Skip to content

Janet Mertz’s response to “The Myth of American Meritocracy”

The following is source material regarding our recent discussion of Jewish admission to Ivy League colleges. I’m posting it for the same reason that I earlier posted a message from Ron Unz, out of a goal to allow the data and arguments to be made as clearly as possible.

Janet Mertz writes:

I became involved in the discussion of Ron Unz’s Meritocracy article because I am a leading expert on the demographics of top-scoring participants in the high school International Mathematical Olympiad (IMO) and the US/Canadian inter-collegiate Putnam Mathematics Competition. I have published three peer-reviewed articles that include data directly related to this topic . . . Had Unz read my 2008 Notices article, he would have known his claim that Jewish achievement in these two competitions had collapsed in the 21st century (which was cited by David Brooks in the New York Times) was simply not true. . . .

The primary questions addressed in this article are the following:
(i) Do the Ivy League colleges use ethnic/racial quotas that discriminate against some ethnic/racial groups (e.g., Asian-Americans; non-Jewish, non-Hispanic whites) in their decisions regarding whom to admit?
(ii) Does the academic performance of very high performing immigrant groups (e.g., Jews) collapse by the time they are 3rd or 4th generation Americans?
I, too, would like to know the answers to these questions. I readily appreciate that Unz has done a huge amount of work compiling and analyzing a large body of relevant data from a variety of sources. Unfortunately, I believe we cannot yet draw definitive conclusions from his study . . . Unz employed a mixture of methodological approaches with different sources of large errors that were additive, including at least one highly subjective one. . . .

Mertz’s full article is a 7-page pdf, “Janet Mertz on Ron Unz’s ‘Meritocracy’ Article.”. She goes into lots of details on the numbers. I won’t repeat the details here because she covers it well enough in her article, also in bits and pieces it was discussed on the blog and in blog comments.

For completeness, I again link to Unz’s original article, Unz’s note to me, and his latest remarks on the subject in which he characterizes this blog post of mine as “a mass of obfuscatory verbiage mostly disputing the accuracy of a couple of my scattered sentences here and there.”

I can’t speak for Mertz, but as for myself let me emphasize once more that I’m just trying to get the numbers right (and also, in some of my bursts of obfuscatory verbiage, trying to understand the process by which we come to our conclusions about the world). Earlier, Unz described Mertz’s reactions as “angry criticism” that “had been floating around the Internet for some time, and had been widely ignored or dismissed.” I think that, for Unz, a good starting point would be to neither ignore nor dismiss constructive criticism coming from well-informed people. Making mistakes is human; what’s important (if the goal is to get closer to the truth) is to recognize those mistakes and move forward.

P.S. Unz responds to Mertz here on his blog. On the plus side, he seems to acknowledge that Mertz knows what she’s talking about on the classification of ethnicities of members of the mathematical olympiad team. He does not dispute her statement that at least 13% of U.S. team members since 2000 are Jewish (that is, have Jewish ancestry, counting fractionally as appropriate) not her best guess that 25-30% of team members in the 1970s were Jewish. Thus, I think Unz is in agreement that his claims of 2.5% and 44% (which made their way into the New York Times as “In the 1970s, for example, 40 percent of top scorers in the Math Olympiad had Jewish names. Now 2.5 percent do”) were in error. He might want to contact David Brooks right away on this. I know that if I had made a mistake that got repeated in the Times, I’d be a bit embarrassed but I’d like to set the record straight.

Given that, in Unz’s words, his estimates were “based on perhaps five minutes of cursory surname analysis,” it should be no surprise that they were wrong. In retrospect, it’s too bad he hadn’t made it clearer that these eye-catching numbers (reporting a 17-fold decline rather than the actual factor of 2 or so) were actually based on nothing more than a cursory 5-minute effort. Then perhaps outsiders such as David Brooks and Steve Hsu wouldn’t have taken Unz’s impressive-looking numbers and graphs as facts.

Given that these high-profile numbers were based on a cursory five-minute effort, it seems more puzzling than ever that Unz had “ignored or dismissed” Mertz’s criticism when it first appeared. Better late than never, though.

