southern Baptist, and that she sometimes wondered whether she was nearly the only one at Yale. America contains some 16 million Southern Baptists, who constitute our nation’s largest Protestant group, so if Yale does indeed only enroll a negligible number of such individuals among its 5500 students, this might raise troubling questions of exactly how the university administrators define their vaunted public commitment to “diversity” and also the nature of their recruitment and admissions policies.
On Wednesday afternoon, I made an hour-long presentation at the Yale Law School, co-sponsored by the Asian-American Law Students Association and the Federalist Society, which drew a remarkable 100 students out of a total enrollment of around 600, filling one of the large lecture halls. In this instance, the research findings and proposals of my article were the central topic under examination, and the law students had many detailed and probing questions, producing a very useful discussion.
But for me, the true highlight of the visit came later that evening, when Amy Chua—of “Tiger Mom” fame—and her husband Jed Rubenfeld, both prominent law professors, hosted a small gathering at their home for faculty members and students, with the resulting discussion of the topics in my article and various other public policy issues continuing on for several hours. I’ve always greatly enjoyed such thoughtful discussions with extremely intelligent, knowledgeable people, and I afterward once again regretted my own late-1980s defection from the academic world.
One of the substantive issues raised by a couple of students during my law school presentation was the accuracy of my analysis of Jewish numbers, and this provides me with a perfect opportunity to discuss this issue. Surprisingly enough, it has not yet been the subject of any detailed published critique, but the matter is obviously crucial to important parts of my analysis.
First, as I indicated in my original text, the Asian numbers I presented were far more solid and reliable than the Jewish ones, and the latter figures inherently contain a great deal of uncertainty. After all, Asian university enrollments are directly reported by the universities themselves in official government-required statistics, while Asian names tend to be extremely distinctive and easy to determine in lists of academic high performers such as NMS semifinalists or Olympiad winners.
By contrast, the only real source of Jewish enrollment figures comes from the Hillel campus organization, and there are serious questions about how accurate or reliable their estimates tend to be. Meanwhile, a substantial fraction of American Jews have non-distinctive or indeed completely Anglo-Saxon surnames, making it difficult to determine their precise numbers on academic achievement lists. Furthermore, Jewish intermarriage rates have been substantial and rising over the last generation or two, further lessening the effectiveness of any numerical estimates based on surname.
Indeed, even as simple a figure as the size and historical growth of America’s total Jewish population is hotly contested among academic experts—even leaving aside the precise definition of “Jewish” itself—and probably uncertain to within 10-20% or perhaps even more, while estimates of intermarriage rates have also provoked huge scholarly or organizational disputes. So given likely errors in several underlying elements, my detailed resultant analysis must be regarded as laced with major uncertainties. However, as I argued in my original article and its quantitative appendices, the approach I followed was probably not unreasonable.
Although the Hillel enrollment figures are mere estimates, they are the only ones available, and thus have been regularly used by the New York Times and other elite MSM outlets, while also constituting the basis for Prof. Jerome Karabel’s award-winning scholarship. Indeed, Hillel’s 1999 claims of a significant decline in Jewish enrollment at Princeton provoked a massive local and national media firestorm, soon persuading the university to investigate and overhaul its entire admissions policy. So I feel comfortable in following the lead of all these other reputable organizations and utilizing the Hillel numbers, while certainly still treating them with some caution.
With regard to estimating Jewish numbers on lists of high academic performers, I already discussed some of the underlying methods and justifications in my Quantitative Appendix E, but should certainly summarize and extend my arguments.
First, the enormous historical over-representation of American Jews on lists of top academic performers led me to generally assume that nearly all distinctively East European or Germanic names were likely or almost certainly Jewish. This over-estimation was intended to partially compensate for the substantial fraction of Jews whose surnames—such as Miller, Gordon, or Brody—would be completely impossible to detect. For entirely similar reasons, I tended to assume that all “Lees” were actually East Asian, even though that surname is also very common among American whites and blacks.
Obviously, such a “direct inspection” approach to ethnic surname analysis is still fraught with error, and should certainly not be regarded as precise. However, after my article appeared, someone brought to my attention a column written a couple of years ago by J.J. Goldberg, editor of The Jewish Forward, which focused on exactly the same question of recent Jewish academic performance, and estimated the number of Jewish names appearing on the finalist lists of the Science Talent Search for this purpose. I therefore compared my own STS finalist estimates with those of Goldberg for the particular handful of years he examined and discovered that the match was almost exact: he counted 100 Jewish names across those ten years, while my own total came to 96. Such estimation methodologies are highly subjective and imprecise, but since my figures were so close to those produced by the editor of America’s leading Jewish newspaper, I doubt my approach is wildly in error.
