In ten days, Breaking Dawn, the final installment in the Twilight movie series, featuring Kristen Stewart, will come to our nation’s cineplexes. Millions of the series’s fans have passionately divided themselves into Team Edward and Team Jacob, arguing as to which prospective beau is better for their heroine. Next week, those who haven’t already read the books will finally learn whom Bella Swan will spend the rest of her unnatural life with. (Hint: It’s not the werewolf.) Tomorrow, the final installment in the reality series called America’s election season will be filmed.
Tens of millions of Americans have divided themselves into Team Barack and Team Mitt, passionately arguing which prospective leader is better for our nation. But we have no book to read to discover who wins. Will we have twilight or breaking dawn for conservatives?
Clearly, supporters of each candidate differ in their predictions. Followers of my writing know which team I am on. (Hint: It’s not the incumbent’s.) But I write as an analyst, not a partisan, and what follows is my best prediction of what will happen, why, and what needs to transpire to prove me wrong.
My Final Answer
Obama: 49.39 percent; 303 EV
Romney: 48.24 percent; 235 EV
Third Parties: 2.37 percent
States Switching to Romney: Florida, Indiana, North Carolina, Nebraska’s second district
States Switching to Obama: None
Senate: D: 52; R: 46; I:2
Seats Switching to Democrats: Maine (I), Indiana, Massachusetts
Seats Switching to Republicans: North Dakota, Nebraska
House: D: 197; R: 238 (+4 D)
Black and White
How did I sort through the polling cacophony to arrive at such precise numbers?
In one sense, I’ve offered a false sense of precision. All of the national polls and weighted averages of state polls have the race within the margin of error. Following that admonition to its logical conclusion would mean I have to call the race as “too close to call.”
But that is both cowardly and inaccurate. Someone has the edge, someone will win, and I need to make my best effort to find out who that someone is.
Pundits and analysts have sparred over the last week over which poll or set of polls to believe. Nate Silver has argued that taking the average of state polls, weighted by each state’s share of the population, provides a more accurate measure of the national picture than the national polls themselves. Republican-leaning analysts have argued that many of the national and state polls are distorted by including higher shares of Democrats versus Republicans in their polls than is warranted by our recent history.
This argument has serious consequences for who you think will win. The weighted state average of polls has been consistently slightly more favorable to Obama than the Real Clear Politics average of national polls. Those polls that show Democrats with only a small partisan edge tend to show Romney doing better, although that pattern has been less pronounced among polls in the last few days.
I could try to cut through this Gordian knot, but ultimately I decided to transcend the whole argument entirely. I am basing my forecast on a factor I believe to be much more stable and hence less subject to partisan shading or statistical variation: race.
I do this for three reasons. First, one’s race is largely not a matter of dispute. A poll respondent is highly likely to accurately give his race; his partisan inclination is less likely to be so firm. Second, the political views of whites and non-whites are sharply different — there are many white Democrats, but there are few non-white Republicans — so unless a subsample of whites or non-whites is unusually imbalanced, looking at race allows you to more accurately determine partisan breakdown.
Finally, I do this because it allows me to run through a series of alternative scenarios and estimate the break-even points for Romney and Obama in both the popular vote and the Electoral College. I provide a variety of options to analyze yourself, and you can see what will have to happen on Election Night for President Obama to win decisively or for Mitt Romney to win at all.
The table below shows the racial divisions for each of the last six national polls to make those breakdowns publicly available (Rasmussen’s are available only by subscription; Sunday’s Wall Street Journal poll’s crosstabs were not released as of this writing). As you can see, there is remarkable agreement among the polls as to where the race stands by race.
If you average the share of the white vote for each candidate, you get Romney taking roughly 56.5 percent of the white vote, Obama taking 39 percent, and 4.5 percent either undecided or supporting third parties. For non-whites I averaged the last three polls, as they are both similar to one another and only slightly different from the 2008 non-white vote breakdown according to the exit polls. That shows Obama leading among non-whites by 77 to 20 percent.