A reader sent the following email:
Good evening, sir! I’ve been enjoying your blog and have a rather random question that I’d love to hear you weigh in on. I recently took a self-scoring IQ test from a book written by the late Victor Serebriakoff. I am curious as to how accurate the scores are. The publishing date was 1996 so I’m wondering how much this would distort the results. A number of online reviewers claim the score they received on this test is identical to the score they received on their WAIS. Would the scores necessarily be grossly inflated? Or could they be relatively accurate? The test consists of 300 questions with subtests for verbal, spatial, and numbers. Just curious to hear your take on it. Thanks for taking the trouble to read this meandering email!
I don’t have a copy of this test so all I can do is provide a meandering speculative reply:
Given Victor’s credentials as the head of Mensa, combined with the fact that the test had 300 questions of diverse type, I would guess it’s a very valid measure of intelligence.
But whether the scores are inflated is hard to say. I’ve seen several self-scoring IQ books in my life and none of them have ever explained how they converted the raw scores into IQs. And in some cases, they don’t even adjust for age, treating all adults as a single age group.
If Victor had done a quality job norming the test, then I’d expect him to have discussed it in the book. If he didn’t, then he may have just used a convenience sample (friends, acquaintances) and since he’s a high IQ man, his circle of acquaintances are likely above average. If so, the test might actually underestimate your intelligence.
On the other hand, the test is over 20 years old, so the Flynn effect might have caused your score to become inflated by as much as 6 points. Although, when tests are normed on the entire country (instead of just the white population) demographic changes can negate or even reverse the Flynn effect.
I also wonder how this test arrived at overall IQ scores since you say it measured several cognitive domains. If it merely averaged your scores on all three domains, it may have underestimated your overall IQ, since for example only 5% of the population maybe smart enough to average in the top 10% across three different abilities, so such people are perhaps in the top 5% (IQ 125+) even though they average in the top 10% (IQ 120+).
A reader sent the following questions by email:
Also, what other traits determine the variation in academic achievement. Achievement and IQ only have a .5 correlation.
Just a hypothetical- what if you come across a vocab word or an information subtest fact a couple days after the test?…
Depends what you mean by academic achievement. If you mean scores on an academic achievement test (which measures your total academic knowledge), the correlation is much higher than 0.5 in the general U.S. population.
If you mean the correlation between IQ and academic success (i.e. school grades, highest level of schooling) then it’s about 0.57 in the general U.S. population, although it used to be 0.7.
The other major trait that predicts academic success is conscientiousness.
Beyond those two major traits are probably a a whole bunch of minor traits, which on their their own explain very little, but collectively explain quite a bit. These might be things like physical heath, conformity, how much your teachers like you, how much you like school, whether you’re a night person or a day person (as commenter Mug of Pee once mentioned) etc.
It’s also worth noting that verbal IQ better predicts academic success than performance IQ, even holding general intelligence constant.
Of course the term “predict” might be a bit of a misnomer. Correlation != causation. There’s evidence that staying in school longer might raise IQ (though it’s unclear if it’s actually raising intelligence itself)
As for your second question, what you know outside the testing room is irrelevant. As David Wechsler noted, the tests are designed to measure your intelligence under a fixed set of standardized conditions. Once you leave that controlled environment, you’ve left the test.
A reader sent the following question by email:
Pumpkin, do most people on processing speed tests go as fast as they possibly can?
Yes, I think they do. Although the term processing speed is a bit of a misnomer. These subtests have among the weakest correlation with chronometric ability. Ironically, it’s the untimed power tests that often enjoy the highest correlation with chronometric ability. People in the field are very familiar with the slow looking nerd who is brilliant at solving really deep problems. Such individuals do relatively poorly on tests of psychometric speed, but extremely well on tests of chronometric speed.
Psychometric speed tests are poor tests of general intelligence, but they seem good for measuring the intelligence of otherwise smart people who just can’t adapt.
I’ve had a couple friends who did well on most g loaded tests, yet I did not consider these friends smart because they couldn’t adapt situations to their advantage in real life. Both of these friends did extremely poorly on the processing speed subtest of the WISC-R.