This paper is worth taking extra time to disassemble: “genopolitics” is part of a growing trend in the academy.
The Mating Game
If you can pass on your political “orientation” through your genes and your goal is national amity and you’re a Republican you should marry a Democrat. Your babies will be a purple blend of red and blue: they’ll subscribe to the New York Times, but won’t read it; they’ll attend NASCAR events, but pack arugula sandwiches; they’ll sign up for Womens Studies courses, but only for the easy As. Why, given three or four generations of selective breeding, we’d arrive finally at Utopia! As it is, given the growing disharmony in our public discourse, it must mean the Eloi and the Morlocks are only mating with their own.
Thoughts like these came from reading the peer-reviewed “Genetic and Environmental Transmission of Political Orientations” in Political Psychology by Carolyn Funk, John Hibbing and Kevin Smith and a bunch of others, an example in a growing genre of research which argues political “orientation” (their word) is somehow—never mind how—selected for and passed on genetically. How to prove it?
In Hibbings et al.‘s case (Hibbing is most well known), they latched onto a bunch of monozygotic (MZ, genetically identical) and dizygotic (DZ, genetically similar but different) twins and asked them a series of questions. They figured that if “environmental determinants of trait similarity are held constant, and MZ twins are observed to be more alike than DZ twins [in their answers to the questions] the greater similarity of MZ twins must logically be due to genetic influences”. Simple, no?
The twins were from Minnesota born between 1947 to 1956. They tossed the boy-girl dizygotic twins, for obvious reasons. The database from which the twins were drawn contained “approximately 8,000 twin pairs”, but the authors never say how many were asked to be part of their study. They admit that everybody contacted did not participate, but all we know is in the end 1,192 people were analyzed. This is 596 pairs: 365 monozygotic, 240 dizygotic. Is that a good or bad response rate?
The questions the twins answered deserve special mention because much “science” is done in the same fashion. Works like this. A group “develops” a question like “On a scale of 1 to 5, how much do you dislike liberal policies?” and asks it of a group of volunteers (i.e., college students). After some tooling, the group calls this question a “validated instrument” and gives it a handy label, like “The Hate Index”. A second group develops a second question, like “On a scale of 1 to 9, how strongly do you support amnesty for illegal aliens?” It too will be validated and given the label “The Human Liberty Scale.”
A third group will ask both questions to a third set of volunteers. It will be found that the answers are crudely correlated, in the technical, statistical sense that those who answered high or low on the first question will more often answer high or low on the second. The third group will write a paper announcing this correlation with the words, “Hate drives dislike of undocumented immigrants.” Finally, a news organization will report on the study: “Republicans despise Mexicans.”
I do not exaggerate. Open any “soft” science journal and prove it to yourself. Same thing here.
“The first personality measure is a 44-item battery…to calculate individual measures for the personality traits extroversion, agreeableness, conscientiousness, neuroticism, and openness.” The word battery maintains its militaristic metaphor, I suppose. “The second personality measure is a right-wing authoritarianism scale created using the average of responses to a shortened 15-item battery”.
Also, “The two value measures are an egalitarianism scale, measured as the average response to five items that is adapted from Feldman and Steenbergen (2001), and a society works best index, which consists of an additive scale of social orientations for 12 items developed by Smith et al.” Lastly, “The two measures of political ideology are a 7-point ideological self-report (1 = extremely liberal, 7 = extremely conservative), and a Wilson-Patterson Index, which is a measure of conservatism based on responses to 27 issues adapted from Wilson and Patterson (1968).”
What is a “conservative,” what a “liberal”? The beliefs and policies of the two biggest parties in the United States have changed radically over the last century. Liberals once preached against “the man” but now embrace the surveillance state. Is the liberal who wants to maintain government growth “conservative” and the conservative who would dismantle bureaucracy “liberal”? Who is more “open to change”, liberals or conservatives? Depends on what is changing, on who you ask, on when you ask.
There is useful information to glean from asking people questions, but the more complicated a subject becomes the less it is likely simple questions suffice. Ask “Which is larger, 15 or 7?” and you reliably peg the mathematically literate. But try and discern the difference between (say) a person’s level of “bravery” and of his “courage” through questions? Could you even design a questionnaire that could reliably distinguish people’s recognition of their shades of meaning, let alone one which discovers if this recognition is genetically heritable?
Anyway, as claimed: a wearying cluster of questions given arbitrary labels and dumped into a statistical chomper to discover “correlations.” All these correlations ever reliably prove is that people answer similarly worded questions similarly and dissimilarly worded questions dissimilarly. Yet that never stops the theorizing, as we’ll see.
Even if the questions are silly and don’t mean what the authors assert, it is still true that questions were put to pairs of genetically identical and genetically different twins, and if differences existed between the groups in the way these questions were answered then something accounted for these differences. That something can only be nature or nurture, biology or environment, or a combination of the two.
Nobody, not even the authors, not even Richard Dawkins, believes genetics predestine one to hold certain political views, but many suspect genetics play a role in “orienting” personalities in this or that direction. Yet which directions? The focus is always strangely narrowed to matters of politics. Nobody asks whether quarrelsome fathers sire quarrelsome daughters or whether odious mothers produce odious sons. Instead, researchers are sure the blue genes of parents are passed down to their children, and vice versa.
