Wherein the Phoenix’ David Bernstein, Flaherty internals in hand goes to war against the Boston Globe.
A little knowledge is a dangerous thing. As I mentioned in my post yesterday, a Globe/UNH Poll has Menino comfortably ahead in the Mayor’s race. It would seem that this bruised Bernsteins feelings somewhat.
From Bernstein’s blog:
The Boston Globe has the results of a new poll in the Sunday paper, which shows Tom Menino cruising to re-election: “Menino holds a lead of 52 percent to 32 percent over Councilor at Large Michael F. Flaherty Jr.,” it says. But a poll conducted at the same time (earlier this week) by Flaherty’s campaign – provided to me by their campaign just now, in response to my inquiry — shows Menino leading by half that gap: 48% to 38%.
So what gives? It’s hard to tell, because the Globe’s story makes it hard to tell what their own (UNH)polling actually shows, and at this hour they have not made the data available.
So, does Bernstein take the time to contact the Globe or the University of New Hampshire Survey Center? No, he just decides to display his ignorance of sampling methodology:
The Globe/UNH poll, however, used 553 “randomly selected Boston residents,” which is frankly a ridiculously overbroad pool — it would include a lot of folks who aren’t even registered to vote. The article says that 438 of them “said they were likely voters,” which shows why you don’t want to use self-identification for that sort of question; if 80 percent of Boston residents are likely to vote on November 3rd then I’m Napolean (sic) Bonaparte.
It is highly unlikely that any reputable polling firm would sample from a raw residents list. The voter list (and voter history list) is computerized and available from both The Secretary of the Commonwealth’s office and the Boston Elections Department, and 553 voters is a reasonable sample (in the context of the 4.2% and 4,7% sample margins of error).
The issue is this: Bernstein is neither a pollster, nor a statistician. He would have done better to discuss the sampling and methodology with someone with the experience and training to understand the models used by John Della Volpe (Flaherty’s pollster) and the UNH Survey Research Center.
…and then it gets worse:
Bottom line? I think the race looks much more like Flaherty’s poll than the Globe’s. But the importance of the Globe poll is not in its accuracy, but in its effect.
If the Globe was reporting today that the gap had closed to within 10
points, that would have produced a major jolt to the final two weeks of the campaign. By reporting that the gap stands at 20, the Globe may have ensured that the city turns its attention elsewhere.
What we have here is a Bernstein implying polling malfeasance on the part of the Globe to re-elect Menino, using his gut feelings and Flaherty’s numbers as his premise. We’ll forget that The Floon receives much better Globe coverage than its component parts deserve, based upon their records, both as Councilors and candidates.
I wouldn’t expect Bernstein to consider the comparative accuracy of Globe and Flaherty polling to be a matter of grave import, to be discussed with input from people in the field: it’s all a conspiracy by the Globe to suppress turnout and disenfranchise the people!
I’ll have more thoughts, I’m sure, when I have details of the Globe/UNH poll.
…that he hadn’t bothered to read the damn thing.