Tuesday, May 27, 2008

VOTING AND GENES--CAUSALITY, PLEASE

I'm back from a relaxing Memorial Day on the beach, but my blood pressure is already rising. And it's not that I have a genetic disposition to hypertension. Rather it's the latest piece of pseudoscience, this time in the form of political consulting masquerading as molecular biology (or vice versa). Thanks to reader MC--an always astute critic of the misuse of genetics in service of social science and punditry.

Today's NYT opines that

"there is tempting evidence of a hereditary component to political choices. There is a strong correlation between the partisan choices of parents and children. Studies comparing identical and fraternal twins suggest that genes are at work alongside the social and psychological influence of parents. Political scientists at the University of California, San Diego have gone another step, identifying specific genes associated with voter participation and partisanship.

It seems as if people with one variant of the MAOA gene are more likely to vote than those with the other version. Among regular churchgoers, those with one type of the gene that make the 5HTT transporter molecule in the walls of neuron cells (don't ask) are substantially more likely to vote than those with the other.


The moral of the story:
If certain genes make us more receptive to political messages, or more or less likely to vote, then we know the next step society must take: Keep the drugs that target the specific genes out of the hands of political consultants
.

The Times has been bamboozled by a very schematic paper by James Fowler and Christopher Dawes at UC San Diego. The paper is full of caveats that serve as red flags. Here's one: "It is crucial to point out at the outset that we cannot test, given our data, the potential causal pathways we suggest. Therefore, the goal of this study is to show association rather than causality."

On the grounds of that "association", they make the huge claim that "these results represent an important step for political science as a discipline. Specifically, they show that incorporating genetic information into our theories and analysis may contribute to a greater understanding of political behavior." Hubris indeed.

In the end, Fowler and Dawes offer an argument by analogy--and a clever, if particularly dubious one in this case. Political scientists have emphasized the role that institutions play in shaping and constraining political behavior. Their argument takes the new institutionalism a step further: "Genes are the institutions of the human body: they constrain individual behavior just as political institutions constrain the behavior of groups of people. In this article we demonstrate that possessing a particular gene is associated with voting activity. Even after controlling for factors known to influence turnout, having a high MAOA allele raises the likelihood of voting by about 5%."

Thanks, but before I come even close to buying the argument that genes determine voting behavior, I want causality, not correlation. And before I begin rethinking the social sciences, we need a lot more hard science, not just speculation. Bringing genetic research to bear on the messy reality of political behavior, especially when the science is still so undeveloped and theoretically unsophisticated, is problematic. Attributing complex human behavior to two genes on the basis of speculative associations is simply not good social science. Our standards should be higher.

3 comments:

Unknown said...

Amen. I definitely agree that reporting about biology determining behavior is annoying because then people (specifically people around me since I'm in academia) start to believe that there's a strong genetic component to all sorts of behavior. Such arguments always seem baffling to me when it generally seems like there are much more direct and understandable sociological arguments (with mechanisms even) for the behavior.

Tom S said...

Amen back to you, CJ. I'm waiting for the follow up article, one that makes the argument that genetic differences explain the higher voter turnout in North Dakota and the lower turnout in New York, or Sweden versus Italy, or most perniciously whites compared to blacks. It's a slippery slope--and one that turns scholarly and public attention away from the well-documented but often still unaddressed problems that are the root causes of low voter participation rates.

Unknown said...

I shiver at the thought, especially since I already feel like I've read articles like that.

What has always frustrated me are the people who accept such genetics-based arguments. Now, with full disclosure I much tend to prefer sociological/psychological arguments for understanding people's patterns and behavior. So I'm honestly curious about the motives/thinking of people who like genetics based arguments. Is it that they live in a very science-centric (and not social science) worldview, to the point that they think phenomena aren't truly explained if you haven't explained them using biological/physical sciences? Or is it the more sinister motive that they want to blame people's problems and behavior on genetics so they can say, in some fundamental way, that they can't really help the person/family/community? The later is why I think racial and class based genetic studies are particularly hurtful. It's like the studies are giving science-oriented educated people a way to shrug off any possible duties to society by saying they really can't help.

I don't know if that's how they're viewed by most people, but I really dislike the idea of people being able to say "biological science has finally settled this particular argument about the differences between two groups." No. It rarely does because such studies are based on indirectly inferring causality from data. (And, unfortunately few people understand how notoriously hard it is to infer the correct things from data.)

Any thoughts on this anyone?