Friday, July 23, 2010

Nature Affects Nurture Politics

As he is so good at doing with his thoughtful posts and wide-range of links, Richard from Alleged Wisdom has timesucked me into pouring over social science articles in order to try to make sense of them.  Instead of, you know, doing the "quantum astrophysics" I would otherwise be required to do by my contract of employment as a physical science dataslave research assistant.  Darn that liberal arts college for making me take things like economics and history, instead of letting me just bury myself in pages and pages of integrals as if I I wouldn't actually hate not to ponder everything else and take an extra semester to graduate.

Note: I have kept this as a draft for a long time, trying to make it intelligent and coherent and Not Wrong as much as possible.  I probably failed at all three points to some degree, but it was sure worth the try.

The idea:  Studies show that political inclinations have a hereditary component.  Hereditary, like, someone determined by the biology you inherent, not just preferring the opinions of your parents 'cause that's how mama raised you.  I must stress that this is the  "gut feeling", "it's just SO!" aspect of political inclination, not any sort of broad statement about your entire thought processes being ruled what genes you inheret, nor is it any kind of statement about the superiority or inferiority of one side or the other.  Actually, it belies many attempts to argue the superiorioty of one side or the other by noting that positions may well be influenced by an essentially random component.

This idea has rocked my world over the past few days because I am inclined (genetically or otherwise) to be skeptical of a) proposed genetic explanations for something as complex as emergent human behavior, b) the potential distortion to which any "genetic" links to human behavior can be put by clumsy science reporters and/or scientists with an agenda. ( You think scientists don't sometimes have an agenda?  Do you then think they ain't as human as the rest of us?)

Richard's link was not to the study but to a commentary at Overcoming Bias, which (as usual) has a decidedly philosophical take:   unless you are convinced you have sufficient experience or evidence that makes you better informed than others who might take a position on a political issue, you should consider shifting your views toward the average because it might be that genes, thus, natural bias, rather than rationality, is guiding your inclination.  Overcoming Bias is, as it sounds, a space for exercising rationality to recognize and surpass bias in search of objective truth, so recognizing "genetically"-induced biases is critically important--if that pursuit is what trips your trigger.  Call me Carol Gilligan, but myself I'm not convinced of the uselessness nor the irradicability of subjective truth (this perhaps being a female trait), but that is a subject for another time and not the thesis of this post.  I do applaud effort to find and remove bias from intellectual discourse, which, sadly, political discourse often is not. 

Though always worthwhile at that blog, to save you the timesuck of reading the comments as well, I can sum them up in three main points:  a) Some protest the numerical description of a political position and the ability to simply adjust it at will.  How, they say, could you just shift your position on some issue from, say, 90 on a 1 to 100 scale to an 81, just because?  How do you numerical-ize political positions anyway, and why is shifting just because useful?  (The answer is to be more likely to have the true opinion, if genetic political bias is a source of misguidance toward less truth)

b) Some are confused or downright skeptical that political inclination could be hereditary at all, and how these scientists actually pinned down "percent caused by heredity" of a whole slue of issues.  Me too, so more on this later.

Some commenters, reflecting the not-entirely-undeserved undertone of intellectual arrogance that nonetheless often irks me about Overcoming Bias, assert that they do not want to adjust their opinions to the average because the average is, by definition of average, influenced by the far-less-intelligent among us, and they don't want their intellectual opinions to be brought down by those of stupider people.

 I admit to feeling that way myself sometimes.

Yet I have observed very intelligent people present on both the right and the left, (and, I hope, the left-center, where I claim to reside) and I have observed some appallingly stupid, or at least, not-too-prone-to-critical-thinking people on both sides.   So clearly, there is more to do with it than IQ.  I would naively posit that that extra factor is experience, so that an intelligent climate scientists is more likely to be in the correct opinion about global warming than an intelligent social scientist, but not as likely to be correct as that same social scientist is about the efficacy of government poverty prevention programs, for example.   However, clever Mr. Hansen covers the experience question with his assertion that yes, if you have reason to suspect you have more experience with something than someone else, than your position is more likely to be correct, but still there is more to the story and that may well be your genetics.  I would then posit with equal naiveté that the non-rational, gut-feeling part of politics, rather than genetics, comes from your moral foundations which in turn come from your upbringing and perhaps even your biology only so much as your moral foundations do.  We know that nature influences you and nurture influences you, a more nuanced view being that nature influences how and to what degree you are influenced by nurture.  It gets complicated so quickly--now he's got me all curious!

So I read the older study mentioned (not having Mr. Hansen's access to pay-service social science journals in order to get the new one), because I did want to know how exactly a quantification of the hereditary nature of each political question had come about.  Being a physical scientist not a social scientist and not particularly schooled in statistical analysis to boot, I don't know if I've been able to grasp the true subtleties of the methodology, but I will do my best.