Now, nothing makes an academic curiouser than wondering why others don’t think like him. He voted for the Democrat, why didn’t his banker neighbor? The academic figures since the truth is out there and some can’t see it, some must be willfully ignoring it or something in their biological makeup causes the obvious blind-spots.
So he “researches.” The results are never pretty.Take the Harvard guys who told us that attending Fourth of July parades turns people into Republicans—only it turns out they never measured parade attendance. Or like those guys who told us that even a brief glimpse of the American flag turns one into a Republican—only their statistical model was so absurd that even a politician wouldn’t rely on it. Interested readers can peruse the The New Mismeasure Of Man: Official List for more examples (it needs updating).
Heritability surely plays a role in personality, but how personality filters up to decide political questions is a mystery. Is there a gene for the kind of gullibility required to swallow the concept of micro-aggression? Can genetics predict prodigal sons or how beliefs change as one ages? Do one’s genes really make him prefer the untested to the known? Who knows.
We can ignore their Table 1 where we learn for example “Right-wing Authoritarianism” correlates (in the technical sense) weakly and negatively with “Openness.” Again, all this really proves is that people answer similarly worded questions similarly, and dissimilarly worded questions dissimilarly.
Hot tip: whenever you read a scientific paper where the actual data is tossed and replaced by statistical soup your Spidey sense should tingle madly. Why?, you should ask yourself, why are they not content with the actual numbers? Are they trying to hide something? Chances are, yes. Though it may only be that the researchers are infatuated with fancy techniques.
Hibbing et al. tossed their data and replaced it with statistical soup.
Specifically, they took answers from each question (“variables”) in order “to partition the variance of a single observed variable (a phenotype) into the latent (unobserved) components associated with genetic influence (A), common environment (C), and unique environment (E).” To observed the unobservable, they used something called the “Falconer approach”, which sounds more like a route up Mt Everest than a method of analysis, but let it pass. This approach “decomposed” the variance of each answer in some fixed way.
This is completely arbitrary but very inventive. The sexy sounding scientific labels “genetic influence (A), common environment (C), and unique environment (E)” were not found in the data. They made them up. They could have just as well have called them the Larry, Moe, and Curly dimensions of variability, and they could have gone on to tell stories about how the Larry dimension was erratic, the Moe dimension forceful, and the Curly dimension strangely adorable.
Forgive me if I worry you did not understand my last point. My powers of explanation are weak and I sometimes rush. The authors claimed that their method proved that so much of the variability in answers was due to “heritability”, so much due to “shared environment”, and so much due to “unique environment.” But there just is no way for them to have known how much of the differences in any answer was due to anything, except twin status.
Why? The only data they had was whether each person was a monozygotic or dizygotic twin and their answers on arbitrary questions. They did not and could not measure how alike each set of twins were raised; there were no environmental variables measured.
What they should have done is something like this. For each question, subtract answers between twins. So that if Mr Smith 1 answered “7” and Mr Smith 2 answered “6” the difference would be 1 (in absolute value; there is no need for numerical signs since there is no ordering of twins). Look at the distribution of subtracted answers for monozygotic and dizygotic twins and seen if there are differences in these distributions (not just means or correlations, which are rotten measures for discrete data like this). We would know whether there was a basis for believing monozygotic twins answer more similarly than dizygotic twins. For instance, if heritability were true (with respect to these questions) we’d expect the distributions for monozygotic twins to cluster around 0, whereas the dizygotic twins would be more spread out.
As it is, the statistical soup hints that the correlation on manipulated answers between monozygotic twins is slightly higher than for dizygotic twins. From this they try to say how heritability and environment “cause” these differences and how correlations between the arbitrary questions are politically meaningful (or meaningful to politics) using an ever-increasing tangle of statistics (all rehashing the same data repeatedly). But all this can be ignored. Grant it to them, even. Suppose it is true that there are differences in the way monozygotic and dizygotic twins answered questions. Does that prove genetic heritability with respect to these questions?
Any good scientific paper includes a list of hey-we-could-be-wrongs. This paper had none. Luckily, its limitations have been highlighted by at least one other researcher, “Evan Charney, an associate professor of public policy and political science at Duke University”, as reported in the Journal Star.
Charney nails the main objection: identical (monozygotic) twins are often raised differently than non-identical twins. Parents, having what they believe to be senses of humor, tend to dress identical twins identically, give them similar names, sign them up for identical activities, and on and on. This is why smart twin studies are keen on analyzing twins raised apart. Much easier—but not easy—to claim that similarities are genetic. Even twins raised apart share many environmental influences.
But studies of twins also tend to neglect the role other biological mechanisms, such as agents that activate certain aspects of genes, could have on imparting characteristics, Charney said. He said genes are only a small part of a much larger system of inheritance.
“Everything that Hibbing and company are working with is a completely static, reductionist, antiquated conception of genes and how they work,” he said.
Their research, Charney said, also reduces all political ideologies to conservative or liberal, neglecting innumerable political ideologies that can’t be considered either liberal or conservative.
“I think this is just completely pseudoscience,” he said.
Can’t say better than that.
Thanks to Tom Hendricks for alerting us to this study.