>>12992810No he hasn't, He used faulty statistics to debunk a straw-man. If you actually fell for it, good job mid-wit.
His first point; The correlation between IQ and SAT scores are linear up to 100, and non predictive preceding that point, Therefor IQ isn't a good measure of 'intellectual capacity/mental powers' (Talib's Wording, not mine). If IQ=SAT, they should map out onto each other linearly, the fact that they don't in the graph he presents, but rather logarithmic-ally, is just the fact IQ and SAT scores are linear up to 100, and non predictive preceding that point. this plus statistical noise (which will always be there) creates the shape of the distribution. (this is what he is arguing, not me)
This is why he is wrong. His data for IQ comes from the ASVAB (the military IQ test) which has a IQ ceiling of 130, and the data for the SAT scores, by definition are already from college bound kids that are predicted to be smarter on average and distribute a heavy positive skewing, because there is an bottom IQ threshold for going to College. by correlating something that is evenly distributed, and something that his heavily, positively skewed will, by its nature, yield a logarithmic association of the two. It is this plus statistical noise (that will always be there) that creates the shape of the distribution (not muh <100 IQ = correlation >100 IQ = no association).
pic rel. is from his article. The red shows a made up model by Talib, which simulates how statistical noise + IQ and SAT scores being linearly correlated up to 100, and non predictive preceding that point, can create the a distribution similar to the bottom graph (whats funny is that it doesn't even look like the bottom graph so his point is stupid anyway. The bottom just looks like a logarithmic association).
The bottom graph is the raw data from actual people.
Part 1
