I’m shocked. Shocked.

by Kieran Healy on May 23, 2011

Oh look, some evidence that inflammatory claims in something written by Satoshi Kanazawa may not rest on the deep structure of reality or spring from his special ability to speak uncomfortable truths, and may instead arise from an inability to analyze AddHealth data properly. I for one am stunned.

{ 65 comments }

1

Antti Nannimus 05.23.11 at 2:25 am

Hi,

What? WTF????

Antti

2

Lemuel Pitkin 05.23.11 at 2:27 am

Wow, the comments on that post are one vile pit of sludge.

3

Dr. Hilarius 05.23.11 at 3:08 am

The only uncomfortable truth Kanazawa represents is the historic obsession with intelligence and race in psychology. There is a need for a new DSM disorder to describe psychologists compelled to rank order humans along racial lines. Burt’s Syndrome?

4

bjk 05.23.11 at 3:18 am

It’s great that we’re getting a careful analysis of a paper that can no longer be read. Clearly a win for science.

5

Zebbidie 05.23.11 at 3:20 am

And the British unis are cutting philosophy so that EvoPsych academics can write trolling articles about race and gender.

Let me be the first to say that I am shocked…shocked that this has occurred under a Tory government.

6

Lemuel Pitkin 05.23.11 at 3:24 am

It’s great that we’re getting a careful analysis of a paper that can no longer be read. Clearly a win for science.

In this case it would be. But have you tried, you know, clicking on the links?

7

BillCinSD 05.23.11 at 3:27 am

Dr. Hilarious

Ontogeny recapitulates philosophy?

Broca’s Brainless?

Morton salts his data?

The G Spot?

8

bjk 05.23.11 at 3:44 am

Yes, all the brave critics of SK are piling on the Article Which Shall Not Be Published. The article which this journal will not publish and you cannot read is wrong – sort of misses the whole point of the marketplace of ideas, doesn’t it?

9

Lemuel Pitkin 05.23.11 at 3:45 am

Troll much do we?

10

bjk 05.23.11 at 3:45 am

Or sunlight as a disinfectant, either or both.

11

Substance McGravitas 05.23.11 at 3:53 am

12

John Quiggin 05.23.11 at 5:32 am

bjk has apparently never heard of the Intertubes, so here’s a hint. If a journal retracts an article, the author now has the option of distributing it using something called a “web page”. If you ask your friends for something called a “Google”, they may be able to help you.

13

Craig Willy 05.23.11 at 6:04 am

I don’t know why the hell this guy is still with either LSE or Psychology today. See this gem from 2007 where he advocated nuking 35 Muslims cities to win the war on terror: http://www.loonwatch.com/2011/05/evolutionary-psychologist-says-black-women-are-scientifically-ugly-advocates-muslim-holocaust/

14

Brussel Sprout 05.23.11 at 8:00 am

I am absolutely no statistician, but I read Kanazawa’s article asserting that atheists are more intelligent than believers. Of course, as an atheist I was quite happy to believe this, but reading the article, it was clear that his statistical analysis is unreliable and his conclusions pretty suspect.

15

Hidari 05.23.11 at 9:37 am

There’s an article in the current issue of New Scientist (yet another in a long series about why the Black Man just can’t cut it in a White Man’s world) which uses the phrase ‘a pair of maverick Evolutionary Psychologists…’. The author of the piece obviously never paused to ask himself whether or not the phrase ‘maverick evolutionary psychologist’ was a contradiction in terms or not.

16

sg 05.23.11 at 10:46 am

He’s from LSE. Presumably Gaddafi told him to write it…?

17

Ben 05.23.11 at 10:52 am

So someone reanalysed his source data and found him to be wrong?

Wasn’t that a better approach than shouting “SHUT UP SHUT UP SHUT UP SHUT UP SHUT UP SHUT UP”?

I think it was.

18

Alex 05.23.11 at 11:30 am

sg: it’s the new “Jesus made me do it”.

