A Bell Curve has that shape with random data. Anything that creates a bias creates a deviation from the nominal which changes the shape of the curve. If the bias created a new value with a different nominal value and its own sharp standard deviation the curve would have two peaks with one likely to be greater than the other.
A Bell Curve has that shape with random data. Anything that creates a bias creates a deviation from the nominal which changes the shape of the curve. If the bias created a new value with a different nominal value and its own sharp standard deviation the curve would have two peaks with one likely to be greater than the other.
It's been years since I read that book, but I think they observed that. The question then becomes, what caused that bias? They did a bunch of controlling for various factors deemed to isolate the cause of the bias to be average IQ for the mean values of the "races" being compared. It looked good. That is standard stuff in methodology, but the question remains, were there factors they did not control? With something as complex as humans and all influences, no doubt they didn't control all biases that lead to a result other than innate "race" attributes, so the question becomes a matter of their significance.
A larger issue in my mind is, is there a single aggerate value for intelligence that is a valid concept? A savant might not be able to tie their shoes but be able to do some "thing" at a level beyond the capability of an IQ genius. Scaled back toward norms there are certainly people with lower IQs than me who I would hire to perform tasks that they have expertise in that I don't. Smarter about what? If the test itself has a bias to favor a certain group, a comparison of the groups is bogus, even if it could mean something valid if unbiased. In my opinion it probably doesn't.
A Bell Curve has that shape with random data. Anything that creates a bias creates a deviation from the nominal which changes the shape of the curve. If the bias created a new value with a different nominal value and its own sharp standard deviation the curve would have two peaks with one likely to be greater than the other.
It's been years since I read that book, but I think they observed that. The question then becomes, what caused that bias? They did a bunch of controlling for various factors deemed to isolate the cause of the bias to be average IQ for the mean values of the "races" being compared. It looked good. That is standard stuff in methodology, but the question remains, were there factors they did not control? With something as complex as humans and all influences, no doubt they didn't control all biases that lead to a result other than innate "race" attributes, so the question becomes a matter of their significance.
A larger issue in my mind is, is there a single aggerate value for intelligence that is a valid concept? A savant might not be able to tie their shoes but be able to do some "thing" at a level beyond the capability of an IQ genius. Scaled back toward norms there are certainly people with lower IQs than me who I would hire to perform tasks that they have expertise in that I don't. Smarter about what? If the test itself has a bias to favor a certain group, a comparison of the groups is bogus, even if it could mean something valid if unbiased. In my opinion it probably doesn't.