Can someone explain regression to the mean for IQ?

[smarterthanone]

Ok. I have an IQ of 140. My girlfriends is about 145. She is better at vocabulary but I am better at math.

How is regression to the mean involved? I understand the super basic version 140/145 would make it more likely our kids are lower than 140s... ok but I am assuming its still more likely for two in their 140s vs two 110s vs 140/90?

I know, I am smart I should be able to look it up online like I do everything else but there is so much scientific lingo and stuff, I am having a hard time finding just a super basic summary so I can grasp the concept. So don't laugh at me. lol

[RichardY]

Just speculation but if intelligence was more dominant genetically, wouldn't IQ increase exponentially quickly. Hence high IQ sometimes skipping generations. Good eyesight perhaps being more dominant genetically and critical for early mans survival.

[Mister Mister]

My amateur understanding of Regression to the Mean, is that with certain traits, not just IQ but height, creativity, aggressiveness, athleticism, productivity, anything which lies on a Bell Curve really, the statistical probability when two people who are on the ends of the bell curve breed, is for their children to be closer to the middle of the bell curve than they are.  A good example is that, Michael Jordan's children are better than most blacks at basketball, but not anywhere close to the elite level their father was.  This is to be expected statistically.Therefore, you and your GF being on the far high end of the Bell Curve of IQ, are more likely to have children who are less intelligent than you, but still smarter than average.  Indeed as RichardY says, this trend explains why there is a Bell Curve at all; there seem to be some stabilizing factor in genetics, rather than an exponential push toward extremes.

 

[ofd]

How valid is the test you have taken?

[smarterthanone]

I have taken multiple IQ tests in many different settings including under trained professionals and score usually between 138-142. This was her first time taking Stanford Binet and got 145. She murders me at scrabble and anything like word puzzles and such. But when it comes to math I am very talented and am very successful at financial modeling. So it seems accurate for both of us.

 

If all regression to the mean is, is the relationship with the bell curve then that is very simple. I thought it was more involved than that. The reason being intelligence is not a specific genetic trait like eye color is. Its not like is blue/brown/green/etc in your dna but its more like x trait delivers 2% extra brain blood flow, y trait makes 1% better use of each brain cell, z trait has 3% more usable brain cells. So some specific combination of these traits will make you intelligent but if you have extra blood flow but no extra cells it won't do anything at all. And so my girlfriends intelligence may be based off traits a b c q while mine are a b q z y and if we have a kid they may be a b c z which may be just as intelligent or not work together well and make the kid a moron.

[ofd]

If we assume that IQ is to 80% heritable, then 80% of an unlikely event (very high IQ) is still much more unlikely than 20% of a likely event (above average IQ).

[engardeknave]

I have a lower IQ than you and your girlfriend, so statistically speaking it would probably make sense to let me impregnate her.

[smarterthanone]

11 hours ago, engardeknave said:

I have a lower IQ than you and your girlfriend, so statistically speaking it would probably make sense to let me impregnate her.

Well actually it wouldn't. 140/145 has a better chance of having a higher IQ than a 120/145 (whatever yours is), although only marginally. I guess on a group level its possible, depending on the members of the group but nobody around here actually believes in Central Planning do they? If they did, logically it would make sense for me to impregnate a large portion out of all the women.  

[engardeknave]

1 hour ago, smarterthanone said:

Well actually it wouldn't.

Yeah, I know.

[lorry]

On 23/06/2017 at 0:15 AM, smarterthanone said:

Ok. I have an IQ of 140. My girlfriends is about 145. She is better at vocabulary but I am better at math.

How is regression to the mean involved? I understand the super basic version 140/145 would make it more likely our kids are lower than 140s... ok but I am assuming its still more likely for two in their 140s vs two 110s vs 140/90?

I know, I am smart I should be able to look it up online like I do everything else but there is so much scientific lingo and stuff, I am having a hard time finding just a super basic summary so I can grasp the concept. So don't laugh at me. lol

 

Say 80% of your child's genetic IQ is 142.5 from the parents and let the other 20% be random. In order for your child to have an IQ > 142.5 it would then have to roll > 142.5, pretty rare, right? If it is random then you would expect the roll to be about 100 (assuming your white). So your child's IQ would be expected to be 0.8 x 142.5 + 0.2 x 100, whatever that is.

The important part of the idea here is that so long as the parents avg IQ is greater than the expected random IQ (the mean population IQ), ie. so long as the parents have an IQ > 100 then you would expect their child to have a lower IQ than them.

Conversely, parents with an IQ < 100 would be expected to have child with an IQ higher than theirs.

Plug in a few numbers and get a feel for whats going on.

P x Avg_Parent_IQ + (1 - P) x Avg_Population_IQ = Child_Expected_IQ

Where P is some number between 0 and 1 representing the percentage of your child's IQ which is a function of parents IQ.

 

[Jsbrads]

The lack of runaway IQ hasn't happened because 3 generations ago, 99.9% of the population did not pair based on intelligence. That has changed drastically to the point that the vast majority in the West are solely pairing based on intelligence. PhDs marry, McDonald's employees pair, etc.

paraphrasing Thomas Sowell