The downside of Unz’s response is that he continues to simultaneously defend his name scale-up estimate and the Hillel reports of the percentage of Jews at Harvard, even though those two estimates are off by a factor of 2.5 when applied to the same dataset (the list of Harvard students). He does not actually seem to dispute this factual claim that the two estimates differ by a factor of 2.5 when applied to this dataset; rather, he just ignores it and defends each of the estimates separately. He still has a way to go before he is in full contact with his data. Again, if Unz’s claim is true that the various factual criticisms do not affect his main points, I think he’d be better off accepting them and moving forward.


  1. jrkrideau says:

    “a mass of obfuscatory verbiage mostly disputing the accuracy of a couple of my scattered sentences here and there.”

    High praise indeed! Unz’s article certainly could not be characterized that way–oh no definitely not.

  2. Sebastian says:

    Two observations about Unz bizarre response:
    1. He really does confirm your “political activist” hypothesis, doesn’t he? My favorite part was calling your readers your “supporters”…

    2. I don’t have any problem with his findings if they turn out to be true (and I’m also neither Asian nor Jewish, so not really personally affected) but Unz’s latest really is more than just a little anti-semitic. The amount of time he spends on “ethnocentrism” as an explanation for people criticizing him is disturbing by itself, and then, to perfect the picture-book anti-semitic trope, he even throws in some anonymous Jewish friends. wow.

    • Wonks Anonymous says:

      Unz identified himself as a Jewish Harvard graduate in the original magazine issue, so I think the Jewish friends angle is less of a “I’m not an anti-semite but” angle than a “silent majority agrees with me” one.

      • nb says:

        I very much doubt a silent majority of Jews agrees with Unz. And Jews can certainly be anti-Semites. Bobby Fischer comes to mind…

    • I don’t think it’s necessary or even helpful to characterize Unz’s discursive style in this matter as that of a “political activist”, and especially that doing so delves fairly deeply into presumed motivations on his part, because it’s sufficient to just note that while Gelman and others are mostly in the academic-discourse mode, Unz and a few others are in debate mode. This is particularly familiar as it’s the style that is nearly universal on the internet, but of course it’s ubiquitous elsewhere, including in living rooms and even in many academic contexts.

      It takes a combination of habituation and some discipline to engage in discourse in the academic mode where there’s disagreement. That mode presumes good faith even when there’s reason to suspect that it doesn’t exist; it’s usually maintained by subcultural norms of behavior. Otherwise, when people disagree — and especially when it involves criticism of personal error and responsibility — the natural instinct of most people is to defend and distract and attack. That is, to debate.

      To whatever degree to which Gelman and others are more dispassioned it’s a combined product of the fact that it’s not they who is the primary target of criticism and that they are more usually in an environment that enforces some discipline on rules of engagement on matters of intellectual disagreement. Unz is the one here who’s behaving more typically.

      That doesn’t make him right to do so — Gelman goes to lengths to relativize this behavior by arguing that it’s expected and productive within the context that Unz usually inhabits. I think that’s well-intentioned but arguably a bit patronizing. Rather, I think it’s counter-intuitively more respectful of Unz to not speculate too deeply about his motives and experiences and recognize that he’s reacting the way that most people do when criticized. And to criticize him for doing so, because he’s defending errors that he ought to be simply acknowledging and taking responsibility for.

  3. […] But Prof. Mertz is hardly a narrow ideologue, focused solely on gender issues.  Having successfully demolished Summers’ male-chauvinistic views by demonstrating that 95% or more of all the world’s top math students—both American and foreign—have always been male, she has recently turned her attention to similarly debunking my own claims regarding the recent pattern of ethnic performance in America.  A few days ago, she produced a 3,500 word rebuttal promoted by Prof. Andrew Gelman on his blogsite. […]

    • nb says:

      My comment on TAC is awaiting moderation, so I will repost it here:
      I would like to refute a couple misleading and false statements in Unz’s rebuttal.
      1. Unz’s reported %ages for the racial background of Harvard undergrads from the Harvard NCES IPEDS data DO include full-time Harvard Extension School Students. The full-time undergraduate enrollment at Harvard in Fall 2011 is listed in the IPEDS data as 3,652 males and 3,555 females, summing to 7,207 students. The enrollment at Harvard College in Fall 2011 was 6,657. The IPEDS data cannot be used to draw any conclusions about the demographics of Harvard College.