19

Jack Strocchi 05.23.11 at 12:12 pm

Kieran Healy said:

some evidence that inflammatory claims in something written by Satoshi Kanazawa may…arise from an inability to analyze AddHealth data properly.

I am not surprised that there is a big outcry against Kanazawa. Sex is a touchy subject, race is a touchy subject, so the relative sexuality of races must be doubly touchy. (Darwin, in the Descent of Man, explained that sexual selection is the fundamental driver of racial differentiation. But he was always very careful to check his reasoning before dropping these intellectual bombshells.)

What surprises me is that the intellectual attack is focusing on the sloppiness in his empirical methodology, which no doubt could be better. But the real problem is in the fallacy of his theoretical biology.

He makes a couple of howlers on genetics, first misunderstanding the criteria for evolutionary success and then confusing overall diversity with deleterious mutability. In a generally over-looked passage he argues that African’s have a higher deleterious mutation load because they have a longer evolutionary history and an associated greater genetic diversity:

There are many biological and genetic differences between the races. However, such race differences usually exist in equal measure for both men and women. For example, because they have existed much longer in human evolutionary history, Africans have more mutations in their genomes than other races. And the mutation loads significantly decrease physical attractiveness (because physical attractiveness is a measure of genetic and developmental health). But since both black women and black men have higher mutation loads, it cannot explain why only black women are less physically attractive, while black men are, if anything, more attractive.

A long evolutionary history does not imply greater genetic diversity. And genetic diversity does not imply a greater average load of deleterious mutations. Most mutations are deleterious but, in the nature of things, are subject to adverse selection.

“All viable breeding populations are equally evolved”, (Haeckel?) is a standard assumption of evolutionary biology. Evolutionary success is relative, measured by “fitness” to an ecological niche. If anything, a given breeding populations longer evolutionary history and greater fertility implies greater evolutionary success. By that standard, African’s are doing okay.

Australian Aboriginal’s have a long evolutionary history without generating a great deal of genetic diversity owing to the uniformity and stability of the Australian ecological niche. They survived quite well until the environment changed drastically with the introduction of European pathogens. African’s are more genealogically diverse because Africa has been an ecologically diverse and somewhat unstable environment.

The profligacy of deleterious mutations in a breeding population depends somewhat on the adversity of the developmental environment (in gametes & in utero). The durability of these deleterious mutations depends on the degree to which the environment weeds them out prior to reproduction. There is no convincing evidence that African gene pool bears a significantly higher deleterious mutation rate.

One hears reports that Africans have higher abortion/still birth rates and lower life expectancies. This could be cited for evidence for Antagonistic Pleiotropy. But these differences in mortality have to be controlled for environmental differences.

The rate of in-breeding will obviously increase the gene expression of deleteriously mutated genes, by increasing the probability that there will be a conjugal match between lethal recessive alleles. There is no evidence that African’s are especially prone to consanguinity.

Such empirical work that exists on the relative rates of deleterious mutations between racial breeding population suggests that Kanazawa’s genetic argument is almost the opposite of the truth. Nature reports on a study which shows that European’s relatively short evolutionary history in a given ecological niche, and their associated genetic uniformity, implies a higher load of deleterious mutations:

A genetic study into Americans of European or African descent finds that the Europeans have a bigger proportion of ‘harmful’ genes than the Africans…The findings confirm the idea that after human groups migrated out of Africa, they experienced a population ‘bottleneck’ in which their overall genetic diversity was reduced. The European population then expanded quickly, accruing new mutations before there was time for the old ones that caused negative consequences to be weeded out…The result is that the European population contains a lower overall genetic diversity than the African population. And a higher proportion of the European genetic diversity is potentially harmful, as the DNA contains mutations that could alter the function of the proteins it encodes, the team says.

Kanazawa is generally a pretty good socio-biological analyst, who often blows a refreshing gust of anthropological realism through the politically correct cobwebs choking social science. But occasionally he over-reaches and sometimes gets way out of his depth. (Not that I am any great shakes at evolutionary biology, but I do know the basics.) If he wants to make bold contentious claims he should make sure they can withstand cursory inspection, perhaps by running them past a few expert colleagues, rather than airing them first in his blog.

