I’ve had a few discussions with Grey Enlightenment on this blog, regarding construct validity. He has now published a response piece on his blog to the arguments put forth in my article, though unfortunately it’s kind of sophomoric.
He calls himself a ‘race realist’yet echoes the same arguments used by those who oppose such realism.
1) One doesn’t have to believe in racial differences in mental traits to be a race realist as I have argued twice before in my articles You Don’t Need Genes to Delineate Race and Differing Race Concepts and the Existence of Race: Biologically Scientific Definitions of Race. It’s perfectly possible to be a race realist—believe in the reality of race—without believing there are differences in mental traits—‘intelligence’, for instance (whatever that is).
2) That I strongly question the usefulness and utility of IQ due to its construction doesn’t mean that I’m not a race realist.
3) I’ve even put forth an analogous argument on an ‘athletic abilities test’ where I gave a hypothetical argument where a test was constructed that wasn’t a true test of athletic ability and that it was constructed on the basis of who is or is not athletic, per the constructors’ presuppositions. In this hypothetical scenario, am I really denying that athletic differences exist between races and individuals? No. I’d just be pointing out flaws in a shitty test.
Just because I question the usefulness and (nonexistent) validity of IQ doesn’t mean that I’m not a race realist, nor that I believe groups or individuals are ‘the same’ in ‘intelligence’ (whatever that may be; which seems to be a common strawman for those who don’t bow to the alter of IQ).
Blood alcohol concentration is very specific and simple; human intelligence by comparison is not . Intelligence is polygenic (as opposed to just a single compound) and is not as easy to delineate, as, say, the concentration of ethanol in the blood.
It’s irrelevant how ‘simple’ blood alcohol concentration is. The point of bringing it up is that it’s a construct valid measure which is then calibrated against an accepted and theoretical biological model. The additive gene assumption is false, that is, genes being independent of the environment giving ‘positive charges’ as Robert Plomin believes.
He says IQ tests are biased because they require some implicit understanding if social constructs, like what 1+1 equals or how to read a word problem, but how is a test that is as simple as digit recall or pattern recognition possibly a social construct.
What is it that allows individuals to be better than others on digit recall or pattern recognition (what kind of pattern recognition?)? The point of my 1+1 statement is that it is construct valid regarding one’s knowledge of that math problem whereas for the word problem, it was a quoted example showing how if the answer isn’t worded correctly it could be indirectly testing something else.
He’s invoking a postmodernist argument that IQ tests do not measure an innate, intrinsic intelligence, but rather a subjective one that is construct of the test creators and society.
I could do without the buzzword (postmodernist) though he is correct. IQ tests test what their constructors assume is ‘intelligence’ and through item analysis they get the results they want, as I’ve shown previously.
If IQ tests are biased, how is then [sic] that Asians and Jews are able to score better than Whiles [sic] on such tests; surely, they should be at a disadvantage due to implicit biases of a test that is created by Whites.
If I had a dollar for every time I’ve heard this ‘argument’… We can just go back to the test construction argument and we can construct a test that, say, blacks and women score higher than whites and men respectively. How well would that ‘predict’ anything then, if the test constructors had a different set of assumptions?
IQ tests aren’t ‘biased’, as much as lower class people aren’t as prepared to take these tests as people in higher classes (which East Asians and Jews are in). IQ tests score enculturation to the middle class, even the Flynn effect can be explained by the rise in the middle class, lending credence to the aforementioned hypothesis (Richardson, 2002).
Regarding the common objection by the left that IQ tests don’t measures [sic] anything useful or that IQ isn’t correlated with success at life, on a practical level, how else can one explain obvious differences in learning speed, income or educational attainment among otherwise homogeneous groups? Why is it in class some kids learn so much faster than others, and many of these fast-learners go to university and get good-paying jobs, while those who learn slowly tend to not go to college, or if they do, drop out and are either permanently unemployed or stuck in low-paying, low-status jobs? In a family with many siblings, is it not evident that some children are smarter than others (and because it’s a shared environment, environmental differences cannot be blamed).
1) I’m not a leftist.
