Scientific misconduct misdiagnosed because of scientific misconduct
Please remember to have no heroes or villains, but this just looks plain bad to be honest. I’m lowering my estimation of the quality of Stephen J. Gould’s work in this area.
The late scientific icon, Stephen Jay Gould, botched and perhaps faked his critique of a racist 19th-Century scientist’s skull collection, suggests a second look at his efforts.
In a 1978 Science paper, Gould (1941 − 2002) , reported that the Samuel George Morton (1799-1851), “a prominent Philadelphia physician,” had mis-measured the cranial capacities of his 1,000-skull “American Golgotha” collection gathered from around the world, to suit his racist beliefs. The finding led to one of Gould’s best-known books, The Mismeasure of Man, a critique of scientific racism.
“Morton is now viewed as a canonical example of scientific misconduct. But did Morton really fudge his data?,” asks a PLoS Biology study led by anthropologist Jason Lewis of Stanford University. “Are studies of human variation inevitably biased, as per Gould, or are objective accounts attainable, as Morton attempted?”
So, the study team remeasured the skulls collected by Morton, now owned largely by the University of Pennsylvania Museum of Archaeology and Anthropology in Philadelphia.
Overall, they find, Morton did make mistakes in measuring skull capacity (he first stuffed them with seeds, and later lead shot to measure their brain size). But the mistakes were random. The random mistakes didn’t favor any racial theory of larger brain sizes for white people over others.
“Given how long Gould’s work has been criticized in this arena, I’m a little surprised that it took this long for the work to be done to write this article,” says the University of Texas’s David Prindle, author of Stephen Jay Gould and the Politics of Evolution. “People who dislike Gould’s work will likely go on disliking him even more after this article. People who are fans of his writing will likely go on supporting his views.”
In reevaluating Morton and Gould, we do not dispute that racist views were unfortunately common in 19th-century science or that bias has inappropriately influenced research in some cases. Furthermore, studies have demonstrated that modern human variation is generally continuous, rather than discrete or ″racial,″ and that most variation in modern humans is within, rather than between, populations. In particular, cranial capacity variation in human populations appears to be largely a function of climate, so, for example, the full range of average capacities is seen in Native American groups, as they historically occupied the full range of latitudes, say the study authors.
...Samuel George Morton, in the hands of Stephen Jay Gould, has served for 30 years as a textbook example of scientific misconduct. The Morton case was used by Gould as the main support for his contention that ″unconscious or dimly perceived finagling is probably endemic in science, since scientists are human beings rooted in cultural contexts, not automatons directed toward external truth″. This view has since achieved substantial popularity in ″science studies″. But our results falsify Gould’s hypothesis that Morton manipulated his data to conform with his a priori views. The data on cranial capacity gathered by Morton are generally reliable, and he reported them fully. Overall, we find that Morton’s initial reputation as the objectivist of his era was well-deserved.
- [Link] Power of Suggestion by 6 Feb 2013 10:04 UTC; 38 points) (
- [Link] The Hyborian Age by 21 Jan 2012 19:00 UTC; 34 points) (
- 5 Sep 2014 18:51 UTC; 14 points)'s comment on “NRx” vs. “Prog” Assumptions: Locating the Sources of Disagreement Between Neoreactionaries and Progressives (Part 1) by (
- 5 Aug 2013 0:09 UTC; 13 points)'s comment on Rationality Quotes August 2013 by (
- 10 Jan 2013 19:30 UTC; 13 points)'s comment on Open Thread, January 1-15, 2013 by (
- 11 Jun 2011 7:59 UTC; 12 points)'s comment on Beware of Stephen J. Gould by (
- 4 Jul 2012 19:11 UTC; 11 points)'s comment on [Link] Why the kids don’t know no algebra by (
- 20 Sep 2012 5:45 UTC; 8 points)'s comment on Open Thread, September 15-30, 2012 by (
- 20 Feb 2012 7:50 UTC; 5 points)'s comment on Brain shrinkage in humans over past ~20 000 years—what did we lose? by (
- 5 Jan 2013 10:30 UTC; 5 points)'s comment on Politics Discussion Thread January 2013 by (
- 14 Nov 2011 21:32 UTC; 3 points)'s comment on 2011 Less Wrong Census / Survey by (
- 6 Jul 2011 10:05 UTC; 0 points)'s comment on What are LessWrong’s thoughts on Venkatesh Rao, Gregory Rader, and Daniel Lemire? by (
Villains do exist in this world. So do heroes, although they’re a lot rarer. That villains have good sides, or heroes have flaws, does not change this point. And yes, Gould is a bad guy. Not Voldemort, but still someone whose scientific works contain lies and misdirections and mis-implications subtle enough that I would consider it to be a foolishly overconfident risk to try to read them.
