A 5 Minute Intelligence Test for Kids

Now, that's disappointing. Newsweek has a fascinating article there, but they missed the best part. Sure, a 5 minute test is interesting, and the correlations are too, but that is fascinating fact. Only 50% of the gifted students are gifted a year and a half later? That really makes you think. What about the other way around? How many students, not originally classified as gifted would be classified that way after a year or two? It also seems to help the claim that school dumbs you down.

I would hazard a guess that at 5, the kids who score high on these "IQ" tests correlates strongly to their environment at home. i.e., the ones with the most involved parents do best.

The child's mind is not properly developed yet, and the best way to continue to develop it is to challenge it at the fastest pace it can handle. That has nothing to do with intelligence. At all.

I call baloney on this. I want to see the peer-reviewed publication that says so. This is the kind of mental test that Galton and James McKeen Cattell did a century ago, and those tests were found to have no valid correlation with intelligence.

http://norvig.com/experiment-design.html

This further paragraph in the submitted article provides better information:

"The team wanted to evaluate several intelligence tests, including their own. So they recruited 77 gifted children through the Parents’ Association for Gifted Children in Switzerland. A previous intelligence test, taken about a year-and-a-half previously, had won them entrance to gifted primary schools. So how many of the kids still classified as gifted just eighteen months later? Only half, no matter what test was used. (And that was using a relaxed cut-off line, to account for standard deviations in testing.)"

This is in accord with many findings by many researchers in many places over decades: preschool IQ tests have remarkably poor reliability for predicting subsequent school-age IQ scores. By the way, the Newsweek blogger's terminology is incorrect. Where Po Bronson wrote, "to account for standard deviations in testing," he should have written, "to account for standard error in estimation," which is a different concept. A "standard error" is something different from a "standard deviation."

The later paragraphs somewhat rescued this casual blog post, but I fear that most news reporting on IQ testing is at the level of excessive credulity found in the earlier paragraphs. It takes time and effort to read and understand the better published literature on IQ testing

http://learninfreedom.org/iqbooks.html

and alas many journalists don't bother to make that effort. Even less do most bloggers bother to check their facts on this contentious issue before posting.

Piaget identified a number of intellectual stages that children go through at fairly predictable ages. Children transition between them during growth spurts. There is variance in when the growth spurts happen, and therefore when those mental leaps happen.

Most children transition from the "intuitive" to the "concrete operation" stages around age 7. Concrete operations include tasks like ordering objects in a logical sequence. This is EXACTLY what the "5 minute IQ test" is testing. The longer test is almost certainly testing a variety of different mental skills that are also enabled by the same transition. Hence the strong correlation.

Furthermore someone who is "gifted" at age 6 on a test like this almost certainly is gifted because he or she hit that growth spurt early. A year and a half later more kids have gone through the same transition, and having acquired those mental skills early hasn't necessarily translated into acquiring them better. That would explain why so many once-gifted kids aren't gifted 18 months later. (Of course the additional time with those skills does help, explaining the fact that many still are gifted relative to their peers.)

In short the results should be unsurprising to anyone who has studied child development.

Slightly OT:

I had hoped that our current economic situation would help obliterate the myth that our best and brightest work on Wall St., but apparently we still have some convincing to do.

A good intelligence test will predict which children will grow up to be doctors and engineers. If you want to find the future financiers, you should put them all in a room and see which kids start bullying the others.

IMO, the bigger issue is that we expect all children to be at the same learning development stage at any given point in their life and to respond equally to a given teaching approach. So, all 5 year olds get lumped into a particular curriculum, and then as 6 year olds move as a group to the next stage. Actually, I think the schools don't REALLY expect that,they just don't care to figure out a better approach.

Something is definitely fishy about this article. Line testing has a R of 0.99 with intelligence tests but intelligence tests themselves have a poor correlation after 18 months. Does that mean line tests are still 0.99 correlated after 18 months? It's hard to see a plausible explanation that fits this data.

"The scale, properly speaking, does not permit the measure of intelligence, because intellectual qualities are not superposable, and therefore cannot be measured as linear surfaces are measured." -Alfred Binet, 1905

Citation needed.

What if you just hold the cars facing each other so the lines are just short of touching? Then you can see the lines pretty much side-by-side.

Yes, I am smarter than a fifth grader.

Shenanigans.

"the two tests have a 99% correlation"

"they recruited 77 gifted children through"

[yes, it could be a rounding error]