20

politicalfootball 05.23.11 at 12:35 pm

I am not a statistician, but the Scott Barry Kaufman rebuttal to Kanazawa seems seriously flawed, in that it accepts Kanazawa’s premise about “objective” measurements of beauty.

Says Kaufman:

The ratings of attractiveness made by the interviewers show extremely large differences in terms of how attractive they found the interviewee.

The implication here and throughout the rebuttal is that had the ratings been more consistent, they would have shown an objective measure of beauty. In fact, what was being measured was the perception of beauty among the raters, which is a different thing.

I’m prepared to believe that there is a legitimate science of evolutionary psychology, but the practitioners always seem, at least to my untrained eye, to make some very unsophisticated judgments about the things they try to measure.

21

Kevin Donoghue 05.23.11 at 12:43 pm

If he wants to make bold contentious claims he should make sure they can withstand cursory inspection….

No, that would be boring for all concerned. He has the potential to be the Jonah Goldberg of his field. He should be given every encouragement.

22

Ian 05.23.11 at 12:44 pm

I particularly enjoyed his claim that “factor analysis has the added advantage of eliminating all random measurement errors that are inherent in any scientific measurement.” If only everybody used factor analysis, we’d never have to worry about measurement error again. This has got to be the most useful innovation in the original post.

23

Walt 05.23.11 at 12:52 pm

pf: Consistency in ratings is a necessary but not sufficient condition of objectivity. If they are objective, then they should be consistent across observers. That doesn’t commit you to believe that if they are consistent in this particular study, then they must be objective. (And it’s easy to think of outside causal factors that would drive consistency.)

24

Alex 05.23.11 at 1:03 pm

The implication here and throughout the rebuttal is that had the ratings been more consistent, they would have shown an objective measure of beauty. In fact, what was being measured was the perception of beauty among the raters, which is a different thing.

No, it’s worse than that. The raters’ perception of beauty was either not being measured meaningfully, or else their perception of beauty was not self-consistent enough to draw conclusions about.

What was happening was that they were not at all likely to rate the same person as beautiful (or ugly) twice, which suggests either that the property of subjective attractiveness isn’t well defined (possible and interesting), the women in question looked unrecognisably different between interviews (unlikely), or the survey design failed to measure subjective attractiveness (my money).

25

sg 05.23.11 at 1:14 pm

Jack, that’s some serious grocer’s apostrophization you’ve got going on there. Sure sign of genetic frailty, that is…

26

Lemuel Pitkin 05.23.11 at 2:00 pm

Wasn’t that a better approach than shouting “SHUT UP SHUT UP SHUT UP SHUT UP SHUT UP SHUT UP”?

I’d like to think they’re complements.

27

roac 05.23.11 at 2:01 pm

@2: In fairness to the readers of Psychology today, it should be pointed out that all the sludge was emitted by a single individual (as far down as I cared to look, anyway).

28

dsquared 05.23.11 at 2:02 pm

Wasn’t that a better approach than shouting “SHUT UP SHUT UP SHUT UP SHUT UP SHUT UP SHUT UP”?

When someone’s got as long a track record as Kanazawa, nope.

29

Walt 05.23.11 at 2:32 pm

His research is pretty uniformly terrible, isn’t it? (Andrew Gelman wrote a paper illustrating elementary mistakes in statistics just using Kanazawa’s research as an example). It’s hard to find that level of consistency in this fallen world.

30

Bexley 05.23.11 at 3:18 pm

@ 19

I’d say your analysis isn’t harsh enough. The section of Kanazawa you quoted makes no sense at all. In particular this sentence:

For example, because they have existed much longer in human evolutionary history, Africans have more mutations in their genomes than other races.

