Hurt Blowout or Tied Race?

We are entering the last month of the election cycle and the whining from the two major candidates has reached historical levels. While Hurt complains about his treatment by the media, Perriello has started to complain about the constant stream of SurveyUSA polls showing him toast. Where's the beef? Not with either one of these so-called men.

Before this, it was Twittergate. Before that was the debate over debates. Complaints over the times of Tommy's Town Halls. Jeff Clark's finances. Threats to Perriello's family. And so forth. And so on.

A lot of style, a lot of drama, and not a lot of substance.

If we actually look at the polling numbers, instead of just throwing a Tommy Temper Tantrum, we might begin to realize why this election has been so emotional, so dramatic, but so substance-less.

Eight months ago, back in February, the first public poll of the race in VA-05 was released by Public Policy Polling. They showed the election, before the Republican primary and much of Hurt's rise in name recognition, at 44% to 44%. TIED. Dead even.

Think about that. On the one hand, Hurt was a virtual unknown to the district outside of the areas he had represented. The primary campaign had barely gotten under way. Certainly this meant he had room to grow?

But this was after a year of even more drama and emotion directed at Perriello. Forged letters. Effigy burnings. Cap and tax. Health care takeover. Town halls. A failed stimulus. After the worst Democrat President since Jimmy Carter, Tommy Boy was still tied? Certainly this meant he could mount a come back once he had a clear opponent to attack?

The biggest problem for the Perriello camp was the makeup of the poll. The February PPP poll was of registered voters, not likely voters. It represented the district as a whole, not the people motivated enough to get off their asses and vote. It showed a district that went narrow for McCain, not heavily Republican like past elections in 2004, and 2000, and 1996 and . . . you get the idea.

As soon as pollsters started screening out unlikely voters, the first time Obama voters like college students at UVA or African-Americans in Danville, a gap was bound to open up between Hurt and Perriello.

But the poll had some good news for Perriello. Throw in a third party candidate with Tea Party values and suddenly Hurt was down to 27% and the Tea Partier got 19%. Perriello was unchanged at 44%. Once a third party candidate announced against Hurt, a clear result of his primary win, and it was possible for Perriello to open up a gap against Hurt. Or at least hold it close.

What has happened?

Let's take a look, without Survey USA. I'll address them later.

The next we have is in late July. Five months later, the Republican polling firm Ayres, McHenry & Associates showed Hurt with a six point lead. Hurt was up to 49% and Perriello down to 43%. The poll's internals show a more conservative, more Republican electorate than the representative sample from PPP. The infamous enthusiasm gap strikes!

But this poll excluded independent candidate Jeff Clark. I know that the betrayal of conservative values by Jim McKelvey and others has hurt Clark's chances, but he should at least have been included in the poll. With Clark polling at around 6 to 4 percent in other polls, this could have cut down Hurt's margin over Perriello in half or even eliminate it.

Also worth noting is that even in this more conservative, more Republican voting sample, Perriello is still at 43%. That leads me to believe that a poll taken at the same time of registered voters, not just likely, would put him over the 44% that PPP found. If the poll included Clark, I suspect it could have shown a small, low digit edge for Perriello. Perriello's ads were making a difference among the public at large, but he's like a salmon swimming up stream against the enthusiasm gap. Right into the mouth of Robert the hungry grizzly bear.

A month later, in late August, we have a Democratic polling firm, Global Strategy Group, showing a two point race. Here Hurt is down to 44%, Perriello is 42%, and Clark is at 6%. This is roughly what I'd expect from the Ayers poll if it had included Clark. Almost no movement.

A few weeks later, in mid-September, the Perriello campaign released their own poll from Benenson Strategy Group showing . . . . you guessed it, a two point race. Hurt is at 46%, Perriello at 44%, and Clark at 4%. This is virtually identical to the Global Strategy Group poll. And virtually identical to what Ayers should have shown with Clark included.

