Hacker News Reader

Comment by tomkin | Hacker News Reader

I hate to say this, but IQ tests are BMI tests for the brain, and about as accurate. If a coffee, bad sleep, or a dose of Ritalin can change the result – at its very nature – it is not a scientific method of measurement (the ability to retest and get the same result).

We could say it's a good read of a spectrum of mental capacity, but to suggest that IQ is capable of anything more is whack science.

View on HN

BMI is a perfectly valid first-order approximation of appropriate weight ranges for the 95% of the population who are not body builders.

Perhaps BMI isn't very accurate for slightly short people. It says I should weigh 115-149 lbs. I currently weigh 190 and when I manage to get down to 170, I look very athletic. The last time I weighed 149, I looked like I was starving.

It seems similarly unreasonable for most of my friends and family.

Your standards are based on living in a society of fat people. At 149 you'd still be sporting copious amounts of flab.

My standards are based on objective reality and a decent knowledge of physical fitness and nutrition.

When I weighed 160, I had slightly defined abs. Similar to what you would see on someone with 10-15% body fat.

The existence of a lot of overweight people in the US doesn't change what the words "overweight" and "athletic" mean.

First you looked "very athletic" at 170, now your abs were "slightly defined" at 160. Uh huh. If that means anything, that means you had a lot of visceral fat, or it's more stored in the legs.

And well, it's WAY past the edit window now. The next time you are at 160, I recommend using the tape measure to measure things like waist to chest circumference, and thigh circumference. Then ask if those measurements are bigger than what can be explained by the muscular performance you get. Another thing you can do is try jumping rope, running in place, or doing jumping jacks. When I do these activities, I get "jiggle." My moobs and stomach get jerked around, because there is some flab there. The effect lessens as you lose weight.

What is not objective is looking in the mirror in some lighting and thinking you look pretty good. That's a subjective measurement based on objective reality. Your body might distributed fat in a way that's relatively aesthetically pleasing -- mine does too, it hides it all in the thighs, and people think I'm way more in shape than I am -- and that's not the same thing as being healthy.

Which BMI are you using? The traditional BMI isn't great for very short people - but for the opposite reason you give.

Traditional BMI makes short people think they're thinner than they are.

My argument is that no one can claim BMI or IQ is a scientific form of measurement.

Both are absolutely scientific: they provide good test-retest reliability and correlate with other measures.

Maybe they are crude. Maybe there are better alternatives. That doesn't make them unscientific.

I really don't know how you got here. BMI has been raked over the coals (new and old methods) for being wildly inaccurate for anyone outside of a conditional range. IQ does not predict what to expect from a human being.

Those who double down on IQ as being infallible have something to lose if it's proven to be hack. And history is not on your side here.

> BMI has been raked over the coals (new and old methods) for being wildly inaccurate for anyone outside of a conditional range.

No it hasn't.

BMI doesn't work for three groups of people: it makes very tall or very athletic people think they're fatter than they are, and it makes very short people think they're thinner than they are. But, as GP says, that's less than 5% of the population.

Look at all the effort that's gone into improving hiring. And what're the only factors that people have ever found that actually correlate with job performance? Work sample and IQ.

History is very much on the side of IQ realism. Ten years ago people were denying that a general factor of intelligence existed at all, now it's down to nitpicking or argument from ignorance.

It's perfectly scientific. It just doesn't mean what people think it means.

My wife grew up in Los Alamos. There was an experiment where some physicists took an IQ test every Saturday for a year. Their scores ranged from Moron to Genius, depending on the day. We presume some bell curve where their 'true IQ' was measured. But take a test once on one day - the result is highly variable.

Not to mention, there are groups like MENSA, of people who have a very high IQ, yet a lot of them are no more productive than the person of average intelligence. We are far away from having a good measure of intelligence.

My two cents: one of our employees is a member of Mensa and participates in a local branch organization. He always mentions how difficult is to reach consensus on basic organizational issues. No different that reaching consensus in a condominium.

Having said that, I think the measure of IQ and related variables such us working memory capacity is very interesting and gives a lot of information about how you can peroform in specific situations. The problem is when you try to infer the individual outcome in a complex environment with a few variables.

But on average, the higher IQ group is more productive than the general population.

I agree with your first sentence but I don't think that argument really supports it. Performance-based testing is not inherently flawed. You just need multiple samples. Bad sleep and stimulants can screw up a lot of other measurements, too, like blood sugar and pressure.

I agree that it's like blood sugar/pressure, but that's sort of my point. I think if you used blood pressure to judge some one's physical capabilities like you would use IQ to judge some one's mental capacity, you would run into unfair outcomes.

"some one with an IQ greater than 120 is definitely intelligent" == "some one with high blood pressure is definitely unable."

I'm fine with IQ so long as we're not using it as a scientific form of measurement...but we currently seem to.

Reply on news.ycombinator.com