The subject line can be read as a statement and as a question.

What is wrong about most of the published discussions about IQ is that they jump to conclusions and policy implications without looking a crucial aspects of the research and models of intelligence. For a nice paragraph-length summary of the problem, see Stephen Pinker's new book "The Stuff of Language" at bottom of page 85.

Basically, intelligence is a fuzzy concept that gets defined in ways that simplify measurement. Harvard psychologist Howard Gardner believes there are six different types of intelligence. Most tests address only two or three aspects, such as verbal, mathematical, and sometimes spatial visualization, and they weight the results so as to boil a complex reality into a single metric.

The problem with applying the metric to groups (such as supposed racial or ethinic groups) is that the results of testing form a curve that is usually bell-shaped. The curves for two groups might be tall and humpy with little overlap, or wide and low with lots of overlap. I believe the results of intelligence tests for different groups have lots of overlap. We say group X is smarter on average than group Y if the middle of the bell curve for X is to the right of the middle of Y's curve. Yet, the averages may differ by a small amount compared to the width of either curve. Thus many of the members of the "smart" group have lower scores than many of the members of the "less smart" group.

On this basis, people with a racial theory, or a policy axe to grind like to ignore the detail and use the supposed results to support their own agenda. The statements made by people quoted in the NY Times article on the reaction to Will Saletan's piece demonstrate this.

It is bad to base policy on group IQ averages for both practical and moral reasons, particularly because the tests themselves are almost always defective to some degree in eliminating cultural bias that gives some groups an advantage and penalizes others (such as the poor and immigrants from cultures that differ from the mainstream culture in which the test is based.)