Looking back I see andyandy asked you for your opinion on the topic, not the extent of your research on the topic. You won't answer my questions. Will you answer his?
Intelligence, Critical Thought Capacity, and Heridability
- Thread starter Dave1001
- Start date
Bruno Putzeys
Thinker
- Joined
- Jul 13, 2006
- Messages
- 155
A quick search yields this study: http://www.michna.com/intelligence.htm (the same study is mirrored on several sites, I just picked the first to pop up on google).
It suggests genetics are responsible for 0.75 of the variation in IQ scores in adults (in children it is markedly lower, 0.45, no news).
So here's a scientific result which some people may like and which others may not like. It's not about liking or not liking though. I don't think this information carries *any* moral import. One takes people for what they are, regardless of the extent to which this is in their genes.
It suggests genetics are responsible for 0.75 of the variation in IQ scores in adults (in children it is markedly lower, 0.45, no news).
So here's a scientific result which some people may like and which others may not like. It's not about liking or not liking though. I don't think this information carries *any* moral import. One takes people for what they are, regardless of the extent to which this is in their genes.
Bruno Putzeys
Thinker
- Joined
- Jul 13, 2006
- Messages
- 155
"OK, you've explained where you got your information that the earth is round and orbits the sun, but would you now tell me your personal opinion?"
Reread my answer. It's an honest answer to his question, which does provide my opinion and what it's based on. By the way, it's a valid opinion to say "I may not know enough yet to know", and certainly one that should be respected on a message board for skeptics.
Bruno Putzeys
Thinker
- Joined
- Jul 13, 2006
- Messages
- 155
@dave
But seriously, if I get it right the whole row (if we can call it that) was based on a suggestion (in another thread) of "eugenic benefit" caused by believers in homeopathy witholding themselves proper medical treatment. It sounded like a bad joke to me but I'm not sure if your position on this has been clarified yet.
But seriously, if I get it right the whole row (if we can call it that) was based on a suggestion (in another thread) of "eugenic benefit" caused by believers in homeopathy witholding themselves proper medical treatment. It sounded like a bad joke to me but I'm not sure if your position on this has been clarified yet.
Can you (or someone else) explain regression to the mean in this context to me? There will probably be several follow-up questions. I think I understand regression to the mean as a general principle, but I'm not sure exactly how it operates in the case of IQ heritability. Does it have to do with the chances involved that one will have a procreative partner closer to the mean, that one's offspring will grow up in an environment closer to the mean than one did, that the raw biological chances are stacked such that even two people with IQs well above the mean or well below the mean will have their dna combine in such a way that their offspring will likelier be genetically closer to the mean IQ, than further away from the mean IQ (perhaps because we as a species are somehow keyed to the mean)? Or some combination of all the preceding reasons? Or for a completely different reason? Here, when talking about IQ, I'm talking about what the offspring's adult IQ will be, not what their IQ will be as a child.
Last edited:
I didn't bring up racial IQ studies in this thread, and that wasn't what you asked my opinion on. I started the thread and you asked my opinion on the the heridability of IQ, intelligence, and critical thought capacity in individuals. Someone else brought up racial populations and I clearly gave my opinion on the utility of studying that.
On the previous page andyandy posted rather a lot more than the one item you mention, which I considered (and still consider) to be in exceedingly poor taste. Pattern indeed.
Bruno Putzeys
Thinker
- Joined
- Jul 13, 2006
- Messages
- 155
Dave,
Regression to the mean has *nothing* to do with genetics. If I take a sample of motor vehicles with the 20% highest horsepower ratings, and I then average their top speeds I'll find the mean of the top speeds of my sample not to be the mean of the 20% highest top speeds of the whole motor car population. It will be closer to the mean top speed of the whole motor car population because horse power ratings don't correlate fully with top speed ratings (take lorries, for instance).
Reading from the terminology of the remainder of your post frankly I must concur with the intuition of Meffy et al. You might wish to change your reading diet a bit.
Regression to the mean has *nothing* to do with genetics. If I take a sample of motor vehicles with the 20% highest horsepower ratings, and I then average their top speeds I'll find the mean of the top speeds of my sample not to be the mean of the 20% highest top speeds of the whole motor car population. It will be closer to the mean top speed of the whole motor car population because horse power ratings don't correlate fully with top speed ratings (take lorries, for instance).
Reading from the terminology of the remainder of your post frankly I must concur with the intuition of Meffy et al. You might wish to change your reading diet a bit.
