Nature: many Ph.D.s produced from few universities, ergo elitism and bias

September 26, 2022 • 12:15 pm

A new report in Nature (click on screenshot below), summarizes a longer study in the same journal (original paper here, free pdf of paper here). The point of the summary, and much of the original paper, is that tenure-track faculty in U.S. tend to come from a small number of what the author, Anna Nowogrodzki, calls “elite universities.” (The rest of the report is on recruiting of women into the professoriate, which is increasing, but only as older male faculty retire. I won’t discuss that part.)

The data: 80% of tenure-track faculty in American universities got their Ph.Ds from about 20% of American universities that grant Ph.D.s (There were 368 such institutions in the U.S, so this means that 80% of faculty come from from about 74 universities.) If you dig further, you find that 13.8% of tenure-track faculty got their Ph.D.s at one of these five schools, representing 1.3% of all Ph.D-granting schools.

University of California, Berkeley
Harvard University
University of Michigan, Ann Arbor
Stanford University
University of Wisconsin, Madison

That, in turn, means that having gone to one of those schools appears to increase your chance of getting a tenure-track job by tenfold over the “average” institution.  To Nowogrodzki, and apparently to the authors of the original paper, this degree of enrichment of faculty is an unacceptable form of elitist bias: a form of bigotry against professors who don’t go to the 20% of “elite” colleges (many of them are not really seen as “elite” schools, but as “good” schools).  And the implication, which is not subtle, is that hiring decisions are made largely on the basis of where you got your Ph.D.  I’m sure you can already see the fallacy in that.

Here’s the figure included, with the schools divided into five groups, each containing an equal number of faculty members at Ph.D.-granting schools. The skew is clear:

There are two tweets below, the first reprouduces data from the paper but is a bit misleading on two counts: it’s not 20% of all professors in the U.S., but 20% of professors employed at Ph.D-granting universities. Nevertheless, the assertion that Strohminger makes about the contribution of the top eight schools (not all mentioned in the article) is correct. It’s clear that there is enrichment and inequality of contributions: the figure above is highly skewed to the left, not flat.

What does the enrichment mean? To writer Nowogrodski, it apparently has to be interpreted as producing a kind of unwarranted elitism that gives the graduates of these schools an undeserved advantage over graduates of others. She also implies that this constitutes a form of racial bias. I’ll give some quotes, but you can interpret them for yourself.

Specifically, the study, published in Nature on 21 September, shows that just 20% of PhD-granting institutions in the United States supplied 80% of tenure-track faculty members to institutions across the country between 2011 and 2020 (see ‘Hiring bias’). No historically Black colleges and universities (HBCUs) or Hispanic-serving institutions (HSIs) were among that 20%, says Hunter Wapman, a computer scientist at the University of Colorado Boulder (CU Boulder) and a co-author of the paper.

I’m not surprised at that: these colleges produce very few Ph.D.s compared to some of the colleges above, which are very large and give out tons of Ph.Ds.  No numbers are given about production of Ph.D.s, but good applicants from these historically minority colleges would be snapped up immediately.

Here’s the bit implying that this disproportionality is an unwarranted sign of bias. First, Nowogrodski changes the term “disproportionality” to “elitism,” and then implies that hiring decisions are often made largely on the basis of where you got your Ph.D. That is simply wrong. But let’s get to what the author really wants to say: the “elitist” disproportionality is, in the end, a cause of racial inequity among faculty. And if you’re a Kendi-an, inequity is a prima facie sign of racism. Now the author doesn’t say that, but it’s clear she’s adhering to the Kendi-an ideology (see below) that the unwarranted hiring concentration on where one got their Ph.D. imposes a form of racial bias, conscious or not.

Hiring committees seem to be using prestige as a proxy for excellence on the job, says Kimberly Griffin, dean of the College of Education at the University of Maryland in College Park. But ‘prestige’ does not necessarily indicate ‘better-qualified’, and prestigious graduate programmes often admit students on the basis of standardized test scores, letters of recommendation and the renown of their undergraduate degree. All of these, research shows, can disadvantage students of colour, says Griffin, who is also editor of the Journal of Diversity in Higher Education.

