Originally Posted by Pennglock
What is with the idea that you need trace every phenotypic trait back to all its coding base pairs in order to draw correct inferences about heritability?
FLM (sorry to pick on him again) implied that genetic predisposition to crime exists that it is correlated with certain nationalities, as did you (effectively, that it would explain varying crime rates in the USA). So you need a metric for this, and some regression score is not really a metric if you believe there is an underlying mechanism -- it should be identifiable given these overwhelming and robust advances in the field of genetics. If you want to use this, in the context of this thread or another other political one, as a means of intervention or policy development, you should be able to score the individuals in a population.
Here's a scenario, then. If a person comes from a region where a regression algorithm shows "heritable violence" is greater and statistically significant, does the INS turn that person away?
If we can take a saliva swab and determine someone's inclination toward violent behavior (even still a probabilistic score, unless you're a fatalist), should that person be denied a visa application based on that ? What about for purchasing weapons?
Take a couple of premises in order:
1. Every human trait that can be measured is heritable. The metabolic stuff, extremely so; the behavioral stuff, still more heritable than most people believe. We have very good estimates of the H2 for every trait under the sun thanks to experiments comparing monozygotic and fraternal twins. This 2015 meta study in Nature was incredible: http://www.nature.com/ng/journal/v47/n7/full/ng.3285.html
That's an interesting study, not least because of the size. But the conclusion boiled down to: heritability cluster strongly within functional domains, and across all traits the reported heritability is 49%.
“If we pool all data, the balance between nature and nurture is near perfect: across all traits the heritability is 49 percent, and environmental influences account for 51 percent”
-Senior author Danielle Posthuma, Ph.D.
The article tells you nothing about gene function, rather that small contributions from many genes add up to certain patterns we are able to detect and draw generalizations based on correlative inference. I have to read through it more carefully but there also doesn't seem to be much in terms of the validity of assuming equal environment in all these studies going back half a century. Seems like an important factor?
2. Certain behavioral traits are robustly predictive of individual life-outcomes. There is a lot of junk in psychology today that doesn't replicate, but psychometrics is not having that problem. The most informative are intelligence, future-time-orientation, and Big 3/5 personality traits. Further, for these most predictive traits, no one has been able to show any impact of shared environment (nurture). It's all genes + stochastic noise.
This seems fine until we go back to the twins study which showed that environment even within twins plays a massive role.
3. Different population groups of people vary in their distribution of traits. In terms of the average and variance within a population. IQ across races is a fun one to bring up if you feel like being socially radioactive. But as Larry Summers showed when he speculated that men's overrepresentation in since was linked to their larger variance in quantitative intelligence--meaning men would have more area under the curve at +3sigma even if men and women had the same mean IQ-- the innocuous stuff can still get you chased out of town by a pitchfork wielding mob.
I agree with that, and it's easily proven for physical traits, perhaps even diseases. But then to go from that to the role these variances place in cognition is a massive leap.
None of the above should be controversial to any numerate person who's bothered to do the homework.
I'm quite the innumerate fool, though
It can tell you a lot about individual outcomes within a single society, especially a meritocratic one like the USA.
Comparing outcomes of societies to each other gets a lot messier, though. You have so many more variables to control, and path-dependence issues. BUT, how this information should be used, and generally isnt, is to control for population differences when studying social and cultural phenomenon. I've read entire tomes on economic development in Africa and South America that made zero reference to population IQ... that seems like something you'd want to account for.
Again, this is predicated on the assumption you've made that there is a direct and identifiable relationship between genetic variance and and the variation in cognitive abilities.Edited by the shah - 7/7/16 at 8:14am