Does he think that other “races” were created from scratch more recently than Africans with no deleterious mutations? Even using his assumptions they will have inherited harmful mutations from their ancestors in just the same way as Africans have.

And as you say, all this ignores the fact that harmful mutations will be selected against. In reality we expect a balance between natural selection (lowering the frequency of harmful mutations) and new mutations.

31

AcademicLurker 05.23.11 at 3:45 pm

the politically correct cobwebs choking social science.

Yes, thank goodness that we have Objective Thinkers like Kanazawa to help out with those and boldly speak the truth to power by using shoddy statistics to tell racists what they want to hear.

Why, I think I hear the jackbooted PC police kicking in my office door right now…

32

Alex 05.23.11 at 3:55 pm

Andrew Gelman wrote a paper illustrating elementary mistakes in statistics just using Kanazawa’s research as an example

Link. Please.

33

Gareth Rees 05.23.11 at 4:05 pm

Andrew Gelman and David Weakliem (2009). Of Beauty, Sex and Power. American Scientist. Vol. 97, pp. 310–316.

34

Walt 05.23.11 at 4:07 pm

Alex: link.

35

Alex 05.23.11 at 4:11 pm

Danke sehr!

36

Sebastian (2) 05.23.11 at 4:15 pm

37

Sebastian (2) 05.23.11 at 4:15 pm

sorry, too late…

38

david 05.23.11 at 4:40 pm

So, which evo psych scholars are being wrongly tarred by this miscreant? I keep hearing how unfortunate it is that the bad apple leads to attacks on the field as a whole…..

39

imajoebob 05.23.11 at 5:16 pm

It’s too bad that Kaufman displays an absolute, gross ignorance of the most basic concept of Statistics in the first few paragraphs: “It must be noted that with so many variables, there are bound to be many statistically significant results in the dataset simply due to chance.” Significance is a measure of the LACK of chance as cause in Statistics. Everything after must, necessarily, be treated as a (bad) joke.

I pray that Dr Kaufman does not next try to lend his “expertise” to the nuclear power industry.

40

grackle 05.23.11 at 5:24 pm

Sorry, I seem to be stuck trying to comprehend exactly what an objective measure of attractiveness might be, explained nowhere as far as I can see in either Kanazawa’s musings nor in the rebuttal linked (the latter of whom seems to have tacitly accepted that the former knew what it might be.)

Perhaps it’s generally known that evolutionary psychologists know objective beauty when they see it? Or is objective beauty a term of art in that most rigorous of sciences?

41

Dr. Hilarius 05.23.11 at 5:40 pm

imajoebob@38 No, Kaufman is correct. Statistical significance can be due to chance. If you have a large number of variables or repeat an analysis sufficient times you are predictably going to obtain statistically significant outcomes by chance. The significance level of 0.05 represents a 5% chance of obtaining a statistically significant result by chance alone. Do enough comparisons and you will get chance significance. Bonferroni’s method for adjusting significance levels with repeated t-tests, for example, is used to minimize such spurious significance.

42

Sebastian (2) 05.23.11 at 6:09 pm

To expand on Dr. Hilarius (who is, of course, right):
If you have a dataset with 1000 randomly and independently generated variables and you are just looking at binary correlations, you would, in expectation, find almost 25,000 pairs with correlations significant at the 95% level (there are 499,500 possible pairings). If you increase the number of variables, the expected number of possible (false) significant pairs goes up even further.
Even if you fix one variable – Kanazawa seems to really like the “beauty” scores – you still have an expectation of 50 variables that will display a statistically significant bivariate correlation.

43

Sebastian (2) 05.23.11 at 6:10 pm

That should read “bivariate” not “binary”

44

Kieran Healy 05.23.11 at 6:13 pm

imajoebob, Statistically “significant” results can be due to chance because, as typically practiced, the threshold for statistical significance (p=0.05) is simply a matter of convention, and the threshold can be met by chance. In fact, with enough variables or a large enough N you’re guaranteed to meet it for some of your variables. There are of course other methods and conventions for assessing the substantive importance of findings in such circumstances.

