Tuesday, September 09, 2008

More on the Dems and Inequality

Andrew Gelman, professor of statistics at Columbia University and coauthor of the new book Red State, Blue State, Rich State, Poor State, has posted a critique of my New York Times article on his website.

To recapitulate for a moment:

My Times article observed that it's a general rule that the more equal a city or county is, the more likely it will vote Republican; the more unequal, the more likely it will vote Democratic. I dramatized this point by looking at the changes in voting behavior in two northern Virginia counties, Fairfax and Prince William.

Gelman makes two main points in rebuttal. Let me deal with them each in turn.

I

Gelman observes: 

Frum writes: "As a general rule, the more unequal a place is, the more Democratic; the more equal, the more Republican." At least at the state level, it's not so clear. ... Overall, the Democrats’ vote share by state is slightly correlated with income inequality, but much less than the correlation with income itself.

My answer:

Gelman is right that the effect I describe is less powerful at the state level than at the city or county level. But it is still present at the state level, as Gelman himself acknowledges.

But there's a problem with using these state numbers - and a reason for finding the city and county numbers more interesting. The high-inequality states (California, Texas, New York, etc.) also tend to have very big populations. The low-inequality states (Iowa, Utah, North Dakota, etc.) tend to have small populations.

The question naturally arises: I want to study the effects of inequality. But when I compare state with state, isn't there a real risk that I'll confuse the effects of inequality with the effects of population size and economic diversification?

In the end, a county-to-county comparison just seemed to offer more "signal" and less "noise" than a state-to-state comparison.

That said (and as Gelman also acknowledges) it is striking that the states with the greatest increases in inequality are Democratic states. Take a look at this report from the (liberal-leaning) Center for Budget and Policy Priorities (especially page 13).

Of the 10 states where the gap between rich and poor has increased fastest since the late 1980s, 7 voted for John Kerry in 2004.

Of the 10 states where the gap between rich and middle has increased fastest since the late 1980s, 8 voted for John Kerry in 2004.

(To be fair, the effects are less one-sided if you look only the most recent years.)

II

Gelman's second criticism is that I have been distracted by my local circumstances.

Frum lives in D.C. and he is naturally attuned to patterns in the northeast. If he were to go to Oklahoma or Texas, he would see that it is the richer areas, and the richer voters, who are more Republican. By focusing on Fairfax County, Virginia, he's missing the big picture.

Not so!

Although I focused on Virginia in the article - principally because I know it best and can write most vividly about it - I first noticed the county-by-county relationship between inequality and Democratic-voting when studying voting in Missouri while researching Comeback.

The state of Missouri commissioned an especially detailed study of local area inequality based on data from the 2000 Census. Based on this research, the state designated certain areas "equality centers" and others "inequality centers," in turn subdividing the inequality centers into "poverty inequality centers" (areas both highly unequal and heavily poor) and "wealth inequality centers" (areas both highly unequal and containing many rich).

To simplify a little: wealth inequality centers were found in suburbs and gentrifying urban areas. Poverty inequality centers were found in rural counties. And the equality centers were the exurbs around Missouri's main cities.

And again: it was the most equal areas that voted Republican, while the most unequal voted Democratic.

***

That said, I'm off to read Gelman's promising looking book.

09/09 10:38 AM