“Accepting that prestige is a good measure of excellence means that we’re not looking into the history of how things became prestigious,” Gonzales says. The founding of elite US universities is “intertwined with exclusion”, she adds. For instance, many institutions have a history of seizing land from Indigenous groups, or originally derived their wealth from or supported their infrastructure with the labour of enslaved Black people.

First, in all my years on the job, I have never seen someone hired over someone else because of where they got their Ph.Ds. That is, I’ve never heard someone say, “Yes, candidate Smith and candidate Jones are roughly equal in their qualifiecation, but candidate Smith went to Harvard!”  Yes, of course the reputation of the institution attracts the best students, and the best students tend to accumulate the best records, and thus get better jobs.  But Nowogrodzki discounts that, for her criteria for the “best” students aren’t the ones that universities use when judging faculty hires:

Hiring committees seem to be using prestige as a proxy for excellence on the job, says Kimberly Griffin, dean of the College of Education at the University of Maryland in College Park. But ‘prestige’ does not necessarily indicate ‘better-qualified’, and prestigious graduate programmes often admit students on the basis of standardized test scores, letters of recommendation and the renown of their undergraduate degree. All of these, research shows, can disadvantage students of colour, says Griffin, who is also editor of the Journal of Diversity in Higher Education.

“Accepting that prestige is a good measure of excellence means that we’re not looking into the history of how things became prestigious,” Gonzales says. The founding of elite US universities is “intertwined with exclusion”, she adds. For instance, many institutions have a history of seizing land from Indigenous groups, or originally derived their wealth from or supported their infrastructure with the labour of enslaved Black people.

No, no, and no. Hiring committees are looking for the best candidates, and the best candidates tend to have gotten their doctorates from the best schools. “Prestige” schools may rarely be a proxy for excellence, but I’ve never seen that happen.  There are many better criteria for excellence that we’ve learned to use. Hiring committees first look at the scientific output of candidates: have they published, and in good places? Is their work good? Do they have a steady record of scientific output? And then they look at the research plans of the candidate: will they be likely to succeed on our faculty? We have the candidates give at least one talk about their work, assessing both the work itself and the candidate’s ability to communicate—giving an idea of how well they might turn out as teachers. There are one-on-one interviews with each faculty member, and we often grill the candidates quite thoroughly. When you have 45 minutes or an hour at most with the candidate, there’s no time to schmooze.

As for letters of recommendation, almost all faculty have learned to discount them, because candidates will ask for letters only from those who are likely to give effusive praise. (I’ve often heard letters of recommendation largely dismissed—unless there’s a bad one, which is so rare that it raises a big red flag and causes more investigation.

All in all, hiring is a rigorous process, and I’ve seen it at both ends, as candidate and faculty hiring agent, many times. To write this off by saying that we’re largely using the “prestige of the school where the candidate got their Ph.D” is deliberately misleading.

As for grad schools themselves admitting students based on standardized test scores, letters of recommendation, and “prestige” of the undergraduate school, that’s virtually irrelevant to the hiring of faculty. Did any universities look at my SATs, GREs, or grades when I applied for jobs? (I got no grades in graduate school.) Nope. They don’t even ask for test scores, and I don’t remember being asked for graduate-school grades.

The part about “many institutions have a history of seizing land from Indigenous groups, or originally derived their wealth from or supported their infrastructure with the labour of enslaved Black people” is also irrelevant, as it’s not just “elite schools” that have done this, and no data are presented showing that the “elite” 79 schools that produce 80% of the faculty are more likely to have seized land from Indigenous people or used the labor of enslaved people than other schools. That accusation is just thrown into the mix to gloss the data with a racial patina—a patina for which there’s no evidence.

In the end, this kind of imbalance is what you expect when schools vary in quality, and students want to go to schools with the best reputation, which, unsurprisingly, give the best opportunities to interact with good faculty and have the resources to do good research. The fact that the existing system does not produce sufficient diversity for people is not the fault of elitism or hiring bias, but because the pipeline itself does not take in enough minority students at the outset. And that’s a residuum of racism in the past. The pipeline problem needs to be fixed, but at the entry end, not at the exist.

