I think you need to remove the unecessary "us" "our" "we" when they suggest several authors. If you say "we can see" it's fine because several single authors use that and I see the "we" as involving both the author and reader(s). But otherwise, avoid that.
I used "we" because there originally were -- or, at least, were supposed to be -- two authors. Replacing "we" with "I" would look silly, so I would have to rewrite a number of sections. If you definitely can not accept "we" let me know. If so, I will have to rewrite a number of sections or "locate" a co-author.
Perhaps you can explain a little bit how you arrived at this value (e.g., instead of dividing by the sum of the SDs, you SQRT the product of the two SDs). But more importantly, is it a common practice to convert CD into Cohen's d ?
The formula is: CD = (Mb-Ma)/(SDa+SDb), where CD=1.28. I assume that the population SDs are about the same. So we have: 1.28 = (Mb-Ma)/2SD, which is:
Mb-Ma = 2.56 SD. No?
and found, based on 71 cranial measures, an average Caucasoid-Negroid CD of less than 0.5; of the traits, only 15% had a CD above 1.0; these results were consistent with those reported by Sarich and Miele (2004), who ended up emphasizing the largeness of the differences relative to those between other primate sub-specific groups.
"largeness of the differences relative to" : which differences ? Maybe you should use "these" instead.
Craniometric -- I will use "these".
This found Hs value corresponds to an F/Gst value of ~0.05 based on our regression line using 24 subspeciated species
I would like you to explain this sentence. It's so obscure...
I rewrote this as: "The average Hs value was 0.72 (and so the maximum possible Fst would be 0.28). When Hs is plotted by F/Gst for the sample of 24 species, a Hs value of 0.72 predicts a F/Gst value of 0.05. Using Heller and Siegismund's (2009) regression line based on 43 species the same result is found. "
What I meant is that among non-human species a Fst of 0.05 is what one would predict given a Hs of 0.72
In general, I find that section hard to read. The whole thing involved answering the question of whether THR meet the 5 criteria listed by Mayr & Ashlock. You come to the conclusion that we cannot say that THR meet the criteria but that we cannot reject THR either, based on these arbitrary criteria. So, what's next ? Was that just to say that scientists shouldn't use these kind of arguments to reject THR ?
It was to give a complete discussion of the matter. Nothing is next. Do you want me to rewrite it? Yes, it was a lengthily discussion, but I was trying to show that no matter how you look at the matter the subspecies issue is undetermined.
(By the way, concerning the formula you applied to convert Fst into d values, I will appreciate if you cite a reference.)
I didn't use "fst", rather I used "%variance", specifically Eta-squared -- see: http://en.wikiversity.org/wiki/Eta-squared
The formula is given in: Cohen J. (1988). Statistical Power Analysis for the Behavioral Sciences (2nd ed.). You can use the "converting effect sizes" excel (at the bottom) here: http://www.stat-help.com/spreadsheets.html
. Note, I used typical interpretation guidelines. See: http://stats.stackexchange.com/questions/15958/how-to-interpret-and-report-eta-squared-partial-eta-squared-in-statistically
. An Eta-squared of 14% is about 0.81 SD.
That said, this sentence is obscure :
That is, to the extent that congenital race differences are questionable, they are questionable not because race is an arbitrary social construct, but because (i) race is delineated in terms of a specific variable (typically geographic ancestry) and (ii) it is questionable that the trait in question would vary congenitally by this variable.