45

Kevin Donoghue 05.23.11 at 6:42 pm

Kieran’s post on Irish roads reminds me that imajoebob’s fallacious reasoning can have tragic consequences. In the days when fewer people had cars and almost all rural Irish roads were narrow and winding, many a drunk driver used to reassure his passengers that the probability of meeting a car coming in the opposite direction around a blind corner at 2:00am was so small that it could safely be neglected.

Alas, as the number of Bernoulli trials increases….

46

Anderson 05.23.11 at 6:50 pm

“So, which evo psych scholars are being wrongly tarred by this miscreant?”

Not sure if that’s meant seriously or snarkily, but I would be interested in that myself. Hard to imagine natural selection affected our bodies but left our minds alone.

47

chris 05.23.11 at 7:08 pm

For example, because they have existed much longer in human evolutionary history,

Er, no. All living species (and subspecies, etc.) have existed the exact same length of time in evolutionary history: from the first lifeform to today.

There’s no way to make this clause even meaningful, let alone true. “They have been reproductively isolated from the rest of the species for longer than anyone else” would work fine — for Australian Aborigines. But it’s a horrible description of Africans, sub-Saharan or otherwise.

Africans have more mutations in their genomes than other races.

Aha — something that resembles an empirical claim! (At least, once you change “mutations” to “alleles”. It may be difficult or even impossible to determine which allele was the original, if that’s even a meaningful question, and in any case it’s probably irrelevant.) Maybe Kanazawa should have stopped and proved this point first, before building chains of speculation on it.

Then, once he got over that little methodological hump of deciding what “attractiveness” means and how to measure it in some way that gives you grounds to believe you are measuring a phenomenon that exists in the subjects being measured, he could tie the greater variation to measures of attractiveness and claim that high-attractiveness genes have somehow been fixed in non-African populations and remain a small part of a larger spectrum of variable ugliness in African ones.

48

politicalfootball 05.23.11 at 8:25 pm

Not sure if that’s meant seriously or snarkily, but I would be interested in that myself. Hard to imagine natural selection affected our bodies but left our minds alone.

The problem with evo psych is not that it’s inherently implausible; it’s that it routinely gets invoked in absurd ways. As we noted above, Kaufman may not be as much of a dope as Kanazawa, but he’s still pretty clueless. And both seem to be employed in the evo psych field. Of scientific disciplines, it seems to me that only economics is simultaneously more influential and similarly disreputable.

You make an observation that’s very typical of evo psych types: “Hard to imagine natural selection affected our bodies but left our minds alone.” It’s typical because it’s a straw man – there’s not much of a constituency in scientific inquiry nowadays for the idea that the mind exists separately from the body. You’d have a problem finding anyone in science who would contend that the brain wasn’t shaped by evolution.

49

JP Stormcrow 05.23.11 at 8:56 pm

But there should be a lot of much more foundational, if less “popularly ambitious” research work that could be done right in that space. I assume it is being vigorously pursued by some area of inquiry (other branches of evo psych? more standard evolutionary biologists?Neurologists? Somebody doing things in conjunction with anthropologists and ethnologists? ) Is the whole thing just too fraught with methodological and “political” disputes to be profitably worked at the current state of these overlapping fields?

50

LizardBreath 05.23.11 at 9:18 pm

I think part of the reason that so much evo psych is, to use a technical term, kinda dumb, is that research solidly connecting a mental or behavioral trait either to any physical structure in the brain or to any identifiable gene or complex of genes is really hard to do — the biology just isn’t there to do more than produce suggestive results. It probably will develop in the future, but the good science being done in this area isn’t yet coming up with cocktail-party ready theories of why people are the way they are. To get that kind of theory from our current state of knowledge, you have to leapfrog over a lot of things that aren’t rigorously established as truths yet (and may not ever be), and then start theorizing about why they’re inescapable truths of human nature, which leads to dopey results.