Though I think that this is a really important counterpoint.
I once tried reading a book of some of Gould’s essays… the parts I read were mostly just boring and I didn’t bother to finish it.
I like many of his essays. In any case, he doesn’t discuss this or the evolution thing in many of them, so it’s fairly irrelevant.
“The Mismeasure of Man” is mostly only good for showing how political correctness messes up people’s ability to think.
In cases of this type, it’s an error akin to relativism. “Suzy is wrong, therefore I should kill her” does a lot of damage, so people start thinking “Nobody is ever wrong”. “Members of outgroup X have negative traits, therefore they’re subhuman” does a lot of damage, so people start thinking “Members of outgroup X can’t have negative traits”. It’s best to stay on the object level and prove “This group’s average brain volume is 95% of that group’s” rather than saying “political correctness”, for the same reason proving particular facts is more effective than getting into “What is truth?”.
In other cases, like changing the color of garbage bags from black to orange, it’s fear of being thought evil and thereby losing status, and attempt to get status by chiding others. This has absolutely nothing to do with the group you’re trying to defend, so it leads to white people telling other white people “This is offensive to black people!” while all black people are saying “But we’re not offended!” and getting ignored. The phrase “political correctness” used to be useful in those cases.
In yet other cases, shouting “political correctness” is just applying a phrase with negative connotations to people who don’t want you to lynch your neighbor.
Would you care to be more specific?
As far as I know, the most thorough criticism of TMOM back in the early eighties was published by Arthur Jensen. Whatever you think about Jensen’s own theories, his criticism of Gould is pretty damning, and it should be mandatory reading for anyone who has read TMOM. (Gould was invited to reply by the journal that published Jensen’s review, but apparently he never did.) For other prominent criticisms of the book, see e.g. the 1983 review by Bernard Davis (Gould’s reply here) or the 1995 retrospective review by John Carroll.
Also, the propagandistic rather than scientific quality of TMOM is especially evident from the fact that Gould republished it 15 years later without a single change in response to the criticisms the first edition received, nor even in response to the relevant scientific developments that occurred in the meantime. (He just tacked on his review of The Bell Curve as an appendix to the original text.)
Well, the whole book is awful.
I read it over a decade, though, and have no plans to revisit it. Update: missing “ago”.
That’s pretty slow. :)
Actually, I’d say it’s a good example of how a passionate ideologue who is at the same time very smart and a great writer can produce an extremely biased propagandistic work that nevertheless looks admirably objective and reasonable to a casual reader. Certainly, the book deserves a place among the most skillfully crafted works of ideological propaganda in history. This is especially evident when you compare it with similarly slanted but much cruder and more obviously propagandistic works by Gould’s less talented co-ideologues (most notably Not in Our Genes by Lewontin et al.).
I laugh at us too GLaDOS. Sometimes, I have to so I don’t cry.
I regard humans as incongruent as well, albeit without crying.
Sometimes I’d really rather be a dolphin, y’know? They’re quite intelligent, and aren’t about to destroy the world by accident. Or maybe a bonobo.
Kinda speciesist, don’tcha think? People in the modern world in large part have learned to be illogical, but it isn’t an inherent quality; in fact, some would argue that the current low level of rational capacity is very difficult to maintain. If people were inherently irrational, why can everyone learn mathematics, why can children sometimes disagree with their parents, and why was a prerequisite for degeneration into the current American political system that 7 corporations should own all major media outlets?
I didn’t down vote this but considering you seem to be new I feel I should explain why it was down voted and probably rightly so.
Currently the consensus on this site and among experts is that humans are inherently irrational on many things. Evolution hacked our brains together and the reasoning part of the brain always having the best possible map of reality was not an end in itself. Search for mention of evolutionary psychology you’ll quickly run into examples, though speculation about the evolutionary origins of said features or from our perspective flaws are often poorly founded.
Besides the older sequences the recent one by Kaj Sotala on Robert Kurzban’s book might be a good place to start reading.