      2. Prof. Mertz did NOT state that the correct figure should be 7% for the % of Jewish NMS semifinalists. She said that when I added in Unz’s new results for MA, I got 7% *based on Unz’s data for each state.* The whole point of emphasizing that Unz underestimated by a factor of over 5 the % of Jewish IMO team members since 2000 is both that Unz’s ability to recognize Jewish names is poor and that many Jews do not have obviously Jewish names. Hence, we believe that Jews represent more than 7% of NMS semifinalists.

      In an upcoming blog post, I will critique Unz’s NMS data in much greater detail.

  4. Brian says:

    “I’m an academic not a political activist” is how Mertz’s paper begins. Right. There is absolutely nothing resembling political activism in her paper demolishing Larry Summers’ claim that men might be better at math than women.

    I have no idea why you take her seriously. Her criticisms of Unz are trivial. Unz was wrong in small ways, yes. His big point about Jewish overrepresentation still seems reasonable.

    • Andrew says:


      As discussed in the original post, there is no disagreement that the percentage of Jewish students at elite universities is much higher than the percentage of Jews among the college-age population. And if you want to stop right there, that’s fine. The debate comes over Unz’s claim that Jewish students are at Harvard etc. in rates that are much greater than their proportion among very high-achieving high school students. This claim might well be reasonable but it is not so clearly supported by the data. Unz’s statistical argument came from comparing a numerator and denominator that came from different methods which, when applied to a common dataset, differed by a factor of 2.5. Others of his claims were based on his subjective estimates which were off at one point by a factor of 5. I agree with Unz that, if people want to study this, they should be careful about making strong claims based on shaky data.

    • nb says:

      It is NOT trivial that Unz underestimated by a factor of over 5 the % of Jewish IMO team members since 2000 (one of Unz’s most notable data points, as his false claim was cited in The New York Times): Unz’s analysis of Jewish academic achievement is predicated on his ability to identify Jews on the basis of their names, which proved spectacularly wrong in the one data set on which there exists confirmed data about the ethnic background of the students, thus calling into question his entire analysis wrt Jews.

      No one is disputing that the percentage of Jews at Harvard and the Ivies is higher than the percentage of Jews among the college-aged population. We are disputing Unz’s claim that Jews are overrepresented at Harvard in proportion to their representation among high-achieving high school students (in Harvard’s applicant pool). In an upcoming blog post, I will critique Unz’s NMS data in much greater detail, though you can read previous discussions about Unz’s NMS data in Prof. Gelman’s blog post and my comment here.

    • “There is absolutely nothing resembling political activism in her paper demolishing Larry Summers’ claim that men might be better at math than women.”

      That statement says more about you than about her. It’s like someone arguing the same thing about a climate scientist’s paper demolishing a claim that there’s no such thing as global warming — to you it’s necessarily “political activism” because it’s a hotly-contested issue in politics, but it also happens to be an academic question. More to the point, while it’s certainly possible that an academic like Mertz or that hypothetical climate scientist would specifically engage in such an issue because they are motivated by their own political activism, it’s also the case that another academic might do exactly the same work who has no partisan political interest or even an partisan interest that is in opposition to the research’s finding. And the point is … you can’t distinguish between these cases on the basis of the work itself. Your argument that the work itself is self-evidently an act of political activism is a fallacious, self-serving argument. It’s an argument that exemplifies a perspective that sees everything, first and foremost, through a political lens and personal motivations.

      Specifically, like Unz, you are both explicitly and implicitly responding to Mertz’s argument by criticizing her personally, impugning her motives. This is not helpful, it’s diversionary. And I make the same criticism of some of those criticizing Unz — discussing his motivations is diversionary.

      • albatross says:

        This. Unz was extremely careless in his calculations, which led him to make some very surprising claims that may or may not be true, but that he just isn’t in a position to make. No claimed bad motives or intentions on the part of Metz, NB, or anyone else has any relevance to that problem.

