Advertisement

The Case of the Missing White Voters, Revisited

By Sean Trende - June 21, 2013

‹‹Previous Page |1 | 2 |

For those with long memories, this stands out as the heart of the “Perot coalition.” That coalition was strongest with secular, blue-collar, often rural voters who were turned off by Bill Clinton’s perceived liberalism and George H.W. Bush’s elitism. They were largely concentrated in the North and Mountain West: Perot’s worst 10 national showings occurred in Southern and border states. His best showings? Maine, Alaska, Utah, Idaho, Kansas, Nevada, Montana, Wyoming, Oregon and Minnesota.

We can flesh this out a bit more by running a regression analysis, which enables us to isolate the effects of particular variables while holding other variables constant.*** We’ll use county-level data, which is granular enough that we can feel more comfortable that we avoided ecological fallacy problems. You can see the overall results here. Almost all of the variables are significant; only the population density variable is of no value.****

For those who didn’t click over to the chart, we’re pretty confident that the voters were more likely to stay home if they resided in states that were hit by Hurricane Sandy, that were targeted by a campaign in 2008, that had higher foreign-born populations, and that had more Hispanic residents. The latter result probably suggests a drop-off in rural Hispanic voters, who are overrepresented in an analysis such as this one.

We’re also pretty confident that the voters were more likely to turn out if they resided in counties with higher median household incomes, high population growth, a competitive Senate race in 2012, or that were a target state in 2012. Counties with higher populations of Mormons, African-Americans, and older voters also had higher turnout, all other things being equal. None of this is all that surprising.

Perhaps most intriguingly, even after all of these controls are in place, the county’s vote for Ross Perot in 1992 comes back statistically significant, and suggests that a higher vote for Perot in a county did, in fact, correlate with a drop-off in voter turnout in 2012.

What does that tell us about these voters? As I noted, they tended to be downscale, blue-collar whites. They weren’t evangelicals; Ross Perot was pro-choice, in favor of gay rights, and in favor of some gun control. You probably didn’t know that, though, and neither did most voters, because that’s not what his campaign was about.

His campaign was focused on his fiercely populist stance on economics. He was a deficit hawk, favoring tax hikes on the rich to help balance the budget. He was staunchly opposed to illegal immigration as well as to free trade (and especially the North American Free Trade Agreement). He advocated more spending on education, and even Medicare-for-all. Given the overall demographic and political orientation of these voters, one can see why they would stay home rather than vote for an urban liberal like President Obama or a severely pro-business venture capitalist like Mitt Romney.

3. These voters were not enough to cost Romney the election, standing alone.

But while this was the most salient demographic change, it was probably not, standing alone, enough to swing the election to Obama. After all, he won the election by almost exactly 5 million votes. If we assume there were 6.5 million “missing” white voters, than means that Romney would have had to win almost 90 percent of their votes to win the election.

Give that whites overall broke roughly 60-40 for Romney, this seems unlikely. In fact, if these voters had shown up and voted like whites overall voted, the president’s margin would have shrunk, but he still would have won by a healthy 2.7 percent margin.

At the same time, if you buy the analysis above, it’s likely that these voters weren’t a representative subsample of white voters. There were probably very few outright liberal voters (though there were certainly some), and they were probably less favorably disposed toward Obama than whites as a whole. Given that people who disapprove of the president rarely vote for him (Obama’s vote share exceeded his favorable ratings in only four states in 2012), my sense is that, if these voters were somehow forced to show up and vote, they’d have broken more along the lines of 70-30 for Romney.

This still only shrinks the president’s margin to 1.8 percent, but now we’re in the ballpark of being able to see a GOP path to victory (we’re also more in line with what the national polls were showing). In fact, if the African-American share of the electorate drops back to its recent average of 11 percent of the electorate and the GOP wins 10 percent of the black vote rather than 6 percent (there are good arguments both for and against this occurring; I am agnostic on the question), the next Republican would win narrowly if he or she can motivate these “missing whites,” even without moving the Hispanic (or Asian) vote.