This mess is a particularly egregious example — the nitwit in question looked at one study (which he misinterpreted, but even if he hadn’t) looking at how people perceived the attractiveness of black women, decided that it stated a fundamental truth about how much of this objectively definable quality called attractiveness that black women had (less than women of other ethnicities), and decided that was a solid enough foundation to go haring off after physical explanations for why black women have less attractiveness than non-black women. But a lot of evo psych I’ve seen follows that same pattern less egregiously: start with some facet of ‘human nature’ that’s really poorly established to exist as an innate/genetically determined quality of people at all, and then build a theoretical structure explaining for why it must have evolved that way.

51

Walt 05.23.11 at 9:33 pm

politicalfootball, your argument that Kaufman is just as clueless doesn’t make any sense to me. A precondition of objectivity is that the ratings are consistent. That doesn’t mean that Kaufman believes that consistent ratings necessarily imply objectivity, and I don’t see anything in the article that suggests that he does believe it. You can believe A implies B without believing B implies A.

52

herr doktor bimler 05.23.11 at 9:34 pm

Kanazawa is assuming that (a) there is some universal (“objective”) standard of attractiveness — some feature of bone structure, perhaps, or of complexion, or whatever — that appeals to onlookers even when their own ethnic origins and cultural backgrounds are factored out;
and (b) the various populations inhabiting Africa have not evolved towards that standard, being somehow exempt from sexual selection for these features.

That’s kinda stupid.

53

Dr. Hilarius 05.23.11 at 10:01 pm

Symmetry has been considered as factor underlying a perception of beauty from Darwin on. Symmetry has been considered as a possible phenotypical signal of underlying developmental health. I have not kept up with the literature since falling into the Great Dismal Swamp of the law, but a quick google check shows ongoing investigation. But if there is some fundamental basis for attractiveness it is likely to be confounded by more transient cultural factors, particularly in the age of mass media.

54

politicalfootball 05.23.11 at 10:06 pm

Walt – Kaufman spent a lot of time discussing beauty ratings while dancing very cautiously around what those ratings measure.

Look at this paragraph:

Note that the data could have come out any which way, and no matter how it turned out we would have reported what we found. We do think this is an interesting and important topic of investigation. Other rigorous peer-reviewed published research (involving a much less representative and smaller sample of the United States) has shown statistically significant mean differences in attractiveness ratings based on ethnicity.

Look at that first sentences: “No matter how the data turned out, we would have reported what we found.” This might be a good time for Kaufman to point out that had the data turned out consistent, the data nonetheless would not have supported Kanazawa’s thesis. And yet, not only did Kaufman fail to make this point, he seems to be allowing for the possibility that the data could have supported Kanazawa.

“We do think this is an interesting and important topic of investigation.” Again, either “this” relates do Kanazawa’s objective beauty standard, or it doesn’t. If it doesn’t, why is Kaufman bringing it up?

55

LizardBreath 05.23.11 at 10:06 pm

I’d have to google for a bit, but I seem to remember recently seeing that researchers were recently having trouble duplicating results showing a strong preference for symmetrical members of the opposite sex.

56

LizardBreath 05.23.11 at 10:15 pm

This is what I was thinking of — it’s the New Yorker, so take it for what it’s worth.:

In the three years following, there were ten independent tests of the role of fluctuating asymmetry in sexual selection, and nine of them found a relationship between symmetry and male reproductive success. It didn’t matter if scientists were looking at the hairs on fruit flies or replicating the swallow studies—females seemed to prefer males with mirrored halves. Before long, the theory was applied to humans. Researchers found, for instance, that women preferred the smell of symmetrical men, but only during the fertile phase of the menstrual cycle. Other studies claimed that females had more orgasms when their partners were symmetrical, while a paper by anthropologists at Rutgers analyzed forty Jamaican dance routines and discovered that symmetrical men were consistently rated as better dancers.