2) I never stated that IQ tests don’t correlate with success in life. They correlate with success in life since achievement tests and IQ tests are different versions of the same test. This, of course, goes back to our good friend test construction. IQ is correlated with income at .4, meaning 16 percent of the variance is explained by IQ and since you shouldn’t attribute causation to correlations (lest you commit the cum hoc, ergo propter hoc fallacy), we cannot even truthfully say that 16 percent of the variation between individuals is due to IQ.
3) Pupils who do well in school tend to not be high-achieving adults whereas children who were not good pupils ended up having good success in life (see the paper Natural Learning in Higher Education by Armstrong, 2011). Furthermore, the role of test motivation could account for low-paying, low-status jobs (Duckworth et al, 2011; though I disagree with their consulting that IQ tests test ‘intelligence’ [whatever that is] they show good evidence that in low scorers, incentives can raise scores, implying that they weren’t as motivated as the high scorers). Lastly, do individuals within the same family experience the same environment the same or differently?
As teachers can attest, some students are just ‘slow’ and cannot grasp the material despite many repetitions; others learn much more quickly.
This is evidence of the uselessness of IQ tests, for if teachers can accurately predict student success then why should we waste time and money to give a kid some test that supposedly ‘predicts’ his success in life (which as I’ve argued is self-fulfilling)? Richardson (1998: 117) quotes Layzer (1973: 238) who writes:
Admirers of IQ tests usually lay great stress on their predictive power. They marvel that a one-hour test administered to a child at the age of eight can predict with considerable accuracy whether he will finish college. But as Burt and his associates have clearly demonstrated, teachers’ subjective assessments afford even more reliable predictors. This is almost a truism.
Because IQ tests test for the skills that are required for learning, such as short term memory, someone who has a low IQ would find learning difficult and be unable to make correct inferences from existing knowledge.
Right, IQ tests test for skills that are required for learning. Though a lot of IQ test questions are general knowledge questions, so how is that testing anything innate if you’ve first got to learn the material, and if you have not you’ll score lower? Richardson (2002) discusses how people in lower classes are differentially prepared for IQ tests which then affects scores, along with psycho-social factors that do so as well. It’s more complicated than ‘low IQ > X’.
All of these sub-tests are positively correlated due to an underlying factor –called g–that accounts for 40-50% of the variation between IQ scores. This suggests that IQ tests measure a certain factor that every individual is endowed with, rather than just being a haphazard collection of questions that have nothing to do with each other. Race realists’ objection is that g is meaningless, but the literature disagrees “… The practical validity of g as a predictor of educational, economic, and social outcomes is more far-ranging and universal than that of any other known psychological variable. The validity of g is greater the complexity of the task.”
I’ve covered this before. It correlates with the aforementioned variables due to test construction. It’s really that easy. If the test constructors have a different set of presuppositions before the test is constructed then completely different outcomes can be had just by constricting a different test.
Then what about ‘g’? What would one say then? Nevertheless, I’ve heavily criticized ‘g’ and its supposed physiology, and if physiologists did study this ‘variable’ and if it truly did exist, 1) it would not be rank ordered because physiologists don’t rank order traits, 2) they don’t assume normal variations, they don’t estimate heritability and attempt to untangle genes from environment, 3) they don’t assume that normal variation is related to genetic variation (except in rare cases, like down syndrome, for instance), and 4) nor do they assume within the normal range of physiological differences that a higher level is ‘better’ than a lower. My go-to example here is BMR (basal metabolic rate). It has a similar heritability range as IQ (.4 to .8; which is most likely overestimated due to the use of the flawed twin method, just like the heritability of IQ), so is one with a higher BMR somehow ‘better’ than one with a lower BMR? This is what logically follows from assuming that ‘g’ is physiological and all of the assumptions that come along with it. It doesn’t make logical, physiological sense! (Jensen, 1998: 92 further notes that “g tells us little if anything about its contents“.)
All in all, I thank Grey Enlightenment for his response to my article, though it leaves a lot to be desired and if he responds to this article then I hope that it’s much more nuanced. IQ has no construct validity, and as I’ve shown, the attempts at giving it validity are circular, and done by correlating it with other IQ tests and achievement tests. That’s not construct validity.