Well, it seems to me that you're saying regression to the mean means that a population that deviates from the mean in one factor will regress to the mean in uncorrelated factors. For example, if you took the 10% of the population with the strongest "curly hair" genes, they'll regress to the mean in terms of their height, presuming that curly hair doesn't correlate fully with height.
But, at least in the wikipedia entry, the author(s) seem to claim to not be switching variables like that. They claim to be measuring adult IQ, claiming that adult IQ is substantially heritable, and to be claiming that offspring adult IQ of parents with very deviant adult IQ do regress to the mean. Perhaps your saying that the author(s) of the wikipedia entry mislabeled what they were describing when they called it regression to the mean? If not, I think the questions I asked about it were quite reasonable, and I'd be interested in answers and explanations from those that may know.
Bruno Putzeys
Thinker
- Joined
- Jul 13, 2006
- Messages
- 155
1) You're defining the concept in terms of itself
2) Curly hair genes? Sometimes you're better off studying phenotypes
3) This is an extreme case. Height does not correlate with hair type at all. As a result, the curly population will have a completely normal height distribution with the mean equal to the mean of the whole population. What regression to the means means is that if two parameters are not 100% correlated, when selecting for one parameter the other is sure to have shifted towards the mean by comparison.
I hope to have explained that you don't need to confuse variables totally in order for this effect to take place. Likewise, the amount of regression to the mean could be used as a measure of correlation.
You asked, I explained. I can't help if the answer is more prozaic and less obvious than you expected.
Signing off for the night now.
Here's my understanding of regression to the mean, as applied here. IQ is not entirely determined by genetics. So, there are some other factors involved, which we can call random. Probably, such random factors are part of the reason that someone with a very high IQ has a very high IQ, and there's no reason to suppose that similar random factors will also improve his children's IQ over what their genetics dictate. So the part of his IQ that's genetic will get passed on to his children, but probably the random part won't. Therefore, his children probably won't have an IQ as high as his.
I'm pretty sure I've understood your points both times now. My core question was if regression to the mean of adult IQ occurs in offspring of adult IQ deviant parents, what is the mechanism? With your motor car population the mechanism seems to me simply to be that horse power ratings and and top speeds of motor vehicles are on their own (not completely correlated with eachother) bell curve type distributions, so the vehicles that have the horse power rating that don't correlate with an equally high top speed statistically are more likely to have top speed that falls towards the middle of the bell curve where most of the inidvidual vehicle top speeds are clustered. Seems reasonable enough.
With adult IQ heritability though, I'm not sure that the 25% nonheritable part is as randomized as I presume the uncorrelated aspects of your motor vehicle example to be, although I'm open to that possibility. Specifically, I'm not sure if it applies in cases where do give two example that informed my earlier questions (1) both parents are equally adult IQ deviant in the same direction, (2) The offspring has had a developmental environment as deviant from the mean in the same adult IQ influencing direction as the parents did.
In particular, if one has equally adult IQ deviant parents (in the same direction), is heritability still 75%? And does regression to the mean still occur? Is it because (as I asked earlier, less thoroughly spelled out) that the adult IQ deviant parents are more likely to benefit from special environmental factors than they're likely to be able to reproduce those factors for their kids? Is it because their genes are just less likely to combine as fortuitously, perhaps because we're keyed as a species to a certain IQ? After all, our offspring don't revert to the mean primate adult IQ (or didn't 20,000 years ago when we weren't almost the entire primate population). I think these are reasonable questions, and explanations from interested parties in the know are apreciated.
Also, I wonder how the Flyn Effect impacts this analysis, cause I think it might indicate a reduced or absent regression to the mean with adult IQ heritability, if the adult IQ mean has been rising with each new generation. But the author(s) of the wikipedia entry seemed to believe both that regression to the mean and the Flyn Effect are both occurring with IQ heritability.
Yup, that I get, and I see how it applies to the motor vehicle example, but I see problems with applying "it's random" and "these IQs are intrinisically high rather than relatively high" to IQ heritability. Unless we as a species are keyed to an IQ, and the randomness is due to factors like how genes combine during reproduction.
Bruno Putzeys
Thinker
- Joined
- Jul 13, 2006
- Messages
- 155
I'm just perusing the articles quoted throughout this thread, but I can't find any references to anyone trying to determine heritability based on parents' scores. Heritability is easiest (and hence most commonly) determined through twin and adoption studies. My guess is that judging how a certain trait is passed from parents to children requires figuring out which genes are involved.
Given that heritability is not 100% as attested in twin studies, you can be pretty sure this randomness has precisely nothing to do with how genes combine during reproduction. This randomness should not be a source of frustration but for the wayward social engineer. Personally I find it a good thing that people are not 100% determined by their genes, even if the remainder is totally random.