        There are some interesting questions raised here, which I very much hope will be pursued. For example, if the kids of European Jews are actually achieving a lot less than they used to, that will matter a lot for America’s future, because they make up a big chunk of our scientists and mathematicians and serious thinkers. A description of how elite colleges’ admissions work out in ethnic terms is worthwhile for a lot of reasons, some spelled out in Unz original article.

        But you can’t answer those questions sensibly without good data. Unz’ methods were flawed enough that some critical numbers in his calculations are known to be off by like a factor of five or something, which makes it hard fo trust any of his other numbers. The solution here isn’t to argue harder, it’s to get better data and show all your work so everyone can see it. But as Keith pointed out above, Unz is in political debate mode, defending his position, rather than acknowledging an error and chasing down a better estimate or answer.

  5. Brian says:

    You might want to compute a 95% confidence interval around that “factor of 5” before making a big deal of it.

    What is the proportion of Jewish students among “very high-achieving high school students”? Here is what Unz says based on a large dataset:

    “In any event, Mertz cites various arguments to suggest that my estimate that Jews constitute about 6% of national NMS semifinalists is too low, and that the correct figure should be 7%. Since I have already stated that I am reasonably comfortable with any figure in the 5.5% to 7.0% range, perhaps our differences are not so enormous in this particular item. But if she accepts that 7% figure, then the over-representation of Jews in elite academic institutions remains just as suspiciously high as I had originally claimed.”

    Are you seriously suggesting that both Unz and Mertz are very wrong here — that the correct percentage might be 30% or more? If Mertz is right, the correct percentage is close to 7%. Which supports Unz’s belief that Jewish students (perhaps a quarter of all Harvard undergrads) are getting many more slots at Harvard than they would be given based on test scores.

    • nb says:

      As I stated above, Unz mis-characterized Prof. Mertz’s discussion of Jewish NMS semifinalists. Prof. Mertz did NOT state that the correct figure should be 7% for the % of Jewish NMS semifinalists, as Unz claimed. She said that when I added in Unz’s new results for MA, I got 7% *based on Unz’s data for each state.* The whole point of emphasizing that Unz underestimated by a factor of over 5 the % of Jewish IMO team members since 2000 is both that Unz’s ability to recognize Jewish names is poor and that many Jews do not have obviously Jewish names. Hence, we believe that Jews represent more than 7% of NMS semifinalists. I will discuss this in greater detail in an upcoming blog post.

    • Andrew says:


      1. I wrote, “his subjective estimates which were off at one point by a factor of 5.” I’m referring to the percentage of Jews among 21st century math olympiad team members. Unz claimed 2.5%. The actual number is at least 13%. 13/2.5 > 5, hence “a factor of 5.” No confidence interval needed. This number may be unimportant, but it made it into the Times. The reason Mertz wrote to me in the first place was that the Times ran this incorrect number and has so far refused to correct it.

      2. When Unz is simultaneously claiming that Jews are 25% of Harvard students and 7% of national merit scholar semifinalists, he’s using two different methods which, when applied to the same dataset, are off by a factor of 2 1/2. If he were to use the same method he uses on the merit semifinalists as on the Harvard students, he’d get 10-11% at Harvard. Also, his 7% is suspect, as other commenters have noted.

  6. Brian says:

    You brought up “a factor of 5” — even though this may “be unimportant” — because the Times made a mistake? I was under the impression that this was a discussion of whether Unz was correct or not.

    If I understand you correctly, the same method that estimates that 6% of NMS semifinalists are Jewish finds that 10-11% of Harvard undergrads are Jewish. This suggests that Jews are overrepresented at Harvard by a factor of roughly 2. That’s a lot, compared to the factors that have gotten people excited in the past. Of course, the 6% is surely too low and the 10-11% is also surely too low. Presumably both should be increased by roughly the same factor. Let’s say both should be increased by 100%. We get 12% and 20-22%. Again, it sounds like a big overrepresentation to me. That’s what Unz concluded.

    • nb says:


      Unz’s factor of 5+ error in estimating the % of recent Jewish IMO participants is important in terms of gauging the accuracy of Unz’s subjective direct inspection method. The fact that Jews represent at least 13% of US IMO team members from 2000-12 may or may not be important.

