Creigh Deeds has been courting them for months, but apparently a sizeable portion of normally solidly Democratic voters and Independents are still not smitten by the Democrat from Bath County. If he has any hopes of winning, Deeds must close the deal with them in the next two weeks.
In a recent survey of likely voters (registered, planning to vote, and following the race closely or very closely) I conducted for The Virginian-Pilot and WVEC-TV, we decided not push those likely voters who said they were undecided to make a decision. The standard practice for pollsters at this stage in an election cycle is to not let undecided respondents remain so, but instead to force them to say to whom they are leaning. The problem with this standard practice is that one loses sight of why those undecided voters are still undecided. Our approach was to let them remain undecided so that we could get a better sense of who they are and why they remain on the fence this late in the game. And what we found is telling.
As of late last week, just over one-in-four (27.1%) African Americans are undecided. Right at one-in-four (24.5%) women are undecided. Just over one-in-five self-identified Democrats (21.7%) and Independents (23.1%) are undecided. Wow!
If I were a Deeds strategist these numbers would make my stomach turn. They suggest that I have not closed the deal with my base, and I have not done a good job speaking to independents. After weeks and weeks of pounding Bob McDonnell on his thesis, a quarter of women voters should not be on the fence.
However, campaigns have to be optimistic or they stand no chance of winning, so on the optimistic side these numbers show Deeds the path to victory. In the remaining days of the campaign, he has to speak to these voters if he hopes to win. Or, he could ask some big-name friends like Bill Clinton and Barack Obama to do it for him.
Quentin Kidd is Director of the Judy Ford Wason Center for Public Policy and Associate Professor of Political Science at Christopher Newport University
I wish CPP’s website would be updated with the survey instrument and other data so someone else could take a look at this.
I am fascinated that Dr. Kidd considers his utter outlier of undecideds to be a result of not pushing them. Evidently, political polling is not his specialty since when a push is done, typically only a few of the undecideds express a preference. That would still leave the undecideds in this race as a total outlier.
Of course, the problem COULDN’T POSSIBLY be faulty methodology. For example, where was the question asked within the survey? Right out of the gate? That would raise undecideds right there if it were since they have not gotten a comfort level yet with the stranger calling them. Or what was the exact wording of the question? Or how was the likely voter screen done? Any of these things (and others) can impact a whole host of response and data issues. I wish this stuff were on the CPP website. Maybe it is, but I can’t find it.
But here’s the kicker. Kidd writes, “After weeks and weeks of pounding Bob McDonnell on his thesis, a quarter of women voters should not be on the fence.” If that were the case, and I suspect that, at least in this, you are right, wouldn’t that make you wonder why you got the result you did? When the outcome so utterly defies logic and is an obvious outlier to boot, doesn’t that make you consider that your poll might be totally flawed? Or even just that 1 in 20 polls that is simply wrong?
But of course, a 506 person sample conducted over 5 days (5 days??? Who does a 500 sample political poll that is not a rolling track over five days???) by the esteemed Dr. Kidd could not possibly be flawed or have errors. That you took that data at face value, don’t seem to question its validity, and then try to analyze it and draw conclusions from it without doing a check poll behind it to reconfirm the obviously odd results, is stunning and more than a little scary.
I apologize about the website delay…the data will be up and available as soon as our IT people get it there.
I would be happy, in the meantime, to send you cross tabs and toplines…just shoot me an e-mail at qkidd@cnu.edu. You can even do it anonymously since you seem not to want to identify yourself.
Of course as with any survey there is always a certain degree of error, and there are certainly bad polls. I make no claims at perfection. My Methodology is clearly spelled out.
But, in talking with several pollsters in the last couple of days a lot of us see the same thing…Deeds is having trouble with a certain proportion of Democratic voters that he shouldn’t (all things being equal) be having trouble with. That is the main point I take from this data, and the one I was making.