Electionate

How to Read the Polls: It's Not Just About Who's Leading Right Now

By

Everyone pays attention to who's winning, but not all leads are created equal: a candidate's standing matters too. If it’s July in a presidential election year, there’s a big difference between leading 46-40 and 50-44. The former could be a toss-up, the latter is probably not. Why? In recent elections, it is unusual for presidential candidates to finish beneath their share of the vote in summer polling.

Consider John McCain in 2008. Other than a Palin-induced sugar high, the final few months of the 2008 campaign couldn’t have gone much worse. The economy collapsed, Palin’s credibility was ultimately crippled, Obama won the debates, and McCain wound up losing by seven points. But despite all of that, McCain managed to pick up a few points over the last few months of the campaign. On August 1, McCain held 43.7 percent in the RCP average, and yet he finished with 45.9 percent of the popular vote. In fact, McCain only briefly exceeded 46 percent of the vote in the immediate aftermath of the RNC, right before the economy collapsed. Obama never polled as high as his final tally.

This was also true state by state. Although McCain lost an overwhelming share of undecided voters, he still picked up additional support over his standing in the RCP average in nearly every battleground.

2004 was similar. In fact, neither Kerry nor Bush ever polled at their eventual share of the popular vote, although intriguingly, both approached it immediately after the conventions. That actually makes a certain degree of sense in a closely divided electorate. Sustained periods of positive media coverage around a convention might just get undecided voters to fleetingly realize who they'll eventually vote for, even if they revert to uncertainty once the battle resumes. To frame it differently: If you didn't support Bush after that successful RNC, when were you ever going to support Bush?

I would expect 2012 to broadly follow this pattern. Even in the best case scenario for Romney, where he secures the undecided vote and locks up everyone with reservations about Obama’s performance, it seems unlikely that Obama will actually lose supporters between now and Election Day. So even if Obama loses by a meaningful margin, I’d be surprised if Obama fell beneath 47 percent of the popular vote, or beneath these tallies in the swing states:

Now, as the McCain numbers show, it’s certainly possible that Obama could finish beneath these numbers. But that would probably require a McCain-esque finish, and that's not an especially likely scenario, although the possibility can’t be dismissed.

So on that basis, Wisconsin, Pennsylvania, and Nevada should edge into the "Lean Obama" column. Depending on the number of third party voters, Obama’s current tallies could win these states right now, and Obama would probably add a point in each state, even in the event that Romney sweeps the preponderance of undecided voters. Obama isn’t assured of winning those states, but when you consider past election results, campaign behavior in Pennsylvania and Wisconsin, demographics in Pennsylvania and Nevada, and the current polls, it would be pretty surprising if he did not.

Focusing on Obama's standing, rather than the margin, also illuminates the varying degrees of volatility across the battlegrounds. For instance, monochromatic Iowa could easily become a big Romney win if he ends up consolidating undecided white working class voters. In contrast, it's hard to see how North Carolina won't come down to the wire. There aren't many undecided voters and it's a state where turnout cleanly trumps persuasion.

While it's tempting to simply follow the margins, the candidate's share of the vote can be just as informative. Based on 2004 and 2008, the odds strongly favor a presidential candidate finishing above their current standing in the polling, even if the intervening months include a failed VP-pick and an economic collapse.

Loading Related Articles...
Article Tools
SHARE YOUR THOUGHTS

Show all 2 comments

You must be a subscriber to post comments. Subscribe today.