Noah Carl Controversy: FAQ (rebuttal)

The colorful pseudoscience Racialism

Hating thy neighbour

Racism Racial pride Nationalism

Divide and conquer

League of the South Fellowship of God's Covenant People Venetic theory Ghetto Men Going Their Own Way

Dog-whistlers

Rush Limbaugh Clyde Winters Bryan Fischer Edward M. Miller Mitch McConnell v - t - e

Noah Carl is a disgraced academic who was sacked from St Edmund’s College, Cambridge University.

On 7 May 2019 he published a blog post on Medium, "Noah Carl Controversy: FAQ" responding to the termination of his research fellowship at Cambridge.

The following rebuts his responses.

Noah Carl

Rebuttal

What is your response to the College’s statement explaining their decision to fire you?

I will be responding to this statement in due course. But for the moment, I am still receiving legal advice. Needless to say, my interpretation of events differs from that of the College.

The College's statement can be read here. Carl hasn't yet responded to this statement.

Last December, 586 academics signed an open letter accusing you of “racist pseudoscience”. That many academics can’t all be wrong, can they?

Given that the open letter demonstrated a basic lack of understanding of the relevant science, it would seem that 586 academics can indeed all be wrong. For example, as Jeff McMahan pointed out in his comments for the first Quillette Editorial: “”One passage in the open letter demands that the various institutions cited “issue a public statement dissociating themselves from research that seeks to establish correlations between race, genes, intelligence and criminality in order to explain one by the other.” This seems to imply that it is illegitimate to seek to explain any one of the four characteristics by reference to any one of the others, and thus that no aspect of intelligence can be explained by an individual’s genes. I would not trust the competence of anyone who endorses a claim that has that implication to judge the work of a candidate for a research fellowship. And Professor McMahan is absolutely correct: the signatories of the open letter were calling for St Edmund’s College to “issue a public statement dissociating themselves” from research backed by overwhelming scientific evidence. In fact, the contribution of genes to variation in human intelligence has been widely accepted by psychologists since at least 1996, when the report ‘Intelligence: Knowns and Unknowns’ was published by the American Psychological Association (APA). This report, co-authored by Professor Ulric Neisser and his colleagues in the aftermath of The Bell Curve debate, concluded that “a sizable part of the variation in intelligence test scores is associated with genetic differences among individuals”. Evidence for a genetic contribution to variation in human intelligence has only strengthened since the publication of the APA report.

This is an obvious misreading of the open letter that denies the validity of linking genes, intelligence and criminality to race, not individuals. At the beginning of the open letter this is clearly clarified: “”A careful consideration of Carl’s published work and public stance on various issues, particularly on the claimed relationship between ‘race’, ‘criminality’ and ‘genetic intelligence’, leads us to conclude that his work is ethically suspect and methodologically flawed. No mention whatsoever of individuals. Later in the letter they write: “”We call on St Edmund’s College, the University of Cambridge, and the Newton Trust to issue a public statement dissociating themselves from research that seeks to establish correlations between race, genes, intelligence and criminality in order to explain one by the other. So when they are talking there about correlations between genes and intelligence, they don't mean individual's genes. The only quibble with the open letter is that "genetic intelligence" should separately read "genetic(s)" and "intelligence". Numerous PhD geneticists and psychologists signed the open letter and it would be foolish to argue they deny the evidence individual differences in human intelligence are partly due to genetic variation, more specifically there has long been a consensus genetic factors underlie somewhere between 30-80% (probably 50%) of the difference in intelligence between individuals. As noted by Genetic Literacy Project: "Typically around 50 per cent of variations in intelligence between individuals can be ascribed to genes although these gene-induced differences become markedly more apparent as we age. As [psychologist Stuart] Ritchie says: like it or not, the debate about whether genes affect intelligence is over."

Okay, so individual differences in intelligence might have a genetic component. But what about differences between groups — they couldn’t possibly have a genetic component, could they?