      I need to clarify an issue with the NMS data. As I stated in a comment on February 13th, Unz’s description of Weyl Analysis was ambiguous since Unz did not clearly describe what he meant by Gold-[]. I interpreted Gold-[] to mean Gold* in my data. I repeatedly sought clarification of this issue both publicly in comments and privately via email, and Unz declined to clarify it. Consequently, I recently searched the 2000 Census to try to determine how Unz performed Weyl analysis, and I discovered that Unz had not meant Gold* but rather {Gold, Golden, Goldberg, Goldman, and Goldstein}. This partially explains why I got higher figures than Unz using Weyl Analysis for the % of Jewish NMS semifinalists for some states, but it also means that Weyl Analysis performed on the Harvard alumni names gives a lower number too. I’m working on finalizing my figures, but right now, my calculations for the % of Jewish NMS semifinalists (*based on Unz’s own data for each state*) and the % of Jewish Harvard College students are much closer (esp when you take into account Unz’s new estimate that 19% of MA NMS semifinalists are Jewish). I’d also like to reiterate my remarks here that the national pool of NMS semifinalists is not a reasonable reflection of the Harvard applicant pool. Once I double-check my data, I will write up all these details in a blog post here.

      • nb says:

        One more thing, I full-counted hyphenated names in both the NMS and Harvard data. I don’t know how Unz treated hyphenated names, so it would be great if he could clarify that issue too.

        • r m adler says:

          nb, I very much appreciate your work on this topic. (along with that of Unz, Gelman, and Mertz) Despite the somewhat adversarial tone, I think that so long as the argument is conducted openly with the evidence presented so that others can verify all claims then all observers are likely to get closer to the truth. This is how science works.

          There is unfortunately a likely outcome which I fear might obscure the truth, namely if the data for either Unz’s conclusions, or your rebuttal of those conclusions, are kept secret. It would thus be very helpful if you could provide references to the where the lists of NMS semifinalists were published for each state. Even better if you could include the lists that you are using in an appendix. Also if you have access to alumni name lists from “elite” universities please publish them or provide references where they can be obtained. If the data are not publicly available then it is impossible for an external observer to verify your claims. Also this makes it relatively easier for others to build on your work, if and when more complete data sets (i.e. from other universities) become available. I request this, not to be a bother, but because it seems from your comments that the conclusions are quite sensitive to the set of names you count as “jewish”, and how you treat hyphenated names.. If the data and counting procedure are publicly available then these name counting procedures can be tested on other data sets where ethnicity is known by other means.

          I fully realize that the alumni list for “elite” universities may be confidential (and for good ethical reasons) – perhaps it would be acceptable for them to be published in a redacted form (i.e. only surnames). If these data are not publicly available then we are stuck, and the truth may be remain hidden. The data unz has presented has its flaws i.e. relying on the hillel stats, but it was the best that was available, and it does give one cause to be suspicious. A true scientist should now seek to find alternative ways for answering the same questions.

          • nb says:

            The lists of NMS semifinalists are in Appendix E of Unz’s original article.

            When I spoke to Unz in person, I asked him to look at the Yale Alumni directory with me, precisely so that he would cease to describe my data as unverifiable (Unz has access to the Harvard alumni directory too). Unz declined my request, but he did say that I should publish my counts for each surname from the alumni directories, so I will do that when I write up my blog post.

          • r m adler says:

            Hi NB,

            Thank you for pointing to the list of MA NMS finalists. I spent a double coffee pause looking over it and out of 411 names, I count 88-89 as “subjectively jewish” due to similarities with people I have known over the years.. this is 21%, similar to what Unz claims to have found. It took me about 15-20 minutes – and during this 20 minutes I kept thinking to myself that this whole process is very doubtful. I would not be surprised if the actual number was as high as 130 (32%). My uncertainty seemed skewed so I also doubt that it is lower than 15%. Okay but this is just my feeling and not really scientific. As mertz pointed out, this should be done in a blinded way. Or actually not at all.

            What is needed is a more rigorous smoothing method where we estimate an ethnic distribution for each surname and then give weight to each ethnic group fractionally. i.e. every miller that occurs we give some fractional counts to “white italian” “white anglo” “white russian”, “white german”, “white jewish” in proportion to their respective fraction of all millers. There is also information from first name / last name pairs which should be used in a controlled way – i.e. eli miller has a higher probability of being jewish than a james miller.