And all of them are about what we would have expected from PPP's February poll if we took their poll of registered voters and modified it for (1) the enthusiasm gap among likely voters favoring Republicans and (2) Clark being on the ballot, but not getting as much support as he could if Jim "Judas" McKelvey and other hadn't changed their mind and broken promises.

So here we are at the start of October. And the race, as far as I can tell, is exactly where it was in February.

I'm reminded of what the good folks at Southside Central predicted right after the Republican primary:

First prediction on the general election vote percentage: Hurt 48%, Perriello 48% Clark 4%. If McKelvey throws his support behind Hurt, it becomes Hurt 49+%, Perriello 49+%, Clark <2%.

First OddsMakers on the general election winner: Hurt 50%, Perriello 50%, Clark 0%. Yep, it’s a fresh start and a toss-up right now.

Ain't that the truth?

Everything that you've read about in the newspapers for the past eight months has been meaningless. The debates, the talk about style, all of it. Everything.

And here is why I'm not mentioning Survey USA. Here's what the campaign looks like if you included Survey USA.

February: The race is tied before polling shifts to likely voters. (PPP)
Mid August: Hurt has surged to a 23 point lead as the polling shifts to likely voters. (Survey USA)
Late July: Perriello has mounted a comeback and is down by 6 without Clark included, potentially down by two or three with Clark included. (Ayers)
Late August: Perriello's comeback settles out at two points behind. (Global)
Later August: Perriello's campaign collapses and he's behind by 26! (Survey USA)
Mid September: The come back kid, Perriello has recovered and is behind by 2. (Beneson)
Late September: Once again race shifts and Hurt is back up by 23 points! (Survey USA)

Seriously? Is the district really swinging back and forth that much?

I don't think so.

Survey USA is clearly polling an entirely different reality in its assumptions about turnout in VA 05 this year. If it was just Survey USA against one polling firm, I could see the argument to split the difference. But four against one?

I think Tommy's a crybaby to go to the press with complaints against Survey USA, but the argument that they are showing an absurdly high number of early voters--a mathematically impossible high number--is persuasive. Survey USA seems to be pre-choosing voters to call and then adding a likely voter screen, something that may come up with a scenario that has an impossibly high enthusiasm gap benefiting the GOP.

If you average together the Survey USA polls (around a 24 point lead for Hurt) and the three other likely voter polls (without correcting for Ayer's excluding Clark) and average it out, you get about a 8.5% gap. Which is, from earlier Tweeting, what Nate Silver had predicted for Perriello's defeat in VA-05.

Nate gives Perriello a 88% chance of being defeated. Not surprising if you assume an 8 point spread. Nate doesn't post the House numbers, but he notes that Gubernatorial and Senatorial candidates ahead by six to eight points end up winning 89% of the time. About what he has for Hurt against Perriello.

I entirely respect a model that treats all pollsters fairly. But I think it's clear that something is clearly off with Survey USA this year in VA 05. Nate hasn't discussed specific House races yet, but I'd love to know if there is any other House race this year that has such a high variance between polls. Survey USA is clearly an outlier and by a huge margin. Is this seen anywhere else in the country?

What if we were to pull it from the batch? I don't know all of the calculations in Nate's House predictions, but I'll just put out this guess. The remaining polls would average out at about 3.3, which would put the race in the 3-6 range and give Perriello a 79% chance of defeat. But if you were to try to fix the Ayers poll to include Clark you'd get closer to 1.7 or 2 depending on how much including Clark narrowed the gap. That puts you within a 55% chance of defeat, if not better for Perriello to hold on. In other words? About even.

This district is narrowly divided right now. The election isn't going to change many minds. It never was. All this talk about independents, moderates, and swing voters is bullshit.

This election was always about who could unite conservatives and keep them energized. Hurt's campaign has failed on two fronts. His vote for the largest tax increase in the history of Virginia alienated true conservatives who weren't drunk of the GOP Kool Aid. His lackluster campaign has failed to energize people outside of his traditional base of supporters. Just being a Republican in a bad year for Democrats may still be enough for Robbie. But he's made this election closer than it should be and Perriello still has a shot at pulling this out.