" Of course, one can only argue against the existence of socially important genetic differences between biological races if one recognizes and defines biological race. The next step is to actually show that the said differences do not exist. It is not infrequently argued, though, that sociological races could not differ on account of being "socially" and not "biologically" constructed (e.g., Alland (2004)). As discussed in section II, this conclusion is an obvious non sequitur. Showing or claiming that sociological races do not correspond with biological ones provides no leverage in arguing that there are no significant genetic differences between them. This is because there obviously are socially important genetic differences between many socially constructed groups, such as social classes. "Social construction", then, necessarily can not be inconsistent with "genetic differences”. Groups obviously can be socially constructed around genetic differences. Knowing nothing else, one would treat socially constructed groups as arbitrarily constructed ones. The a priori probability that any differences between arbitrary constructed groups is a function of the heritability of the trait in the population (see Tal, 2009 for proof concerning individual differences).This is merely a restatement of the behavioral genetic default according to which group differences should be, knowing nothing else, assumed to have a genetic basis since genes condition differences between individuals within the population at large. To argue against this probabilistic reasoning, one must provide evidence that ethnic groups are not arbitrarily constructed -- that they are constructed in a manner such that there are no such differences or such that the existence of such differences is a priori implausible.
Were we to socially construct groups and then select ones for which there were appreciable differences in some highly heritable trait, it would be more likely than not that the between-group differences in that trait would be partially congenital. In the case of skin color: Argentinians versus Colombians; North Hemispherians versus South Hemispherians; Theravada Buddhists versus Mahayana Buddhists; wealthy Mexicans versus poor Mexicans. In some instances the group differences would be completely unrelated to genetics. Whether the majority of the pairs would exhibit genetic differences would depend on the precise population heritability estimate of the trait in question. But it is clear that social constructionism per se is not inconsistent with between-group genetic differences. Thus, when it comes to “race”, what is the argument? It can not merely be that races are socially constructed. It would have to be that races are socially constructed only around non-behavioral differences. But this begs the question. "
Not just because of the unecessary use of "us" (or "ours") but more because the sentence suggests you have mentioned Braces' argument earlier, which wasn't the case, as far as I can see. I also believe the following sentence is badly written :
"Another way to escape the reverse of the "too little variation" argument would be to reiterate a version of Loring Braces’ (1999) argument. According to this, since each population has an equal ability to use language and to develop culture they must necessarily not differ in behavioral traits such as intelligence. This type of argument, of course, is ridiculous when applied in context to normally distributed traits because within each and every population innumerous subpopulations exist which do differ in these traits. If these subpopulation which differ can exist then populations which exist can differ. Worse, it is already known that human populations do differ in the said traits; the debate is over “why” not “whether”. In short, the “too little variance” argument cannot be salvaged. To the extent that it is deemed to be valid it clearly fails to support the position in defense of which it has been enlisted."
This problems occurs quite frequently throughout the entire text. And not just section 4.
Point the passages out please and I will rewrite them.
The argument is that the results depends on sampling. Spencer (The unnatural racial naturalism, 2014) on the other hand said it does not do any harm to this view.. It seems to me that the results from STRUCTURE may remind us of what you can get from factor analysis, including CFA, where the modeled latent factors depend on the sampling of subtests. Yet, no one would never say that there is no speed factor because it fails to emerge when the battery of tests is composed of 7 verbal, 3 performance, 2 memory and 1 speed subtests.
Yes, there are at least two different arguments. You are correct that I didn't directly discuss this issue. How about:
"Box 4.1. Critiques of Unsupervised Cluster Analysis
A number of critiques have been made against the use of unsupervised cluster analysis to objectively delineate racial classifications. One is that cluster-analyses can not establish a correct level of genetic granularity; thus any level of analysis must be subjectively chosen. Another is that global genetic and morphological data bases such as 1000 Genomes, HapMap3, and W.W. Howells’ Craniometric Data Set are based on biased samples, ones which were collected with traditional racial classifications in mind. A third is that cluster analysis outputs heavily depend on the population data inputted; when one inputs data from different sets of populations one gets different results.
