IQ Testing

I found an interesting old post and conversation a while ago about a sad article in the New Yorker. The New Yorker article was about a young gifted child who committed suicide. The brief facts are that a young 14 year old boy committed suicide with the gun at his parents’ farm. He had been homeschooled because he was gifted, and his parents had had advice that school would do nothing for him, because he was too smart for school.

One of the commenters, Jason Smith, said:

“Unfortunately my conclusion is that much of the popular writing, including advocacy groups and programs, on the gifted is distorted to emphasize their exceptionality as much or more than the mainstream education movement distorts the facts to justify the cookie cutter educational model. “

This over identification of extremely gifted children is, in my inexpert view, largely for technical reasons around the IQ test.

The original IQ tests were fairly approximate. They started out being approximations of the idea that your IQ is the ratio of your “mental age” and your “actual age”, particularly if you were a child. So if you were 10 years old, but could largely do school work of a 15 year old, your IQ was 150. Then that got conflated into IQ tests being statistical measures, with a mean of 100 and a standard deviation of 16, and a normal distribution. The trouble is that those two things don’t actually go together. The ratio thing was only a way of helping to understand the concept, it was never particularly scientifically based; and it would be a miracle if the human population neatly fitted those two concepts together with such nice round numbers.

I’m cribbing from this article, which is mainly about the new SB V test.

So until two or three years ago, there was one test for testing IQs for gifted children (the others tended to have too low a ceiling – if they were too gifted, they got full marks, which didn’t help work out just how gifted they were) – the SB LM. It was old, and had a number of other issues (like cultural biases), but had discriminatory power (that is even when you got to the very smart kids, you could tell which one was more likely to be a genius). That could give results up to 200, neatly fitting the mental age ratio idea (a 10 year old with a mental age of 20 is almost conceivable – see Terence Tao as an example). But nearly everyone who wrote studied gifted children commented that there were more of them than you would expect. But instead of concluding that the test wasn’t calibrated in the way you would expect – either the distribution of the population wasn’t normal, or  the standard deviation of the population was bigger than everyone thought on that particular test – testers still categorised children tested with IQs of (say) 160 as 1 in 10,000. But actually, it was probably more like 1 in 500 to 1,000.

The most likely explanation for this (in my non expert view) is the Flynn effect. The Flynn effect essentially says that the IQ of the population has increased by 3 IQ per year per decade for the last 30 years or so.The SB LM was introduced in 1972, so it is 30 years old. So the mean IQ of the population has gone up from 100 to 109 in that time. Doesn’t sound like much. But that alone, assuming that the distribution changes uniformly, would increase the proportion of children scoring above 160 from 1 in 11,000 to 1 in 1,300. Put that together with the commonly held view that the distribution of IQs has fatter tails than a normal distribution (ie there are more people at both extremes than suggested by the standard measures), suggests that there are a lot more children with IQ’s above 160 than the statistics suggest.

So Stanford Binet introduced a new test – the SB 5. It was designed to fix a whole lot of problems around IQ testing, among them, the ceiling effects on previous IQ tests, and the weird distributions that seemed to suggest we had “an epidemic of geniuses”.

So the new SB 5 really has a standard deviation of 15. I’ve been trying to find a sensible comparison for that child with an IQ of 160 on the old tests. The answer appears to be an IQ of somewhere around 135 – 145. (also see Table 7 here). But many people don’t like that, as our previous genius now appears to be merely smart. And if you are a parent with a gifted child, it’s easier to get your school to pay attention if you’ve got a very big IQ number to wave at them, as this kind of subtlety, particularly something involving standard deviations, is not what your average primary school teacher is trained to understand.

So anybody who has been tested on the old SB LM tends to rubbish the new SB V as incorrect – sometimes because it doesn’t give the old ratio IQ answer, and sometimes because it implies that the high IQ they have been tested with isn’t as rare as you might think.

Miraca Gross, studies 15 exceptionally gifted children in her book Exceptionally Gifted Children. They have been identified as children with IQs (on the SB LM test) of 160 or over. And they certainly seem to do a lot better with radical intervention than they do if teachers try to teach them pretending that they are no different to other children. So even if the old IQ tests overstate the rarity of gifted children at this level, they still identify children who will do better with radical intervention (such as skipping several years of school) that feeds their hungry brain at a faster rate than your average child.

Do I have any conclusions? Not yet. The main one is that just writing this out has made me even more uncomfortable with pushing myself down a path of caring about the result on an IQ test of either of my children. But yet; if it helps me advocate for them to get a more fulfilling and interesting education than they might otherwise get; is that worth while? Or will it just turn me into that nightmare pushy parent who I have rarely come across, but seems to exist just out of reach in all the stories about gifted children?

 

Advertisement