Then the theory started to fall apart. In 1994, there were fourteen published tests of symmetry and sexual selection, and only eight found a correlation. In 1995, there were eight papers on the subject, and only four got a positive result. By 1998, when there were twelve additional investigations of fluctuating asymmetry, only a third of them confirmed the theory. Worse still, even the studies that yielded some positive result showed a steadily declining effect size. Between 1992 and 1997, the average effect size shrank by eighty per cent.

57

imajoebob 05.23.11 at 10:15 pm

Dr Hilarius@41 et al – Sorry, but Kaufman is obviously saying you can “prove” this is due to random chance, when he should know that you can’t. All you can do is reject the original hypothesis for lacking credible significance. He can claim unreported correlations will reject the hypothesis, that the data simply doesn’t support the hypothesis, or a number of other things, but he can’t say there is any way to use statistics to prove the negative.

58

herr doktor bimler 05.23.11 at 10:48 pm

Between 1992 and 1997, the average effect size shrank by eighty per cent.
I blame the growing popularity of amputee fetishism.

imajoebob: Have you considered the possibility that the statisticians with whom you are arguing actually know some basic statistics?

59

Norwegian Guy 05.23.11 at 10:59 pm

“I don’t know why the hell this guy is still with either LSE or Psychology today. See this gem from 2007 where he advocated nuking 35 Muslims cities to win the war on terror”

A bit odd for someone from Japan, post-1945, to be so enthusiastic for nuclear war. He also recommended Ann Coulter for US president. This man makes wingnuts look sane.

60

ckc (not kc) 05.23.11 at 11:13 pm

Kaufman is obviously saying you can “prove” this is due to random chance, when he should know that you can’t.

– citation/quote to back up what Kaufman is obviously saying?
– explanation for the scare quotes around “prove”? (this is the one I’m really looking forward to)
– clarification of how you know what Kaufman should know?
– bonus giggles for definition of non-random chance

61

hopkin 05.24.11 at 12:51 pm

Wow, a post and 60 comments on Kanazawa’s ‘research’, and only one feeble attempt to tie it to the Gaddafi money.

Looks like we’re over the hump.

62

leederick 05.24.11 at 7:03 pm

“imajoebob: Have you considered the possibility that the statisticians with whom you are arguing actually know some basic statistics?”

imajoebob is absolutely right, and I doubt he’s arguing with any statisticans.

Kaufman couldn’t be clearer: “there is no difference between the ethnicities in terms of ratings of physical attractiveness… Differences in the distributions … can be attributed to chance”. This is a open and shut case. He’s accepting the null, rather than failing to reject it. If he wanted to do this he should have used an equivalence test. While we’re on the topic it’s strange that he used Pearson’s Chi Squared with this data, as he’s ignoring the structure of the rating scale. The test can’t tell that ‘very attractive’ is higher rated than ‘unattractive’ – it just sees undifferentiated classes – strange thing to do if you want to test for differences in attractiveness. He should have used Mann-Whitney or Kruskal-Wallis. Aside from that he’s also got 4 out of 20 cells with expected frequencies less than 5, so even the test he did is suspect as the approximation to the chi-square distribution won’t be valid.

63

daelm 05.25.11 at 10:51 am

it’s maybe not Gaddafi – thanks hopkin – but i’m surprised no-one has mentioned that the LSE has a track record with this stuff.

see Oliver Curry, for example, in 2006.

http://www.badscience.net/2006/10/%E2%80%9Call-men-will-have-big-willies%E2%80%9D/

d

64

JL 05.25.11 at 9:44 pm

I will just note that Kaufman did his reanalysis of the Add Health data with Jelte Wicherts. I think Wicherts was probably responsible for most of the analysis, because he specializes in vetting psychological studies, e.g. http://wicherts.socsci.uva.nl/BakkerWicherts2011.pdf.

65

Norwegian Guy 05.29.11 at 11:10 pm

I see you noticed Kanazawa’s stupidity already three years ago:

http://crookedtimber.org/2008/03/11/in-the-eea-tactical-nuclear-weapons-were-highly-fitness-enhancing/

Comments on this entry are closed.