Public Opinion Watch

(covering polls and related articles from the week of May 30–June 5, 2005)

In this edition of Public Opinion Watch:

  • Democrats and the Middle Class
  • How Big Is the White Working-Class Vote?

Democrats and the Middle Class

Third Way is an organization that describes itself as “Modernizing the Progressive Cause to Connect with Mainstream America,” certainly a worthy endeavor. The founders of the organization include Jonathan Cowan, Jim Kessler, Matt Bennett, Nancy Hale, and Nancy Jacobson, all serious, thoughtful Democrats. They have now entered the what’s-wrong-with-the-Democrats debate with a new report, “Unrequited Love: Middle Class Voters Reject Democrats at the Ballot Box,” based primarily on 2004 exit poll data. The report has been circulating around Washington (though no public link is available for the report as yet) and has been written up in the Washington Post, so it is getting a fair amount of exposure.

Here are my observations on the report, which, while useful, tends to overstate its case and make some questionable claims.

The report is organized around five basic findings. The first is:

“White middle income voters (who constitute three-quarters of the middle class and one-third of the entire electorate), delivered landslide margins to Republicans. The economic tipping point—the income level at which whites were more likely to vote Republican than Democrat—was $23,700, not far above the poverty level….

There is no difference in the preferences of white middle class and white wealthy class voters….”

The claim that the Democrats lost the white middle class (defined here as between $30,000 and $75,000 in family income) by a wide margin is correct. It is also correct that there was little difference between the political preferences of these voters and those of white “wealthy class” (defined here those with over $75,000 in family income) voters in this election. Bush carried both of these groups by 22–23 points.

On the other hand, this situation (white middle class and more affluent—it seems silly to call them wealthy—voters having similar voting patterns) also obtained in the 2000 election. Bush carried both groups by about 15 points in that election. So both groups moved toward Bush by similar amounts in 2004.

As for the claim that $23,700 is some kind of “tipping point” among white voters, I wouldn’t take that too seriously. It’s not in the exit polls directly, of course, which contain only income data by category. What the report’s authors do is assume that, since whites earning below $15,000 supported Kerry and whites earning $15,000 to $30,000 gave Bush a small margin, that the tipping point is somewhere in the latter category. Maybe so, but there’s little methodological justification for ginning up such a precise estimate of that tipping point and their confusing explanation for how they arrived at that precise $23,700 estimate does not inspire confidence.

The report’s second basic finding is:

“Contrary to other voters, blacks conferred overwhelming majorities to Democrats, regardless of income level.”

No argument there. That contention is solid.

The report’s third basic finding is:

“A rapidly growing Hispanic middle class is leaving the Democratic Party.”

The report kind of goes off the rails here. One of the reasons for this is their use of the now-discredited Hispanic sample in the national exit poll, which leads them to report that Kerry defeated Bush among the Hispanic middle class by only 10 points. If you use the more reliable Hispanic sample from the combined state exit polls, you find that low-income Hispanics supported Kerry by 40 points, middle-class Hispanics supported him by 17 points and affluent Hispanics by 4 points.

“Leaving the Democratic Party” seems a bit strong. On the other hand, it is true that Hispanic middle-class voters moved more toward Bush between 2000 and 2004 than other Hispanic income groups and that is cause for concern. At this point, the middle-class Hispanic vote drives the overall Hispanic vote and, therefore, the key to moving the Hispanic vote toward the Democrats is building up their margins among middle-class Hispanics.

As for the “rapidly-growing Hispanic middle class,” one should be cautious about this. While the exit polls do show a big shift toward middle- and high-income Hispanic voters between 1996 and 2000, they show no such shift between 2000 and 2004. In fact, they show middle-class Hispanics actually declining slightly to 45 percent of Hispanic voters in 2004, from 47 percent in 2000, while high-income Hispanics go up a single point, from 23 percent to 24 percent of Hispanic voters. This should perhaps not be so surprising: the real median income of Hispanic households declined by about $2,500 between 2000 and 2003 (2004 data not yet available).