Nevertheless, Nowogrodzki’s remedy for the situation is pretty much okay with me:

There are ways in which academia could de-emphasize prestige and reduce inequalities. The first, basic step is questioning prestige and where it comes from, Gonzales says. She advises hiring committees to list all the places they plan to advertise a position, including their personal connections; examine the institutional diversity of the list; and add HBCUs, HSIs and regional institutions if they are not already included.

Not much wrong with that. The problem, though,  is that widely advertising jobs to huge lists of Ph.D.-granting schools is already being done: emails are sent out to schools, jobs are placed online in both big venues (e.g. Science and Nature), and there are group-specific sites where jobs can be seen. It is the fault of the candidate, not of the hiring school, if candidates are not aware of the jobs that are on offer. (By the way, many schools actively discourage their faculty from using personal connections to ask for recommendations of candidates, for that already imposes an unfair bias on a search.)

As for “questioning prestige and where it comes from”, well, we don’t have time for that: we’re too busy advertising widely, deciding what kind of job we have to offer, and which candidates are the most desirable. We’ll leave the “questioning prestige” bit to the sociologists of science.

This disproportionality seems to me a tempest in a teapot: an inevitable result of an inequality in the quality of Ph.D.-granting schools. The remedy, which hiring schools already use, is to put very little weight on where a candidate got their Ph.D. and a lot of weight on the research and knowledge they’ve produced while getting their Ph.D. and doing their postdocs. Advertising is already very wide (what university wouldn’t want as many candidates as possible to know about a job?), and “questioning prestige” is an ineffectual substitute for widening the opening of the pipeline.

45 thoughts on “Nature: many Ph.D.s produced from few universities, ergo elitism and bias

  1. Awww, I wish I could choose one variable and attribute everything to that one variable! That looks like so much fun! You could put in what they have for breakfast everyday, too! I mean, that’s way more entertaining than, oh, I don’t know, academic publication records – or explaining precisely how to know what good research arises from – instrumentation, hospitals, or resources, for example. I mean, that’s too worldly.

    Football team names though – catchy!

  2. So the best students get themselves into the most prestigious universities; and they’re the ones who go on to faculty positions. Isn’t this a sign of things working as they should?

    1. You assume that the best students necessarily go to the elite schools and, hence, are most qualified to be hired as professors. Certainly, some graduates of doctoral programs at fine universities, but not elite ones, are every bit as qualified as at least some of the graduates of elite universities. But, these graduates are competing with one strike against them because they didn’t attend a “name” school. Moreover, in today’s higher education environment, at least in the humanities and perhaps other fields, colleges have greatly increased the hiring of adjunct faculty at effective minimum wage, making it even more difficult for newly minted Ph.D.s to get on a tenure track.

      1. I’m not so much assuming that, I’m saying that if it were true, then you’d expect the pattern in the above paper, therefore these findings are not evidence that something is wrong. And, as Jerry says in the article, once you’ve got to the PhD stage, people judge you on your individual merits and publications, not the school you went to.

      2. Historian, Do you have any evidence for your claim that when two candidates applying for jobs have roughly equal qualifications, there is a much higher chance of the one being hired who came from the better school? Or are you just saying that because you think it’s true? Surely you must have had evidence to say that BEFORE you wrote your comment. But sadly, I see no evidence.

    2. So the best students get themselves into the most prestigious universities; and they’re the ones who go on to faculty positions. Isn’t this a sign of things working as they should?

      Providing that there’s equal opportunity, then yes. If “the best students” are able to buy a better education then probably no.

      PS – there seems to be a delay in my comments appearing on WEIT today. Is it just me? I hope that our host isn’t having to moderate them all because there’s a glitch.

    1. Interesting – it was originally from linguistucs – quoth Wikipedia :

      “the frequency of any word is inversely proportional to its rank in the frequency table. Thus the most frequent word will occur approximately twice as often as the second most frequent word, three times as often as the third most frequent word, etc.”

      It explains that “the”, “of”, and “and” are very frequent words, and also account overall for, say, less than 10% of all words in a certain corpus.

      That makes a lot of sense.

      Great call!… not sure if it explains the observation here, but Zipf’s law is used for all sorts of phenomena.

