First, let me say how PSYCHED I am that three polls just came out, and they all show Deval Patrick with a lead of at least 24 points. “Single digits” my eyeball, Governor. That’s what happens when you use push polls as your internals.
Second, here’s a nice rundown of how the trends are running in the Suffolk polls, prepared by Stealth:
Previous polls are 10/12 and 10/05.
Patrick: 53% (46%, 49%)
Healey: 26% (33%, 28%)
Mihos: 9% (7%, 6%)
Ross: 2% (1%, 1%)
Undecideds: 11% (12%, 16%)
Favorables:
Patrick: 60-24 (50-31, 55-24)
Healey: 30-53 (40-44, 34-43)
Healey tough on crime?
Tough: 24% (36%)
Talk: 59% (40%)
Deval soft on crime?
Yes: 26% (32%)
No: 53% (41%)
Has Healey’s recent tone made you more or less likely to vote for her?
More likely: ?? (22%)
Less likely: 61% (53%)
Right track: 28% (28%)
Wrong track: 49% (53%)
I’m struck by a lot of these in these numbers. Particularly noteworthy, of course, is the fav/unfav ratings, which have substantially improved for Patrick (from +19 to +36) but substantially worsened for Healey (from -4 to -23). An astonishing change in less than two weeks, and it doesn’t take a rocket scientist to figure out what caused it.
I’m also fascinated by the “soft/tough on crime” numbers. On “really tough” vs. “talk tough,” Healey has gone from 36-40 (-4) to 24-59 (-35). Yet Patrick, who has been the subject of an unremitting barrage of negative ads on precisely this subject, showed a dramatic improvement on “not soft on crime vs. soft on crime”: from 41-32 (+9) to 53-26 (+27). Phenomenal. Talk about a strategy that backfired.
Also, I love the SurveyUSA polls because they let you draw all the spiffy graphs you want, broken down by your choice of subgroup. Particularly excellent are the graphs for men (in which Patrick has gone from a slight deficit to a 15-point lead), young voters (+10 to +34), and independents (+5 to +12). Even among Republicans, Patrick has improved from -66 (76-10) to -52 (71-19). Not sayin’ that’s going to win him the election, but it’s still significant.
OK, that’s enough. Back to your phonebanks. These numbers should be exactly the shot in the arm that the amazing operation built by Deval Patrick and John Walsh & Co. needs (if it needed one at all) to redouble its efforts over the next two weeks.
Please contact the Healey Committee and express your gratitude for the negative advertisements, which in recent weeks have helped Patrick/Murray so much! I did!
<
p>
Seriously, I phonebanked Lancaster tonight. I agree with the sentiment that a “Anyone but Healey” ID code is needed. I heard it so many times tonight it wasn’t even funny. Thanks Kerry!
<
p>
Disclaimer Yes, I am being a smartass. Yes, I am OK with this. Just blowing off some steam…
Lancaster is part of my district. I sincerely thank you for your hard work. đŸ™‚
<
p>
Are you part of Hank Naughton’s group?
I’ve been working out of the Carpenter’s Local 107 here in Worcester.
I looked at all the demographic subgroups and it seems only four show Healey getting a majority of votes: Republicans, Bush-approvers, conservatives and…Hispanics. Hispanics? Does this surprise anyone else?
3% of 623 is about 19 people.
<
p>
But yeah, it’s weird, especially since that number has trended up sharply for her.
Okay, 19 people is not even a remotely statistically significant sample, so it seems likely that’s just a fluke. 26% to 59% would mean an additional 6 people said they’d vote for her.
if you look at the MOE, it’s something like 25%. Tiny sample size = totally unreliable result for that subgroup.
I just ran a quick “chi-square” test for statistical significance on today’s numbers for hispanics (no significance at all) and the change between last time and today (slight significance: 20% chance that the difference was totally random).
<
p>
Stop me before I compute again!
I ran it through Kevin Drum’s handy MoE calculator, and it looks like there were 15 Hispanics surveyed, which gives a MoE of 24.25 percent (which jibes with the stated MoE of 24.3%), and an 83.88% chance that Healey really leads among that subgroup.
<
p>
I don’t know if that makes it totally unreliable, though I’d like to see more than just 15 polled!
My honest answer to that would have to be no, since the chance of my voting for her both before and after her latest ads was exactly the same. But I suppose they don’t necessarily want you to parse these questions carefully.