++Addition++In Mississippi, Hillary again won (by five points) among those who made their decision in the last three days.
My first extinct upon waking up to Hillary's victories was to look at evidence for Obama's over-polling. It seems plausible, has been alluded to in much of the major media fare as well as in more meticulous places, and the apparent Texas primary/caucus split we'd been anticipating suggests it might be happening.
Yet there has been little in the campaign to point to as much actually happening on the ground. Obama has for the most part been performing on par or exceeding expectations arrived at through polling information gathered nearest to the vote. Click on a state via the previous link to see the final results as well as what polls showed just days before the vote. California and Massachusetts superficially show the "Bradley effect", but that's really all there is to go on from the forty states that have gone thus far. Obama has even been out-performing the last polls in many of the states he's lost. If anything, it seems to be the Bradley effect in reverse.
The difficulty seems to rest in making sense of how much push Obama gets from the way he tends to continuously do better and better in polls (and by extension in electoral support) all the way up to the time the actual contest takes place, and how much opposing pull comes from the dissonance of white voters, who tell the pollsters one thing about how they are going to vote that doesn't match what they actually end up doing.
Looking at the exit poll results for those who made the decision within the last three days prior to the vote, prior to yesterday, Obama outdid Hillary.
Of the 25 states with exit polling data, Hillary took the last-three-days vote in ten of them, all states that she won overall. The candidates broke even in New York, and Obama took the other 14, including a couple of states Hillary came out on top in overall.
In only one state did Hillary's last-three-days performance exceed her overall margin of victory. In the other nine, she topped Obama among last-three-days voters, but by margins less impressive than her actual victory in the state was. In California, for example, she edged Obama 48%-47% among last-three-days voters, while winning the entire state by ten points. The one exception was New Mexico, where whites voted against her 55%-43%. If anything, this suggests that voters (at least non-Hispanic voters!) claim to be less, not more, supportive of the non-white candidate than they actually are.
Well, until yesterday, that is. The racial/ethnic results in Ohio and Texas are standard for the campaign. Blacks overwhelmingly back Obama; whites and especially Hispanics get behind Hillary. The big change this time around is that Hillary cleaned Obama's clock among those who held out until the end to make their decision.
Among last-three-days voters, she not only beat him in Ohio, Texas, and Rhode Island, she did so enjoying wider margins than she enjoyed in the overall votes in all three states. Again, she'd only done that one other time before, in New Mexico, where she had to rely on her Hispanic support to overcome her white opposition. And in Vermont, she outdid Obama 52%-44% in the last-three-days category even while losing the state 39%-59%. Up to this point, she'd not pulled that off a single time.
So is the Bradley Effect finally making a fashionably late entrance? Maybe the African pictures did it?!
Or might Obama really be losing his momentum? Are Democrat's eyes starting to glaze over when they hear their presumptive nominee speak passionately but predictably about "change" and "hope" for the umpteenth time? Whatever the exact political definition of 'momentum' is, the ability to absorb fence-sitters seems to approach it. And he clearly did not have that yesterday.
In all likelihood, Obama will win Wyoming and Mississippi, as the demographics of both states play strongly to his advantage. But it will be interesting to see how those late-deciders shakeout. If the Obama ship is running out of steam, it should also show up in forthcoming Pennsylvanian polls.