The study uses twins, both the "egg" kind in which a fertilized egg splits into two copies with identical genetic information*, and the "fraternal" kind in which two distinct eggs were hanging around to get fertilized instead of one.  In the case of egg twins, genetic material is 100% the same, because the zygote copied itself.  In the case of fraternal twins genetic material is just like normal siblings:  50% from the mother, 50% from the father but the what-from-each is randomized, still equaling 50% shared material overall.  Whatever influence the uterine environment has on development, fraternal twins get to share that--but their different genetic material might mean they respond differently to it.

Yet fraternal twins make better comparison for genetic differences than siblings because their environment, while still very different due to ever-present subtle socialization, parental preference, experience differences starting from when you are physically able to occupy spaces other than the mother's womb, etc, is still more the same than it is for non like-aged siblings.  Or, if you want to control for environment, then you can look at raised-apart fraternal or egg twins.  To really say something you really ought to get a large sample size comprised of both.

The researchers looked at 30,000 twins and some of their close relatives in Virginia.  They gave them a political survey designed to assess conservatism or lack thereof by providing words and phrases like "disarmament", "socialism", "patriotism" and asking for gut reactions either positive, negative, or don't know/neutral.  They looked for correlation** in these surveys between egg twins, between fraternal twins, and also between a twin and another non-twin family member like a parent or sibling for comparison.  If I'm reading it right, the idea is this:  since egg twins have 100% identical genetic material, then an overall correlation of 1 between egg twin pairs would indicate a completely genetic-driven result.  Since fraternal twins have only 50% shared genetic material, then an overall correlation of .5 between fraternal twin pairs would indicate something genetic-driven.  Of course you don't get results that are 1 or .5, because politics is NOT purely a result of genetics.  But you can differentiate genetic and environmental influences on twins giving similar answers using the following math:

To find the genetic component, subtract fraternal twin numbers from egg twin number and multiply by two.  So if purely hereditary, a 1.0 for egg twins minus a 0.5 for fraternal will yeild 1.0-0.5=0.5*2 =1.0, or 100% genetic.   To find the non-genetic, thus environmental, component, double the fraternal twin correlation then subtract the egg twin correlation, so that in a purely genetics-caused situation, 0.5 for fraternal twins *2 = 1.0-1.0=0.  Thus you're scaling your result to reflect level of genetic or environmental influence on a scale of 0 to 1 for each, and you can, as they did, report results on genetic scale and on environment scale for a variety of political issues.

Since I'm reminded of the cliche "correlation does not equal causation"** and don't have a solid grip on the operative definition of either of those words as social scientists use them, I do worry that I'm missing something, because it seems to me that they are missing something.  How can we be gauranteed that a correlation of 1 between egg twins and 0.5 between fraternal twins is necessarily a result of shared genetics?  It does match the percentage of shared genetic material...but is it that simple?  

Regardless, because of the survey itself, it is imposable to take culture completely out of it.  Ideas such as "Republicans" (remember, respondents responded positively, negatively, or neutrally in each case), "Women's Liberation", "Modern Art" are very much culturally defined.   The way to interpret these results thus is not "my genetics make me prone to having such and such gut feeling related to such and such political opinion."  Rather, genetic variability is the underlying wildcard paired with cultural variability, perhaps determining how or to what extent we react to the cultural variable we are dealt.  It is then more accurate to say that "my genetics make me more prone to reacting in such and such way in response to such and such cultural stimulus which translates into such and such gut feeling."  The end result is the same:  we sometimes have strong opinions based on non-rational components of our psyche.  The fact that genetics can be pinpointed is both interesting and yet does not add anything particularly new to one's quest to overcome innate biases.  The study is not providing a roadmap which takes one from particular alleles to particular tendencies for responding to particular environmental factors in particular ways that lead to particular political gut feelings, thus "genetics" may not even be a good word to use.  Such a study may not be possible, and this one established fraction influenced by heritability of specific political inclinations.  The hereditary fraction was never 1, the average heritability of all issues was 0.32 in the old study.  Toward the quest to Overcome Bias: consciously avoiding decisions based on gut feelings as opposed to reasoning and experience should be an obvious course of action, no matter the source of the gut feelings.  Is that harder to do if "genetics" is your enemy, as opposed to straight acculturation?

We physcists love to doubt and disdain the difficulty of social sciences, it's true.  But we physicsts like to take situation and simplify them into models and Do Math on those models--and how can you create any simplified model of the human brain? 

*I'm talking the part of our genome that makes us humans and individually expressed humans at that.  Yes, we share 99% of our total genome with pretty much everything else on this planet.

*The authors note that "correlation" is technically "polychoric correlation analysis", which is standardly used on self-repsonse surveys where things are rated on a scale. I do not know any more about it than that, and since I'm supposed to be doing a very physical-science-y "cross-correlation" of spectra, I haven't got the time teach myself another barrage of statistical methods.  Sorry.

No comments:

Post a Comment