Given that heritability is not 100% as attested in twin studies, you can be pretty sure this randomness has precisely nothing to do with how genes combine during reproduction. This randomness should not be a source of frustration but for the wayward social engineer. Personally I find it a good thing that people are not 100% determined by their genes, even if the remainder is totally random.
Last edited:
Regression to the mean is a purely statistical phenomenon, based on the properties of the normal distribution. As you approach the tail of a normal distribution the chance of sampling a score even more extreme descreases, and the chance of sampling a score less extreme increases.
If you start with parents that are already at the tail of a distribution on some measure (for example much taller than average), the chances are that their children will be shorter rather than taller than their parents because height is normally distributed. The more extreme the parents' scores, the smaller the chance of sampling a score even more extreme in their offspring. On the other hand, children born to parents of average height are equally likely to be taller or shorter than their parents because the proportion of taller and shorter heights is roughly the same either side of the mean in a normal distribution.
At least that is how I was taught it.
Also, heritability isn't a fixed factor. It describes the percentage of variance in a population attributable to genetic factors in a trait that is partly heritable. If everyone had identical experiences from conception, heritability of such a trait would be 100% because genetic factors would be the only possible source of variation.
Within a 'group' that shares a similar environment heritability may be high, but that says little or nothing about genetic factors contributing to differences between 'groups' that experience different environments.
bpesta22
Cereal Killer
- Joined
- Jul 31, 2001
- Messages
- 4,942
Regression to the mean is just the fact that rare events tend to be followed by more typical or average ones.
A simple example would be the lotto winner today who plays again tomorrow. Chances are, tomorrow's ticket isn't gonna hit.
This is not because the "dice have memory" (that would be the gambler's fallacy), but just the opposite. Since playing the lottery today and tomorrow are independent events, and since the most common result when one plays is to lose, the rare event of actually winning yesterday will likely be followed by the more comment event of losing today. This is true even though winning yesterday has nothing to do with the odds of winning again today.
Perhaps a simpler example is to flip 10 coins all at once. Keep flipping them til all ten land heads.
Once they finally all land heads (the rare event...) flip em again just one more time. Chances are, you'll get about 5 heads / 5 tails (... will likely be followed by a more common or average one).
***
I initially thought that regression to the mean would not occur with IQ, because the test's reliabilities are so large (upper .90s where 1.0 is perfect). 1- the reliability squared is the amount of error in a test (variance not due to the construct the test is measuring) so there's not much room for regression to the mean-- At least, I think.
I guess though you'd get regression to the mean when the parents are either very low or very high in IQ (the rare events). Even with a large chunk of the variance in IQ being genetic, the best bet would be the kids of very bright parents would have higher than average IQs, but IQs lower than their parents (due to regression to the mean).
Still though I think the effect would be small due to the high reliability and high hereditability of IQ.
***
We've had many debates here on IQ and genes, I won't open that can o worms, but just some points for interested posters to consider:
1) One cannot deny that race differences exist on IQ tests. The data go back to WW I, and the differences have to be among the most replicated effects in all psychology. The differences are not trivial either (1 standard deviation, roughly, for blacks versus whites).
2) All I'm saying in 1 is that races differ in mean IQ. This is not a racist statement. It's a fact.
3) Racism potentially comes in to play when you try explaining WHY races differ in mean IQ.
It could be that IQ tests don't measure intelligence at all, but instead measure culture-- and thus are biased. This is a potential explanation for the gap that doesn't seem at all racist.
On the other hand, the GAP could be entirely driven by genes, with some races clearly "superior" to others (a potentially racist explanation for the difference).
4) The fairest conclusion right now is that no one knows why races differ in IQ.
Evidence for the genetic view is anything but conclusive.
That said, no one has been able to find variables in the environment that trully matter (meaning when one controls for these variables, the difference between races goes away). This isn't for lack of trying either. Just about everything one can think of in the environment that might matter has been studied / controlled for, and yet the race differences remain.
5) The differences are not due to test bias. "Test bias" has precise meaning in psychometrics, and is something that can be easily measured. In study after study after study, there is no bias-- IQ tests predict many important things, and they do so just as accurately for whites as for blacks (so much so that the EEOC allows employers to use IQ tests in selection, even though they exclude black applicants at a disproportionate rate!).
6) If you buy the methodology / logic behind identical twin studies, and the correlations (e.g.,) between foster kids and adopted parents (r=00), then the evidence for the hereditability of IQ is overhwelming. Again, perhaps one of the most replicated effects in psych, with between 50-75% of your IQ being determined by genes.
