Talk:Race and intelligence/Archive 39

From Wikipedia, the free encyclopedia

Archive This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.

Contents

1981 assessment

Was the last racial IQ assessment really taken in 1981? If so, why was the Reynolds piece about it from 6 years later? Is that some kind of typo?

The 1986 Minnesota Transracial Adoption Study results weren't published until 1992 either. You might want to sign your comments btw. --Zero g 23:10, 17 September 2006 (UTC)

== Uh, just I was reffering to the chart on the main page of the article. Could you please confirm to me if was it published in 1981 or not?

Expert Opinion*s*

Just as a pointer here (please ignore if you wish), I've tried to summarize what I understand is the mainstream "opinion" of different "expert groups" on relevant issues that touch the subject. Please feel free to add.

  1. Psychometricians: IQ is measurable, in good part genetically-defined and the rest is environmental. Racial differences in IQ are most probably partly genetic. (Snyderman and Rothman survey, for example)
  2. Psychologists: (in general): same opinion, but less certainty on the possibility of partly-genetic contribution to racial differences. (APA statement)
  3. Anthropologists: "Race" exists as a social construct which correlates poorly with biological categories of humans, i.e. if they exits "biological races" aren't the same as "social races". (AAA statement, mostly)
  4. Neuroscientists: Given the known plasticity of the brain, small variations in general brain volume poorly explain differing IQ results. In any case, the only racial-difference studies conducted so far in this field are about facial recognition, gender and race recognition thus far. Overall, "intelligence" is still poorly understood in a mechanistic sense, but great strides are being made rapidly. (still looking for the best possible reference, but the consensus is clear so far).

I'll add more as I come across them.

This kind of classification is NOR unless it has been made elsewhere. --Rikurzhen 02:53, 8 October 2006 (UTC)

Rik, this is just a pointer, and definitely not for inclusion in the article. So, not sure what you're concerned about???--Ramdrake 15:57, 8 October 2006 (UTC)
Oh --Rikurzhen 18:36, 8 October 2006 (UTC)

The first picture

However made that is dumb or ignorant "white" "black" "hispanic" "asian" are American street words and inaccurate and incorrect categories. This should really be messured with the proper "-oid" names. --guest

i agree. also the categories are very general. like what does asian denote? chinese? indian?

It generally denotes Eastern Asian, such as Chinese, Japanaese, Korean, and most nationalities of Southeastern Asia.--Ramdrake 14:24, 21 October 2006 (UTC)

disputed tags

"still disputed" -- really? most of the discussions have been archived at this point, and there doesn't appear to be any live debate on the talk page. the purpose of those tags is to attract readers attention to an actual conflict or controversy, not hypothetical or archived ones. --Rikurzhen 19:57, 8 October 2006 (UTC)

There has been a lot of pressure to reduce the size of this talk page through archiving, which has been done. Personnally, I was taking a bit of a breather before continuing the work on the article proper. But the disputed tags still belong up there as far as I'm concerned.--Ramdrake 20:08, 8 October 2006 (UTC)
I ask the question of whether the tags are serving their intended purpose. Note that they say "Please see the relevant discussion on the talk page." Perhaps you or JK could construct a short list of which archived discussion(s) each tag refers to. In the current state of affairs, they are of little value: I've seen no improvement in the quantity or quality of contributions with them in place. If I were new to this article, I would find nothing helpful on the talk page and probably leave without comment. --Rikurzhen 20:29, 8 October 2006 (UTC)
Good idea, Rikurzhen - each one of them probably deserves a significant treatment. I think though, we should try to continue our original idea from several weeks ago to pick on topic and put it to rest at a time - often we get side tracked, and fail to come to reasonable compromise simply because of the distraction factor. --JereKrischel 02:04, 9 October 2006 (UTC)

race and climate

article says > "pointing out that the racial correlation was merely a by-product of the correlation with climate"

this is truly at the heart of the racial cline/cluster debate, and suggests a conflict where none seems to exist. races (of any species) are largely a product of climate (plus other factors like geographical boundaries -- mountains, deserts, oceans) by the standard thinking on the question of animal biogeography. the questions about race should be discussed in that section where they can be fleshed out in a balanced way, not on a phenotype-by-phenotype basis throughout the article. in each section of the article, the data should be described as it is by the researchers (whether clines or clusters) without reengaging that debate. --Rikurzhen 20:41, 8 October 2006 (UTC)

the actual conflict is over how to describe human biogeographical variation. a cline is a geographical gradient of phenotypes/alleles. clines exist undoubtable as they can be easily seen for many phenotypes/alleles. and the cluster view, that the geographical variation of many alleles/phenotypes can be reduced in dimensions to a smaller number of clusters. the current race section describes that various constrasting opinions on the appropraiteness and existence of clusters in human biogeography. --Rikurzhen 20:47, 8 October 2006 (UTC)

