Conventional wisdom says Dianne Wilkerson is a long-time incumbent, and the only reason she nearly lost two years ago is that she failed to get her name on the ballot so had to run a write-in campaign. Plus, she’s got lots of powerful friends — Senator Kerry, Governor Patrick, Senate President Murray, Speaker DiMasi, Mayor Menino, and a host of prominent lefty interest groups (including MassEquality and the Boston Teachers Union) — in her corner. All of that should translate into an easy win, right?
Well, maybe wrong. Check out the results of a tracking poll just released by the Chang-Diaz campaign:
The numbers, for those of you not graphically inclined (counting “probably” and “lean” voters):
Chang-Diaz: 47.2%
Wilkerson: 29.7%
Undecided: 23%
(Data collected 8/10-12. 417 likely Democratic primary voters. MoE: +/- 4.8%)
If those numbers are good, then even if every single undecided voter broke for Wilkerson, she would barely eke out a win. And that, of course, is not going to happen — late-deciders seem at least as likely to favor the challenger as the incumbent.
Also of interest are the “Wilkerson deserves reelection/time for someone new” numbers, which have heavily favored Chang-Diaz since January:
Deserves reelection — time for someone new
January: 28% — 47.5%
August: 25.1% — 46.9%
This is huge news for Sonia Chang-Diaz, who appears to have a real chance at unseating Wilkerson. It’s also huge news for progressives tired of having to explain away bad behavior by people who advocate for positions they like.
But polls released by campaigns are never meaningful unless they show that campaign behind. It’s precisely why folks like Pollster, 538, and electoral-vote don’t use ’em.
The poll has a margin of error of 4.8%. She’s not making this up. In any event, this race is very competitive and Sonia can win it, there’s only so much a campaign can make the poll look good for the campaign without sacrificing accuracy. This poll does them no good if it weren’t accurate.
Who says it’s accurate? Who says the internal poll matches the poll released? Who says they didn’t take five polls, each with five different turnout models, and release only the most favorable one?
<
p>Publicly released polls do plenty of good when not accurate. They serve to motivate volunteers. They serve to keep campaign donations rolling in. They serve to demoralize the opponent, her supporters, and her volunteers. They serve to generate press coverage.
<
p>They even serve to get discussions brewing at blogs. Fancy that.
I agree that the poll likely wouldn’t have been put out with a big splashy press release if it hadn’t shown good results for Chang-Diaz. However…
<
p>Who says it’s accurate? Sure, maybe they’re just lying. I seriously doubt it. Pollsters have reputations that they want to keep intact.
<
p>Who says the internal poll matches the poll released? Ditto — I don’t really understand how this question is different from your first one.
<
p>Who says they didn’t take five polls, each with five different turnout models, and release only the most favorable one? This is possible, but highly unlikely, because the campaign would have to pay for each poll. Polling is wicked expensive, and it seems very unlikely that the Chang-Diaz campaign has the money to commission multiple polls over a single time period.
<
p>It’s fair to be skeptical of a campaign’s own polling, but it’s not fair to assume that they’re just making shit up.
Who says she would have the money for two polls anyway???
<
p>I mean, how viable would it be for a local candidate to commission two polls, one that’s “favorable” and one that is more accurate to be kept private?
<
p>Yeesh…paranoid much…?
I can’t even imagine two polls, though, nevermind five.
<
p>Those things can be real expensive.
But consider:
<
p>1. Guaranteed sample bias, since polls that don’t look good simply don’t get released. This shifts the mean substantially, and guarantees that the average campaign-released poll will look better for that candidate than polls not commissioned by a particular candidate [and therefore guaranteed to be released].
<
p>2. The details of the poll matter, and they dramatically impact results.
<
p>Give me any poll and I’ll tweak turnout models by a smidge here and there and swing the results 10 points or more. Am I lying? Nope. Just changing the parameters.
<
p>I might use one set of parameters for the campaign’s turnout model, but then release a different set of parameters for public consumption. Maybe I think that, based on my ground game, turnout will be
<
p>W/M: 55/45
B/H/W: 40/35/25
$$$$/$$$/$$/$: 5/25/50/20
D/I/R: 70/20/10
<
p>But the poll I release weighs the responses so that the turnout model is
W/M: 50/50
B/H/W: 50/25/25
$$$$/$$$/$$/$: 10/40/40/10
D/I/R: 60/25/15
<
p>By changing the turnout model, I can skew the results of the poll dramatically. Am I lying? Nope. Am I “making shit up”? Not really. As long as the turnout models are somewhere within the realm of reason, my model doesn’t stink like a skunk. But, I can tweak it to say what I want since I’ve already seen the results.