          • nb says:

            r m adler, how did you determine there were 411 names on the MA NMS list? Unz calculated his %ages using a total of 355, the # of MA NMS semifinalists in 2011, as reported by NMSC.

            I spent some time on the MA NMS list too, and you came to the same conclusion that I did: that it is hard to pin down a single number, and I could only list my findings as a range. My lower bound is in fact higher than Unz’s claim that 19% of the MA NMS semifinalists are Jewish. For example, I was able to determine that MA NMS semifinalists with the following surnames are Jewish based on publicly available info and/or the fact that they attended a Jewish school (Maimonides or Gann Academy – I wonder if Unz knows that Gann Academy is a Jewish school): Windmueller, Gossels, Vale, Bunis, Karr, Chiel, Snow, Haber, and Unger. (I’m not listing their first names so that Google searches don’t involve them in this brouhaha.) I wonder how many of those names Unz counted as Jewish. I imagine the average American would not consider most of those to be obviously Jewish names.

            You also raise another issue: how to count possibly Jewish names like Miller. Even if 25% of Millers in the US are Jewish, it’s not necessarily the case that 25% of the Millers who are MA NMS semifinalists are Jewish. For example, I’m from the Boston area and attended Harvard, and most of the Millers I know are Jewish, yet most, if not all, Millers in Amish country are presumably not Jewish. I inferred from Unz’s statement here that he did not count any students with the surname Miller as Jewish, which is one of many Unz’s sources of error.

            btw, Emily Unger (whose permission I have to use her full name) is on both the MA NMS semifinalist list and the Harvard Class of 2013 junior Phi Beta Kappa list. I wonder if Unz counted her as Jewish. I could not have included her on a list of likely Jews had I not googled her name and found that she is active in the Harvard Jewish community (most of Harvard’s Jewish students do not have leadership positions in Jewish student groups, so such Google searches are unlikely to capture most Jewish students at Harvard). The point is that many Jews do not have obviously Jewish names, as one can also see from Prof. Mertz’s partial list of Jewish or part-Jewish US IMO team members.

          • r m adler says:

            nb, sorry to break the nesting convention here:

            The list that I used was obtained at the following URL: presumably 2008

            Here is a link to my classification for what it’s worth (and I now feel that it is not worth much!) –
   (this link should not be indexed by search engines, to at least make a gesture toward protecting the privacy of the names on the list) I am really putting my prejudices out in public, which I suppose is where they belong.

            I indeed missed some of the names that you have better information on. still I think at least for this list an upper bound of 30% 130/411 names is reasonable and something around 20% is a good point estimate for the jewish fraction.

            I look forward to seeing your results for the university directories.

          • nb says:

            r m adler, I do not feel comfortable publicly speculating on whether or not an individual is Jewish – as I am concerned this could cause discomfort to the individuals involved. I listed the surnames above as examples of Jews with non-obviously Jewish names because they attended Jewish schools or because they or members of their family have publicly stated that they are Jewish; i.e. it’s already public info that these individuals are Jewish. I think it’s important to illustrate that many Jews do not have obviously Jewish names (whereas it’s much less likely for non-Jews to have distinctively Jewish surnames), as it indicates that Unz’s estimates of Jewish representation among high academic achievers are likely underestimates (though curiously he substantially overestimated the % of Jews among 70s IMO participants); hence, comparing them to Harvard Hillel’s Jewish enrollment figures (which I believe are overestimates), as Unz has done, yields erroneous conclusions.

            Regarding your list, you did omit some of the Jewish students I noted above, though you also included some students I ruled out as Jewish. I’m curious as to why you listed one of the NMS semifinalists as 0.5. I had him in my J? category – as in I would not include him as part of my lower bound, but I would include him as part of my upper bound, as that name is ambiguous.