The fourth basic finding of the report is:

“With the exception of those with graduate degrees, education level does not predict voting behavior. Education level predicts income, which predicts voting behavior.”

This just isn’t right. In fact, if you look carefully at the data in their own report you can see that education does have an effect on level of Democratic support, even controlling for level of income. But the report’s authors are intent on showing that, at any given level of education, income has an important effect on Democratic support. This is undeniably true, but they appear to believe that establishing that fact somehow proves education has no independent effect on income. Wrong. Both relationships can and do exist: income has an effect on Democratic support at any given level of education and education has an effect on Democratic support at any given level of income.

Take the white middle class, on whom the report tends to focus. In 2004, Bush beat Kerry by 33 points among non-college-educated middle-class whites, but only by 3 points among college-educated (a four-year degree or more) middle-class whites. Moreover, between 2000 and 2004, Bush’s margin among non-college-educated middle-class whites increased by 15 points, while his margin among college-educated middle-class whites increased by just 7 points.

Lest one think that the differences between college-educated and non-college-educated middle-class whites in 2004 were all driven by postgraduate middle-class whites, those middle-class whites with a four-year degree only were still markedly less pro-Bush (an 18-point-smaller margin) than the non-college-educated.

Conclusion: yes, income matters—but so does education.

The final basic finding of the report is:

“The entrance of married women into the middle class led to a dramatic increase in Republican support.”

This is awkwardly phrased, making it sound like there’s some sort of social trend unfolding, with married women “entering” the middle class and then voting Republican. What they’re really saying here—what their data show—is that married women, particularly white married women with moderate to high incomes, voted Republican in 2004, while unmarried women with those incomes still leaned Democratic (though less so, the higher the income level).

But we knew that.

Anyway, I don’t want these criticisms to lead people away from the report. On the contrary, I think people should grapple with it. The authors of the report and Third Way as an organization are to be commended for making an empirically based case for their political views, instead of simply asserting that their views are correct. The debate in and around the Democratic Party could use more, not less, of this kind of serious data analysis.

And I certainly don’t disagree with the thrust of the some of the final remarks in the report:

“Democrats talk and legislate a great deal about issues that they believe are of concern to the middle class, such as better schools, affordable health care, and job security. This has not translated into middle class votes. Assuming these issues are truly important to middle class voters (and there is no reason to believe they are not), it could be that Democrats have a set of flawed messages that do not reach the middle class. Or, the middle class may simply believe that their schools will not be better, their health care will not be more affordable, and their jobs will not be more secure should Democrats run the Congress and control the White House.”

Either way, the so-called party of the middle class has some serious work to do. Democrats should be grateful to the authors of this report for calling their attention to this challenge.

How Big Is the White Working-Class Vote?

I’ve documented how poorly Democrats have been faring with white working-class voters (defined here as whites without a four-year college degree). In 2004, Bush beat Kerry by 23 points among these voters, according to the National Election Pool (NEP) exit poll, up from the 17-point margin Bush enjoyed among these voters in 2000. Also according the NEP exit poll, white working-class voters were 43.3 percent of all voters in 2004.

That certainly sounds like Democrats have to improve their performance among these voters, and quickly, if they hope to build a majority coalition.

Last week, I presented some data from the newly released Census voter supplement data on the age, race, and education distribution of voters in 2004. Further analysis of these data to look at the specific question of white working-class representation among voters in 2004 reveals that the Democrats’ white working-class problem isn’t as bad as suggested by the NEP data. It’s worse.

That’s because the NEP data underestimate the proportion of non-college-educated people in the voting pool and, therefore, the proportion of white working-class voters. The Census voter supplement data indicate that white working-class voters are actually a majority (51.5 percent) of all voters, rather than the 43.3 percent indicated by the NEP exit poll.

A big challenge for the Democrats just got a little bigger.

Ruy Teixeira is a joint fellow at the Center for American Progress and The Century Foundation.