I likewise do not dispute the colour orange has no clear discrete border to the colour red, and that indeed both are a social construct. This seems an implicit appeal to Lewontin’s fallacy. Though in this case it seems almost like a half-hearted ritual denunciation put there as safety precaution because they are criticising the, in debates oft cited, saint Gould.
When thinking about the climate and cranial capacity connection the most likley explanations seems to be simply that cold clime, all else being equal, requires, more smarts, but please note that it is also possible that cranial capacities vary due to the problem of temperature regulation of the brain (the relationship between surface and volume matters in this sort of thinking).
Mr Johnson, sir, there you are! The lab boys have been looking for you. They say they’ve figured out where the missing personality core got to!
Another alternative explanation that has surfaced (paper) is that both bigger eyes and bigger brains developed in order to deal with the low light condition. Commentary on the study, by Peter Frost:
What I find surprising is that human eyes size increases further from the equator, this is something I think I’ve never heard of before.
Temperature regulation aspects might rate more highly as an influencing factor than one would think. Large bodies (and probably the head in particular) would be more resistant to hypothermia, whereas small bodies would be more resistant to heat-stroke.
Regardless of why, animals definitely do become larger further north and the brain size seems to follow the body size quite closely without much impact on intelligence. I don’t know if arctic animals are quite on the scaling line. They do seem a bit smarter.
(The point of this comment is just to disentangle theory from observation.)
If I’m reading this right, the brain-to-body mass ratio dosen’t change?
I was not claiming that. That is the thing I said I don’t know: “I don’t know if arctic animals are quite on the scaling line.” This is a precise question about data is that has been collected. I just don’t know what the data says. I’m not sure what I meant by “quite.” When animals diverge from the scaling line, like primates, corvids, and dolphins, they move to parallel scaling line, not far from the main line.
Incidentally, the scaling line is not a constant brain to body mass ratio, but that the brain mass is a constant multiple of the 3/4th power of the body mass.
Ok than you for clearing that up (up vote), I hope you didn’t mind me asking since I wasn’t sure if I understood the comment properly or not. :)
See now, this layman couldn’t tell from Wikipedia why Edwards’ critique actually contradicts what the intro calls the main point of Lewontin. Edit: I mean the section on Lewontin’s argument.
It would seem very odd if a sufficiently knowledgeable geneticist couldn’t tell a person’s natural skin color from their genes with near 100% reliability. Melanin clearly has a strong genetic component, as do other physical features that correlate with melanin. We want to know if it correlates with any interesting genetic differences.
Well, rather obviously it correlates with all sorts of things: not having red hair, or blue eyes, or blond hair, or straight hair, not being an Ashkenazi Jew, and not being able to digest milk. What would you find “interesting”, though?
A priori unlikely—skin colour is strongly selected for just on its own—near the equator you need melanin for protection from the sun, in high latitudes you need a lack of it to get Vitamin D—so Africans have lots of the stuff despite having the widest genetic variance per area of all humans. (That linked paper uses this fact to support their assertion that east Africa is where humans come from - it’s a standard expectation that variance is greatest near the origin).
“Race” is a magical category that does not carve the gene pool at its joints. It’s a social categorisation pretending to be a genetic one. e.g. Anyone who regards “negro” as a genetic grouping, on a par with Ashkenazim or Icelanders, has just invoked a magical category and needs to be led gently through charts of genetic variance per area.
When measuring effects of melanin, did they measure Indians as well?
That doesn’t matter so long as people use “race” to refer to “phenotypic” categorizations rather than genetic categorizations. People don’t care if differences between races have a genetic origin, they just care about having roughly correct cached priors for predicting stereotypical behaviors/traits based on skin color, presence or absence of epicanthic folds, et cetera. In practice these priors are roughly as informative as those for clothing choice and the environment of interaction for predicting behaviors. The vast majority of people have no reason to care that Africa has a ton of genetic diversity, it’s practically useless information. Race categorization, on the other hand, is not useless.
(It is very important to note that the above is descriptive and not normative.)
That’s what I meant by it being a social category. A social category with a visible marker! At least the Burakumin can hide in plain sight.
The other problem with the usual question “race and intelligence” (which usually seems to start at “black Americans and intelligence”) is that our tool to measure intelligence is IQ tests. Although a 10-point IQ difference within one social group that e.g. correlates with lead in paint is something to worry about lots, it’s ridiculously easy to get 15 points’ difference between groups for really obviously social and cultural factors (e.g. Burakumin). So if you’re measuring IQ between groups, a difference of 15 points or less may well be cultural. And then there’s the Flynn effect, which could be culture or food.
