**Primaflora**, your points are all valid. I didn’t realize that you were talking about such extreme scores. It’s possible (and far more likely) to get ceiling effects with more average scores when the examinee’s age falls in the area of overlap of the tests. The age overlap of the Wechsler scales is more than adequate for most of these cases.

You were talking about truly extreme scores, though. It’s important to remember that much of what is true about the middle of the distribution doesn’t apply at the tails (the Flynn effect, for one thing). One reason it’s so difficult to measure extreme IQ scores is that standardization samples are, by definition, limited in size. The largest sample I’ve ever seen is around 3000 total, for all age groups combined. As a practical matter, what this means is that if I earn a standard score of 80, I know that there were a fair number of people in the standardization sample who performed at or near my range, and I can be confident that my score accurately reflects my position in the general population. If I get a score of 180, on the other hand, I know that there was likely no one in my age group in the standardization sample who scored anywhere close to that, so my relative position in the general population, as indicated by my IQ score, is more speculative.

Scores beyond 155 or so are pretty much extrapolated, regardless of the test used. This doesn’t mean that they’re useless, and a psychologist who gets a WISC-III Full Scale IQ of 145 should probably give a test with a higher ceiling like the SB:LM (even though the norms are out of date), to find out exactly what kind of 145 he’s dealing with. And with a score that high, you would want to do some cross-battery assessment to find out what the intra-individual strengths and weaknesses are; Wechsler isn’t much good for that.

However, we need to be careful when we talk about the differences between scores of, say, 160 and 190, on the one hand, and scores of 85 and 115, on the other. It’s a discrepancy of 30 points either way, but standard deviation means something different at the tails than it does at the mean. Personally, I’m unconvinced that there’s any value in measuring IQs beyond 170. That represents a score of one in a million. We don’t understand giftedness well enough to know what scores higher than that even mean.

However, getting back to the original point… (Remember Alice? It’s a song about Alice.) Nothing in the foregoing discussion changes the fact that IQs are considered stable after about age seven. Variance in IQ *scores* over time, due to ceiling or floor effects, is a function of the psychometric properties of the test used, not the underlying intellectual functioning of the child. Barring trauma or educational deprivation, IQ remains relatively stable throughout life.