**I recently was debating a neuroscientist on this, who pointed my to a study suggesting that h2 estimates are flawed because they ignore the prenatal environment.
Her argument was it's the womb that matters. Clearly, this is an environmental variable, but (i still need to think about this) in the typical "identical twins reared apart" studies, the environmental effects of the womb would be confounded with the measure of H2. So, h2 would vastly overestimate the effects of genes on iq, were indeed the womb the key thing that matters when it comes to determining IQ.
Interesting argument-- I need to read more about it, but my initial reaction is that it's such a potentially compelling wrench in the engine to the standard genetic model, that unless there were some strong paradigm-shift inertia going on, I don't understand why this research isn't getting paramount attention in the area.
7) Damn this post is long. I wonder if anyone read it. Percocets make me chatty.
A simple example would be the lotto winner today who plays again tomorrow. Chances are, tomorrow's ticket isn't gonna hit.
This is not because the "dice have memory" (that would be the gambler's fallacy), but just the opposite. Since playing the lottery today and tomorrow are independent events, and since the most common result when one plays is to lose, the rare event of actually winning yesterday will likely be followed by the more comment event of losing today. This is true even though winning yesterday has nothing to do with the odds of winning again today.
Perhaps a simpler example is to flip 10 coins all at once. Keep flipping them til all ten land heads.
Once they finally all land heads (the rare event...) flip em again just one more time. Chances are, you'll get about 5 heads / 5 tails (... will likely be followed by a more common or average one).
***
I initially thought that regression to the mean would not occur with IQ, because the test's reliabilities are so large (upper .90s where 1.0 is perfect). 1- the reliability squared is the amount of error in a test (variance not due to the construct the test is measuring) so there's not much room for regression to the mean-- At least, I think.
I guess though you'd get regression to the mean when the parents are either very low or very high in IQ (the rare events). Even with a large chunk of the variance in IQ being genetic, the best bet would be the kids of very bright parents would have higher than average IQs, but IQs lower than their parents (due to regression to the mean).
Still though I think the effect would be small due to the high reliability and high hereditability of IQ.
***
We've had many debates here on IQ and genes, I won't open that can o worms, but just some points for interested posters to consider:
1) One cannot deny that race differences exist on IQ tests. The data go back to WW I, and the differences have to be among the most replicated effects in all psychology. The differences are not trivial either (1 standard deviation, roughly, for blacks versus whites).
2) All I'm saying in 1 is that races differ in mean IQ. This is not a racist statement. It's a fact.
3) Racism potentially comes in to play when you try explaining WHY races differ in mean IQ.
It could be that IQ tests don't measure intelligence at all, but instead measure culture-- and thus are biased. This is a potential explanation for the gap that doesn't seem at all racist.
On the other hand, the GAP could be entirely driven by genes, with some races clearly "superior" to others (a potentially racist explanation for the difference).
4) The fairest conclusion right now is that no one knows why races differ in IQ.
Evidence for the genetic view is anything but conclusive.
That said, no one has been able to find variables in the environment that trully matter (meaning when one controls for these variables, the difference between races goes away). This isn't for lack of trying either. Just about everything one can think of in the environment that might matter has been studied / controlled for, and yet the race differences remain.
5) The differences are not due to test bias. "Test bias" has precise meaning in psychometrics, and is something that can be easily measured. In study after study after study, there is no bias-- IQ tests predict many important things, and they do so just as accurately for whites as for blacks (so much so that the EEOC allows employers to use IQ tests in selection, even though they exclude black applicants at a disproportionate rate!).
6) If you buy the methodology / logic behind identical twin studies, and the correlations (e.g.,) between foster kids and adopted parents (r=00), then the evidence for the hereditability of IQ is overhwelming. Again, perhaps one of the most replicated effects in psych, with between 50-75% of your IQ being determined by genes.
**I recently was debating a neuroscientist on this, who pointed my to a study suggesting that h2 estimates are flawed because they ignore the prenatal environment.
Her argument was it's the womb that matters. Clearly, this is an environmental variable, but (i still need to think about this) in the typical "identical twins reared apart" studies, the environmental effects of the womb would be confounded with the measure of H2. So, h2 would vastly overestimate the effects of genes on iq, were indeed the womb the key thing that matters when it comes to determining IQ.
Interesting argument-- I need to read more about it, but my initial reaction is that it's such a potentially compelling wrench in the engine to the standard genetic model, that unless there were some strong paradigm-shift inertia going on, I don't understand why this research isn't getting paramount attention in the area.
7) Damn this post is long. I wonder if anyone read it. Percocets make me chatty.