Infant IQ Tests Predict Scores in School

Monday, April 14th, 2014

Twenty-five years ago, Gina Kolata of the New York Times reported on infant IQ tests that predicted later school performance:

The tests, which attempt to measure what babies remember, are based on the assumption that they will be more interested in stimuli they have not previously encountered. One test involves showing infants photographs or pictures and measuring how long they look at them. In theory, babies will look at a new stimulus for a longer time than one they remember having seen before. Babies that are likely to be below average in intelligence will remember fewer of the stimuli they have seen before.

Working independently, Dr. Bornstein and Susan Rose of Albert Einstein College of Medicine in New York have found that a baby’s performance at age 4 months and 6 months on tests of visual memory correlate with I.Q. at 4 and 6 years of age. They used tests that they developed themselves but that are similar to those developed by Dr. Fagan. The predictions are independent of the parents’ education and income group, which also are correlated with I.Q.

The original article is full of fears that such tests might give already-advantaged children even more advantages.

Kenneth Change, also of the Times, revisits the original article:

For the most part, the validity of the Fagan test holds up. Indeed, Dr. Fagan (who died last August) and Dr. Holland revisited infants they had tested in the 1980s, and found that the Fagan scores were predictive of the I.Q. and academic achievement two decades later when these babies turned 21.

“It’s really good science,” said Scott Barry Kaufman, scientific director of the Imagination Institute at the University of Pennsylvania and the author of “Ungifted: Intelligence Redefined.”

But Dr. Fagan’s hope for widespread screening of infants has not come to pass. “There are some centers that have it,” Dr. Holland said. “It never came to be the kind of thing where it’s widely available.”

The trend is perhaps in the other direction, away from dividing young children by I.Q. and its surrogates out of concerns that the labels become self-fulfilling prophecies. Private schools in New York City, for example, have agreed to abandon intelligence tests for 4- and 5-year-old applicants.

This stands out:

For the last decade of his life, Dr. Fagan was unexpectedly drawn into the “genes versus environment” debate over intelligence after he found that babies from widely different cultural backgrounds performed equally well on his test. That, he argued, undercut the argument for a biological basis for the stark “achievement gap” between white and black children, or rich and poor.


  1. Guy says:

    …babies from widely different cultural backgrounds performed equally well on his test.

    In the same percentages? If not, “Individual smart baby is smart regardless of race,” isn’t all that controversial is it?

  2. Gwern says:

    Guy, it’s hard to say without reading the original research. There’s a lot that could be going on, including insufficient power. I mean: typically, it’s harder statistically to conclude anything group differences than to conclude something about the whole sample; that is, one would expect it to be easier to confirm that the infant test predict future school scores than to confirm that, say, East Asian babies score higher than white babies (either on the original test or at the followup).

    So what could easily be happening here is that the data turns in the usual Asian > white > hispanic > black ranking but there’s not enough babies in the sample to find group differences at p < 0.05, and the researchers are simply ignoring the point-values because they're not forced to admit the differences are there.

    This is completely consistent with the claims made in the article, and just one of the potential tricks or stories going on here.

Leave a Reply