One of the most popular and political papers in academia right now is “Rich state, poor state, red state, blue state: What’s the Matter with Connecticut?”, co-authored by Columbia School of Social Work Prof. Andrew Gelman. Gelman used statistical models that break down many of the prevailing ideas about political polarization and income in the recent presidential elections. The Blue & White’s Ethan Pack sat down with Dr. Gelman to talk about which party really represents the wealthy.

What motivated your research?
At first, it was more of us reacting to what is in the news media. There has been so much talk and maps of red and blue states. The media has been repeating that Democrats were the party of the rich. But every poll that comes up shows that richer Americans are more likely to vote Republican. The idea that it was only elite Democrats, and working class Republicans—we knew it wasn’t like that.
But there is a history to all this talk. It goes back to the 1930’s, labeling the parties by a fight between the common man and the rich. At the time, many considered the New Deal to be an imposition of the elites, and fought against it as such. This trend strengthened in the Nixon-Kennedy election, where Nixon was the common man and Kennedy was wealthy and powerful.
The Democrats also have made the counter-claim, that the Republicans are the party of the rich. And then I started to see non-ideologically-based journalists making this claim. When I read [New York Times columnist] David Brooks say this, I realized that this is how they see it. Brooks is a Republican, but he’s not a spinster. It would be one thing if I read William F. Buckley propounding this, someone with a clearly directed ideological agenda. But it was beyond that. It had really taken hold in the mainstream media. And part of this is because the red-blue map is so misleading.

When in the research process did you first get a sense of some of the more original conclusions your study?
Only at the end really. We needed lots of data: state-by-state data, public opinion surveys, exit polls of course. We really had to do a lot before we could graph the conclusions. I am embarrassed, it took six months, really almost a year, to fit the right models.
The conclusions only came up after looking at all the sets of raw data; for example, the disparity between the slopes for income and voting Republican in Connecticut and Mississippi.
Right now, I’m writing a book about regression models. This study changed the whole way I look at that, and now I have had to change the book. Interesting phenomena came up.

What do you think of the public reaction to your paper in the press, academia, and on the web?
It all happened on spring break. I was away from school. I grew up in a Washington D.C. suburb, but my parents moved. It would have been fun for them to see me in the morning paper [in Washington Post writer E.J. Dionne’s column], but it didn’t happen.

The political discourse on these stereotypes is today, perhaps more pervasively than ever, often reduced to bombastic polemics and simplistic “parodies”, as E.J. Dionne wrote. How substantial is the contrast between your principally statistics-minded approach and much of the wider discourse? And how does that contrast effect its reception?
We used statistics because we wanted people to get a sense of the real subtlety.
Of the four authors of the study, we had two Democrats and two Republicans. And we were working all across the country [Chicago, St. Louis, Andover, New Hampshire, and New York]. It gave us a good overall perspective.
Some people think income and politics should be deterministic. Others don’t want to admit that they correlate at all. Democrats want to know why low-income voters don’t all vote for them. Republicans don’t want income to be predictive of voting patterns at all. Of course, they like the rich to vote for them and be active, but being identified solely with the rich is dangerous to either party. It suggests you aren’t the party of the people.
I didn’t like the discourse on both sides. Each one delegitimizes the other side’s voters. People spout ideas, like that Democrat voters are drug dealers, welfare queens, college professors. And that Republicans are all church-going holy rollers or fat cats in corporate suites. That isn’t how people decide their vote. Half the country is voting Democrat and half the country is voting Republican, so clearly there is some overlap in the categories commonly imagined.

You mention that the media misperceives things, which you correlate to their location in Washington D.C. and New York.
Journalists are disproportionately Democrats, to the extent they identify with a political party. And then you have the elite journalists, who are far richer than the average American. I’m saying if you compare them to the waitress or service worker; journalists are poor compared to some doctors and CEO’s.
But journalists exist largely in this quadrant, wealthy and Democrat. And then they see the country voting Republican, which is different from their quadrant, so they assume everyone else in the population who isn’t like them in voting patterns isn’t like them economically. So they think the poor vote Republican.
It is natural for journalists to feel that correlation. It is called the Availability Bias in psychology studies. You infer about the whole from what you have experienced. It causes errors by failing to see the more subtle areas at the correlation level.
If a few journalists and people see our study, maybe when it comes to writing about 2008, they will correct this and change.

Are academics making the same bad assumptions as journalists? Both groups are concentrated on the coasts.
I wouldn’t say so, as much. Political scientists, economists, and social scientists are very quantitative. We are very cautious. We don’t say much until we see the data, and even then we don’t say much.
I teach this in terms of the “auto mechanic’s payoff.” You bring your broken car in to the mechanic, and he tells you the problem is your engine. You pay him to fix it, get your car back, and it still doesn’t work. “Oh, it was your muffler,” he says the next time, and you pay him again. Whether he diagnoses the problem right or wrong on his first guess, he gets the payoff. It’s worth it for him to guess.
We don’t get a raise for being in the news. But our standing is lowered if we get something wrong. Sure, it is nice to be in the news. But we won’t risk getting something wrong the first time for that. Now, academics do say stupid things. But that is at the theoretical level. That’s just from their perspective, not factually stupid. You see all these stupid quotes by academics, but they aren’t saying things that could be verified.
Journalists have more of the auto mechanic’s payoff incentive. They are paid to attract attention. If it turns out they were wrong, they will just guess again, and still get the money for the first guess, the first time they grabbed your attention. Like the auto mechanic, if they are too cautious to guess wrong, only then do they lose the payoff.

Does the Columbia community mischaracterize political polarization in America? Are we worse off because we are in expensive, private higher education?
Probably. It would be natural to think our community is richer and more liberal. Students are more likely to come from higher income families.
We proposed to the Psychology Department to do studies of political perceptions through little stories or vignettes. You say, “this person is a NASCAR dad; this is a waitress-mom,” and have them predict which party that person would vote for.
We both realized it wouldn’t work that well at Columbia because it would be hard to get conservatives. We could find other colleges. But in any state, the university town is going to be different from the rest, and is going to be the most liberal.

Are you a Democrat? Do you think the Democrats should run Hillary Clinton in 2008, or a southerner like John Edwards or Mark Warner?
There is a general belief that the more centrist policy will get more votes. Candidates want to be perceived to be at the center, and not be at the center at the same time.
I won’t say what party I affiliate with. I will say our study had two members from each party. We have common goals: peace, prosperity, and so forth. We choose different means to the same end. But, as to my party, I’ll make you do the work.
You can find out who anyone votes for. That is what we did in our study. You could guess based on a predictive model, by my profession and where I live. But predictive models aren’t always right. It would be the same if I asked you.

I am from upper-middle class suburban Kansas, but I’m liberal, so I don’t fit the slope models of your study.
Yes, but as a Kansan who went to university in New York City, the predictions begin to change.