    • Andrew says:


      Unz’s incorrect numbers appeared in the Times, which gave them greater credibility and a much wider circulation than they would’ve received had they only appeared in Unz’s self-published magazines and in various blogs. The mistake the Times made was (1) to take Unz’s numbers at face value, and (2) to not run a correction once the mistakes were revealed. Although, to be fair to the Times on item 1, Unz’s article had good graphics, it appeared to be thoroughly documented, and it did not mention that some of the numbers were constructed based on a casual five-minute effort. But, really, that’s all understandable. Mistakes happen. As I noted at the end of the above post, the point is to acknowledge one’s mistakes. Unz has had difficulty doing this so far but there’s always hope for the future.

      • Brian says:

        What about my more important comment (#2)?

        • Andrew says:


          See Mertz’s report and various comments including NB’s above. That 6% and 7% are Unz’s numbers based on his counts which are not so trustworthy. Nor do I have huge confidence in the particular scale-up method being used. It may very well be that Jews are overrepresented or underrepresented compared to the pool of high-achieving high school students who are interested in going to Harvard. I suspect much depends on how these pools are defined. As I’ve written many times, I’m not trying to confirm or refute Unz’s larger claims. What happened was that he originally presented what looked like very convincing data (as so it appeared to me, David Brooks, Steve Hsu, and others) but many of the numbers were way off. Given that Unz’s claims had appeared in the Times, and I’d presented them uncritically myself, I thought it was appropriate for me to present the criticisms.

          • Mickey Colman says:

            Unz says you have been invited to define your own framework so that you and he can work out areas of agreement and disagreement in a scholarly way that would make it hard for any side to use underhanded tactics for unstated motivations. Why don’t you do this? You shouldn’t say that you don’t wish to challenge Unz’s wider claims. In effect you are challenging those wider claims, but currently the way you are going about it is not made explicit by you and so is at least theoretically open to misuse by you and others.
            This is not a good situation. You’ve become involved in a question that will be important to large sections of the American population. Fairness in ivy league admissions has consequences, not just for future academic performance of the nation, but also the political institutions and elites that govern it, both which draw heavily on ivy league graduates.
            It’s time for you to put up or shut up. Define your framework and begin engaging Unz in a profession, above board, explicit manner.

          • Andrew says:


            We can each contribute to this discussion in our own way. Doing the original research is hard. Mertz has done some, and it’s taken a lot of work. Another way to contribute is to raise issues more speculatively. Unz did that. Some of his comparisons were way off, but, as he said himself, he obtained some of his estimates based on a few minutes of casual effort, so that’s OK too. My own contribution has been to post and comment on some of the numbers provided by Unz and others. If you or anyone else would like to do more research on the topic, that’s fine. No need to wait for me to do it. In the meantime, I believe my engagement with Unz, Mertz, and others (including you!) have been completely professional, etc.

          • Mickey Colman says:

            “We can each contribute to this discussion in our own way”
            You are a statistician. Yet at no point have you made a statement as to what size the errors are compared to the work overall. Are they of sufficient size to actually draw into doubt whether there is any over-representation of Jews at all allowing for grades?
            Would it be possible for you to provide clarity on this question? The reason I ask is because the most common way to attack the findings of another person in an underhand way is to make a lot of noise about small errors as a way to discredit the author and work.
            In no way am I saying you are doing this, but it happens a lot. So for example, can you help by explicitly stating whether in your view the errors are large or small relative to the overall numbers.
            Put differently, Are his conclusions basically safe, the only question being how much Jewish admissions are overrepresented for actual grades? Or are the errors large enough relative to the whole work that the conclusions themselves are not safe?

          • Andrew says:


            This is what I wrote in my earlier post:

            My point here is not to slam or “fisk” Unz. Rather, I’m giving all these details because, as a statistician, I think details are important. One reason I suspect that Unz’s article was originally received so uncritically by many bloggers and journalists is that Unz presented lots of numbers and described where he got them. There were no other easily accessible numbers on the topic, so we were inclined to go with what Unz had. So I think it’s important to pick on the details here, so that it’s clear that these are not minor technical criticisms but rather get to the center of Unz’s argument. I read his article as claiming that Jewish students are overrepresented in Harvard’s admissions by roughly a factor of 4 (compared to other whites) compared to their academic achievements and abilities. But when you make the comparisons carefully, this disparity goes away. There still is arguably an underrepresentation of high-achieving Asian students, but some of Unz’s comparisons there are off too, in that he at times is lumping U.S. and foreign Asians into a single category, whereas it would seem more appropriate to focus on Asian-Americans when comparing to other college applicants.

