State House News Service’s survey is the latest of three gubernatorial polls issued within the last week. Each of the three polls has Tom Reilly on top (although the CBS4 poll has all three candidates within the margin of error). Two of the three polls show Chris Gabrieli in the second spot. This is the second poll that has Deval Patrick as the only Democrat losing to Kerry Healey in a head-to-head, although both losses were within the margin of error.
Attorney General Tom Reilly continues to hold a substantial lead in a Democratic gubernatorial primary trial heat three-way race against Deval Patrick and Chris Gabrieli, with Reilly receiving 37% vs. 25% for Gabrieli and 15% for Patrick. Gabrieliâs surge in vote support and favorable rating is no doubt due to his television advertising campaign. In general election trial heats pitting each of the three Democrats against Republican Lieutenant Governor Kerry Healey and Independent Christy Mihos, Healey now runs statistically even with all three Democrats
DEMOCRATIC PRIMARY:
REILLY v PATRICK v GABRIELI
37% 15% 25% Now
43% 21% 8% March
GENERAL:
HEALEY v REILLY v MIHOS
30% 38% 13% Now
31% 34% 18% March
26% 45% 11% January
HEALEY v PATRICK v MIHOS
31% 29% 15% Now
32% 25% 18% March
32% 32% 13% January
HEALEY v GABRIELI v MIHOS
31% 37% 10% Now
36% 21% 18% March
Use these links to see the Suffolk University and CBS4 polls.
I can’t get Gabrieli off my tube. He follows me up and down the dial. Clearly, he’s spending bucks to build momentum toward the convention in June, something Patrick and Reilly don’t have to do.
<
p>
This race starts in June.
http://www.boston.com/news/local/articles/2006/05/09/patrick_pushes_outsider_profile/
<
p>
So, the sample is 201 likely voters? That means that the margin of error (at 95% confidence) is about 7%! No serious, statistically significant poll – especially state-wide – should have less than 400 respondents, twice the number they have here. Forget this poll… Gabrieli might be picking up speed – in fact, it’s likely given his self-financed TV-based attempt to buy the nomination – but I’ll need to see another poll before I buy any numbers as a result. This was is just way too flawed.
<
p>
Sorry, maverick.
I didn’t conduct the poll. I just posted the results. Many people like to dissect polls, just as you did. It’s like high school biology, minus the frogs.
<
p>
I agree that a 200 person sampling is too small for this poll to stand on its own, but the results are quite similar to the 400 person sampling used by Suffolk. Of course, CBS4 used a 400 person sampling, and they have the race closer, although in the same order: Reilly, Gabrieli, and Patrick.
<
p>
None of these polls tell the whole story, but if you piece them together, they certainly are telling part of the story.
“High school biology minus the frogs.” Good one!
Just to clarify: When it says 6.8% +/-, does that mean that, for example, Reilly could be down 6.8 and Patrick could be up 6.8% (or some other combo)?
it means that, according to this poll, Reilly’s support could be anywhere between 30-44% and Patrick’s could be anywhere between 8-22%. Most polls rely on a sample of about 400 so that they can reduce the +/- by half of what you see in the SHNS poll.
I believe that the Margin of Error is the same as twice the standard deviation of a bell curve (so if 6.8% is the margin of error, there’s 67% chance of being being in +/-3.4%, 95% chance within +/-6.8%, and 99% chance within +/-10.2%). What I’m not sure about is if the error is in absolute points (Reilly having 37% +/- 6.8%) or in relative points (Reilly having 37% +/- 6.8% of 37%, ie 37% +/- 2.5%), which might make more sense IMO.