Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Most Reported Genetic Associations with Intelligence are Probably False

dash.harvard.edu

Post reply on HN
Print view View original

41–50 of 57 posts

Re: Most Reported Genetic Associations with Intelligence are Probably False

#41

Post by Throwadev »

btilly wrote: post #7

This whole field seems likely to be a case study supporting http://xkcd.com/882/ . But the research does have a questionable assumption. On p 5 it says, "Such a result would not likely be due to differences in the methods used to generate g in the various datasets under comparison, since g is consistently measured by a wide variety of tests." This is true. However it is also true that, as http://vserver1.cscs.lsa.umi…

I don't know enough about STATs to get the XKCD comic. Anyone care to explain?

Something about how P < 0.05 still doesn't mean there is a link between green Jelly Beans and acne?

Re: Most Reported Genetic Associations with Intelligence are Probably False

#42

Post by yummyfajitas »

bayesianhorse wrote: post #25
JackFr wrote: post #21

Earlier quoted context omitted.

I'm not sure if you are misreading the article or disagreeing with it, but in either case you're wrong. The fact that a 'wrongness gene' hasn't been, and may not ever be, identified does not undermine my assessment, but rather indicates current models of the interaction of genes and wrongness might be naive, and the interactions are probably deeper and more complex.

Please tell me one "practical purpose" were "Intelligence is largely genetic" as a use...

If we know intelligence is largely genetic and differs across genetic phenotypes, it would tell us that public policies designed to alter group outcomes by altering the environment might be counterproductive.

To take an extreme example where genetics is the clear cause, cats are underrepresented in college. Because we know that cats are genetically less intelligent than humans, we can conclude that affirmative action/college for cats education programs/etc are a waste of money.

Another use would be knowing that if only 50% of the population has the intrinsic intelligence for college, college prep programs should not serve more than 50% of the population (modulo technical details).

Re: Most Reported Genetic Associations with Intelligence are Probably False

#43

Post by yummyfajitas »

btilly wrote: post #7

This whole field seems likely to be a case study supporting http://xkcd.com/882/ . But the research does have a questionable assumption. On p 5 it says, "Such a result would not likely be due to differences in the methods used to generate g in the various datasets under comparison, since g is consistently measured by a wide variety of tests." This is true. However it is also true that, as http://vserver1.cscs.lsa.umi…

Also, for anyone reading Shalizi's article, please note that he uses the term "statistical myth" in a non-standard way. By his definition, any aggregate quantity derived from a microstructure model (e.g. temperature, pressure, climate, electric current) is a "statistical myth".

This is not to dispute his calculations or model - they seem solid. I'm just pointing out that his definition is probably not what most people think when they hear the term.

Re: Most Reported Genetic Associations with Intelligence are Probably False

#44

Post by samatman »

Heritability is related to the degree to which a trait is genetically determined. The relationship is far from trivial.

Skin color makes this much easier to understand. Imagine a couple, where the man and the woman each have one Scandinavian parent and one Bantu parent. Their children could have almost any skin shade.

If they have identical twins, those twins will have the same shade of skin. Skin color is genetically determined but the heritability varies.

Imagine two people who have the same skin color as our first two, but they come from an island population founded by Norwegians and Nigerians exclusively. They are 20th generation, or 50th. Now, it is likely the skin color of their children will be the same as their own.

Intelligence inherits upon many factors. Even identifying those factors proves to be elusive and hard. The good evidence that intelligence is genetic is twin studies: the good evidence that it is variable in heritability is every family you've ever known.

Re: Most Reported Genetic Associations with Intelligence are Probably False

#45

Post by gwern »

Boring. The effect size of the variants is too small to find with n=10,000, hence any positive findings are almost certainly false positives which will fail to replicate. That's why the 3 SNPs found recently (http://infoproc.blogspot.com/2013/05/first-gwas-hits-for-cog...) were in a sample of n=101,000.

There's nothing here to undermine the view that genetics causes much of variation in intelligence, it's just that the variation wasn't as clustered on a few genes as people hoped because it would make the research so much easier.

