Tuesday, May 15, 2012

Fake and faulty polls deny Barrett the lead he has

Apparently I'm getting a reputation, as one of the readers asked if I have a breakdown of today's PPP poll which showed a Walker 49, Barrett 45 result. (I use people who can, you know, vote. "Likely voters" is a near-worthless filter). Well, glad you asked, because I do have a breakdown. And like other polls, it's understates Tom Barrett's support.

When you go into the cross-tabs, there's no secret why Walker's winning. More right-wingers answered the poll than left-wingers, and not by a little.

Respondents to May 11-13 PPP Wisc. poll
Republican 35%
Democrat 28%
Independent 37%

Conservative 41%
Moderate 41%
Liberal 17%

Well, if that's who actually shows up to the polls on June 5, then yeah, it looks pretty good for Scott Walker. Too bad for Scotty that this is NOT the breakdown that happens in actual Wisconsin elections, and I'll use the on-target 2010 CBS exit poll as a proxy. Those numnbers were 36-37-28 GOP/Dem/Indy and 36-43-21 Con/Mod/Lib. Change the party and ideology numbers to match that, and check out what you get.

By party- Barrett 48.77, Walker 47.14
By ideology- Barrett 48.41, Walker 45.92

Barrett by 2 or 3 makes a very different scenario than Walker by 5, doesn't it? Again, Kos makes the same mistake they made last month, where they assume Dems and liberals won't vote because it's a "special" election. This is absurd on its face, because the only reason we're having this "special election" is largely because of the efforts of left-leaning groups and individuals (with an assist to moderates and conservatives with a sense of decency). After all the work we've done and engagement that we've had, we're gonna shut it down at money time? THE HELL WE WILL!

And the scary part is that the 2010 turnout model was high for GOPs. As noted in my article last month, Kos's correct predictions for the Senate recalls ended up the way they did because the results went right between the Obama/Dem landslide in 2008 and the Walker/GOP landslide in 2010. So if anything, Barrett could well be up by 4 or 5 with an accurate voter turnout model.

There's also been 2 BS polls by right-wing propagandists in the last week showing Walker with a lead. One of those making the poll was Rasmussen reports, who is hired by Fox do to do polls that fit the station's agenda, and Nate Silver pointed out that Raz was GOP +4 vs. reality in his final polls in 2010, the worst of the major pollsters he tracked. Silver adds that Rasmussen also tried to rig a poll to make it look like Walker had more support than he did after dropping the bomb in Feburary 2011, and if you've followed Razzy over the years, you know this is the point of a campaign where he tries to do "agenda-setting" hammering his fist on the scale for the GOP. So when he says Walker +5, it's probably closer to Barrett +5.

The last poll that has to be called out is "We Ask America", who did a poll for the FIB version of WMC- the Illinois Manufacturers Association. Madison-based Ruth Coniff does a great takedown of this garbage poll in the Progressive.
The text that introduces the "Recall Fever" poll on We Ask America's web site tips the pollsters' hand, calling the law that allows Wisconsin law "loopy" because it's "comparatively easy to recall a governor" (tell that to the 30,000 volunteers who stood out in the snow to gather 1 million signatures).

The numbers are "fluid" and the race is "far from over" the same text says. "Still, some of Walker’s detractors have been vocal about the fact that Walker was duly elected in 2010 and in most states would not be able to be challenged with a recall without greater cause." (Ed. Note: Total weasal words, allowing these gys to double back when reality strikes and Barrett "takes" the lead he already had.)

The poll, of 1,219 likely voters, came in for some harsh criticism by Daily Kos which challenged its cheap methodology, small sample size, difficulty checking the group's polling track record, and obvious Republican bias.
We Ask America also has a record of producing horribly biased polls in Wisconsin, as they said last August that crazy Northwoods Teabagger Kim Simac was within 2 points of Jim Holperin, and she lost the election by 10. Well, if they're GOP +8, then it means that Barrett-Walker is a toss-up, which matches most polls when you use real turout models. Also interestingly, "We Ask America" tried to throw out a poll during the Wisconsin Uprising, and despite a a snarky and pro-right wing narrative that tried to blame President Obama for saying something about the events, still couldn't find a majority of Wisconsinites to support Act 10.

But here's the part that bothers me about these faulty polls. When they come out saying Walker leads when he doesn't, lazy media reports Walker as being ahead without looking behind the numbers, they slant their coverage with that in mind, and casual bystanders may become likely to want to vote for Scotty solely for the selfish reason of wanting to say they supported the winner. I certainly believe bad polls that overstated Ron Johnson's and Walker's leads in 2010 helped them greatly, and both Tom Barrett and Russ Feingold overperformed final polls by 2-3 points on Election Day, even with Dems discouraged because of "GOP landslide" projections.

On a scarier point, inaccruate pro-Walker polls open the door for the Kathy Nickolauses of the world to plausibly cheat and throw the election to Walker. That's why we have to be pushing on these companies to get these polls accurate and matching real Wisconsin voter turnouts, because if they're showing Barrett +5 (like it could conceivably be)and all of a sudden Walker is reported to win by 3 n Election Night despite exit polls showing a Barrett win, a lot of people might find it fishy and not allow the fraud to happen. They might be more likely to shrug their shoulders if the polls show Walker up or even, and allow the cheating to go unpunished.

Bottom line- MAKE DAMN SURE YOU GET OUT TO VOTE ON JUNE 5, and that every decent person you know does the same. If we do that, I think we'll be throwing a helluva party 3 weeks from tonight.


  1. So I guess the next question would be, why are two consecutive polls by a Democratic-leaning pollster so skewed towards a GOP turnout? Allegedly they have ways to correct for cell-phone only households, so it couldn't be that.

  2. FWIW, I agree completely with Jake. Forget what the polls say (and I don't say that easily, being a numbers-driven stat person myself); when Wisconsin turns out to vote in numbers, Democrats win. The models can predict a lot of things, but they haven't quite figured out how to adjust for the 318,000 voters in Milwaukee County (213,000 in the urban core) who only show up when motivated. That's a HUGE contingent (+/- .5%) that decides Wis elections -- as it did in both 2008 (when they showed up to vote for Obama) and 2010, when 108,738 of them (75,675 in the urban core) stayed home. Barrett lost the election by 124,638 votes statewide.

    It's almost as if you have to assign a 'probability index' to their engagement in an election (ie, their likelihood of voting) in order to forecast an election result. Maybe PPP does this already.

    Barrett doesn't have Obama's appeal. And, despite all the ads, etc, we are running into some 'low information' voters in the city. But there's a massive GOTV underway to get the people left behind by Walker's policies to take out their anger at the ballot box. They hated him when he was Milwaukee County Executive, but that didn't motivate them to deny him the Governor's office. Observers feel it will be different this time around.

    Most insiders feel Walker is on shaky ground.

    One final point: Jake is dead on about election fraud, too. Election Integrity experts warn that it's hard to steal an election when pre-election polls indicate a big gap. If people believe it's going to be a 'close race', the vote totals can be cooked, just like Walker's job numbers.

    This whole 'Marquette poll' was set up a year ago to fool people. It's right out of Karl Rove's campaign manual.

  3. Felling Blue- Thanks for the good words, and I like the report from Milwaukee. That part and the HWy 29 corridor will be key. Which makes it all the more stupid that Obama isn't making an appearance in Milwaukee and/or somewhere else in the state in the next 10 days.

    Good call on the Marquette poll, I just called them out today for having their sample be 47% conservative and filled with people from the 262. Franklin's whole WPRI-funded background is sketchy.