Much of next Tuesday’s election outcome depends on which electorate shows up for the election. Will it be the “likely voter” electorate assumed by many national polls or will it be an electorate that looks more like the actual pool of eligible voters? A lot rides on the answer to this question.
Start with likely voters. Likely voters are a subset of registered voters generally selected in a survey sample through screening questions on voters’ intentions, interest and enthusiasm. These are the voters the pollster believes provides the best approximation of the actual voting electorate this year. Determining this subset is a laudable goal but the key word here is “approximation” since pollsters don’t yet know who the real voters are going to be. Nor are past practices in selecting likely voters a particularly reliable guide, which is why there seem to be as many ways of selecting likely voters as there are pollsters.
Gallup has one of the most complicated methods for selecting likely voters, using a battery of seven questions. Based on this method, Gallup has been selecting likely voter samples that have about 20 percent minority voters. This is on the low end of estimates, but other well-known national pollsters like Pew and Washington Post/ABC have drawn recent likely voter samples with about 23 percent minorities.
How plausible is this projection? Well, the exit polls in 2008 pegged the minority voter share at 25.7 percent. That was a 2.8 percentage point increase over 2004. Was that an anomalously high increase, due to the excitement of Obama’s historic campaign? No. The increase from 2000 to 2004 was actually larger: 3.5 points. And the increase from 1996 to 2000, 2.4 points, was only slightly smaller than the 2004-2008 increase.
The reason for this trend is well-known: Every year, there are more minorities and fewer whites, both in the overall population and among eligible voters. The most recent data, from demographer William Frey of the Brookings Institution, indicate that minority eligible voters increased by 2.5 percentage points to 29.1 percent of all eligible voters from 2008 to 2012.
So, if we are to believe some of these likely voter polls, despite the consistent historical trend among actual voters and continuing gains among eligible voters, this year not only will we see no gains in minority voters but an actual decline in these voters compared to the previous presidential election. To the extent anyone bothers to rationalize these findings, the explanation is that minority voters are so much less enthusiastic than in 2008 that their low turnout will undercut ongoing demographic shifts. That’s technically possible but there’s little actual evidence for this hypothesized fall-off in minority voter enthusiasm, either among African-Americans or among Hispanics (Hispanics, if anything, appear to bemore enthusiastic).
This likely voter universe therefore doesn’t appear so likely. But if it did happen we should be clear about its implications. If we accept Gallup’s 20 percent minority vote share scenario, and assume that Obama’s minority vote support is the same this year as it was 2008 (a pretty safe assumption since Obama is holding has black support and overachieving among Hispanics in pre-election polls), then Romney would only need to increase the Republican margin among whites to 16 points, from 12 in 2008, to win the popular vote. If 23 percent turns out to be an accurate estimate of minority vote share, as other polls predict, then Romney could get by with a 19 point margin among whites—an eminently achievable margin.
But what if the minority vote share in 2012 merely matches its 26 percent share in 2008—which would still be off historical trends and inconsistent with the observed increase in minority eligible voters since 2008? Then Romney needs a 22 point margin among whites just to nose out Obama by two-tenths of a percentage point in the popular vote. And Obama likely wins the popular vote if he can get just 39 percent of the white vote (he got 43 percent in 2008).
It’s worth dwelling a little bit on what a 22 point margin among whites for Romney would have to look like. If Romney does no better than McCain did in 2008 among college-educated whites (a 4 point margin) then he would need an astounding 40 point margin among working class whites to produce an overall 22 point margin among whites. In the latest Pew poll—one of the only national polls to provide a breakdown by college and noncollege whites—Romney is still far away from this level, with a 29 point lead among noncollege whites. To be fair, the Pew poll does have him ahead a bit more among college whites (7 points) which would drop his requirement to a 37 point white working class advantage under this scenario. But he’s still not there.
And what if minority vote share follows trend and goes up to 28 percent? Then Romney would need a 25 point margin among whites to prevail in the popular vote. That 25 point Romney margin among whites would require a gargantuan 46 point margin (probably around 72-26) among working class whites to be realized, assuming that his margin among college whites does not improve over McCain’s in 2008. But even if the latter does improve—say to George W. Bush’s margin in 2004 (11 points)—Romney would still need a 39 point margin among working class whites to carry to carry the popular vote in this scenario.
Which scenario is the most probable? I'd simply underscore the basic grounds for skepticism about the demographic scenarios for “likely voters” implied by many national polls: minorities' share of the total voting population has never decreased from one presidential election to the next, at least not in recent history. And if these polls do indeed prove wrong, and the voters on election day look more like the actual pool of eligible voters, then the demographic scenarios that play out will be much more favorable to President Obama and much more challenging for Governor Romney.