No one could blame a Massachusetts voter for being confused about which way the electorate is leaning these days. A raft of polls over the past two weeks have offered widely varying snapshots of voter sentiments looking ahead to Tuesday, when they go to the ballot box to choose between Attorney General Martha Coakley, state Senator Scott P. Brown and independent Joseph L. Kennedy.
The polls are all attempting to measure the same thing, but they use widely different methodologies. Some are more scientific than others, and some are sponsored by partisan groups with an interest in the outcome.
"God, what a puzzle," said Mark Blumenthal, editor and publisher of pollster.com, who has been sifting through the results of eight polls released this month. "My advice to a voter would be to say that the only thing we can say with any scientific precision is that it's looking like it will be a close race, and go vote if you want your voice to be heard."
Pollsters love to argue, in any race, about which methods best determine who is likely to vote. An added concern this time around is how to fix on likely voters in a rare special election held in mid-winter.
“No one really knows anything about the electorate that is going to show up,” Blumenthal said. “We don’t have any experience with special elections held in Massachusetts in early January.”
In the past week, one poll showed Coakley, the Democrat, leading Brown, the Republican, by 14 points. But that poll was conducted by a longtime Democratic pollster, Mark Mellman.
Another poll showed Coakley trailing Brown by 15 points, but was conducted by Pajamas Media, a conservative website with ties to Republican consultants. It used automated phone calls, which some pollsters say are not as reliable as live interviews.
Yet another poll, by Suffolk University, showed Brown ahead by 4 points, while still another, commissioned by the liberal blog BlueMassGroup, showed Coakley ahead by 8 points.
Even before all those polls had been released, longtime poll watchers were struggling to make sense of the results.
"Are you seeing a consistent pattern here? I'm not,” Nate Silver wrote in a post about the race last Sunday on his popular polling blog, fivethirtyeight.com. “All of the polls have positives and negatives. And any of them could be right.”
Blumenthal said the challenge for pollsters is to determine who will vote on Tuesday, particularly now that the race has intensified.
“Really small differences in pollster methods can make big differences in the final numbers, and it’s not at all clear what the perfect poll would look like,” Blumenthal said. "There’s a lot of art and there’s a lot of guesswork and there’s a moving target."
Most polls, like Suffolk's, use a formula, or screen, to focus on those most likely to vote on Tuesday.
Also common are adjustments to reflect the predicted composition of the Election Day turnout. For instance, the Survey Center at the University of New Hampshire, which in a poll commissioned by the Globe between Jan. 2 and Jan. 6, showed Coakley ahead by 15 points, used data from the Census and past elections to try to make sure the population surveyed reflected the demographics of those expected to vote.
“The one thing that is clear across all these surveys is that the most enthusiastic, most interested, and most likely to vote are the conservatives who are supporting Brown,” Blumenthal said.
David W. Moore, a former vice president of Gallup and managing editor of the Gallup Poll, said the Suffolk poll used reliable methods and should be a "wake-up call" to Democrats who might have thought Coakley would trounce Brown. Still, he said, “It’s hard to know how to interpret the various polls."
"I think there are a lot of undecided voters and they’re not measuring that,'' Moore said. "And I’m skeptical about the robo-polls. So I would say it’s a close race and I wouldn’t feel comfortable, if I were Martha Coakley, and I wouldn’t feel panicked, either.”
Blumenthal was also cautious in interpreting the results.
“Setting aside gut instinct and guesswork and judgments about where things are going, this looks like a race that is too close to call, though it may not end up close,” he said. “It may end up being 4 or 5 points apart. Or more.”
On the beat
Columnist Shirley Leung says Boston mayor-elect Martin J. Walsh should focus on middle-class housing. Read more