I believe we've been over this before: If we are to say that "brain size varies according to race", we should present the counter-argument brought by Beals et al (1984) and reviewed among others by Lieberman (2001) that this variation has nothing to do with race, which Lieberman shows is a view shared by many researchers in the field. This counter-argument rightfully belongs side by side with the original statement, otherwise we will end up putting more weight on the statement than on the counter-argument.--Ramdrake 20:59, 8 October 2006 (UTC)
What about what I'm saying doesn't make sense? Lieberman's argument is that race (clusters) doesn't exist, and what we're seeing is merely variation by cline. That brain size varies by cline is noncontroversial. But to put the race doesn't exist argument everywhere you find a description by cline is to overemphasize that debate. Right? Much of the data on brain size, for example, comes from people in the U.S. who are merely identifiable by "race". Thus, description by race is what we have. But to bring up disagreement about the discreteness of "races" in this context is not to refute that brain size varies by "race" when race is the classification you have at hand. By analogy, to argue that skin color variation is clinal is not to argue that skin color doesn't differ by racial groups. It's to prefer to describe the variation in terms of clines rather than clusters. --Rikurzhen 21:06, 8 October 2006 (UTC)
Lieberman does not support or deny the existence of races. He criticizes the definition of races that Rushton makes and uses as being too vague and not based on precise, definable criteria. I think that especially in the United States, it is evident that the definition of "race" used there is first and foremost a social construct, as a very large proportion of "Blacks" there have a significant "White" admixture. I'm not saying that what you're saying doesn't make sense; I think what you're saying does indeed make sense; however, to me, it's just not the best argument. For example, in the situation we're discussing, race is NOT the only classification you have at hand.--Ramdrake 21:37, 8 October 2006 (UTC)
Maybe I'm confused about something, but I've made a change to the text based on my current understanding on the literature at hand. --Rikurzhen 21:47, 8 October 2006 (UTC)

My inability to fathom that Lieberman would object to the terms "cline" and "cluster" is substantiated by the fact that he uses the terms in the 2001 paper. Here are two sections:

  • "Clines provided a concrete alternative to thinking in terms of races. Identifiable traits were not confined to one “race” and were not uniform in frequency within a geographic area."
  • "There is little doubt that there are variations in [brain size] and in other biological traits from one population to another. However, they are not clustered in races but vary from one population to another, forming discordant geographic clines. While Rushton acknowledges variations, he persists in aggregating all the available data on diverse populations into the three traditional races that were politically correct for the past three centuries."

--Rikurzhen 22:42, 8 October 2006 (UTC)

The bolded part (my format addition) is the major issue in the cline/cluster disagreement. One view is that the discordant clines are truly discordant and the other view is that taken as a whole there is an underlying clustering amongst the clines. --Rikurzhen 22:47, 8 October 2006 (UTC)

