Gifted Issues Discussion homepage Forums Identification, Testing & Assessment A score of 160 -- what does it really mean?

When a test mean is 100 and the standard deviation is 15, then a score of 160 should appear in approximately 1 out of 31,560 people. Correct?

A few questions:
1. How is it possible to assess the rarity of scores above, say 145, when the normative samples for these tests are only a few thousand?

2. Why is it that achievement scores of 160 are anecdotally way more common than IQ scores of 160?

3. On achievement tests like the WIAT and WJ, is it possible to interpret a raw score that exceeds the minimum required for a 160? (some kind of 'extended norms'?)

-Ul.H.

That's the idea.

I don't know, but I tend to think that the higher the results are, the less accurate they may be, partly because of difficulties in crafting test questions for more and more rarefied and refined sorts of intelligence (I'm a layperson so go easy on me), but also because of lack of testing samples. I also wonder whether having a bad or good test day might not influence results much more in the upper ranges-- I'd be curious to know how much a single extra right answer would influence the final score on various tests. A person who can score an average of 180 on a modern IQ test might fluctuate more wildly than someone who averages lower.

I've wondered that myself. I think it's a combination of factors. IQ tests may be harder to construct so that they accurately test what they're designed to test, especially in the upper ranges, while intelligence that doesn't show well on a particular IQ test might easily still result in high achievement; hothousing effects may be much more exaggerated in younger testees; and there may be high achievers who simply aren't as natively intelligent, but "play up" due to determination and hard work.

I don't know about the WIAT, but the WJ-III (Ach) is already able to generate scores above 160 without the need for extended norms or the like.

I think that our entire concept of measuring intelligence is a bit too simplistic because our models aren't giving an accurate picture of reality.

It's like we've figured out that atoms can have a bunch of electrons, and we might be able to guess an approximate number of electrons, but we don't understand that there are multiple orbitals than can be filled in multiple ways depending on the individual in question.

Yep, I totally agree with you on this one. I guess I was asking if we can even statistically interpret a 160 on these tests (compared to, say, a 150) or if we just have to wave our hands and say 'it's way out there'.

It seems more straightforward to me with achievement tests for children, since you can measure whether the child has attained certain skills, and then compare that to a normal age of attainment. Then again, that's old-style ratio thinking-- really the way it goes is that a child's attainment is compared for rarity against age and/or grade peers.

From that perspective, as long as you can construct a question where some people get it right and others don't, it's of a general type previously deemed to measure IQ, and the results don't seem backwards-oriented to the other measurements, it's a question you can use for differentiating IQ. A 160-scoring person is just rarer by virtue of getting X more questions correct than a 150-scoring person. That part seems pretty straightforward to me-- what's not obvious is how well IQ tests really measure all of what I consider to be intelligence.

Empirically speaking, though, since they are correlated strongly with certain measures of success, they are measuring something useful. I've seen quotes here recently about physics Ph.Ds generally having IQ scores of 145+, with Nobel laureates etc. higher still, so there must be something to it.

This is a topic that I've spend some time thinking about. I believe that "160" doesn't really mean what it seems to mean and the tests just aren't good in the tails.

First, there's the problem of the test itself. To get a math score with a GE, there are a few problems at each grade level -- not many and many of them are quite easy so that vague and passing familiarity would allow a child to answer the question despite not knowing that much. A score that returns "11th grade GE" does not mean that a child knows the intricacies of pre-calc and trig. So I think the tests are pretty weak.

Next, there's the issue of norming. I looked harder at IQ tests than at achievement when researching this, but a norming sample of 2000+ children may not include any children scoring very high. If there's no norming sample in the tails, it's much harder to get an accurate measurement of what number right would be something = to the rarity of 4 sd above the mean. For IQ, none of the original WISC-IV sample scored above 150 for full scale. I understood this to mean that there was a lot of extrapolation and guessing in determining the raw score differences that would give 150, 155, or 160 for full IQ. If your sample is miniscule, the error bars are huge. If the error bars are huge, a random guess or miss makes way too much impact on the score.

I agree that GE helps with the "out there" factor for a 160 at a young age. At late elem, there's probably not enough ceiling to tell, but a 6 yo who scores 160 with a GE of >12.9 has answered more questions than a 6 yo with 160 and a GE of 6.1.

Finally, I personally have seen way too many crazy out there scores to believe that 160 really indicates the rarity of 4 sd above the mean. I've seen multiple scores with all subtests at 160 indicating even more rarity and although the children are definitely smart, they are not that unusual.

I'm not sure what good such tests are except for another clue in putting together an understanding of a child. If a kid hates school at 6 and complains of boredom and the WIAT shows >12.9 in multiple subtests, the kid is probably bored. If a kid has 160 achievement test scores overall and an IQ of 105, I'd wonder if there was some missing information to explain the disconnect. Similarly, if an IQ score is 160 and achievement averages 110, that seems to be a mismatch worth investigating. I'm not convinced that an achievement score of 150 is really worse than 160 or that there is any useful information about what a child actually knows from the WIAT or WJ achievement.

Really?

160 is the ceiling for the WISC IV. What I don't understand is how the extended norms work. Our tester said there are no further questions for the extended norms, it's just a comparison with a different sample with a score supposedly going up to 210. But a score in the 150s represents so many maxed out subtests already that I don't understand how they can distinguish that much over that. I'd love to hear an explanation from people who really understand it.

Tallulah - the maximum score for a subtest with standard norms is 19. The raw score is scaled to 19 or below. So picking numbers out of the air NOT real numbers, you might need a minimum of 23 questions right to get a scaled score of 19 at a certain age. So assuming all the children are the same age, and remembering that my numbers are not at all real...

Child A gets 23 right and gets a 19, this score would not necessarily go up with extended norms. Child B gets 29 right, their scaled score is 19 too, but their score might go up a few points with extended norms. Child C gets 35 right and their scaled score is also 19, but their score would go up more again than child B with extended norms.

I do wonder how rare kids in the tail really are. BUT 1/30000 is still a fair number of people in the world. And I don't know about everyone else but although all of our IRL friends' kids are very bright, and many I am sure are at least MG, the families I know with kids who are out there enough to be tested and who have very high scores - I know their parents because we went seeking each other, mostly online. So I feel like I know a disproportionately high number of them, but I didn't just bump into them at the local shops. Oh and my own HG+ kid is NOT 1/30000, she's 146, theoretically there should be 15 odd children in her grade in our rather small city.

Ah, good. So the extended norm thing happens between testing and scaling the score? That makes me feel better.

wrt to everyone you know being bright, it's the same for us. All of my friends (and their kids) would probably test at least 120 or 130 IQ. Which isn't surprising because 130 is 1 in 44 people. But, you do self-select for people who are like you, have similar interests, professions, etc. My best girl friends are all really interested in stuff and having deep discussions, and their kids are all ahead of their classes at school.

But, if you look at that IQ frequency chart http://www.iqcomparisonsite.com/IQtable.aspx 150 is 1 in 741 but 150 is 1 in 2331. And going from 154 to 156 takes you from roughly 1 in 6,000 to 1 in 10,000. That's certainly rare enough to be a problem finding peers at school. For my child, I don't give a toss about the world in total, I want children who live near us and are a similar age and aren't homeschooled. And by that chart there would be 28 people in our county with a score from 150 to the top end of the range. Luckily it's a college town, so there's not a random distribution and those frequencies don't hold true. But this is why they have magnet schools.

Any particular reason you don't want your child to associate with kids who are homeschooling, even if they might be intellectual peers? A lot of families with highly or profoundly gifted kids homeschool because they have such a hard time getting an appropriate education in schools.