Massachusetts voters woke up this past Sunday to read a Boston Globe headline proclaiming “Senate poll: Coakley up 15 points”. The news that Attorney General Martha Coakley, the Democratic candidate to fill Ted Kennedy’s vacant Senate seat in Massachusetts, was leading her Republican opponent Scott Brown in a Globe poll by 50%-35% nine days before the special election must have struck most readers as remarkably unremarkable. This is Massachusetts, after all, where Republican officeholders are an endangered species – the last Republican Senator elected here served more than a quarter century ago and it is inconceivable that voters in this staunchly liberal state would replace a Kennedy with a Republican. You might as well tell me a Tom Brady-led football team would lose a home playoff game.
Then again – did anyone see Sunday’s game? Or, more to the point, see a more recent poll by Public Policy Polling that has…drum roll please….Brown leading Coakley by 1%, 48%-47%, with 6% undecided? To further complicate matters, Rasmussen issued their own poll earlier last week that had Coakley up by 9%, 50%-41%, with 7% undecided.
What is going on here? All three polling firms are quite reputable, so how do we explain the seemingly divergent results? One explanation is that Brown is closing fast – that the Globe poll, which was taken Jan 2-6, did not pick up on the movement toward Brown that the more recent PPP poll, which was taken Jan. 7-8, did capture. For reasons I discuss below, I don’t think this is the explanation. Indeed, I think Coakley is likely to win this race.
Note, however, that if Brown wins – and this is not inconceivable – it would be a significant upset, with national implications. Massachusetts is a state that recently passed a system of universal health care, and Brown has campaigned on a promise to vote against the current health care/insurance reform bill now being debated in Congress. His election would undoubtedly be spun as an indication of softening support for health care reform. At the very least, it would cost the Democrats their 60-vote “filibuster proof” majority and might even jeopardize passage of the current health care bill.
I no longer live in Massachusetts, so my read of the local political situation there is far more impressionistic than I’d like. But I would be very very surprised if Kennedy’s seat went Republican. In looking at the two most recent polls, the more likely explanation for the disparity in results centers on differences in turnout projections. Mark Blumenthal has a very nice discussion at Pollster.com, but the gist of the argument is that Brown’s chances go up as turnout goes down. That is, the difference between the PPP and the Globe poll is that PPP’s survey methodology screens out less committed voters, while the Globe’s keeps them in. This may reflect differences in polling technology – PPP uses an automated voice system, while the Globe uses live interviewers in their surveys. To see why the voter screens matter, consider the following tables that show the breakdown of voters by intensity in the Globe poll and the earlier Rassmussen poll that has Coakley up by 9%. (The table is posted at Pollster.com):
We see that the race is essentially a tossup among those who say they are “extremely” interested or “certain” to vote. However, among the less interested or those not certain to vote, Coakley has a large lead. This suggests that the election will turn on whether Coakley can get these less passionate voters to show up nine days from now. If not, Brown has a chance to pull an upset. In this sense, the Globe – a staunchly Democratic paper – may be doing Coakley a disservice by suggesting that she’s coasting to victory. Because the Globe, like all media outlets, acts as if its own poll is the only one to show results, Globe readers aren’t likely to know the race may actually be closer.
These dueling polls are a reminder of the difficulty in predicting special elections, something that became apparent in the special election in NY’s 23rd congressional district. Turnout tends to be lower and thus there is often a greater proportion of more committed voters in the voting pool. PPP’s results may be more accurate not because they are more recent, but because they more accurately model the likely turnout. We may know more as more recent polls come in – but not necessarily. In the New York election, polls published on election eve turned out not to be particularly accurate.
I repeat: I expect Coakley to win this race – but then, I expected the Patriots to prevail yesterday.
Final score: Ravens 33, Patriots 14.
That’s why they hold elections.
Thanks for the blog post about this race – I saw that headline the other day about the “closeness” of the race and wondered what was up with the polls.
Matt- You’re right, this race has is developed more interest than you would think for Massachusetts politics. My first impression here is that Coakley has not done much to impress people. Her ads are everywhere now, but she hasn’t said much of consequence, preferring instead to coast into the seat (This was also her approach in the primary, and that worked fine).
But Brown has been a beacon for those Massachusetts Republicans that elected Romney (they look like brothers), and further back Bill Weld and other governors. Will Brown win? Probably not. Are the Massachusetts Republicans gaining traction? A little. Will we have an interesting gubernatorial battle this fall with Charlie Baker and Deval Patrick? I would think so.
Alex,
I hadn’t thought much about the governor’s race, but you are right that it would be the more likely one for a Republican to win, given Patrick’s dismal approval ratings and the fact that Massachusetts voters have shown a propensity for choosing Republican governors.