Contrary to the implications of the open letter, I have never actually done any original research on racial or population differences in intelligence. The only contribution I have made to this area of study is a research ethics paper arguing that “it cannot simply be taken for granted that, when in doubt, stifling debate around taboo topics is the ethical thing to do”. While this paper does not claim that genes do contribute to group differences in intelligence, it does entertain the possibility that they could contribute to such differences. I consider this to be a perfectly defensible scientific position. We know that there are group differences in intelligence, both across countries, and between groups within a country. The question is why. And there is no good reason to rule out the possibility that genes do make some contribution to these differences. It may turn out that genes make zero contribution, or it may turn out that they make a contribution greater than zero. Deciding in advance that they make zero contribution is not science. It is proof by assertion. As James Flynn has noted, the hypothesis that genes contribute to group differences “is intelligible and subject to scientific investigation”. I trust 1 James Flynn a lot more than 586 petitioners.

While Carl may claim to sit on the fence regarding the cause of (mean) group differences in intelligence, his paper is clearly committed to defending racialism i.e. the belief human races are valid biological categories. Carl for example writes: "To deny the existence of ‘race’, or to insist that it is a wholly ‘social construct’, as many critics of the ‘hereditarian’ view of human nature do (see Wade, 2014,Ch.5; Winegard et al., 2017; Sesardić, 2005, Ch. 4), is to commit oneself to an erroneous view of modern medicine, one which could conceivably come at the cost of people’s lives." Carl is wrong about race being useful to biomedical research, for example Tishkoff & Kidd (2004) warn: "Although information about ethnicity can be informative for biomedical research, it is imperative to move away from describing populations according to racial classifications such as 'black', 'white' or 'Asian'" because "there can be considerable genetic heterogeneity within a region, it is most useful to be specific as possible about geographic origins, ethnicity, or tribal affiliation." In other words, while some local populations may be of use to biomedical research such as the Amish (Study of Major Affective Disorders), races in the sense of continental populations certainly aren't. Despite not openly supporting hereditarianism i.e. group-differences in intelligence are substantially genetic in origin, Carl has made some posts on The Unz Review and OpenPsych forum that arguably show he is at least sympathetic to this viewpoint.

But if genes contribute to group differences in intelligence, wouldn’t that mean the Nazis were right… or something?

No, it absolutely would not. First and most importantly, “political equality is a moral stance, not an empirical hypothesis”, as Steven Pinker has noted. Please read my research ethics paper for a longer explanation (I really would recommend it). Second, from what we can tell, the Nazis actually opposed intelligence research. To quote from Heiner Rindermann’s recent book "Contradicting common beliefs, National Socialists were opposed to intelligence research (Becker, 1938; Jaensch, 1938): in their view, intelligence research would represent a ‘supremacy of Bourgeoisie spirit’ (Jaensch, 1938, p. 2); intelligence measurement would be an instrument ‘of Jewry’ to ‘fortify its hegemony’ (p. 3); selection in schools according to intelligence would stand for a ‘system of examination of Jewish origin’ (p. 4), especially the concept of intelligence as a ‘one-dimensional dimension’ (p. 3) and ‘one common central factor’ (Becker, 1938, p. 24). Because people differ and therefore intelligence differs (p. 4) they called for an ‘intelligence measurement according to a national and typological point of view’ (p. 15); for Germans they asked for a measurement of ‘realism’, ‘conscientiousness’ and ‘actually of the character value of intelligence’. They were opposed to a measurement solely of ‘theoretical intelligence’, of ‘intellectualism’ (Becker, 1938, p. 22); instead they favoured ‘practical intelligence’ (p. 18)" So it seems that the Nazis’ opposition to intelligence research stemmed in part from their anti-Semitism. By the logic of my critics, this would imply that opposing intelligence research is racist…

The Nazis may have had a different concept of intelligence, but they still used IQ tests and didn't oppose intelligence research or testing e.g. the Nazi Law for the Prevention of Progeny with Hereditary Diseases used IQ testing to determine whether an individual had "feeble-mindedness" that resulted in selecting candidates for sterilisation.

Didn’t you go to a secret “eugenics conference” with “Neo-Nazi links”?