The x axis is a genetic magical category, the y axis is incredibly shaky and people are way invested in the answers. What could possibly go wrong?
Didn’t the eggheads say back in 1994 that it wasn’t due to cultural bias?
Wow its sounds dumber when I quote them.
Heck, what do I know? I can’t be bothered to read more than one page on this stuff. Got my info from this flyer.
But, American “blacks” have a 15 point difference because they’re [EDIT] about 20% European. Africans have average IQs of 70.
[EDIT]: Citation (see also), though I should note that it is questioned.
The best evidence available today indicates that this is simply false. A recent systematic literature review of Africans’ performance on IQ tests other than Raven’s Progressive Matrices finds an average IQ of around 82. An analogous review of results from Africans who took the Raven’s Progressive Matrices finds a similar average of about 80.
I think that’s false. African Americans are about 20% European in ancestry if memory serves. Of course if eugenic selection did its work, some high IQ alleles may have escaped the “neutral” baggage from the Native American and European gene pool.
Lynn’s exact numbers, while spot on the general pattern, have been criticized by others. There are a some studies that point to an actual IQ (that try to estimate the effects of things like people being unused to using a pencil or solving the test in a nonnative lanugage) nearer to that of 75 and a handful even something as high as 80.
Regression to the mean of the children of African immigrants in Europe, points to a 80ish IQ once environmental advantages kick in.
Citation? If this were a well established finding without known flaws in its methodology, I would expect it to come up a lot in race/intelligence discussions, but this is the first I’ve heard of it.
American “blacks” aren’t about half European anyway, they’re less than 20% European on average.
This seems extremely implausible. Would a group with intelligence as low as an average IQ of 70 connotes be able to maintain language at the same level as most human groups?
I think you don’t have a clear picture of how highly functional low IQ people can be. This is in sense natural, since I mean how many people with an IQ of 70 do you know? Its silly but I sometimes have to remind myself they are by definition as common as people with an IQ of 130. They hold jobs and they manage to reproduce just fine, so they aren’t exactly the helpless drooling passives that popular imagination paints them as.
Also I think its often underestimated how different low IQ people of different types can be from each other.
Can a 13 year old learn to drive a car safely? Yes. Can he work in a factory? Yes. Can he run a farm? Yes. Can he be a soldier? Yes. Can a group of 13 year old children use lanugage at basically the same level as most human groups? Yes.
But how well would a 13 year old do compared to 18 year old on a IQ test? Not that well I suspect.
If the test got mislabelled as that of a 18 year old, what would be his estimated IQ? In abstract thinking he may not be that better than a 18 year old with a 70ish IQ but in many many other regards the test wouldn’t do him justice by putting him in the same category.
Black job performance is slightly better than what would be expected going just by their IQs. And Blacks with IQs in the 70s are generally on average more functional than Whites or East Asians with the same IQ. This has been the basis of some speculation that low IQ Whites and East Asians do worse on average because the sample contains not only people who are plain dumb, but a greater fraction people who suffer from other disabilities and more general brain damage (and/or underdevelopment) too, than is found in the group of IQ 70 Blacks.
Funnily enough, I recently posted some comments on exactly the IQ 70 Africa claim against the usual view that that is impossible, quoting a LWer’s blog post about working in Haiti:
You’re right. I was posting from a position of considerable ignorance about what low IQs might mean in practice.
What would you expect to see from a civilization with an average IQ of 70?
Farming, simple irrigation systems, pottery and other crafts (with objects of great beauty made for the upper class), walled cities, some siege engines, long distance trade, domestication of animals, iron working, sailing, class divisions, chariots and writing.
This they could all eventually develop with no input from the outside in favourable climactic conditions.
Exposed to outside influence I think they can pick up things like clockwork, ocean worthy ships, machine tools, the internal combustion engine, radio and television (and improve it on their own as well). Even basic nuclear technology, computers and some complex medical equipment isn’t completely out of the picture, though don’t expect any refinements. As a society they can probably get clean water, electrification, good roads, decent hospitals, reasonable safety from crime and quality primary school education.
This is what I’d expect of a civilization with an average IQ of 70. But please remember this civilization requires some adaptation, what I wrote goes for a “civilized” people (or at least a people that practised farming for some time) with an IQ of 70. Not all hunter gatherer populations with that IQ would be able to make this work. Self-domestication in humans isn’t really primarily about raising IQ as some people assume it is.
