The Iron Law of Heritability
From Thomas Bouchard 2004:
“As Rutter (2002) noted, ‘Any dispassionate reading of the evidence leads to the inescapable conclusion that genetic factors play a substantial role in the origins of individual differences with respect to all psychological traits, both normal and abnormal’ (p. 2). Put concisely, all psychological traits are heritable.”
Bouchard then points to the general heritabilities of commonly used psychological traits:
I call this “The Iron Law of Heritability”, though this is not a commonly used term. It is meant as a label for the common understanding among psychologists that all general psychological traits are to some degree heritable.
So when we explore any kind of psychological variation, the question is not “is that variation down to genetics”, the question is “HOW MUCH of the variation we see is down to genetics”, because it is always going to be some.
Heritability of Political Views Within the United States
There is a general misunderstanding of how twin studies are done. The common conception of twin studies is that identical twins are separated at birth, and then you see how similar they are in various traits when they grow up in different environments.
This has the obvious problem in that the range of environments of adopted twins may not reflect the range of environments for the whole population.
Another, better, way to do twin studies is simply to compare identical and non-identical twins reared in the same environment. Non-identical twins share roughly 50% of their genes + additional similarity from assortive mating, whereas identical twins share almost 100% of their genes.
So instead of controlling for genes and looking at the impact of environmental variation, twin studies can control for environment and look at the impact of an increase in genetic similarity.
For example, IQ scores. What they do is look at how similar the IQs of non-identical twins are, and then compare that to how much MORE similar the IQs of identical twins are; to see how much of an impact an increase of 50% genetic similarity has.
For example, if the IQs of non-identical twins correlate with each other at 0.5, and the IQs of identical twins correlate at 0.8, then that implies the general heritability of IQ in the populations examined is 0.6, or 60%. This is because a 50% increase in genetic similarity produced a 30% increase in IQ similarity.
But because of assortive mating, this is going to be an underestimate, because couples are not randomly selected; couples tend to be more genetically similar to each other than they are to the general population.
If, for example, non-identical twins aren’t 50% genetically similar, but are 60% genetically similar, then identical twins are only going to be 40% more similar than non-identical twins. If non-identical twins were 60% genetically similar in the previous example, then the estimated heritability of IQ would be 75% instead of 60% – this is because the jump from non-identical to identical is only 40% more genetic similarity this time, and this 40% increase produces a 30% increase in IQ score similarity. And 0.3 / 0.4 is 0.75.
But then there is another problem, which is that assortive mating is based on traits in a person, not their genes. And so lets say non-identical twins are 55% genetically similar across the whole genome; well, for genes relevant to IQ, they may be 70 % genetically similar. And so in this case, identical twins only represent a 30% increase in genetic similarity, but produce a 30% increase in similarity of IQ scores, which would mean the real heritability of IQ in this population is 100%.
That digression aside, if we just go with the most basic and nurture-friendly approach, assuming a non-identical twins have 50% genetic similarity, and identical twins have 100% similarity, we find in most studies the heritability of “political views” to be around 0.4:
But keep in mind that this “0.4 heritability of political views” is like says “there is a 15 point IQ gap between blacks and whites”. The heritability of political views increases with age, and there’s no industry-standard age when studies on the heritability of political views start asking.
According to a Virginia Twin study, the heritability of political views is around 0.57 by age 50:
The heritability of political views also varies by issue. A study on Swedish twins, which had 2,338 identical twins and 4,868 non-identical twins, found that the heritability of immigration views was higher than any other issue:
Heritability of Political Opinions in Swedish Twins
|Foreign Policy Opinions||0.417|
|Opinion on Feminism||0.414|
|Locus of Control||0.281|
|Left vs. Right Self-Placement||0.154|
This was also found back in a 2001 study which looked at heritability of various psychological traits. Views on immigration had a heritability of 0.46, the second highest heritability of any political item, and was the third most heritable out of thirty psychological items, behind opinion on the death penalty and enjoyment of roller coaster rides.
And remember, this is assuming a genetic similarity of non-identical twins of 0.5, when it could be higher across the whole genome, and higher still for those genes associated with psychological traits, which would mean that the increase in psychology-relevant genetic similarity you get going from non-identical to identical twins is probably less than 50%, and so all these heritabilities from all of these studies would be higher.
Molecular Genetic Data
Regarding group differences, researchers are starting to find some genes associated with social sensitivity and collectivism, and how those genes differ by race.
A 2010 paper by Way and Lieberman looked at country differences in allele frequencies two gene locations: A118G and MAOA-uVNTR, both of which have been independently found to correlate with social sensitivity WITHIN populations. For example, a Swede with the “G” allele at the gene location A118G is more sensitive to social pressure than a Swede with a different allele.
(You need to within-group validate the gene otherwise you’ll just have correlations with genes between groups, which could be coincidental. I.e. – if Japan has any purely cultural difference with Europeans, AND some genetic difference with Europeans, even if that genetic difference has zero causal impact on the cultural difference, the cultural difference will correlate with the genetic difference. This is why they must be validated WITHIN populations before comparisons between populations are made.)
