From the University of Massachusetts polling site:
The results of a new UMass Poll released today by the University of Massachusetts Amherst shows state Rep. Martin J. Walsh with a seven-point lead over Boston City Councilor at Large John R. Connolly among registered Boston voters in advance of the Boston mayoral election on Nov. 5. The UMass Poll confirms that voter support has swung from Connolly to Walsh in the final days of the campaign.
With leaners included, likely voters prefer Walsh over Connolly 47 percent to 40 percent, a lead that is outside the 5.9 percent margin of error for the poll. With less than a week to go in the campaign, 13 percent of likely voters remain undecided.
“No matter how you define ‘likely voters,’ our poll shows that Walsh has capitalized on a couple of strong weeks of campaigning to vault out in front as the clear leader in this race,” said Brian Schaffner, chair of the political science department at UMass Amherst and director of the UMass Poll.
The telephone poll conducted Oct. 22–26 included interviews with 600 registered voters sampled from a list of registered voters in the city of Boston. The sample included both landline and cell phones.
Toplines here
Crosstabs here
David posted this information before I got this up. But it would be interesting to get people’s take on the data, so forgive the redundancy. At a guess, two things are happening:
Undecideds are swinging Marty Walsh’s way.
The “likely voter” contact universe is beginning to mirror the probable electorate.
had Connolly up several last week but their polling’s been a little suspect in recent elections. The Globe, unsurprisingly since they commissioned the poll, ran a story on it that didn’t mention any other polling as I recall.
With that one included it seems like a huge shift but the other poll had it pretty much tied, so the shift (if it’s real at all) might be smaller than it seems.
…that jumps out at me is this (chronilogically by release):
The MassInc Poll (Oct 19-20 in field) has:
• White: 39-33 Connolly
• African American: 35-32 Connolly
• “All Others”: 21-38 Walsh
The Globe poll (Oct 17-22) has:
• White: 46-38 Connolly
• African Americans: 41-31 Connolly
• Latinos: 40-39 Connolly
• “Other”: 56-19 Connolly
The UMass Poll (Oct 22-26) has:
• White: 48-32 Walsh
• NonWhite: 44-32 Walsh
It is rather remarkable to see a shift that dramatic in all racial groups. I have to admit that the Globe looks just a little outlierish given the trend here, though the extended 5-day field time of that poll may have been in the middle of the changing dynamic.
.
…than whether it is evidence of shifting support to Walsh on the ground (however imprecisely measured).
I have problems with the UMass poll in isolation, because (among other reasons) “nonwhite” wasn’t separated by race or ethnicity. Nor is there a cross tabulation for geography or gender for respondants, irrespective of race.
In short, there was no way I could look at the data in the context of existing (and preexisting) political organization on the ground.
It may be an indication of a trend, however. The consensus by both campaigns that the race is a dead heat is what makes this particular poll interesting in light of two things:
Walsh’s history of outperforming the polls; and
The possible evidence of Walsh momentum, however imprecisely measured.
It was in that context that I submitted the post and request for feedback.
I hear you. It was just the first thing that really caught my eye.
I think this poll feels like too big a leap from the Globe poll, and maybe both are a bit off. But I do think that undecideds knew Connolly better a month ago, and are being persuaded towards Walsh as they’ve gotten to know him. I’m not sure issues is what’s doing it, so much as his story.
And I think there comes a tipping point when every endorsing elected of color, nearly every elected progressive, and nearly every leader in the communities of color come down on one side, it can’t help but have an impact.
…which reinforces your point.
isn’t it harder to get accurate polls from smaller groups? How probable is it that none of the polls is very predictive?
…the smaller the sample the greater the margin of error.
I assume that most of them are internal polls run by the campaigns, but I have gotten so sick of them that I have started answering randomly.
You do have to wonder if the easy of automated polling might cause “poll fatigue” that could degrade the accuracy of results.
I have noticed a lot more Walsh canvassers in our neighborhood than Connolly ones, so perhaps Walsh’s ground game has been having an effect.