Re: Most Reported Genetic Associations with Intelligence are Probably False

#46

Post by nialo »

Throwadev wrote: post #41
btilly wrote: post #7

This whole field seems likely to be a case study supporting http://xkcd.com/882/ . But the research does have a questionable assumption. On p 5 it says, "Such a result would not likely be due to differences in the methods used to generate g in the various datasets under comparison, since g is consistently measured by a wide variety of tests." This is true. However it is also true that, as http://vserver1.cscs.lsa.umi…

I don't know enough about STATs to get the XKCD comic. Anyone care to explain? Something about how P < 0.05 still doesn't mean there is a link between green Jelly Beans and acne?

P < 0.05 means that the probability of seeing the observed results if there is no difference is less than 1 in 20. So, if you run the experiment 20 times, you'd expect to see a link once, even if there is really no link.

Re: Most Reported Genetic Associations with Intelligence are Probably False

#47

Post by clarkm »

bayesianhorse wrote: post #25
JackFr wrote: post #21

Earlier quoted context omitted.

I'm not sure if you are misreading the article or disagreeing with it, but in either case you're wrong. The fact that a 'wrongness gene' hasn't been, and may not ever be, identified does not undermine my assessment, but rather indicates current models of the interaction of genes and wrongness might be naive, and the interactions are probably deeper and more complex.

Please tell me one "practical purpose" were "Intelligence is largely genetic" as a use...

It's useful in convincing parents to stop wasting their money on Baby Einstein CDs and other unscientific marketing ploys.

Re: Most Reported Genetic Associations with Intelligence are Probably False

#48

Post by anaphor »

Xcelerate wrote: post #5

Note that the article is not saying that intelligence is not heritable. In fact it makes it quite clear that despite the controversy, the opposite is true: > Although the exact figures have been the topic of much debate, the claim that IQ is at least moderately heritable is widely accepted. IQ may in fact be similar in heritability to the physical trait of height What the article is saying is that intelligence cannot…

IQ becomes more heritable the more homogeneous an education and development a group has. So it is probably less heritable in a very poor country where there is a large disparity in the kind of nutrition and education someone has access to (lots of very rich and very poor people).

Re: Most Reported Genetic Associations with Intelligence are Probably False

#49

Post by anaphor »

thetruthhurts wrote: post #29
thetruthhurts wrote: post #23

"African Americans currently score lower than European Americans on vocabulary, reading, and mathematics tests, as well as on tests that claim to measure scholastic aptitude and intelligence. This gap appears before children enter kindergarten (figure 1-1), and it persists into adulthood. It has narrowed since 1970, but the typical American black still scores below 75 percent of American whites on most standardized t…

Continued: I think it's pretty obvious that genes have quiet a bit to do with intelligence and achievement. There have been too many studies done on race and iq, race and crime, race and achievement to just throw out. I'm not the smartest but at least I'm living in reality. And I'm proud to see so many HNers here doing the same and acknowledging the important role genetics has in our lives. It's good to know there's…

The effect a gene has on a trait isn't equal in all environments. It plays a larger part in groups of people who have very similar environmental conditions, conversely it plays a much smaller role in populations with a high variation in environmental conditions. We know that there is a fairly large variation in the environments of the groups you mentioned so it stands to reason the differences have very little to do with genetics. tl;dr go learn what heritability is.

Re: Most Reported Genetic Associations with Intelligence are Probably False

#50

Post by bayesianhorse »

clarkm wrote: post #47
bayesianhorse wrote: post #25

Earlier quoted context omitted.

Please tell me one "practical purpose" were "Intelligence is largely genetic" as a use...

It's useful in convincing parents to stop wasting their money on Baby Einstein CDs and other unscientific marketing ploys.

Would telling them "Your child is stupid because you are stupid!" really be more successful than pointing out a lack of effectiveness? And would they really ignore the left-over variability? And will they stop at Baby Einstein CDs or will they just neglect any IQ improvement method, like regular and complete school attendance?
Post reply on HN