No. But I did attend and speak at a meeting of individual differences researchers called the London Conference on Intelligence (LCI). This conference was widely mischaracterised in the media, and some of us who attended responded via a peer-reviewed correspondence in the journal Intelligence. To quote from our correspondence: Contrary to allegations, the annual LCI conference was not secret but invitation only (like many small conferences). The attendees had a range of theoretical orientations and research interests, and their attendance does not imply agreement with the views of all of the other attendees, be they political, moral or scientific. The conference program covered many topics related to the fields of intelligence and personality research and there was no exclusive focus on ‘eugenics’ or IQ differences among populations (although both issues were discussed). Furthermore: The overwhelming preponderance of talks dealt exclusively with data or substantive theory. Moreover 48% of talks were associated with (either based on or in most cases yielding) ‘mainstream’ publications over four years. Thus, LCI’s productivity is comparable to that of conferences in biomedical science — a field in which, according to one meta-analysis, 44.5% of conference presentations yield publications (Scherer, Langenberg, & von Elm, 2008). Finally, the speakers originated from 13 different countries in total, including Japan, China, Brazil and Slovakia, thus the conference can reasonably be described as cosmopolitan as opposed to “white supremacist” in character. So the LCI’s productivity, measured by the percentage of presentations associated with publications in ‘mainstream’ journals, was about average for conferences in the field of biomedical science: The ‘mainstream’ journals in which articles have appeared include (in no particular order) Intelligence, Personality and Individual Differences, Learning and Individual Differences, Frontiers in Psychology, Frontiers in Human Neuroscience, Journal of Experimental Psychology: General, Evolutionary Psychological Science, Twins Research and Human Genetics, Cortex, and Evolutionary Behavioral Sciences. Academic monographs that either formed the basis of presentations or incorporated results presented at LCI have been published with Cambridge University Press, Palgrave Macmillan, and as part of the Journal of Social, Political, and Economic Studies occasional monograph series. Hence if the LCI constitutes “racist pseudoscience”, then all the above-mentioned journals and publishers would presumably fall into the same category. Either our critics are mistaken or there’s an awful lot of “racist pseudoscience” out there… I am not aware of any attendee at the conference who holds “Neo-Nazi” views.

The LCI was clearly secretive, i.e the university it took place was not informed in advance about the speakers and content of the conference(s); this is highly unusual. Had the university been shown in advance a list of talk abstracts and speaker-list, they almost certainly wouldn't have given permission. So the organisers of the LCI were very deceptive and breached the universities room bookings process. As for LCI being attended by neo-Nazis this was never claimed, rather the London Student said individuals had Nazi links. This is sourced in the article i.e. 82% of LCI speakers (including Carl himself) have either published in the Mankind Quarterly or sit on its editorial board. The racist pseudojournal MQ is bankrolled by the far-right Pioneer Fund, a Southern Poverty Law Center-listed hate group founded by Nazi sympathisers. While Carl is correct a very low % of talks at the LCI were about eugenics, virtually everyone who attended (even Toby Young who wasn't a speaker) are known proponents of eugenics; at least all individuals who attended the LCI based on the 2015 and 2016 talk abstracts (including its organiser) are pro-eugenics. In other words, there was little to no media misrepresentation: the media knew two dozen or more eugenicists were meeting in a room and so some newspapers decided to describe the LCI gatherings as a "eugenics conference."

But don’t some of the researchers who attended the LCI hold “far-right” views?

Some of them may indeed hold such views. However, as a rational adult in control of his faculties, I am capable of interacting with people who hold different views from me. (I realise this may come as a shock to some readers.) Nobody at the conference tried to coax me into adopting any kind of “far-right” agenda.

It is a lot more than "some". Rather the vast majority of speakers who have attended LCI are far-right or white nationalists e.g. Richard Lynn, Helmuth Nyborg, Edward Dutton, Adam Perkins, James Thompson and Emil Kirkegaard; Aurelio J. Figueredo another speaker, has been funded by the far-right Pioneer Fund. Additionally co-authors of paper abstracts who didn't attend, include Donald Templer who was a regular speaker at white nationalist American Renaissance conferences and John Fuerst who calls himself an "ethnonationalist" and is an anti-Semitic conspiracy theorist. At least one speaker, Dalibor Jurášek is a self-described Holocaust revisionist and the LCI was also attended by many others who hold racist and sexist views, for example Dimitri van der Linden and Paul Irwing who argue women score substantially lower than men in IQ tests when their studies haven't been replicated and have been widely criticised as sexist pseudoscience. There is also the fact Thompson advertised the LCI conferences on the alt-right website The Unz Review and VDARE supported the conferences.

Aren’t you a member of the “alt-right”?

No, I am definitely not a member of the “alt-right”. My political views vary from one issue to another — on some issues I take more left-wing stances and on some issues more libertarian stances — but overall they reflect those of a moderate conservative.

