''The Bell Curve'' inflamed readers when it was published three years ago by arguing that economic and social success in America had become largely a matter of genes, not education, environment or other factors over which society might exert control. The chilling genes-are-destiny thesis, laced with racial overtones, was greeted with furious criticism. But much of the initial criticism was ill informed and driven by ideology.

It could hardly have been otherwise. The book's authors, Richard Herrnstein and Charles Murray, did not release their statistical findings -- the only important original contributions in the book -- for formal review by scholars before publication. Their runaround obstructed response by other social scientists, who needed time to appraise hundreds of pages of statistical analysis. Now, three years later, scholars have caught up, shattering the book's core claims.
. . . [T]he book's evidence is riddled with mistakes. Two stand out.

The first error flows from biased statistics. The book tries to determine whether I.Q. or family background is a better predictor of success. I.Q. is easily measured. But family background is not. The authors' simplistic index incorporates parental income, education and job prestige, but leaves out numerous components of a child's upbringing.

That creates a statistical mirage, or bias, because statistical tests inevitably underestimate the impact of factors that are hard to measure. Mistakes in measuring family background obliterate the ability of statisticians to detect its impact on future success. Thus, as James Heckman of the University of Chicago has convincingly argued, the book's finding that family background is a weak precursor of success reflects its biased methods rather than the workings of American society.

Also compelling is evidence about the second notable error -- that the authors' measure of intelligence is by no means immutable, as their thesis requires. Prof. Derek Neal of the University of Chicago and Prof. William Johnson of the University of Virginia have shown that scores on the measurement used by Mr. Herrnstein and Mr. Murray, the Armed Forces Qualification Test, depend on how much schooling individuals have completed. Put simply, the more students study in school, the better they do on the test. So what the authors call immutable intelligence turns out to be what others call skills -- indeed, teachable skills.

This mistake turns the message of the book on its head. Instead of its sighing surrender to supposed genetic destiny for poor children, there's a corrected message: Teach them.

Andrew Sullivan remains a shameful figure in our public discourse.