      Zipf’s … that is tricky to type out… Zipf’s…

      1. This would be a very tricky issue to sort out. Not all PhDs want to go into academia. Would vary with discipline. In my discipline, psychology, many (majority?) of PhDs plan for clinical practice … much more lucrative than academia! Perhaps similar in other disciplines. I also expect that distribution of disciplines and applied vs academic would vary across institutions, perhaps related to prestige or whatever less evaluative term is preferred. Could also be influenced by mobility of grads. Someone mentioned adjuncts … again, some people seek PhDs but aren’t really free to pursue academia which generally requires moving (e.g., spouses). Then there is the issue of postdocs and hiring applicants for academic positions. Certainly too messy to allow facile answers.

  3. Jerry, you *are* misreading the graph, I think. The coloring is done on quintiles; so, 20% of faculty at PhD-granting institutions do indeed come from just 8 schools. 40% come from 21 schools, etc.

    1. The headline and data show that 80% of the faculty at Ph.D-granting schools come from 20% of the sampled schools. Yes, 20% could come from 8 schools, and I’ll fix that, but the claim of the article is that 80% of the faculty come 20% of the schools. 20% of the schools is not eight. The tweet, however, does seem correct.

      1. From my reading of the data shown in the graph, both of these statements are true:

        *) 20% of the faculty come from 8 schools
        *) 80% of the faculty come from 79 schools; out of the total of 387 schools surveyed, that amounts to 20%

        Maybe I’m missing your point…?

  4. It would be interesting to see what the data look like when normalized by the number of PhDs minted annually by each school. Large land-grant universities are over-represented in the top-8 — so, I imagine their success is in part due to their high output.

  5. When I was on faculty hiring committees we looked at publications, grants, ability to give a good research presentation, interviews with all the faculty, an interview with the Dean of the college, conversations with faculty members from other universities who have similar expertise to the candidate, ability to mentor graduate students (usually via a student-oriented get-together or social event), how the candidate might complement other faculty already working in the department, how the candidate’s interests might mesh with those in other departments (as in a paleontologist working with folks in the biology department or with folks in statistics, or elsewhere), and perhaps more.

    It was a comprehensive review and we agonized over the different candidates. There were disagreements and tussles over which candidate was best but, at the end of the day, we usually achieved unanimity over whom to hire. (That wasn’t easy for a faculty of over 20 members, but we wanted the new professor to have full support of the faculty so that he or she could get the best possible start.)

    For junior candidates, we did pay attention to where the person got his or her degree, but mostly because we wanted to be sure that the candidate got the training needed to be a good (at this stage) assistant professor. Most of our successful candidates came from well-known schools. These schools are well known because they tend to be large and they tend to graduate many PhD’s. So it’s hardly surprising that most Ph.D.’s in Ph.D. granting institutions come from well-known, even elite, Ph.D. granting institutions. We definitely wanted folks from a range of graduate schools, not from just a few, and we also seriously considered (and even hired) candidates from other countries. I was in a geology department after all, and the earth includes places other than North America.

    If we were guilty of bias, it was because we favored candidates who demonstrated excellence.

    1. What? You mean faculty hiring committees don’t specifically seek out candidates with degrees from institutions with “a history of seizing land from Indigenous groups”? This could be hard to rank properly. I suppose sub-committees would have to investigate whether Harvard, Stanford, Berkeley, or U. Michigan had seized the most land.

    2. “For junior candidates, we did pay attention to where the person got his or her degree, but mostly because we wanted to be sure that the candidate got the training needed to be a good (at this stage) assistant professor. Most of our successful candidates came from well-known schools.”

      Here you are confirming a point that I made in a comment above. You assumed that a candidate for a junior position was more likely to succeed if she were a graduate from an elite school or at a least well-known one (although I’m not clear what you mean by well-known). This means that a highly qualified and talented candidate from a non-elite school would start out at a disadvantage vis-à-vis the elite school graduate.

