Racial IQ gap narrows: blacks gain 4 to 7 points on whites.

A statistical analysis of four national intelligence tests indicates that the difference in scores between blacks and whites decreased by about a third between 1972 and 2002. The findings challenge a century-old argument that the racial gap in performance on IQ tests is primarily genetic and therefore invulnerable to social change, say the researchers who performed the new study.

They examined data that have only recently become available to researchers, says William Dickens of the Brookings Institute in Washington, D.C. Using test results from a random distribution of people in the United States, he and James R. Flynn of the University of Otago in New Zealand tallied the increases in IQ scores of blacks and whites over 3 decades. Each of the four tests analyzed included two or three groups of people that took the test at different times.

Previous measures of the intelligence gap, which had used localized populations tested only once, found blacks 15 to 18 IQ points lower than whites, Dickens says.

In the new analysis, all four tests reflected a similar gap in 1972 but indicated that blacks have since gained ground in IQ.

"The whole distribution of black cognitive ability is moving up relative to whites," says Dickens. "There's no reason to believe [the gap] isn't going to get more narrow as we move forward and as measures of social equality continue to improve."

Neither increased rates of mixed ancestry nor changes in test content explain the narrowing gap, Dickens and Flynn argue in the October Psychological Science.

Last year, a review by J. Philippe Rushton of the University of Western Ontario in Canada and Arthur Jensen of the University of California, Berkeley concluded that intelligence is determined predominantly by genetics. The researchers argued that the IQ gap had held steady for a century, despite social-equality efforts, and noted that studies of adopted children and twins have attributed 80 percent of the gap to genetics.

The new data, by contrast, instead indicate that environmental factors contribute greatly to IQ scores, says Flynn. "It's exciting to show that the gap isn't written in the stars,' he says.

But the new findings contain many holes, Rushton and Jensen maintain in a rebuttal published with the study. For example, Dickens and Flynn "cherry-picked" their results by leaving out four tests that don't support their conclusions, Rushton says.

Dickens answers that the tests that he and Flynn omitted were less reliable than the ones that they included.

A 4-to-7-point closure in the gap would be "truly astounding" says Linda Gottfredson of the Delaware-Johns Hopkins Project for the Study of Intelligence and Society.

However, the conclusion "just doesn't fit" with data showing that in the past 30 years, U.S. black children between the ages of 9 and 17 have approached white children in reading scores but not in science and math, she says. If general intelligence, as measured by IQ tests, were really rising, scores would have increased in all subjects.

A fundamental problem with studies of IQ-test results is that "nobody really knows what intelligence is,' says psychometrician Peter H. Schonemann of Purdue University in West Lafayette, Ind.

The new findings "won't change that many minds;' says Doug Detterman of Case Western Reserve University in Cleveland, who is editor of the journal Intelligence. More important than measuring a gap, he says, is identifying the specific environmental factors or genes that contribute to intelligence.