Next, I validated my methods by applying two rounds of Weyl Analysis, an ethnic estimation technique developed by Nathaniel Weyl some fifty years ago. His ingenious idea was simply to restrict analysis to those surnames which are absolutely distinctive to an ethnic group, and then estimate the total number by considering the fraction of group members who bear those distinctive surnames, based on Census and Social Security lists. For some groups, the task is particularly easy, since a large fraction of all Vietnamese are named Nguyen and so many Koreans are named Kim or Park, but for Jews and most other ethnicities a set of several different distinctive names must be employed for statistical validity. Furthermore, the technique would only be statistically valid when applied to very large datasets such as accumulated NMS semifinalist lists, which contain tens of thousands of names; such sampling would be useless for lists of Olympiad winners or even STS finalists.
With regard to Jewish surnames, I performed Weyl analysis based on two separate sets, a small group of distinctive surnames which had appeared on Internet discussions of the NMS semifinalists and a larger set which Weyl himself had used for his own ethnic analyses. As I mentioned on Prof. Andrew Gelman’s statistics blog, my estimate of recent Jewish NMS semifinalists was 5.95% based on direct inspection, and 5.92% and 6.03% for the two Weyl analyses. These three results were quite consistent and thereby tend to validate the approximate accuracy of the direct inspection methodology, which strengthens the case for its use on the STS, Olympiad, and other lists, for which Weyl Analysis is statistically inapplicable.
The possible impact of intermarriage rates is a separate one. In general, intermarriages between Jews and non-Jews tend to be roughly gender-neutral, so a simple surname analysis would tend to identify one-half of the half-Jews, one-quarter of the quarter-Jews, and so forth. However, it appears plausible that Hillel’s estimates of Jewish numbers might include a much larger fraction of part-Jews, possibly resulting in an inconsistency between estimates of Jewish elite enrollments and separate estimates of high Jewish academic performance. However, I would argue that the impact of this potential mismatch would probably be relatively small.
Although the figures are hardly certain, it appears that at most only 40-50% of identified Jews today enrolled at colleges come from intermarriages, while a generation or so ago the figure was probably closer to 20%. The gender-neutrality of such intermarriages means that the surname estimation error would have thus increased by 10-15% during these decades, leading to an increased relative underestimate of high Jewish academic performance in that same range. However, the apparent decline in such Jewish performance during those decades has been closer to 70-80%, implying that only a small portion might be due to mismatch, and for the same reason, only a relatively small part of any inconsistency between performance and enrollment might have the same cause.
These obviously represent several separate points of possible error—reliance on the Hillel estimates, the uncertainty in surname analysis, and the impact of intermarriage—and such errors might easily compound in the final result. For that reason, if the Jewish enrollment anomalies I noticed in the Ivy League had merely been in the 50% or even the 100% range, I probably would have ignored them, as quite possibly being due to such measurement errors. However, the actual anomaly was in the 1,000% range, and seems very unlikely to be due to such simple measurement-error causes.
These same challenges to my Jewish surname estimation methods had previously been discussed in the comments of a somewhat obscure rightwing Jewish blog.
On a different matter, I should mention that Slate recently ran a rather eccentric article on a topic somewhat related to mine, and the author briefly referenced my work, though he misidentified me as “Run Unz.”
Finally, one of the issues raised in all my Yale events was the considerable skepticism surrounding my suggestion that Harvard and its top national peers should use their gigantic financial endowments to eliminate undergraduate tuition, as I had discussed at length in my sidebar Paying Tuition to a Giant Hedge Fund. As it happens, today’s NYT carried a story on leading university endowments, and the figures indicated that eliminating all tuition would require merely reallocating only about 10% of yearly endowment expenditures, or perhaps a bit more if room and board were included; meanwhile, this would also allow the elimination of an enormously complex and costly financial aid bureaucracy.
I would argue that such a change might attract massive positive media attention, with beneficial reverberations throughout our economically-polarized society. Given the tiny financial cost, there seems no logical reason for those school to resist such an idea.