      This article from Slate details the problem for aspiring Ph.D.’s from non-elite universities is that they can’t build up a track record of accomplishment in the first place since they are not hired. For those lucky few that are hired, after a few years of accomplishment, only at that point can they compete with elite Ph.D. on a somewhat even basis. The article notes: “For a university, the easiest way to burnish your reputation is to hire graduates from top schools, thereby importing a bit of what made these institutions elite in the first place, while signaling to prospective students and faculty that you attract top talent.” In fields where there are many candidates for a single position (quite common these days and for many years in the past), the poor Ph.D. from a non-elite university could very well end up working as an adjunct, with little hope of ever getting a job with a tenure track, earning starvation wages wondering why she ever wasted her time and earning potential by studying for a dead end Ph.D. Then fed up with academia, she looks for job outside her field that will earn her a living wage. Her ability to contribute to society with applying her knowledge in gaining the Ph.D. goes out the window unless perhaps she can get a job in a related field in the private sector. The only possibility of this is if the Ph.D. is in a STEM field. If it is in the humanities, she can think of Shakespeare while taking orders at the local diner.

      Bottom line: if you’re thinking of entering a Ph.D. program with the goal of a career in academia, you better go to an elite university. Otherwise, you’ll be wasting your time and accruing many thousands of dollars in student debt.

      https://slate.com/human-interest/2015/02/university-hiring-if-you-didn-t-get-your-ph-d-at-an-elite-university-good-luck-finding-an-academic-job.html

      1. Teaching skills were important—as far as they could be evaluated. It was noted if the person served as a teaching fellow or assistant and, if so, what the person’s student ratings were like. Also, the seminar that all job candidates are required to make was for purposes of evaluating presentation skills, which serve as a proxy for teaching skills. There often isn’t a lot of direct information about teaching skills for junior candidates.

  6. I’m not really into the prestigiousness of US universities (ignorant in other words), I know Harvard or Stanford are considered highly prestigious, but I’m surprised there is no Yale, Princeton or MIT in the top mix (nor Chicago, where our host resides), I always thought those were considered well within the top, the elite. I never heard the same thing about, say, U Michigan or U Wisconsin-Madison. I’m not dissing the latter two, just pointing out that I never heard them mentioned as so very highly prestigious.
    Am I missing something? I’m just mystified, can someone enlighten me?

    1. Michigan and Madison (and a few others) are sometimes called “public ivies”. They are very large schools, and produce large numbers of PhD’s, which I suspect contributes to the frequency of their graduates in professorial positions at research universities. Also contributing is the fact that, for the sciences, they are very well funded, and thus they attract the best students who do the best work.

      GCM

    2. As Jerry says, UM and UW are large programs with well known faculty. I got my Ph.D. at the former when Peter and Rosemary Grant, Bill Hamilton, Wes Brown, and other high profile biologists were on the faculty of a single very large, very well funded biology department (that has since split in two) with many grad students. Just on the numbers we are well represented in academia. Also, like Jerry I’ve been on a lot of hiring committees, and no one EVER talks about what university a candidate went to for their degrees–it’s who they worked with and what they got done.

  7. The uniformity fallacy afflicts the Left with virtually perfect, uhhh, uniformity. In the late lamented USSR, attacks on the “elitism” of the old, pre-revolutionary professional class had already taken a lethal form with the 1928 Shakhty trial of engineers, then went on to the celebration of simple peasant agronomists like Lysenko. Similar instincts could be seen in Mao’s cultural revolution of the 1960s. Now we have campaigns against features of “meritocracy” such as tests and advanced courses. And, of course, the imposition of Diversity Statements in academia will help substitute doctrinal uniformity (either real or pretended) for ability.

    A general theory of this phenomenon is obvious: ideological uniformity as a substitute for competence is championed by the incompetent. This might have something to do with the overall productivity of societies which impose this way of doing things.

    1. On the topic of “Diversity Statements”, Tablet posted, three weeks ago, a piece entitled “Higher Ed’s New Woke Loyalty Oaths” written by John Sailer.

  8. I could not disagree more with your assessment. I find it naïve to assert that “the best candidates are chosen”. In fields such as the social sciences where there is no objective “best” and even publications are at the merit of two or three editors who themselves are part of reciprocal networks successful publications are to a large part the reflection of networks.