4. The GOP faces a tough choice.

Of course, it isn’t that easy. Obama won’t be on the ticket in 2016, and the likely Democratic nominee, Hillary Clinton, could have a greater appeal to these voters (current polling suggests that she does). But there are always tradeoffs, and Clinton’s greater appeal to blue-collar whites, to the extent it holds through 2016, could be offset by a less visceral attachment with young voters, college-educated whites and to nonwhites than the president enjoys.

But the GOP still has something of a choice to make. One option is to go after these downscale whites. As I’ll show in Part 2, it can probably build a fairly strong coalition this way. Doing so would likely mean nominating a candidate who is more Bush-like in personality, and to some degree on policy. This doesn’t mean embracing “big government” economics or redistribution full bore; suspicion of government is a strain in American populism dating back at least to Andrew Jackson. It means abandoning some of its more pro-corporate stances. This GOP would have to be more "America first" on trade, immigration and foreign policy; less pro-Wall Street and big business in its rhetoric; more Main Street/populist on economics. 

For now, the GOP seems to be taking a different route, trying to appeal to Hispanics through immigration reform and to upscale whites by relaxing its stance on some social issues. I think this is a tricky road to travel, and the GOP has rarely been successful at the national level with this approach. It certainly has to do more than Mitt Romney did, who at times seemed to think that he could win the election just by corralling the small business vote. That said, with the right candidate it could be doable. It’s certainly the route that most pundits and journalists are encouraging the GOP to travel, although that might tell us more about the socioeconomic standing and background of pundits and journalists than anything else.

Of course, the most successful Republican politicians have been those who can thread a needle between these stances: Richard Nixon, Ronald Reagan and (to a lesser degree) Bush 43 have all been able to talk about conservative economic stances without horrifying downscale voters. These politicians are rarities, however, and the GOP will most likely have to make a choice the next few cycles about which road it wants to travel.

---

* Ruy Teixeira has mostly convinced me that the correct final exit numbers for 2008 were 74.3 percent white, 12.6 percent black, 8.5 percent Hispanic, 2 percent Asian and 2.6 percent “other.”

**I also note that Hispanic participation probably exceeded projections when you consider that a disproportionate chunk of the Latino population growth consists of non-citizens who are therefore ineligible to vote. Also note the disproportionately large drop-off in “other”; I suspect this is mostly a function of the “rounding issue” I describe above.

***As my independent variable, I used the percentage change in turnout in each county from 2008 to 2012. Since we have over 3,000 observations using this technique, we can run a large number of variables. I went with 13. Five of them were meant to control for basic external effects: population growth, whether the county was in a state targeted in 2008 or 2012, whether it was in a state affected by Hurricane Sandy, whether there was a competitive Senate race in 2012 (the states that had competitive Senate races in 2008 were almost all swing states).

I ran a variety of demographic controls: the percentage of the county that was above age 65, that was African-American, that was foreign-born, and that was Mormon. I also included population density and median household income.

Finally, I included the percentage of the vote cast for Ross Perot in 1992.

**** The r-square is a bit low at 0.3, but we’re trying to explain a vast amount of data that probably relied on thousands of variables (local weather, differing amounts of money spent, other statewide contests). Moreover, a lot of these counties are so small that “quantum effects” -- random individual decisions -- can start to skew things. An extended family afflicted with food poisoning at Sunday dinner can materially affect turnout in some counties in western Kansas. If you exclude the 29 worst outliers (in geek speak, the ones whose standardized residuals exceed 3), the r-square jumps to 0.4. 

‹‹Previous Page |1 | 2 |

Sean Trende is senior elections analyst for RealClearPolitics. He is a co-author of the 2014 Almanac of American Politics and author of The Lost Majority. He can be reached at strende@realclearpolitics.com. Follow him on Twitter @SeanTrende.

Did We Vote for War?
Pat Buchanan · November 18, 2014
Why We Can't (Or Won't) Govern
Robert Samuelson · November 17, 2014
Wanting Desperately to Matter
Erick Erickson · November 14, 2014

Sean Trende

Author Archive

Follow Real Clear Politics

Latest On Twitter