I read this interesting quote today, and I couldn’t help but remember your car example:
--Howard 2001 Searching the Real World for Signs of Rising Population Intelligence
(This leads to a provocative thesis which I find amusing just to contemplate.)
But how highly functional are low-IQ people on average? Looking at only the most functional low-IQ people wouldn’t give a good idea of what to expect in general.
The 13-year-old’s raw score would generally be lower, but their IQ should be about the same as the 18-year-old’s; IQ tests are age-normed nowadays.
My point was that different groups of people that might score low on IQ tests sometimes differ systematically from each other. In my last paragraph I suggest quite directly IQ 70 Black people are different from IQ 70 Whites and East Asians.
Well aware of this. Hence:
Fair enough. I misread the bolded part of your post as just amplifying a more general point about younger people scoring lower on IQ tests.
Knowing that someone has substantial Ashkenazim ancestry lets you make many probabilistic predictions about both neutral and significant genetic variations (e.g. diseases, lactose tolerance, etc). This is because of the fact that historically mating behavior was highly nonrandom across geographical and ethnic lines. Since selection pressures and new mutations varied by region (see lactose tolerance, malaria resistance, fast-twitch muscle, salt retention, etc, etc) the differences predicted are enriched for differences due to natural selection, i.e. interesting ones that actually made life-or-death differences in the past.
Learning that someone is of European Ashkenazi descent lets us (probabilistically) predict a variety of genetic differences from the European average at each of many loci, and to very accurately predict, in the aggregate, a systematic skew across many loci to the Ashkenazim distribution.
This is just factually wrong. You can use many, many different traits, or just neutral variation to easily cluster humans in a high-dimensional space of genetic variance. The oceans, the Sahara desert, the steppes, and other geographical features were major factors in historical reproductive isolation, and the largest (and quite clear) clusters correspond pretty well to the old anthropological classifications. If you tell me some facts about bone structure, body fat ratio, malaria resistance genes, whatever, I can quickly assign high probability to a given mix of recent continental ancestry for an individual and make much better predictions about other traits using that info than I could without using that clustering. In the U.S., self-identified race is itself a very strong predictor of continental ancestry, and if one supplements with further questions about grandparents’ race or asks about multiracial ancestry it gets even better.
This post by physicist Steve Hsu discusses the topic well, also see this paper by Risch et al:
Yes, that’s what I meant—they’re an actually close-knit genetic group. “Negro” is not. “Black American” is a little closer, but not much (their African ancestors having been taken as slaves mostly from the west coast) - you’ll see obviously highly selected-for factors, like black skin and sickle-cell.
There are people who identify as black or who are identified as black under a “one drop” rule. This doesn’t seem as though it would give huge amounts of information about associated traits.
Aren’t clams that race is not genetic just plain silly?
Slightly but not just plain so.
For example, a lot of phenotypes would get classified as “black” in the American race structure, and “white, somewhat mixed” in the West European one. In the US, there’s a category called “Asian” with covers both East and South Asians. Two people in the US with the same ancestry and physical appearance may get classified as either white or Native American depending on their ties with a tribe. And so on.
Sure, “race” divides people according to physical appearance, ancestors, and culture, so it has a strong genetic component. But it’s not solely genetic, and inasmuch as it is, it’s still bad genetics, like the distinction between fish-and-dolphins and mammals is bad taxonomy.
Reminds me somewhat of the study that took another look at some of the work of Franz Boas. It was first brought to my attention by Nicholas Wade’s article about it in the NYT:
Very intent apparently.
Abstract of the 2002 study:
That does interest me. Why did people citing Boas think it made sense to talk about “the American environment”? Even the USA had 48 states by the end of February 1912.
From the OP:
Think of it as a category like “developed world”.
At the same time, some other anthropologists dispute Sparks & Jantz’s conclusion that Boas was incorrect. From the abstract of the paper I’m linking:
Somehow this post has actually increased my confidence in Gould’s claim here.
Further reading suggests Gould is not representative of scientists. My confidence has gone back down.
This happens frequently, and we don’t see these questions resolved because the scientific method is far from bulletproof. Doubtless many of our modern ideas will be proven incorrect by the next generation; others will learn to make more accurate predictions using more advanced analysis; some paradigms will seem ludicrous in rhetrospect (as some models which were accepted only decades ago seem today). Just how frequently such an obvious problem happens, for the same reasons this case went unnoticed, it is very difficult to estimate.