    I encountered this over and over. The rational of hiring a candidate from a top place is often to get access to favor networks. Heck, I was even explicitly asked during a job interview what network I would be bringing (coming from a university out of this list). Also in econ hardly any candidate has publications…

    A nice example of how naturally people at the top of the econ profession tab into this reciprocal mindset when it comes to hiring “If you do not publish my paper, I will no longer hire your job market candidates”: https://statmodeling.stat.columbia.edu/2022/01/25/how-do-things-work-at-top-econ-journals-exactly-this-is-one-weird-ass-story/

    Also here: https://www.aeaweb.org/articles?id=10.1257/jel.20191574
    ~ 40 percent of the the publication in one of the top econ outlets (QJE run by Harvard) are from Harvard and MIT faculty (well, add to this former PhD students…). So better hire one of the folks from there if you want to ever publish there — preferably an PhD student of one of the editors.

    First question at conferences: “Where did you do your PhD?” immediate disinterest if not from the top (or part of the top) … maybe there are differences across disciplines yet I could go on with many many examples of this reciprocal networks.

    1. First of all, read the posting Roolz so you can be civil. Look at your first two sentences above.

      And perhaps I was speaking only from the vantage of the “harder” sciences, and that things are different in sociology, econ, and so on. I wouldn’t know because I’m familiar only with hiring in the sciences.
      That first question at biology meetings is not the same as at econ meetings. But I’ve canvassed five or six scientists before this post, and vetted it by one before it was posted. So I guess we’re all “naive”.

  9. For me, the strange part is that the study is looking at the professors and where they come from and then saying that it is disproportionate and so going to certain universities increases your chances of being hired, but that seems to be the wrong comparison to make. Surely, if you were interested in bias in hiring, you would need to compare the distribution of hired professors against the distribution of applicants to the position. If 80% of the applicants come from 20% of the universities then there is no bias in hiring, it perfectly represents the applicant pool.

  10. Jerry, would you be up to writing a rebuttal to their paper as a Correspondence? I think Nature does that. Basically a more criticized summary of what you wrote here. The academia must hear your valuable voice!

    1. Yeah, that’s what I was thinking. Yay, they rediscovered the Pareto principle, which applies just about every field of human activity where there are differences in talent and productivity!
      Seriously, I would have been shocked of they hadn’t found something like this.

      1. I read recently that this was from old Italy, someone observed that 80% of the land was owned by 20% of the population….

        [ digs through pile ]

        … oh .. Victor Pareto – turn of 20th c. – economist… held for money too…

        Memory cells!

  11. I always enjoy reading your blog Jerry, but I usually don’t comment. I’ll leave one here though.

    I think the argumentation in the article you summarize is hideous, but at least in my field (formal theoretical high energy physics) there is a steel-manned version of the claim that passes the smell test. Namely, that whether or not one is strongly supported by the luminaries of the field (who are often at the most prestigious institutions) has an O(1) effect on getting high-profile postdocs and getting shortlisted for junior faculty searches. A young researcher who has strong support letters from certain people at Harvard, wrote excellent papers with some of the best people in the field while there, but struggled to distinguish himself afterward generally has a better shot at future success than someone with the same or stronger publication record who is not supported by the gods of the field.

    At least for us this can be made somewhat quantitative, since postdoc hires and most shortlists for junior faculty appear in the public domain, and all one has to do from there is compare publication records. The field is not so big either so it doesn’t take too much work. It’s only somewhat quantitative, since publication record is a coarse measure of productivity and impact, but it ain’t nothing.

  12. I suspect the best way of presenting the data would be plotting a scatter graph of tenured professors from a school, against the number of PhDs produced by that school. (Within a given time).

    1. I want to add … in 2018, 20% of the universities produced 67 % of the PhDs, and 30 % of the universities produced a little over 80 % of the PhDs. So I can’t be that shocked by 80 % of tenured profs being produced by 20 % of the universities.

      If I wanted to be a tenured prof, I would have done my best to get to an “elite” school. As it happens, the postdocs I had an opportunity to share a lab with persuaded me to get out of academia asap.

  13. How does someone get a bad letter of recommendation?

    Shouldn’t they ask the professor if the professor can write a positive letter? And shouldn’t the professor warn them if the letter is not likely to be positive?

    1. Yes, you should ask the professor if he can write a positive letter, but that’s the student’s responsibility. If they just ask “will you write me a letter?” can decide whether or not to say that he cannot write a good one. Fortunately, I haven’t been in that situation, as all my students have been good, but I have SEEN some pretty bad letters of recommendation. And, to be sure, if all letters are good then of what use are they. They really should be OBJECTIVE>

Leave a Reply