Larry Bartels wrote an excellent op-ed on rich and poor voters, ringing many of the bells that we strike in our forthcoming book. Bartels writes:
Do small-town, working-class voters cast ballots on the basis of social issues? Yes, but less than other voters do. Among these voters, those who are anti-abortion were only 6 percentage points more likely than those who favor abortion rights to vote for President Bush in 2004. The corresponding difference for the rest of the electorate was 27 points, and for cosmopolitan voters it was a remarkable 58 points. Similarly, the votes cast by the cosmopolitan crowd in 2004 were much more likely to reflect voters’ positions on gun control and gay marriage.
Small-town, working-class voters were also less likely to connect religion and politics. Support for President Bush was only 5 percentage points higher among the 39 percent of small-town voters who said they attended religious services every week or almost every week than among those who seldom or never attended religious services. The corresponding difference among cosmopolitan voters (34 percent of whom said they attended religious services regularly) was 29 percentage points.
It is true that American voters attach significantly more weight to social issues than they did 20 years ago. It is also true that church attendance has become a stronger predictor of voting behavior. But both of those changes are concentrated primarily among people who are affluent and well educated, not among the working class.
Well put, and nicely backed up by statistical evidence.
One little thing . . .
I don’t, however, follow what Larry is saying in the conclusion of his op-ed:
John Kerry received a slender plurality of [rural, working-class] votes in 2004, while John F. Kennedy and Hubert Humphrey, in the close elections of 1960 and 1968, lost them narrowly.
Mr. Obama should do as well or better among these voters if he is the Democratic candidate in November. If he doesn’t, it won’t be because he has offended the tender sensitivities of small-town Americans. It will be because he has embraced a misleading stereotype of who they are and what they care about.
First, I don’t see why Larry says that Obama “should do as well or better” among rural, working-class voters than Kerry. Unless he’s just referring to the general expectation that Obama is predicted to do better than Kerry overall–Doug Hibbs predicts a 53-54% Democratic vote share in 2008 [typo fixed]. Second, I don’t see why Larry says that, if Obama doesn’t outperform Kerry among those voters, that “it will be because he has embraced a misleading stereotype…” My guess is, if Obama doesn’t do well in this group, it’ll be because the economy is going better than expected.
In any case, I don’t see how Larry’s last paragraph follows from everything that came before. Otherwise, though, I like the article a lot. Maybe there’s just pressure in an op-ed to come to a ringing conclusion?
"Dogma voters" is the more fitting label. "Values voters" is a label invented by people who like to think of themselves as championing good human values. What many of them are pushing actually is dogma. "Values" are "the principles that help you to decide what is right and wrong, and how to act in various situations." Cambridge Dictionary of American English. "Dogma" is "a fixed, esp. religious, belief or set of beliefs that people are expected to accept without any doubts." Id. The two, we can only hope, overlap to some extent, but they are hardly the same. Some of what religious fundamentalists hold up as values others find plainly wrongheaded and even immoral. Labels count. Those pushing the "values voters" label hope it will help them pass off their dogma as values. If they want to push their dogma, that's their right. But "dogma voters" they are, and that's what I'll call them.
Is measuring the percentage point difference in Bush voting among pro/anti abortion voters of a particular class the best way to measure the strength of the effect? If small-town working class voters had a large majority for Bush, and cosmopolitan voters had a very small minority for Bush, then the increase in percentage points for being anti-abortion is almost certain to be far greater in the latter case. How do things look if you compare the ratios of the percentages instead of the differences, or use some other measure of increased support?
Isn't Doug Hibbs predicting a 53-54% Democratic vote share for 2008, not 2004?
Bartels writes "Among these voters, those who are anti-abortion were only 6 percentage points more likely than those who favor abortion rights to vote for President Bush in 2004. The corresponding difference for the rest of the electorate was 27 points, and for cosmopolitan voters it was a remarkable 58 points."
I tried to produce this result from the ANES 2004 dataset and failed. Either I don't know the precise question he asked or I bungled the analysis. Or both. Or neither.
Does Bartels's result come from ANES 2004, and if so, what exactly is the question it answers? I looked at V045132 (abortion position self-placement, trying various interpretations of anti-abortion and favoring rights), V043254 (education level, both counting and not counting a 2-year degree as a college degree), and V043409a (household income) to define subsets of "small-town working class voters", "other voters", and "cosmopolitan."
Is there another, larger post-election dataset Bartels might have used? There are only 227 rural cases in ANES 2004 to begin with, and it's not clear to me that the figure of 6 percentage points difference (between very small subsets of cases) is statistically significantly different from the 27 and 58.
Can anyone regenerate the exact statistic in the article?
Steve