Monday, March 07, 2016

"Voting self-interest" -- you're kidding, right?

Some people claim that you can take a bunch of demographic data and decide who is -- and who is not -- voting for their "self-interest", whatever that might mean, based on that data alone.

Several questionable implicit assumptions are built in, such as that people would always want government handouts over the freedom that comes with economic and personal liberty. Or that people are rational actors 100% of the time - is there anybody that really believes that everybody lives their life logically, all the time, with no emotions driving anything? That’s laughable.

They might as well come out and say "I've analyzed things I can measure about you and now I know more about what is good for you than you do."

Shorter version: “I'm smarter than you, because you’re a dumbass.”

This is patently stupid on its face, and immensely troubling for the following reasons, and probably more I haven't thought of yet:
  1. It is dangerous to assume you can derive somebody's self-interest given a bunch of demographic data. We are more than the sum of things that the Census bureau and tele-marketers know about us. 
  2. It is dangerous to assume a priori that everybody wants handouts over economic and personal liberty. Not everybody likes dependence on others - the history of America comes to mind.
  3. It is dangerous to assume you can build accurate models of human behavior from measuring things ... that we measure ... because we can measure them. Which leaves out things we don't measure. Because we can't. Self-select much?
  4. It is dangerous to assume that voters pay much attention to policy - because they quite clearly do not. Who gets to decide what an individual's self-interest "should" be? What does that even mean? Voters respond in various emotional ways to voice, manner, appearance, all kinds of things that have nothing to do with policy. JFK got millions of women's votes because he was dreamy … and Nixon was Nixon. Was that in their self-interest, or not? I don’t know, and don't care. Is a woman voting in her self-interest if she, contra the "experts", loves herself a dreamy JFK for president and couldn't care less about economics or any stupid b.s. because, hey, look at him! This is not news. People vote with their gut a lot more than we’d like to think, which is probably not ideal, but there it is.
  5. Likewise, Obama got millions of votes because he is "well-spoken" as his running mate famously said. Whose self-interest was met there? The people who like well-spoken black men, apparently. Also, whites feeling racial guilt. And blacks, who turned out almost 100% for the guy. Those black voters got to live to see a black president. This is an accomplishment, at least for them, no? Even though, objectively speaking, that is just as racist as whites voting only for whites, the fact remains, for lots of black folks, that’s all they cared about in 2008. This would meet the criteria of satisfying their self-interest, by definition. Even if others don’t understand it.
  6. Even if you could properly model behavior based on all the crucial factors, and collect and organize all of that, then what? You've got a 40-way (or 4000-way) matrix with a bunch of tiny numbers in it. Yay! All cleared up now, right? Some things are just not measurable, and determining whether others are voting in their own self-interest is surely one of them.
I have this same disagreement with baseball stat geeks too.

People who are obsessed with numbers fall into a couple of traps: just because we can measure something does not mean it is more important than something we can't measure, or that the thing we cannot measure is not a factor; and a belief that we understand completely what drives our behavior so that we can build accurate behavioral models from things we can currently measure.

If all of that were true, we should stop researching it, shouldn't we?