At pollster.com, Charles Franklin has a spectacular plot of MA-Sen town-by-town voting data. Town-by-town he compares the McCain absolute vote to the Obama absolute vote.
The turnout in Massachusetts was 55% of eligible voters. Typical special elections get half as much; this turnout was as larger or larger than the last statewide governor's election and at the very upper end of what was considered plausible. The turnout was not quite Presidential level, but it wasn't far off. My stark conclusion from the plots: McCain's voters showed up--every single one of them. And Obama's didn't.
When you see a result like 52-47, you might get the idea that a few swing voters moved from one side to the other, making it a close race. In the most recent statewide election for federal office, Obama walked all over McCain in Massachusetts, by nearly a 2-to-1 margin. If you're a Republican in Massachusetts, getting a few swing votes doesn't get you home.
What gets you home, if you're a Republican in Massachusetts, is if the other side just doesn't show up. And that's what happened. The early reports of heavy turnout, which gave Coakley supporters some hope, were absolutely correct. McCain voters were unprecedentedly likely to go to the polls.
Here's what I wrote Monday morning before the election:
I live in a Boston suburb that Obama won by thirty points.
If you judge by yard signs, rallies, street corners, conversations, and dedication this morning (people were out on overpasses in a fugly ice/sleet/snowstorm), this election won't even be close. There are twenty or more visible Brown supporters for each one of Coakley's. They've been everywhere, consistently, for weeks, it's not a one-off, not something dismissed as just a few malcontent teabaggers.
Those may not be the best metrics, but they are all pegged on "blowout". It takes a hell of an iceberg organization to overcome such a visible enthusiasm gap, and Coakley hasn't shown much flair for putting together that kind of organization.
So does the data support the enthusiasm gap premise, or a voter-preference-shift preference?
That's where you look at the data Franklin provides. At the town-by-town level, the Brown vote total (i.e. absolute number of votes) is essentially always equal to that town's McCain votes (in absolute number of votes.) Overall, it averages about 5% higher. Perhaps those are the swing votes. And Coakley's vote total? In absolute numbers, town-by-town, it's 56% of Obama's vote total.
Broadly speaking, these numbers actually accord reasonably well with those in a recent PPP poll of MA-Sen
Republicans continue to show much more enthusiasm about the election than Democrats, with 89% of them saying they’re ‘very excited’ to go vote compared to 63% of Dems who express that sentiment.
As Franklin carefully and correctly notes, his data are an aggregate number, and may not hold down to the level of individual voters. (I.e. it's possible that various kinds of individual shifts could add up to this.) But I would note that a "voter-preference-shift" would show overperformance of the McCain absolute vote in areas that were strongly Obama to start, and underperformance of the McCain absolute vote in areas that were strongly McCain to start. To explain across-the-board near-equality of the vote (over several decades of magnitude, as shown in the plot) the most straightforward model is simply: McCain voters showed up, while Obama voters flipped a coin and showed up if it came up heads.
The national party isn't going to win many 2010 races if Obama voters feel they have nothing more at stake than what they'd be willing to put on a coin toss.