Carl is part of the "human-biodiversity" online network and blogsphere, that is strongly linked to the alt-right and hence most the people he associates himself with are white nationalists. Of course, most individuals who espouse HBD views online deny they are alt-right or far-right, which is expected and nothing new. Carl is little different to his colleagues Nathan Cofnas and Kirkegaard who also pretend to be political moderates. As noted in the NewStatesman, "like many in the academic far-right, Noah Carl lives a double-life."

Haven’t you published several papers in a non-peer reviewed “pseudojournal”?

I have published several papers in the OpenPsych journals, which use a form of open peer review. This review system is clearly laid out on the journals’ website. Hence it is false to claim that the journals are not peer-reviewed. Nor can it be claimed that any attempt has been made to conceal the journals’ review system: upon reaching the homepage, you only have to click ‘About’ to find out how it works. In addition, the names of the reviewers and a link to the review thread are provided next to every published paper. Open peer review is one of the key principles of the ‘open science’ movement; the others being ‘open methodology’, ‘open source’, ‘open data’, ‘open access’, and ‘open education’. The essence of this system is that reviewers’ names are disclosed to authors. Open peer review systems vary in several different ways, such as whether peer review happens before or after publication; whether reviewers are blind to authors’ names; whether editors have discretion to reject papers; and whether reviewers’ comments are published alongside the article itself. Perhaps the most radical form is ‘post-publication peer review’, where open peer review takes place following instant publication. This is the system used by platforms such as F1000Research and The Winnower. Of course, there are advantages and disadvantages to every system of peer review. For example, the main advantage of double-blind peer review is that, assuming anonymity is preserved, reviewers cannot be influenced by irrelevant characteristics of the authors themselves (e.g., race, gender, personal connections). However, an obvious disadvantage of double-blind peer review is that authors cannot hold reviewers to account for biased or incompetent reviews. The problem of unaccountability is particularly serious in ‘controversial’ areas of research because — due to the massive left-wing skew of the social sciences — papers often get rejected for ideological reasons. In addition, as the use of pre-print archives (e.g., OSF, arXiv, SSRN etc.) becomes more and more common, author anonymity will be increasingly difficult to preserve. Since starting my DPhil in 2010, I have submitted manuscripts to more than 20 different journals, and served as a reviewer for about the same number. Based on this experience, I would say that OpenPsych ranks about average in terms of rigour. I have certainly dealt with journals where the review system was more rigorous than at OpenPsych, but I have also dealt with journals where it was less rigorous. For example, one of my papers in a ‘mainstream’ journal was accepted for publication by a single reviewer following a single round of review. Hence even if you remain unimpressed by the quality of the reviews at OpenPsych, dismissing them as “pseudojournals” would imply doing the same to a substantial number of ‘mainstream’ journals as well. Finally, we wrote an Editorial responding to many of the specific criticisms of OpenPsych, but almost nobody has attempted to engage with it.

OpenPsych journals aren't peer-reviewed. Carl and Kirkegaard can keep repeating the lie they are, but that doesn't change the facts. Meng Hu a former referee for an OpenPsych journal, even quit after raising concerns about the lack of peer-review. His main criticism is referees aren't impartial: "One problem with OP is that most (if not all) reviewers have an 'hereditarian side'. I would like to see reviewers hostile to the genetic theory. Otherwise, people may think OP looks like a Mankind Quarterly bis." Another issue with OpenPsych is most referees don't hold any relevant qualifications or credentials to competently review submissions; for example Kenya Kura holds a PhD in economics but bizarrely reviews papers submitted to Open Behavioral Genetics. Emil Kirkegaard, the co-founder of OpenPsych only has a BA in linguistics, while Gerhard Meisenberg is a biochemist, but reviews submissions to Open Differential Psychology. It is true OpenPsych published an editorial with Carl as a co-author, responding to various criticisms, however it never responded to the main criticism about lack of impartiality and all referees being hereditarians or sympathetic to hereditarianism. It's hard to take Carl's claim serious that "OpenPsych ranks about average in terms of rigour". For example Carl's controversial paper "An analysis of Islamist terrorism across Western countries" was reviewed by Kirkegaard and John Fuerst. Emil Kirkegaard has claimed he wants to ban Muslim immigration to the West because it is "self-destruction", while also describing Muslims immigrants as "terrible to get". Fuerst has similarly posted a bunch of rants against Muslim immigrants on his Facebook. These are hardly the sort of people who can impartially review a submission on the subject of Islamist terrorism; Carl also reviewed Kirkegaard's papers on immigrants.

