In an embarrassment of riches for poll junkies, Suffolk University and Channel 7 have released a new poll on the same day as Survey USA/Channel 4. And the Suffolk poll is wildly different from Survey USA. The Suffolk press release, however, leaves out a lot of important information – like the size of the subset and the MOE for the Dem primary numbers. I talked to David Paleologos, a Suffolk adjunct professor who was involved in the poll, to clarify these points. Many thanks to Prof. Paleologos for his time.
Suffolk poll: Overall, 600 registered voters who voted in the 2002 general election. Overall MOE: +/- 4%. More on interpreting the MOE. Confidence: 95%. Data collected 8/17-21. Since the 600 voter set includes Republicans and independents leaning Republican, they weren’t all asked about the Democratic primary.
Of the 600 voters, 369 are either “core Democrats” or independents leaning Democratic. The MOE for the 369 subset is +/- 5.1%. The widely–publicized numbers for that subset are:
Gabrieli: 32%
Patrick: 24%
Reilly: 20%
Undecided: 24% (wow!)
Of those 369 voters, 214 actually voted in the 2002 primary. The MOE for the 214 subset is +/- 6.7%. Here are the numbers for the 214 subset:
Gabrieli: 30%
Patrick: 27%
Reilly: 20%
Of the 369 subset, 122 did not vote in the 2002 primary (the other 33 didn’t remember or refused to answer) [these numbers have been updated from my original post]. The MOE for the 122 subset is +/- 8.9%. Here are the numbers for the 122 subset:
Gabrieli: 35%
Patrick: 21%
Reilly: 20%
So, as Prof. Paleologos said in our conversation, it’s all about turnout. The most likely voters in the Suffolk poll are the 214 subset, and that’s practically a dead heat. Gabs’ biggest lead is among the voters who self-identify as Dems, but who didn’t vote in the 2002 primary (they did vote in the general election). Note, of course, that among these smaller subsets the margin of error becomes quite large. Interesting, also, that Reilly’s numbers are stuck at 20%, regardless of which subset of Democratic voters you look at.
Whatever the sample sizes, these numbers are of course great news for Chris Gabrieli, since he is ahead in each subset. One does have to wonder, though, how they can be so wildly different from Survey USA. I mean, look at just the undecided numbers. Survey USA has 6%, while Suffolk has 24%. Does this mean that all horserace polling is worthless, even just 4 weeks before the election? I asked Prof. Paleologos about the difference. He hadn’t had a chance to look at the Survey USA poll, but said he would check it out.
Suffolk also did some November horserace results. They look remarkably different from the Healey campaign’s internals:
The 7NEWS/Suffolk University poll also revealed that Gabrieli will be the strongest Democratic candidate to defeat Lt. Governor Kerry Healey in the final election. Gabrielis margin over Healey was a whopping 21% (46%-25%), while Reilly topped Healey by 9% (38%-29%), and Patrick led Healey by 8% (38%-30%).*
Literally hundreds of pages of crosstabs and other detailed info on the Suffolk poll is scheduled to be posted by 5 pm today at this site. I’ll update this post, or put up a new one, depending on what we can find in there. Hopefully I’ve gotten most of the basics right.
UPDATE: The detailed results are now posted.
*Note [updated]: I corrected a typo in the Suffolk press release, which had the Patrick-Healey numbers at 38%-39%. The real numbers are 38%-30%.
It’s not terribly surprising to see the Gabber moving up in the polls. He was on the air pretty much nonstop in the time period leading up to the dates of this one.
<
p>
Your subgroup analysis is especially revealing. People who aren’t likely to vote in the Primary may not yet be paying all that much attention to the race, so react by citing the name they are most familiar with. Makes sense. The Patrick TV ads should counter that.
<
p>
Your point about turnout, of course, is the critical one. It doesn’t much matter who people prefer unless they go to their polling station on September 19.
<
p>
I’ll look at this more carefully when the crosstabs come out, but just from the little that you’ve been able to furnish, I note at least one glaring inconsistency. With Patrick leading Reilly by a good margin, how is it that he trails in the Healey match-up? Different groups of voters is my guess. Still, this squares with my intuition that the winner of the Dem Primary will be the next Guv.
<
p>
Wha’ happened to Mihos? Oh, wait, here’s the footnote:
** Another inconsistency yet to be explained.
<
p>
More anon, I’m sure.
<
p>
I’m not sure that’s so much of an inconsistency as it is sloppy wording. I’m guessing that they mean “Regardless of who you are supporting, who do you THINK is going to be the next Governor?”
<
p>
Curious that they only listed Buffy and Gabs, when they only accounted for 39% of the response. Just by basic math, that would mean one or several of the other candidates would have had at least 15%.
15% Reilly, 15% Patrick, 2% Mihos, 29% Undecided
Deval has hit the ceiling. This is where he pays the price for calling into his house parties.
Gabs has been doing a lot of the right things. Including making thousands of personal contacts. (Much different than Deval singingthe same ole same ole to the same ole same ole)
<
p>
Deval will die by the sword.
Why on earth does he put Bark Obama in his tv ads.
Deval doesn’t get it.
Gabs has learned from his past campaigns. He is proving himself to me. Reilly has become such a joke that I might vote for him out of pity.
<
p>
We may after all put a Dem in the corner office.
Except it won’t be Deval.
Nothing personal. Just the facts as I see them.
Preaching to the converted won’t win new converts.
How cool is that – personal follow up? Love it.
David,
<
p>
I must say good job in disecting the early poll returns from Suffolk Univerity. I would also say that Suffolk and Ch 7 have done some great work in the past elections. My take on this poll is that Gabrieli certainly has shown tremendous growth over the last Ch7 poll. In that poll Patrick 31%, Reilly 25% and Gabrieli 22%. This new poll shows Gabrieli gaining 10 points which is signaficant, It is also interesting to see the poss/neg results for the candidates.
<
p>
Gabrieli 50/13
Patrick 36/14
Reilly 41/33
<
p>
This is really bad news for Reilly and can be directly tied to David’s point that he does not move beyond 20 in any of the breakdowns he pointed out.
11% Murray
6% Goldberg
5% Silbert
<
p>
Murray gets 26% in the “West”
Not surprising, but depressing. The press really ought to have given more attention to this race, but I suspect they will after Labor Day.
I think he pretty much explained why the polls are so different. On the SUSA, it essentially asked ‘if the election were today, who would you vote for?’ On the Suffolk poll it was something more along the lines of ‘who are you likely to vote for?’
<
p>
Those are very different questions and would likely account for why there are way more independents and why Patrick’s numbers are down in the Suffolk as compared to the SUSA.
<
p>
Regardless, the first poll that is radically different I always take with a grain of salt. The SUSA poll is much more like polls we’ve been seeing for the past month or two and seems to be the more reliable at this point. If Suffolk follows up with another poll showing similar results and other polling companies to do, then the race has shifted and Deval will have some definate things to work on. It’s a bad time to lose momentum.
I didn’t mean independents, I meant undecideds. Sorry.