<
p>Therein lies the rub. The campaign bought the poll, and they can alter the turnout model. By altering the turnout model, they can make the poll seem better [or worse!] than what an unbiased turnout prediction might look like. Of course, if tweaking the turnout model doesn’t swing the numbers, they just don’t publish the poll.
<
p>
<
p>Given that poll results can be swung dramatically after the poll has been conducted by changing turnout models, and given that campaigns would never release an unfavorable poll, polls released by campaigns are worthless to folks interested in really getting a sense for the race.
is an awfully big margin to ‘tweak.’ The most important thing with a poll like this is accuracy. A candidate wants to know where they stand. That’s far, far more important than a flashy press release.
Everything you say about “tweaking the turnout model” is true; whether the Chang-Diaz folks did that in this case, I have no clue, nor do you.
<
p>But this —
<
p>
<
p>in no way follows from what came before. Of course you’re right that unfavorable internal polls don’t get released. But this one was favorable, so it did get released. And while fiddling with turnout models may be able to affect the results to some degree, 17% is an awfully big spread, as Ryan notes elsewhere.
<
p>To declare this poll definitive would be silly — it’s too early, and it’s an internal, among other issues. But to declare it “worthless” is equally silly, IMHO. Even if you are suspicious of the precise numbers, it surely gives at least “a sense of the race.”
especially if you combine the sampling bias of poll releases and manipulating turnout models. In my opinion, the true MOE on polls released is around 25%, especially if the poll exists in a vacuum like this one.
<
p>If another poll comes out [neutral, or Wilkerson] and shows data that starts to show a trend, that’s one thing. But a single poll released by a candidate showing her winning is worth so little that it might as well be worthless, because if it shows a margin over the “stomv true MOE” you should have known one candidate had a solid lead already.
<
p>If you want me to retract worthless, fine. I’ll refine: candidate polls, when considered without corroborating polls not released by the candidate, and when showing a margin of 25% or less, provide extremely little useful information and that usefulness decreases as the size of the district decreases.
Seems that it would just involve picking up the yellow pages and making a couple hundred phone calls, enough to get an adequate sample size for the population you’re dealing with. Where’s the big expense come in?
Also out of curiosity, Ryan, I notice that in the last week or so you’ve added a new semi-disclaimer to all your comments. Are you actively working for Sonia Chang-Diaz now, or is there anything else in particular that prompted the addition of the disclaimer-type-thingy?
I “have sources.” These polls are good, coming from people I trust on the subject and know.
<
p>Then again, by all means, don’t trust them and work harder for Sonia. I don’t want any Sonia supporters resting on their laurels.
get off me.
Wilkerson needs to go. The fact that someone so fantastic as Sonia is on the cusp of victory is so much for the better. We’ll have not only a progressive in office, but one who we can trust and is worthy of our trust.
why is it that Wilkerson’s “powerful friends” feel the need to constantly endorse her? Especially knowing she has these problems, indescretions, legal issues, whatever you want to call them?
<
p>Including MassEquality, which could have bowed out and endorsed no one because of what Wilkerson did do for LGBT people — in other words sending her a kind message. I just don’t get it — one wonders if Howie Carr’s book about “connected” or “made” Democrats in this Commonwealth may just have some truth to it afterall.
That if they don’t support her, she’ll stop supporting their issues and giving them the time of day. And, then again, there’s always other people… with far different priorities… that she could be getting her support from. Although, if I were these endorsing orgs and that was the implied threat, I’d just tell her to go get her support from them. It would make her less, not more, electable in her district.
However, what does it take for them to understand that as far as the voters of Massachusetts go, we believe that when MEQ said that it was “implied” that “we need to help those who helped with marriage unless you do something illegal.”
<
p>Seriously folks, be careful, one can judge the character of a person by the character of his/her friends. I would suggest it possible to judge the character of the endorsers by the character of the ones being endorsed.
it’s simply rational herd behavior.
<
p>most interest groups support incumbents. therefore when it’s time for your individual interest group to make a choice, you have 2 options:
<
p>a. also endorse incumbent. if incumbent wins, no sweat. if challenger lives, will forgive all interest groups, in hopes of getting them all to line up forevermore.