I think the harsh criticism of Lieberman regarding race (who mocks the idea of how "Caucasoids" got big crania, but then they shrunk), is contained in the following part of your quote, Rikurzhen: he persists in aggregating all the available data on diverse populations into the three traditional races that were politically correct for the past three centuries.
So the cluster/cline issue is really a red herring - the critique being made is whether or not "politically correct" races make any sense in regards to the current data. That is to say, to say that "brain size varies by race" is inherently flawed because of the lack of a clear definition of "race". Maybe this is another assumption that should be made clear? That most racialist hereditarians depend on the validity of archaic and simplistic definitions of racial groups? --JereKrischel 02:14, 9 October 2006 (UTC)
We're back to not making any sense. I hope Ramdrake gets it and can weigh in here. --Rikurzhen 04:03, 9 October 2006 (UTC)
Here are the full quotations from Lieberman about Beals' data and its interpretation by Rushton [1]
(p.6)
Rushton cites anthropologists including Beals, Montagu, Tobias, and others, taking their data out of context and using them in a manner contrary to the conclusions these authors reached and without informing his readers of this crucial omission.
(p.8)
Rushton (1990:786) takes cranial measurements from a study by Beals, Smith, and Dodd (1984) without mentioning that study’s finding that while climate variables were strongly correlated with cranial variation, “race” and cranial variation had low correlations.
(p.8-9)
The relationship between latitude and cranial size is an example of Bergmann’s principle that crania are more spherical in cold climates because mass increases relative to surface area to conserve core temperatures: “A slight increase in head size combined with a rounder cranium has a disproportionate effect upon volume” (Beals, Smith, and Dodd 1984:312). “The closer a structure approaches a spherical shape, the lower willbe the surface-to-volume ratio” affecting radiation of metabolic heat and temperature regulation, which is especially important in colder climates because as much as “80 percent of body heat may be lost through our heads on cold days” (Molnar 1998:202). Beals, Smith, and Dodd emphasize that this relationship is independent of “race.” “In fact, several of their climatological-cranial correlations reach .60, much higher than any relationship Rushton has been able to report for race, except for one study” (Weizmann et al. 1996:196).
(p.9)
Rushton argues that “Mongoloids” have superior, larger brains because in their evolution they had to adapt to a cognitively demanding but predictable cold Pleistocene climate (1997a). An alternative scenario is provided by Brace (1998:112): “the mode of subsistence of all human populations was essentially the same throughout the entire range of human occupation over the past 200,000 years. This was conditioned by adaptation to the selective pressure engendered by the cultural ecological niche. For these reasons, then, cognitive capabilities should . . . be the same in all the living populations of the world.” Brace points out (p. 4) that 100,000 years ago early modernsat Qafza “were making the same tools, hunting the same animals . . . as their Neanderthal contemporaries,”and therefore we can conclude that human cognitive capabilities are distributed in a nonclinal way.
Emphasis is mine. So, in all those quotations, not only do we see that Lieberman uses a word related to "cline" once (and nothing about "cluster"), but he quotes another researcher, agreeing with him that cognitive abilities are distributed in a nonclinal way. And we see that even if he discusses clines to a small extent in the article, the sections which reference Beals' studies specifically make no mention of the cline/cluster debate and actually go as far as rejecting a clinal distribution for cognitive abilities. I believe that this fully demonstrates that the reference in the Wikipedia article should not be directly associated with the cline/cluster debate, as according to the very authors we cite, it isn't. Saying otherwise is purely OR, IMHO.--Ramdrake 12:28, 9 October 2006 (UTC)
. Actually, I wouldn't object to taking any of these quotes and using them instead of the sentence we are currently using in the article. However, I do object to saying that this difference of opinion is all about the cline/cluster debate.--Ramdrake 12:46, 9 October 2006 (UTC)


There. I've gone and replaced the disputed passage with an actual quotation from Lieberman et al. That way, nobody can reinterpret it to mean this or that.--Ramdrake 14:26, 9 October 2006 (UTC)

effect size of brain size variation

does anyone know the effect size (measured by a statistic called d) of brain size differences by race? the effect size (for the B-W IQ gap) is d = ~1 among adults. it would be a good addition to include that statisitic for brain size as well. another statistic of interest is the within family, within race, between race ANOVA table for brain size. we have this data for IQ in the signficance section. --Rikurzhen 21:13, 8 October 2006 (UTC)

Most variance occurs within families

If say, Jensen finds a 20-point average difference between the IQ of a Black and a White person chosen at random (I believe some find a bit less), and there is a 12-point gap between two random members of a same family, then 12 out of those 20 points of difference are attributable to within-family differences, thus proving that most variance does occur within families. It is not taken to mean that the variance between two members of the same family is larger than the B-W variance, which seems to be the way you read it.--Ramdrake 12:57, 17 October 2006 (UTC)