But aren’t all the reviewers at OpenPsych “hereditarians”?

The term ‘hereditarian’ presumably refers to someone who believes that genes make a non-zero contribution to group differences in intelligence. This does not seem to me to be an unreasonable scientific position. The alternative is to believe that genes do not make any contribution to group differences in intelligence. Note that ‘hereditarian’ does not mean someone who believes that genes explain 100% of all group differences (a highly untenable position). Neither Charles Murray and Richard Herrnstein, nor Philippe Rushton and Arthur Jensen, believed that. I am not aware of how many of the reviewers at OpenPsych are “hereditarians”, but even if all of them were, I would not consider this to be a particularly egregious failing on the part of the journals. It would simply mean that every reviewer believes that genes explain some proportion of group differences in intelligence that is greater than zero. And it would not prevent there from being considerable diversity of opinion as to the extent of that genetic contribution. For example, some reviewers might believe it was closer to 10%, while others might believe it was closer to 50%. Expert surveys reveal that a substantial percentage of individual differences researchers are “hereditarians” in the sense given above (i.e., believing that genes explain >0% of group differences in intelligence). Hence it would not be particularly remarkable if most of the reviewers at OpenPsych were in fact “hereditarians”.

Carl uses a straw man argument often used by proponents of hereditarianism that redefines the term to mean "genes explain >0% of group differences in intelligence". This was never how the hereditarianism hypothesis was defined by Arthur Jensen, who instead defined hereditarianism as the view genes explain >50-80% of group differences in intelligence, although elsewhere he defined it more vaguely as the view "genetic factors are strongly implicated in the average Negro-white intelligence difference" or "a substantial part of group differences in human behavioural traits is genetic". While this leaves the question of what cut-off percentage of genes "substantial" or "strongly implicated" means, perhaps as low as 25% or even 10%, someone who argues genes explain a mere 0.1% of group differences in intelligence can hardly be described as a hereditarian since 0.1% is not at all substantial, although Carl would like to redefine hereditarianism to include psychologists who favour the view genes are negligibly involved in explaining the black-white IQ gap (0.1% genes, 99.9% environment). The problem with this redefinition is proponents of the latter view do not consider themselves hereditarians and hereditarianism since Jensen (1969) has always meant the view genes are substantially, not negligibly involved in explaining group differences in intelligence. Therefore when Carl says hereditarianism is not an "unreasonable scientific position" he needs to be more specific with what percentage of genes he has in mind. As for referees for OpenPsych being hereditarians, as already mentioned, Meng Hu who quit an OP journal and is familiar with the views of his former colleagues noted "most (if not all) reviewers have an 'hereditarian side', furthermore all of them (including Carl) are racialists. Therefore OpenPsych is extremely biased towards one viewpoint and lacks impartial referees. Contrary to Carl's claims none of OpenPsych's other referees are known to hold "diversity of opinion as to the extent of genetic contribution" in regards to group differences in intelligence, rather they all defend Jensen's and Richard Lynn's hypothesis that genes explain at least 50% of the black-white IQ gap.

But aren’t all the reviewers at OpenPsych “far-right”?

I cannot speak for every reviewer at OpenPsych, but I know that my own political views are not “far-right”. I also happen to know that the views of several other reviewers are not “far-right” either.

Almost no one associated with the "human-biodiversity" online network and blogsphere admits to being far-right or alt-right (despite they all blatantly are), it's a highly deceptive crypto-politics movement. So no one is expecting any honesty from Carl. And it is easy to prove almost everyone who is associated with Kirkegaard's OpenPsych is far-right.

But don’t the OpenPsych journals have an incredibly controversial editor named Emil Kirkegaard?

I will let Emil Kirkegaard speak for himself. But suffice it to say that he has demonstrated competence in the area of individual differences research by publishing in a variety of ‘mainstream’ journals, including Intelligence, Journal of Individual Differences, Journal of Intelligence, Psych and Evolutionary Behavioural Sciences. I would also reiterate that it should be possible for scholars to collaborate with people who hold different views from themselves.