<
p>b. endorse challenger (or no endorsement). if challenger wins, you have curried a bit more favor than the other interest groups, but not much. but if incumbent wins, now has a grudge, because there’s only a few interest groups who bucked the trend.
Sonia and her campaign team used the “release a good poll to generate buzz” tactic last time too:
<
p>Taling Politics
Monday, August 07, 2006
Second Suffolk poll, heavy on the salt
<
p>Got the following press release from Sonia Chang-DÃaz’s campaign yesterday. Long story short: when Chang-DÃaz partisans talk up their candidate and hammer Dianne Wilkerson, voters think Chang-DÃaz sounds good and Wilkerson sounds bad.
<
p>http://thephoenix.com/TalkingP…
<
p>This lack of creativity is a little surprising. I hope she’s not going to run the same campaign she lost with last time & is not paying too much for recycled advice.
That poll was completely different than today’s news Today’s poll included a direct comparison between Wilk and Sonia. Sonia won in a landslide, making your point absolutely moot (nice try, though). I’m sure the election will be closer, because things always tighten up, but it’s its safe to say Sonia has a very good shot at winning and has run an excellent campaign.
I said it was the same tactic as two years ago, not the same poll results…the same tactic- as in sending out a press release with news of a surprisingly positive poll that you paid for. It’s done by the same entity, “Connection Strategies”, and even released at the media at a similar period in the election cycle Aug 7, 2006, then Aug 18, 2008. And yes, I do read what I linked to.
<
p>BTW- since you seem to be in the know, do you know if Connection Strategies by any chance serving as Sonia’s strategist as well as pollster?
…that poll was not totally out of whack in 2006. Now with yet another scandal under her belt, would it be any wonder that Diane’s numbers are even worse?
<
p>Ya, I’ve never heard of Camapign Connections either and their web site seems to be down. But given the baggage Diane has, I can believe the numbers.
<
p>Of course the numbers are based on the voters knowing about Diane’s problems and that means Sonia needs money for media and field. I don’t know if this is any indication, but she’s raised about $20,000 off her actblue efforts. She’ll need a lot more, but if that is just online (and some campaings use Actblue for offline fundraising also) then I’ll bet she’s got some money to tell the voters the truth about Diane.
…the poll results publicized in 2006 were not totally out of whack. Perhaps that’s a good omen for Sonia, it’s really impossible to say. I have doubts that Wilkerson’s latest tangle with campaign finance authorities will make a huge difference- they never did in the past.
<
p>It’s fine to release a poll (or commission a poll for release) as a campaign tactic, but it’s weird to do it so similarly in back to back campaigns.
<
p>It’s inherently a conflict for the pollster and strategist be the same person/group (I don’t know if that’s the case here)- all to easy for them to recommend more polling and a pr strategy based on highlighting their polls.
…and a state senate is a small campaign, who else but the pollster your paying would be a chief strategist? They can recommend all the additional polling and PR services they want…but you can’t get water out of a stone. (But to support your point, I’ve seen plenty of good money pissed away to consultants who don’t know their ass from their elbow.) As for repeating the gimmick of releasing polls…hey, we bit once and now we bit again. If someone is “stupid” it ain’t the campaign.
who make up “weird” problems. It’s weird? Says who? I don’t know if Sonia will win, but no pollster just makes shit up and gets away with it. Furthermore, as others have pointed out, there’s no way Sonia could afford more than one poll – a tracking poll goes for thousands and thousands and thousands of dollars a pop. Even if Sonia did have that cash, there’s so much better ways to do it than score one good press release.
Disclosure, please.
If you’re going to essentially call this poll bs or say that it’s a gimmick, and display an awful lot of knowledge about aspects of it that the average reader wouldn’t know, it leads me to believe you’re a part of the campaign. If you are, per the rules of the road, you should cough it up. Are you a Wilkerson staffer? Campaign Worker? Friend? Volunteer? Enquiring minds want to know.
I’m not a part of either campaign, and likely couldn’t pick either candidate out of a lineup. I just call ’em as I see ’em, especially as things relate to math, science, and engineering.
<
p>It’s not that I think the poll is fishy; it’s that statistically it isn’t likely to be accurate precisely because of the selection problem.
<
p>In any case,
Ryepower12: what is your connection to this race?
jpfernando: what is your connection to this race?
that’s a given (it’s a campaign after all), my point was that it’s a recycled gimmick run from the same playbook as last time.
