crossposted from Blue Commonwealth
Yesterday we got the results of two straw polls and one public poll from SUSA. Those paying attention also know there is a PPP poll in the field whose results are likely to be out in next few days.
Each of the 3 gubernatorial campaigns and/or their supporters are trumpeting the results that make them look good while ignoring those that could be warning signs. So perhaps there should be some reality checking going on.
I come to this not as a statistician, although I have a fair amount of statistical training from my doctoral work in education. Also, since I teach government, I spend the better part of a week teaching my Advanced Placement students about the different types of polls, including how to look at cross-tabs where available.
And I add to that the words I heard last night at Brigades, where someone urging people to a Jeff-Jack dinner reminded the people present there would be a straw poll, so get your supporters out to do well.
Let me note again I am supporting Creigh Deeds and Mike Signer in the contested races.
Let's start with the two straw polls, in Fredericksburg and Winchester. Moran kicked ass in Fredericksburg, Deeds did so in Winchester. In general, straw polls do not tell us all that much we do not already know. There are exceptions, such as Webb surprisingly winning the straw poll at Connolly's St. Paddy's Day event in 2006, when most people assumed that Miller had the poll rapped up. The Webb campaign did not have the money to buy a lot of tickets for supporters, so the results then were a reflection of the support he was drawing on the ground, from ordinary people.
But most straw polls tend to be reflective of the support one has from party regulars in that area. In a low turnout primary, that may be fairly indicative of the kind of support in that area one will see when people vote, because not only will these regulars vote, they are the key to turning out others.
But when the turnout begins to increase beyond the party base, you get a higher percentage of people who do not react the way party regulars do, so the straw polls may carry less weight.
As a Deeds and Signer report, let me thus comment first on the recent 5th CD straw poll and the one in Winchester. In both cases candidates I support did quite well - Deeds over 90% in the 5th and almost 80% in Winchester. Deeds lives in the 6th, but has strong support throughout the Valley (Winchester) and in much of the 5th (his Senate district overlaps a chunk, and among Democrats he was known for his strong support of Perriello). Thus he would be expected to win a real primary vote, but almost certainly not by such obscene margins. As for Signer, he has personal roots in the 5th - especially in Charlottesville - and he was present for the poll while Wagner was not. Mike is an effective speaker. Thus his 66% margin may or may not be indicative. Further, his campaign chair is Dem. party chair of a County in the 5th (Lunenberg), which further amplifies Mike's support among party regulars. It is possible he could win 2/3 of the vote in a 2-way, but that straw poll took place when Bowerbank was still in the race, and thus was in a 3way contest. actually Bowerbank had dropped out, as someone noted over at BC, so the major impact was probably Mike's presence at the event.
About polls. One always needs to see the crosstabs to see if the sample might be biased towards groups or regions that would disproportionally favor one candidate. Of greater importance is to compare the crosstabs from one version to the next of the same poll - movement in top level numbers from or to a candidate may be entirely or largely explained by changes in the demographic makeup of the sample.
Thus in the recent SUSA poll, the model has significantly reduced the percentage of the sample that is women from 58% to 52%. In this case the top level results may actually understate how well McAuliffe is doing, since he has a strong advantage among females, winning 39-23-22 over Moran and Deeds respectively, while among men his margin is 34-31-21 with Deeds in second. And since we know that traditional females react more negatively than do males to what is perceived as attack ads, that change in distribution may have two important effects on top level numbers - it may overstate how much ground Deeds has made up relative to Moran, and it may well understate how badly the perception that Moran is responsible for the negativity is having upon his overall numbers.
There are still a substantial number of voters whose either have not decided (15%) or who could still change their minds (57%), although people are beginning to fix in their choices - that 57% is down from 64% in the last poll 3 weeks earlier, although again that difference could largely be explained by the demographic differences between the two polls.
Deeds can rightly claim some momentum, but if he could not one would be totally writing him off, as he is now advertising, whereas McAuliffe had had the advertising field to himself before the last poll.
There are, to my mind, key things to note in the most recent SUSA poll, most of which are bad news for Moran.
- He is now in 3rd place overall.
- He trails badly among African-American voters, with those numbers being 41-27-15 for McAuliffe, Deeds and moran respectively
- Moran's margin in NoVa has dropped from 23 to 10 points in three weeks - and no one is advertising on air in NoVa, although McAuliffe is quite aggressive using direct mail.
The second point above may be why Moran is now running ads that are fairly negative on Black radio stations - it appears that he may have to try to depress Black turnout to have any hope of being competitive.
And given that McAuliffe will continue to have the advantage in advertising, it will be interesting to see whether positions are beginning to harden - there are no more debates, and some people will soon be voting.
Yesterday Tom Jensen of PPP made this post about a poll they have in the field, which should be released either today or tomorrow. Let me quote two paragraphs from the post and then comment:
A pretty clear divide is emerging. Among the frequent primary voters Terry McAuliffe, Brian Moran, and Creigh Deeds are basically in a three way statistical tie based on the interviews conducted so far. But among the more casual primary voters who did not find Webb-Miller compelling enough to head to the polls but who are intending to come out this time McAuliffe has a substantial lead.
It looks like it would be in Deeds and Moran's best interests for turnout to be as low as possible because if it does get up over 200,000 or so it appears, at least three weeks out, that McAuliffe would be the beneficiary.
I read yesterday that Steve Jarding, a top strategist for Moran (a role he also played for Webb in 2006) is now predicting a turnout around 200,000. It has always been my sense that were the turnout at 130,000 or below Moran was the likely winner, but once it got above 175,000 he would have real difficulties. I have heard from a key figure in the campaign of Creigh Deeds that his original prediction had been well over 300,000, but he now expected around 150,000. Deeds might have his best shot in the range or 150,000-175,000, because then it depends how around the state the votes are distributed. A high percentage of the vote in NoVa clearly is to Moran's advantage, even given how much that advantage has shrunk. But a turnout of 160,000 with less than 30% of the turnout from NoVa would be a big help to Deeds. The current SUSA model has 28% from NoVa - one notes that such a proportion is somewhat lower than we have seen in several recent primary and general elections.
Once the numbers get over 200,000, it is hard to see anyone but McAuliffe winning, regardless of how the votes are distributed.
As to NoVa turnout - I think the phenomenon of election fatigue may be largely confined to that part of the state. We have had a series of special elections, and at some point it becomes difficult to keep voters engaged for so long. But that has not been the experience in most of the rest of the Commonwealth.
So here are my takeaways.
- Don't pay a lot of attention to straw polls.
- Remember that a campaign has to give its supporters reasons to feel positive, to stay engaged, so each will emphasize what shows them in the best light
- Raw topline numbers often hide the real picture, for which one needs both the cross-tabs and the same poll over time to see if any trends are indicated.
On Tuesday, Ben Tribbett of Not Larry Sabato made his prediction for the gubernatorial primary in this post:
McAuliffe 42
Deeds 31
Moran 27
Ben has a pretty good track record of calling races in Virginia. And it is interesting that his numbers reflect the order and general spread of the PPP poll (hmmm....).
For me? I still think turnout will be over 200,000, in which case I have little doubt that McAuliffe will win.
Are there things that could still change the dynamics? Perhaps. But there are no more debates, and hence few occasions for gaffes. McAuliffe will continue to have a spending advantage, both for advertising, and in having paid staff on the ground around the state.
I'm not ready to call the race yet. But it becomes increasingly difficult to see how either Deeds or Moran can find a game-changer.
But don't worry, we only have 19 days more, and then on Jun e 10th instead of obsessing about who will win, we can start the process of determining how and why one candidate did.
Peace.