Today I want to focus on a publication which came out from the Macdonald-Laurier Institute called “The equity excellence trade-off: A study of EDI in the Canada Research Chairs programs” which appeared on September 1, 2026. I have already commented on the document on social media, calling it “the worst piece of social science I have seen in a long time”. But I thought those of you who don’t follow me on Bluesky might be interested in a critique as well, so here it is.
At a very high level, the document purports to document two things. The first is that “equity hires” for the Canada Research Chairs (CRCs) have reduced the quality of academics being funded, and the second is that this loss of quality has actual financial implications to the tunes of hundreds of millions of dollars of “lost productivity”. Both of these claims are nonsense, but I will focus my remarks on just the first one.
To make claims about scholarly quality – who has it and who doesn’t – the paper relies on something called the “H-index”. An H-index, as you perhaps know, is a single-integer measure of scholarly output: a researcher has an H-index of h if they have published h papers that have each been cited at least h times. Basically, it’s a single integer that captures both some aspects of output (publications) and some aspects of quality (citations). It’s not a useless number by any means (many moons ago, we at HESA Towers used it to make comparisons of research intensity across Canadian institutions across different fields of study), but like all bibliometric measures, it has its limitations and only clowns use it as the sole measure of scholarly quality, whether of an individual or a group of individuals.
You will be shocked, dear reader, to learn that this paper does assume H-Index is everything. In fact, it does its best to make sure that you think that H-Index is everything. In the executive summary, the only measure of quality the authors want you to think about with respect to CRCs is the following:
“The average h-index of publication impact for Black or Indigenous chairs is 18, for female chairs is 28, and for White or Asian males is 38.”
Got that? Big integers that say, in the loudest possible terms, “Men > women” and “Whites/Asians > Black/Indigenous”. Subtle, it is not.
Now, let’s talk a little bit about the biases of the H-Index. As should be blindingly obvious, H-index is inextricably correlated with age. An individual’s H-index ceiling is simply the number of papers published. Older scholars have had more time to accumulate publications than younger ones, therefore their H-index ceiling is higher. Ceteris paribus, given two scholars of identical talent and publication habits, the older one is always going to have a higher H-Index. Comparing two groups of scholars would therefore only be appropriate if i) the two groups were identical in terms of age distribution or ii) age differences are explicitly controlled for.
You will be shocked, dear reader, to learn that this paper does not control for age.
Oh, it acknowledges the age confound. It purports to test for the age confound (figure 11) by including a test for “publishing career length” (although the methodology section is completely silent about how this data term is defined exactly and how the data was collected) and finds via this that even controlling for this, there is a difference between white mean and “equity deserving groups”. It’s actually not all that large a difference – only about 10% or so. Absolutely nothing like the magnitude of the 100% or more difference in raw H-indexes that the authors choose to harp upon in the executive summary, and in their op-eds. But having shown via this test that there is a difference between white males and the rest of the pack, they handwave away all objections to using those raw H-index numbers. They do not report what the differences in H-index average would be with such a control in place. Or, to put it another way: they know the raw H-index numbers are bullshit but shout them to the world as loudly as possible anyway.
(It should not, by the way, be all that surprising that women and visible minority scholars would have lower H-indexes holding age constant. The former are disproportionately impacted by breaks for parental leaves, and the latter tend to finish their PhDs at a later age for a variety of mostly structural reasons. Hence, even though they might be the same age, they might not have had the same amount of time to produce the requisite number of papers. Not that MLI cares about any of that).
There’s another facet of H-indexes which is relevant to thinking about the validity of MLI’s conclusions and it has to do with how H-indexes reflect the different patterns of scholarly publication that exist in different disciplines. In the humanities, for instance, scholars are more likely to communicate via monographs (not captured by standard bibliometrics) than by articles. Scholars in the humanities are also much less likely to cite other scholarly works than the sciences; in history, for example, most citations will be to primary documentary sources than to other articles on a given subject. Something similar is at work in fields like architecture. Scholars in the hard sciences, on the other hand, also tend to publish shorter articles more frequently, and do so in larger teams. These tend to drive up both publications and citations in these fields.
The result of these different scholarly cultures translates directly to H-indices. Back in 2012, we at HESA Towers did an analysis of the H-indices about 60,000 full- and part-time faculty members at Canadian universities which showed that the mean H-index for all scholars in the humanities was 2.3, in social sciences 5.2 and in the sciences 10.6. Using H-indices (or any citation-related measure for that matter) without controlling for field of study simply is not measuring apples-to-apples.
You will be shocked, dear reader, to learn that this paper does not control for field of study.
It does acknowledge that field of study is an issue. And it does test for the field of study confound and it does finds that even controlling for field, there is a difference between white men and equity deserving groups. As with age, it’s actually not all that large a difference – only about 10% or so. They, once again, do not report what the differences in H-index average would be with such a control in place.
To be perfectly clear: the paper could have chosen to report race- and gender-based differences in a way that controlled for differences. The authors clearly know that confounds exist, and they seem to have the statistical chops that would have allowed them to show racial and gender differences in age- and field-adjusted terms. So why didn’t they do it? Simple: it wouldn’t have shown big, cavernous differences between genders and races that the raw numbers they choose to report prominently do. Instead of “White male scholars = 2x Black and Indigenous Scholars! What a huge gap! Why are we sacrificing quality??!?”, it would look something like “adjusting for age and field of study, white males have an H-index which is 1.2x those of non-white scholars”. To which, I suspect, most people would have shrugged and said, yeah, that’s about right, that’s the difference privilege makes.
So: the authors knew that a careful, scholarly presentation of the properly adjusted data would result in headlines which are less sensational than the raw data. They knew this and yet chose to keep repeating the deeply misleading raw data anyways because they – and by extension the MacDonald-Laurier Institute – preferred to generate headlines. Headlines which were ultimately prejudicial to scholars from equity-deserving groups.
Why might they have done this? They might have slightly different motives. I would guess that MacDonald-Laurier has spotted a fundraising opportunity in being the instigator of MAGA-style campus culture wars in Canada. But the study’s lead author, Eric Kaufmann, might have a slightly different set of motives.
The Lemkin Institute for Genocide Prevention puts Kaufman on the advisory board of a magazine called Aporia, which it says publishes race science. Hope not Hate, a British anti-extremist campaign group published an exposé of Aporia’s activities, which places Kaufman at a meeting in central London, sponsored by the magazine, in which he addressed such topics as “the problems with… black culture” and the need to “become more comfortable with a natural level of inequality”. Dude is free to hold such views, I suppose, but I think if one does so, they forfeit the benefit of the doubt with respect to good intentions when publishing something that deliberately exaggerates differences by race and sex.
In any event, it’s utterly reprehensible that MLI would publish this. Treat it – and them – with the contempt this paper deserves.
