Bryan Caplan  

Quality Comment

College and Inequality... DeLong vs. Boldrin...
Chris Auld just posted an excellent comment on Brad DeLong's blog.  Brad showed his readers a few graphs on the income-voting connection.  Chris went to the original data:

Basically, the poor were more likely to vote for Obama, but once household income hits $30k the income/Obama gradient is basically flat, so I don't think "the rich vote Republican" is very accurate. Controlling for race diminishes the effect of being poor by roughly a third. After controlling for race, controlling for age, education, sex, and religion have practically no effect on the income gradient.

Controlling for race and other demographics or not, the highest income group ( > $150k ) were slightly more likely to vote for Obama than all groups between $40k and $150k, that is, the rich voted for Obama at slightly greater rates than the middle class (I do not know how to reconcile this result with the maps. I did not weight anything, perhaps that explains the difference.)

This is very much in line with my critique of Andrew Gelman's work.  The real news isn't that "the rich vote Republican."  It's that the effect of income on voting is small, especially after adjusting for race.

Comments and Sharing

TRACKBACKS (1 to date)
TrackBack URL:
The author at Maggie's Farm in a related article titled Friday morning links writes:
    One last weekend of skiing for me, but already planning for summer. My group just leased our usual 345 Conquest for July and August, which we keep docked in Rhode Island. Thinking about beaches (that's a Theo beach gal) and fishing and clams, an [Tracked on March 6, 2009 5:45 AM]
COMMENTS (12 to date)
El Presidente writes:

That's interesting to once-upon-a-time political scientists, like me. Thanks for posting it.

Peter writes:

So what are the odds on Brad deleting the post?

h-dawg writes:

Time to deletion: 4...3...2...

Jesse writes:

OK, I followed the link to Bryan's critique. I don't know if Gelman has ever responded to you, but having read a lot of his stuff I'd guess that the response you imagine him giving ("I show that *within* states, income is important") is EXACTLY what he would say; that's why he's always showing those electoral college maps with bottom-third, middle-third, and top-third (within each state) voters. Those are his most vivid graphs, and indeed that's exactly the picture DeLong used in his post. In your critique, you respond:

Then your real message isn't that "Income is important." Instead, your real message is that "Income is important adjusting for state culture." Or even "In poor states, income is important adjusting for state culture." You sound a lot more like Thomas Frank than you'd like to admit.

But this doesn't seem to be an accurate characterization of what he says. If you look at his book's webpage, you get stuff like:

Myth: The rich vote based on economics, the poor vote "God, guns, and gays."
Fact: Church attendance predicts Republican voting much more among rich than poor.

Myth: A political divide exists between working-class "red America" and rich "blue America."
Fact: Within any state, more rich people vote Republican. The real divide is between higher-income voters in red and blue states.

This doesn't sound like Thomas Frank at all.

hamilton writes:

The real issue is with economists' expectations, not the actual impact of income on voting. As Caplan himself says in the previous post, if one starts from the very strong position that income does not matter at all, or any position weaker than "virtually all poor people [are] Democrats, and virtually all rich people [are] Republicans", then Gelman's results are very relevant. Caplan, pointing to Gelman, also notes that, conditional on other cultural factors ("state culture"), income is also more powerful than without such controls.

The bottom line is this: even in the unconditional graphs Caplan supplies in his previous post, the Democrats win elections by landslide margins held among the poorest Americans (even the poorest whites!), and lose elections by landslide margins held among the wealthiest (again, among the whites). That he sees this twenty-point swing as "small" is a function of his economist expectations, which he himself points out quite cogently:

"As an economist, I was raised to expect virtually all poor people to be Democrats, and virtually all rich people to be Republicans. From this starting point, Gelman's data show that income is practically irrelevant."

Yeah, going from near-certain victory to near-certain defeat is clearly irrelevant. Of course. How silly of me to think otherwise.

I had the same thought as Peter; would DeLong delete the comment.

For those who don't know Chris Auld, he had some notoriety a few years back with his Drink-Income Puzzle paper. It showed, in a study of Canadian males, moderate drinkers earned about 10% more than complete teetotalers, but that heavy drinkers only earned about 4% more than non-drinkers.

He's also the polar opposite of DeLong in another respect; I participated in econ discussions on usenet's sci.econ with him for five years, and I don't have a clue as to Chris' politics.

Steve Sailer writes:

What drives some states red and other states blue is that Republican "family values" appeals don't work in states where white people are less likely to be able to afford to get married and have kids when they are young due to expensive housing prices.

Jesse writes:

Sailer, correlation does not equal causation. Your labeling of certain variables as fundamental drivers (housing prices, cost of living) of other secondary variables (age at marriage, number of kids, responsiveness to Republican Party appeals) is not based on any evidence. You could just as easily tell a thousand other stories to explain the correlations. In particular, housing prices do not drop out of the sky, but are themselves reflections of individual decisions, job markets, and government policy.

I also wonder why you bothered to construct your contrived "Years Married for White Women" statistic, when the much simpler "Median Age at Marriage for All Women (or Men)" would have produced the same stratospherically high correlation with the presidential vote shares.

Methinks writes:

I'm too late! My first thought when I read the comment was that DeLong hasn't gotten around to deleting it yet. I see I'm not the only one familiar with his blog.

Neil West writes:

DeLong's title may be the biggest culprit. What may be a more accurate title is "Yes, Virginia, the Rich Vote Republican in Some States." When looking at it state by state, the income affect may be completely different. In other words, being rich in California may be positively correlated with voting Democratic and the rich of Texas may be positively correlated with voting Republican. If you notice the rich in the largest population centers voted Democrat possible offsetting the states where the rich voted Republican.

Or I could not have properly read Chris Auld's comment.

Greg Ransom writes:

Is it gone yet?

Galileo writes:

Steve Sailer said:
"What drives some states red and other states blue is that Republican "family values" appeals don't work in states where white people are less likely to be able to afford to get married and have kids when they are young due to expensive housing prices."

This may be valid. However, the red / blue states comparison appears to miss the main point. If you look at the local distribution of Republican / Democrat votes, what you see is that the larger the city/town, the more likely it is to be Democrat. And the Blue states then turn out to be the ones most dominated by large cities.

Whether this is because people of a Democratic disposition are attracted to large urban centres, or the large urban centres produce a disposition to support liberal (US definition) attitudes I don't know.

One might argue that localities with smaller populations are more likely to have meaningful communities able to manage their own situation and that may encourage conservative values and attitudes. This is just speculation on my part.

What is not speculation is that if you want to explore the differences in voting direction in the US, you have to start at the local level, not the state level (the states being just an aggregate of the local).

Comments for this entry have been closed
Return to top