            In his article, Unz claims to have found that elite college admissions underrepresent Asian-Americans (in comparison to their academic talent achievements) and overrepresent Jews, leaving non-Jewish whites squeezed out. Looking at the statistics more carefully, we see no evidence that Jews are admitted preferentially compared to other whites. Unz’s error arose because he used different sorts of information with different biases that did not cancel out but actually reinforced each other, underestimating the proportion of high-achieving Jews and overestimating the Jewish presence among Ivy League students.

            I have long argued that meritocracy can’t work (for more recent discussion, see here and here) and so I’m sympathetic to Unz’s general concerns. But it looks like he garbled the analysis for one of his main points.

            So, to answer your question, I think Unz’s errors are big, not small. A lot of uncertainty remains.

  7. Leah says:

    Sorry Andrew, but Unz has nailed it. Even if you take a more lenient approach you still get over a 100% ethnic bias in favor of Jews.

    I think it’s slightly amusing that you’re essentially taking the same defensive position as gilded WASPs did in the 1950s when the roles were reversed.

    I think affirmative action can be defended on socio-economic grounds for blacks/hispanics. Do you seriously think Jews need affirmative action? Considering the very high median income, college rates etc?

    • Leah:

      I’m not taking any defensive position at all, I’m just reporting the numbers. I have no idea what is meant by “100% ethnic bias” but I distrust quantitative claims that are based on shaky statistics. I have no desire to “take a lenient approach,” I just want to get the numbers right and to get a sense of uncertainty.

    • nb says:

      In order for me to reproduce Unz’s result that Jews represent only 6-7% of NMS semifinalists in the 25 state aggregate according to Weyl Analysis, I had to interpret {Gold-} as {Gold, Goldberg, Golden, Goldman, Goldstein} and either half-count or zero-count hyphenated names (where half of the hyphenated surname was one of the Weyl distinctive Jewish surnames). Using this same methodology on the Harvard College alumni directory, I obtained the result that Harvard College was 7-9% Jewish in Fall 2008. Where is this 100% ethnic bias in favor of Jews?

      Also, performing Weyl Analysis on the publicly available Stanford directory (as stated above) yields the result that Jews represent 3-5% of Stanford undergrads. Does this mean Stanford is discriminating against Jews? (Hint: no.)

  8. […] *Let me just be clear about my opinion of Gelman’s contribution. He’s been clear before that 1) he recognises the tricky moral  terrain of apology, and 2) that ideally the approach should look a little like this: […]

    • Andrew says:

      I followed the links, where Johann Koehler writes of me, “sometimes, as in the PS comments in posts such as this one, I’m not sure how receptive he really is to apology.”

      I followed the link, which refers to a case from the Retraction Watch blog, describing “Adam Savine, a former graduate student at Washington University in St. Louis who was found by the Office of Research Integrity (ORI) to have committed misconduct.” I did not follow all the links in this story but I hadn’t seen anything where Savine had apologized. All I saw was that he’d been caught and had admitted wrongdoing. If Savine has indeed apologized, I agree that’s a good thing to do. Not that it would mean that I’d trust the guy right away—apologizing after getting caught seems to me like a first step but only minimal—but, sure, you gotta start somewhere.

  9. Westcoast says:

    So Unz is off by a little bit. Jews are overrepresented by 150% instead of 300%. Still the bottom line is, American meritocracy has died, but its death was less bloody than initially thought.

    • nb says:

      No, Weyl analysis on the current Harvard directory yields the result that Harvard College is 6-7% Jewish, which coincides with the results from Weyl analysis on the list of NMS semifinalists.

  10. szopen says:

    Does Janet Mertz have her own blog? “Janet Mertz blog” in google does not give any directions. I’ve just read her paper “debunking myths on gender performance” and I have plethora of questions; I’d prefer talking in blog instead of sending an email.

    e.g. given that high math results correlate highly with high income, and that technological positions require high math, how can she take an index in which some of components are high income and more technological position for women and be surprised that there is a correlation between this index and math abilities?