Sorry, but I think you've got it wrong. The family variance becomes meaningless when comparing two groups, because members of both groups suffer, or benefit, from family variance, hence nullifying the influence. And even if the family variance was greater it would be too much of an absolute statement to say that 'most' variance occurs within families, like it would be too much of an absolute statement to say that the jews are the 'most' intelligent race. --Zero g 13:12, 17 October 2006 (UTC)
No actually, it is not an absolute at all. The fact that there is a family variance for both Blacks and Whites doesn't nullify its role in the average B-W gap. Again, using Jensen's numbers, if you were to take two pairs of siblings of the same race, the average variance between each pair member (assuming non-twins) would be 12 points, and the average variance between the two pairs would be 17 points or so. The difference between B-W picked at random would be 20 points. Thus, you could say that of the 20 points between them, the greatest source of the variation is found within families, at the individual level. It is NOT taken to mean that within-family IQ variation is the greatest that can be found; obviously, it isn't.--Ramdrake 13:22, 17 October 2006 (UTC)
I see what you mean now, still it's oddly worded in the article and unclear to the readers. The statement also does not go in the case of a brother and sister with an IQ of 150 and 138, in which case most of the variance doesn't come from family variance with strangers. It's also known there's an IQ gap of 5 points between males and females, which can make sibbling studies like these unreliable if you interpret them freely and don't take into account that this difference will have a lesser impact on strangers. So I'd like to see a source claiming most of the variance is from family variance, otherwise the statement is original research. --Zero g 15:15, 17 October 2006 (UTC)
Jensen's (1998) book The g Factor includes an ANOVA on a sample of Black and White Californians. A fuller description is in the text here: Race_and_intelligence#Scope. The numbers make sense if you think in terms of (a) the high heritability of IQ would make siblings quite dissimilar (they are only ~50% related), (b) that the average difference between measurements of the same person is going to be ~4 points, and (c) that for the average trait only 15% of the variation occurs between groups and 85% within, but for IQ there's actually more between group variation. --Rikurzhen 15:40, 17 October 2006 (UTC)
The difference of ~4 would balance out if the sampling group is big enough. One issue however is that the IQ difference between males and females would increase the gap within families, while this would be balanced out in a large group of (black/white) strangers of mixed gender.
As I already said, to follow Wikipedia standards a citation is required of (preferably) an expert making this claim to not make this original research. Next I still think the current wording of the statement is confusing and should be improved. Out of curiosity, what's the average gap between 2 random white persons in the US? And asuming it's close to 16, should this be seen as blacks having a 4 points differing IQ, or as blacks having a 20% differing IQ? --Zero g 23:11, 17 October 2006 (UTC)
A few things: 1)Apart from maybe Rushton's latest studies, I seem to remember that most studies find no significant difference in IQ between men and women, just a much higher IQ variability in men (more geniuses and more idiots, very bluntly put). At this point, I would hesitate to put much credence in Rushton's findings, in the light of the consensus in the rest of the literature. 2)And no, you do not need an exact quote to meet Wikipedia's standard, you just need for the statement to be sourced from a reliable source, and that would be among others, the Neisser report from 1996 which makes this statement just about word for word. 3)Also, to answer your further question, I believe Jensen found that the average gap between any 2 random people of the same race is 12 points, not 16 points. 4)And the answer to the last question is neither: the difference in average IQ is what should be used, and that difference is estimated to be between 7 and 15 IQ points, depending on which researcher's literature you happen to read. The difference cannot be expressed percentage-wise, as that's not how the IQ scale works.--Ramdrake 23:25, 17 October 2006 (UTC)
Jensen (1998) states something like 'within-family is single largest source of variation'. The average diff btw 2 stangers is 17 points in Jensen's dataset upon which this is based. This can be computed directly from this formula: |\bar{x}| = \frac{2 \sigma} { \sqrt{\pi} }. Where |x| is the average difference and σ is the SD. If σ=15 then |x|=17. Re-reading Jensen (1998) the within family |x| he reports is 11, not 12, but 12 is the number you see more often. You can predict the |x| within families based on h^2, c^2 and e^2, but I forget the formula right now. Suffice it to say that because c^2~=0, e^2 is very small and h^2 is very large, you expect large differences between full siblings. Small diffs between siblings are expected when c^2 is large and h^2 is small. Siblings by definition come from the same social class and race. Thus, the within race, within social class |x| should be around the same as the sibling |x|. -- MF diffs balance out in all comparisons because |x| is the average diff between two people, and half of all such pairs are same-sex or opposite-sex, so long as families are not non-randomly sex-biased. --Rikurzhen 23:44, 17 October 2006 (UTC)
I recall Rushton tested college students, which might explain the results since all the idiots would have been filtered out of the sample group. Since it looks like a valid source is backing the claim the OR part is covered.
I'm still not convinced of the actual relevance and the presentation of these statistics, since it's possible to have the same variance both within and between races, while still having different means. --Zero g 00:43, 18 October 2006 (UTC)
That's true. The relevance according to Jensen is (to paraphrase) that IQ tests are unlikely to be mostly measures of "bias" or privledge or race or whatever if the measure something that mostly varies within familes and within social classes. He wrote "The pernicious notion that IQ discriminates mainly along racial lines... is utterly false." --Rikurzhen 00:53, 18 October 2006 (UTC)