<
p>I didn’t know displaying more knowledge than the “average reader” (whatever that means) was so suspicious. All I did was read this post, remember a similar story from last time and then google “wilkerson sonia poll” to find the 2006 Pheonix story.
<
p>Anyway, I have no connection to Wilkerson or her campaign in any way. I only asked you if you knew whether “Connection Strategies” was doing double duty as pollster and strategist because you said you “have sources” about it. Do you know?
I asked a very specific question and got a very specific answer: was the poll any good? Yes, the poll’s good. That’s the only thing I know about coming from where I asked. Let’s not blow this up into something that it’s not – I wouldn’t presume to know the strategy or strategists of Sonia’s campaign, because I’m not a part of it. I’m a podcaster and blogger and take my sources as seriously as any Boston Globe writer would. My email is listed on my BMG profile if you’d like to continue this conversation.
to know enough to vouch that the poll is good. Sorry for thinking that might be based on some actual knowledge about it, my bad.
<
p>I was just noting a similarity between this event and one from the last campaign, and giving my opinion about that. There’s no need to continue this by email, it’s not personal.
… is that he knew his source well enough to trust their claims.
Just wondering. It sounds like you may be a supporter of Wilkerson. It would be interesting to hear why people are still supporting her… given all her troubles. So if you are, in fact, a supporter, why? (Or maybe you just hate polls and how they are used by politicians.)
Sure, they should have reasons, and they should be able to explain why the legal troubles don’t tip the scales against her. But I don’t like the attempt by her detractors here to create the illusion of a consensus that the default position should be against her. This poll announcement seems part of such a strategy. It’s a smart one, I’ll grant.
If anything, the poll’s out there because it’s a) good news and b) shows Sonia could very well win this. I can’t tell you how many people have been told by many people and organizations supporting Wilkerson that Sonia has no chance.
<
p>That actually matters in politics, because a lot of people – for whatever reason that fails to make sense to me – won’t vote for people who they think have no shot. Sometimes, they just won’t vote at all. Other times, they’ll vote for who they think is inevitable. That’s the meme that’s been repeated by Wilkerson’s camp – which is another “smart” strategy.
<
p>On your first point, I don’t think anyone should have to explain why they’d support Wilkerson. I get it. I’d support her too, if she didn’t have all this other baggage. That doesn’t mean I shouldn’t try to convince people that they’re wrong – because it sends a terrible message across the state when we elect people who have pleaded guilty or have been convicted of serious crimes, repeatedly. I’m even a man of lessons learned and second chances (note how I only hopped on this bandwagon after Wilkerson’s latest legal problem), but this is a tad absurd.
<
p>On the other hand, we ask lots of people who they support what they support and never question their intentions. If someone’s curious, I think that’s an important question to ask of any supporter. I’d be happy to talk about why I support Sonia any time.
I but I do have a grudging respect for her. I agree that her many (seemingly endless) travails with various authorities are embarrassing, unseemly, and even a little ridiculous. However, I think she also plays a role at the State House that needs to be played and is played by few others. I respect she seems to play by her own rules, which may be connected to the well documented irresponsibility with her finances. I think I do actually have some idea why Wilkerson’s people like her, but I’d have to organize my thoughts better to give an all out defense of Wilkerson.
I’ve seen a number of times in politics where a reasonably popular incumbent finally gets a challenger that would appear to be a better match for the district. The kind of situation where there’s not much question who would win an open seat.
<
p>Many voters will end up choosing the incumbent out of loyalty or the fear of the unknown (in the challenger). Often low-information voters go with the candidate they feel they know.
<
p>The next time around, a repeat challenger starts as much less of an unknown. If she’s been smart, she’s identified areas where she can improve the vote and gets out to meet those communities.
<
p>Frequently the incumbent decides to retire as she hears from key supporters that their constituencies that once were solidly behind the incumbent are now split. There are often meetings among leaders in the district right after the first election where the discussion amounts to “she’s had a nice run, but it’s time for her to go.”
<
p>It’s not unusual to see at 10-15 point swing between elections, especially if the incumbent mails it in.
It’s within the realm of plausibility. But no one would allow it a shred of credibility if it were released by a candidate they opposed.
It doesn’t amount to a hill of beans when it comes down to it. Pollsters are hired in order to phrase questions in a way that makes their client look good.
<
p>Jeez, Dan Cohen could produce a poll next week saying that McCain would beat Obama in Massachusetts. And it would be true.
that I fully support Chang-Diaz in her bid for Wilkerson’s seat.