They also used a combination of indexes of collectivism and individualism from four databases, and plotted those results with allele frequency data.
Way and Lieberman’s results for A118G
Way and Lieberman’s results for MAOA-uVNTR
Chiao and Blizinsky did the same thing in 2009 with the location 5-HTTLPR, plotting population differences between populations in the proportion with the “S” allele with individualism and collectivism in those countries:
Out of curiosity, I plotted Chiao and Blizinsky’s numbers by country with country results from a Pew Survey results on various free speech questions:
First the proportion of percentage of people who support the right to criticize government and the percentage of the population with the “S” variant of the 5-HTTLPR gene:
Next is the right to criticize one’s own religion:
And then there is views on the right to say things that are “offensive” to “minorities”:
Of course we don’t know if “speech offensive to minorities” means the same thing in China as it does in Germany. For example, I have personal experience of people who support banning “hate speech” but don’t support banning sites like Truth is Justice. Even though there is plenty on this site that “minorities” will be offended by. I suspect a typical Chinaman who opposes “hate speech” would be against banning this site, however a German who similarly says he is “against hate speech” would be more likely to support banning this site.
So if there is a bit of fuzziness between countries as to what these questions are actually asking, that will dampen any correlation because the effect of different countries having slightly different conceptions of what the questions are asking would be to introduce randomness.
That said, we still see a meaningful correlation between frequency of the “S” allele and various free speech issues. If you think these correlations are weak, remember that the correlation between an individual parent and their child’s height is about 0.45. And also keep in mind that these are just individual genes.
It’s also important to know that support for laws that ban speech critical of “minorities” is something that distinguishes the first and the third world. If you click above the links and look into the data, you will see that Latin American countries are about as supportive of free speech as European countries, even though “hispanics” in the United States are less supportive of free speech than European-Americans are. However, one of the distinguishing traits between the first and the third world is that the third world is more likely to support laws that prohibit speech “offensive” to “minorities”, and there are internally-validated genetic correlates that explain part of this. Obviously we have a long way to go to show the full extent of genetic causation, but it can’t just be hand-waved away as totally environmental.
You may find instances where the correlations between allele frequencies and a given trait add up to more than 1. This just means that “genes for” that trait tend to evolve together in sets. I.e. the G allele of A118G probably has its own independent effect, but its presence also indicates that you probably have other alleles that CAUSE an individual to be more collectivist within a given environment. And a .4 correlation between that G allele and collectivism takes into account both the direct effects of that single allele, AND the effects of other alleles that a person who has the G allele also tends to have.
But the point is that the twin studies find heritability estimates of around 40% for political views in general, and slightly above 50% for views on immigration.
Environment ain’t all it’s cracked up to be
Now, when someone says “X trait is 50% heritable”, they’re not only referring just to a specific environmental range, but the “non-heritable” or “environmental” proportion of the variance may not mean what you think it means.
For example, a twin study from 2011 found the heritability of independent reading to be 0.62 at age 10 and 0.55 at age 11.
There have also been studies on the heritability of diet. For example, one study estimated the heritability of diet to be 0.32.
However, that’s just an aggregate. Another study found for example the heritability in men of how many potatoes you eat was 0.68, what we call vegetables 0.24 heritable, red meat 0.34, etc. If you want to see the complete breakdown, you can look through the tables in that study.
Exercise and sports participation also have high sbustantial heritabilities. This study put the heritability of voluntary non-sports exercise at 0.63 for males and 0.32 for females, and the heritability of sports exercise at 0.684 for males and 0.398 for females.
This replicated an older study that found the heritablity of sports exercise at 0.83 for males and 0.35 for females, and non-sports exercise at 0.62 for males and 0.29 for females. This study gave an overall heritability of exercise of 0.49.
Intuitively, it seems that “independent reading” would be a good proxy for the intellectual environment one creates for themselves. Which is to say that, in the United States at age 10 and 11, the heritability of a person’s intellectual environment is about 50% to 60%?
The next thing to note is the difference between “shared” and “unshared” environment. The term “shared environment”, in the context of twin studies, is the environment that twins share with each other as a result of being in the family they are in.
In short, “shared” environment is the environment that was given to you, while “unshared” environment is the environment you create.
Which then calls into question what “environmental” really means. Let me give two examples: one involving strength training and the other involving vocabulary.
1. Bob and Bill, in an untrained state, lift almost exactly the same amount of weights in any lift. However Bob, for genetic reasons, goes to the gym, works out and builds muscles. Bill doesn’t. Thus there develops a gap in strength and muscle mass between Bob and Bill. So is the difference in muscle mass and strength between Bob and Bill due to a genetic or environmental difference? Well, it’s kind of environmental, but the environmental difference stemmed from the genetic difference.
2. The same thing could happen with independent reading; Bob and Bill, if they both read the same amount, would have the same active vocabulary. But Bill, for genetic reasons, does independent reading, and thus has a more active vocabulary.
So when someone says the heritability of a trait is 0.5, and the rest is down to “environment”, unless otherwise specified, they are almost always talking about direct heritability only.
But that “environment” proportion is itself always a function of genetic variation to some degree. And this is important when getting into the next part of this series on civilizational tendencies.