Logo IMG
HOME > PAST ISSUE > Article Detail


Rising Scores on Intelligence Tests

Test scores are certainly going up all over the world, but whether intelligence itself has risen remains controversial

Ulric Neisser


In the countries where IQ scores are rising, people are also staying in school much longer than their parents and grandparents did. Could sheer duration of schooling be responsible for the gains? This hypothesis is plausible because the general effect of schooling on IQ is very well established. Many studies indicate that children who do not attend school for one reason or another score lower on the tests than their regularly attending peers. One especially unfortunate example of that principle appeared in the 1960s, when some Virginia counties closed their public schools to avoid racial integration. Compensatory private schooling was available only for white children. On average, the African-American children who received no formal education during that period fell back at a rate of about six IQ points per year.

Fortunately, such episodes are rare. School attendance through the elementary grades is virtually universal in all modern industrialized countries. In the United States, for example, more than three-quarters of the population goes on to finish high school. This universality makes it difficult to separate the contributions of age and schooling to mental development, because the "average eight-year-old" is an eight-year-old who has been in school for two or three years. Some separation is possible, however, because admission to first grade in most school systems is governed by an arbitrary age cutoff, such as six years old by September 1 of a given year. This practice ensures that the children in any given grade will vary in age by up to a year and that there will be children in different grades who are very nearly the same age.

In 1987, Sorel Cahan and Nora Cohen of the Hebrew University in Jerusalem took advantage of these birth-date distributions in an ingeniously designed study. They administered 12 different brief tests—presented together in a single session—to some 10,000 fourth-, fifth- and sixth-grade children in the Jerusalem schools. Using a complex statistical analysis based on birth dates and school admission, Cahan and Cohen compared the effects of a year of school (controlling for age) with those of a year of age (controlling for school) on each test separately. As one might expect, schooling mattered more than age for every test of verbal or numerical skills. More surprising, perhaps, is that schooling also made a contribution—albeit smaller—to performance on several nonverbal tests of abstract and visual reasoning. One of those nonverbal tests consisted entirely of items from the Raven.

Despite those data, schooling is not an altogether satisfactory explanation of the secular rise in test scores. For one thing, elementary school children tested with the WISC show gains comparable to those of adults who take the WAIS. Since elementary education was already universal in the 1930s, the WISC gains cannot be attributed to increased years of schooling. Another argument is based on further analysis of the Dutch Raven data. Flynn reports that grouping the subjects by educational level makes very little difference: The gains appear almost undiminished in each such group considered individually. Finally, the effects of schooling do not fit the overall pattern of test-score gains; schooling affects tests of content more than tests of reasoning, and the rise in test scores shows exactly the opposite pattern.

comments powered by Disqus


Of Possible Interest

Computing Science: Belles lettres Meets Big Data

Feature Article: Engines Powered by the Forces Between Atoms

Technologue: Quantum Randomness

Subscribe to American Scientist