Emil Kirkegaard on the link about RationalWiki Carl provides doesn't respond to any actual criticisms about himself, rather spends all his time attacking a RationalWiki editor of his article. As for Kirkegaard demonstrating competence by journals he's published in, Psych that Carl lists is a very poor-quality journal, widely regarded as OpenPsych's successor. While Kirkegaard has published in Intelligence, it has controversially allowed hereditarians to sit on its editorial board, although Richard Lynn was later removed.

Didn’t you publish a paper claiming that “racist stereotypes” are “rational”?

No. I published a paper showing that “in the UK, net opposition to immigrants of different nationalities correlates strongly with the log of immigrant arrests rates”. And I concluded by noting that the study’s findings were “consistent with a model of immigration preferences in which individuals’ expressed support or opposition to immigrants from different nationalities is informed by rational beliefs about the respective characteristics of those immigrant groups”. It has been observed in a number of social surveys that public attitudes to immigration vary substantially across national-origin groups. This is an empirical regularity in need of explanation. One possibility is that public support or opposition to different national-origin groups is partly informed by rational beliefs about the respective characteristics of those groups. In other words, it is possible that the public are more opposed to certain national-origin groups, at least in part, because those national-origin groups have average characteristics which the public deems less desirable (e.g., higher crime rates). My study provided tentative evidence that this is the case, although it is far from definitive. Hence I was not claiming that it is rational for people to hold “racist stereotypes”. Rather, I was using the word ‘stereotype’ (specifically, ‘consensual stereotype’) in the technical sense that it is used in psychology, namely to refer to people’s average beliefs about the respective characteristics different groups. Moreover, my conjecture that “public beliefs about the relative positions of different immigrants may be reasonably accurate” is hardly revolutionary, given that there is already a large literature on stereotype accuracy. According to Lee Jussim and his colleagues, “stereotype accuracy is one of the largest and most replicable findings in social psychology”.

Carl's paper was published at OpenPsych. It is criticised in detail by an academic, Niko Yiannakoulias. Yiannakoulias points out that the paper was "published in an bottom-tier online journal" and understandably goes on to note the paper suffers from methodological flaws and "research this bad should never be published in any form".

Referring to your paper, didn’t an “external reviewer” claim that “research this bad should never be published in any form”?

The person who made that claim was not an “external reviewer” (I’m not even sure what that would be), but rather a Professor of Geography at McMaster University, who wrote about my paper on his personal blog (which he is perfectly entitled to do). Professor Yiannakoulias largely misunderstood the analysis, and I responded to his criticisms more than a year ago. Moreover, as Jonatan Pallesen pointed out, Professor Yiannakoulias’ criticisms are undermined by the fact that, if true, they would apply to other analyses that he himself has carried out. While I have no doubt that Professor Yiannakoulias is highly eminent in his field, he does not appear to possess any expertise in the relevant subject matter. By contrast, the three scholars who did review the paper have all published extensively in psychology.

Laughably Carl here is describing Emil Kirkegaard as a scholar since he was one of three individuals at OpenPsych to review Carl's paper; Kirkegaard's only qualification is a BA in linguistics and Aarhus University where he graduated have made a statement warning he isn't a scientist. While Yiannakoulias is not a psychologist, he is arguably far more knowledgeable in the relevant subject matter than Kirkegaard. And one of the other reviewers of Carl's paper is Heiner Rindermann who in 2016 controversially spoke at a Property and Freedom Society conference delivering an anti-immigration talk: "Cognitive and Cultural 'Enrichment' of Europe by Immigration". Annual conferences by the PFC have been described as "serious academic racist events" by the SPLC.

14. Okay, but in your paper n = 23. I mean, come on!

This point is to a large extent already answered in my response to the McMaster Professor’s criticisms. But I will add a bit more here in the interest of public engagement. Using small samples is very common when doing aggregate level analyses, including within the psychology of estimation and belief-formation (e.g., see Figure 2C in this paper, n = 20.) An important thing to remember is that measurement error is typically much lower when doing aggregate level analysis, meaning that n = 23 at the aggregate level is not equivalent to n = 23 at the individual level. For example, the YouGov poll from which the group-level means in my study were computed had a sample size of n = 1,668. Moreover, if one objects to using small samples when doing aggregate level analysis, then one should dismiss a lot of cross-country and cross-regional studies too. Such a stance is not totally indefensible, but it would entail throwing out a large number of studies published in ‘mainstream’ journals.

The small sample size is a very minor criticism and not worth disputing.

See also

• Noah Carl