Regarding the first, it is correct that cluster-analyses can not establish a correct level of genetic analysis. This is as one would expect since there is no such level. One can look at genetic propinquity on a broad continental level or on a small regional one. In context to race, this has always been recognized. The same consideration holds with respect to taxonomic categories. For example, the level of genetic analysis that corresponds to "genus" is no more correct than that which corresponds to "species". Nonetheless, unsupervised cluster-analyses can provide objective grounds for preferring one level of intrapecific analysis to another given some notion of what makes for an ideal race, for example: degree of clusteredness, amount of genetic differentiation, and the genetic coherence of the divisions picked out. When unsupervised cluster analysis is run which automatically picks out a "best" level of analysis (e.g., fineSTRUCTURE and DAPC) or which generates results which allow one to do so (e.g., STRUCTURE), the TRCL has generally been shown to cut out divisions which are preferable given these pre-specified criteria (Dienekes, 2005; 2014). Regarding the second argument, one can not directly refute a hard form of this, according to which subtle and difficult to detect biases inevitably shape sampling decisions. One can only request that the argument is applied equally with respect to other scientific endeavors. If done, this would unveil the epistemic nihilism on which it is based. In response to a moderate version of the same argument, it can be pointed out that multiple data bases developed by different research teams for different purposes generate similar results. Moreover, the results are as one would expect given known historic geographic barriers to movement, for example, the presence of large deserts, mountain ranges, oceans, and so forth. The third point is superficially correct; the population data which one inputs into a cluster analysis program flavors the results. But this point only works as an argument against the TRCL insofar as one maintains, in accordance with the second argument, that global data bases are riddled with sampling-bias. As discussed above, there are good reasons for concluding otherwise."
It's curious you're citing that study, since it says that environments cause genetic effects by way of mutual causality.
What Flynn emphasized and what his model allows for are two different things.
frequency differences in specific behaviorally associated alleles can (statistically) explain some of a number of interesting national, regional, and ethnic sociocultural differences
I will appreciate if you say explicitly if these relationships are strong, or just small/modest.
This is tricky because the effect sizes of the alleles themselves explain very little. The cross population frequency differences, though, can explain a lot. Maybe I should clarify this. I am not sure how I can be explicit because it depends on the traits in question. I cited a bunch of different studies. I will see what I can do though. How about the following footnote:
"The effect sizes of the alleles themselves explain little; but the patterns of frequency differences can explain a substantial portion of the phenotypic ones in the sense that the squared correlation between allelic scores based on multiple alleles and phenotypic scores is often moderate to high for some traits.
Section IV-L :
readers are cautiously referred to Lynn (2008) (cautiously because the work is badly in need of updating 20)
20 We make this claim based on our own investigation of the data.
What's the use of saying this when the reader has no way of checking this out ? Also, remove the "our".
You are the one that asked me to provide a reference. Lynn's work is
in need of updating. My reference is my own judgment based on my own analyses.
Generally speaking, global variance in intelligence exhibits a north south clinal pattern for indigenous populations. This pattern does not hold when it comes to recent (post-1500) global migrants (e.g., Europeans in Australia, N.E. Asians in Brazil, and S.S. Africans in North America). Since there is a high correlation between skin color/reflectance and measured ability (typically around 0.9), the distribution of the two can be thought of similarly.
I don't understand the whole thing, i.e., in what way the last sentence relates to the first/second sentence(s).
The skin color distribution is clinal just like the IQ distribution. I changed this to: "Generally speaking, global variance in intelligence exhibits a north south clinal pattern for indigenous populations. This pattern does not hold when it comes to recent (post-1500) global migrants (e.g., Europeans in Australia, N.E. Asians in Brazil, and S.S. Africans in North America). Since there is a high correlation between skin color/reflectance and measured ability (typically around 0.9), the distribution of the two traits can be thought of similarly (i.e., both are clinal)."
Secondly, the migrant and intranational data, while generally consistent with a non-trivial hereditarian hypothesis, is not compelling; groups which one would expect to do poorly not infrequently fair well.
Can you cite a reference or two ?
In the text, I cite De Philippis (2013) and Fuerst (2014) which disccuss the matter. I added:
"groups which one would expect to do poorly not infrequently fair well (e.g., Fuerst (2014))"