to demonstrate how the average within family difference being ~12 makes sense, here's an snippet of R code that simulates 100,000 familes with 2 children each that matches expected criteria

n<-100000;
mu<-rnorm(n,mean=100,sd=10.7); # sd=10.7 if you want avg diff to be around 12
x<-rnorm(n,mean=mu,sd=10.7);   # sd=10.7 if you want avg diff to be around 12
y<-rnorm(n,mean=mu,sd=10.7);   # sd=10.7 if you want avg diff to be around 12
mean(abs(x-y))                 # average diff between siblings
[1] 12.07189 
cor(x,y)                       # correlation between siblings
[1] 0.4986535
sd(c(x,y))                     # overall standard deviation for pooled population
[1] 15.11559
mean(abs(sample(x)-sample(y))) # average diff between random strangers
[1] 17.06024

--Rikurzhen 00:03, 18 October 2006 (UTC)

Data used for bell curve image

What data was used for the Asian curve? As Flynn outlines in his book, Japanese and Chinese Americans actually have slightly lower IQs than whites. As for Asians in Asia, the American Psychological Association's Task Report on Intelligence says the following:

"It may be worth noting that the interpretation of test scores obtained by Asians in Asia has been controversial in its own right. Lynn (1982) reported a mean Japanese IQ of 111, Flynn (1991) estimated it to be between 101 and 105; Stevenson et al (1985), comparing the intelligence-test performance of children in Japan, Taiwan and the United States, found no substantive differences at all. Given the general problems of cross-cultural comparison, there is no reason to expect precision or stability in such estimates."

"Nevertheless some interest attaches to these particular comparisons: they show that the well-established differences in school achievement among the same three groups (Chinese and Japanese children are much better at math than American children) do not simply reflect differences in psychometric intelligence."

"the average IQ for African Americans was lower than those for Latino, White, Asian, and Jewish Americans (85, 89, 103, 106, and 113, respectively; Herrnstein & Murray, 1994, pp. 273-278). " (Rushton and Jensen 2005)
I believe there was some skepticism about the White-Asian IQ gap in the 80s and early 90s, but you don't see that in literature now AFAIK. Historically, the average IQs coming from East Asian populations have trended upwards, pulling away from Europeans, so this may account for some of the earlier disagreement. --Rikurzhen 22:01, 19 October 2006 (UTC)

Referencing within Wikipedia

RE: Reference 31 - 10-23-06

I'm a bit confused. I referenced from one Wikipdia article to another, and the replicated but condensed text taken from an article of the same volume to another should not need further attribution, as the validity of the referenced article should stand on its own. Nor should it create any issues regarding “copying,” especially since by definition no copyright applies to Wikipedia text.

I'm not seeing the problem here. Unless you site a specific issue, I'm at a loss to redress your concern.

With all due respect.


Kevin Murray 03:44, 24 October 2006 (UTC)

Kevin, referencing within Wikipedia is only as good as the material you reference. And, as I explained to you on your talk page, the material you added to the article from the William Shockley article is unfortunately unreferenced (that I could see) in the original article itself, so that's not a good reference. What you need is to find the reference from which this material was originally taken.--Ramdrake 13:19, 24 October 2006 (UTC)

Referencing Shockley

Ramdrake has convinced me that there is a citation problem regarding the paragraph regarding William Shockley; however, I believe that the problem lies with the quality of references at the Shockley article. I have placed a citation needed marker at that article, and within the the citation to the Shockley article here, I have placed a disclaimer questioning this issue. The section at Shockley has a reference under the title-header to a PBS article which substantiates the paragraph, but not all information there. However, I remeber the period of time and the issues; this all rings true but we need to dig up much more solid documentation than my feeble grey cells.

Cheers,

Kevin Murray 15:58, 24 October 2006 (UTC)