Late this afternoon, Brian Moran’s campaign released internal poll numbers showing Terry McAuliffe with a statistically insignificant two point advantage over Moran, 31% to 29%. Creigh Deeds was in third place, according to the Moran poll, with 18%. Undecided votes constituted 22%.
These results contrast with two public polls released during the past week showing McAuliffe with a double digit lead. McAuliffe told at least one individual last week that his own internal polls showed him up by 12.
What do we make of this?
First, in primary campaigns, I tend to believe that the internal pollsters have a better grasp on “likely” and “possible” voters than the public pollsters.
This isn’t a matter of competence, but of resources and urgency.
The public polling, especially last week Survey USA’s release, have tended to overestimate the number of Democrats who will likely participate, sometimes by a very wide margin.
The campaign’s own pollsters are far more likely to spend the money necessary to identify who a likely voter is when only three to eight percent of registered voters will actually show up.
There is a special urgency for the campaigns to do this because electoral success is predicated on identifying and then bringing to the polls a small subgroup of the total number of registered voters.
Second, until we see the model of likely voters that the campaigns themselves are using, we cannot make complete sense of why the results differ from McAuliffe to Moran.
The Moran release does not provide a full accounting of how it identifies likely voters other than to say that the results are based on a “carefully screened survey of 606 likely primary voters…based both on vote history in Virginia and self reported likelihood to vote in the upcoming Democratic gubernatorial primary.”
Since the McAuliffe campaign has not released its internal numbers as of this writing, I cannot say how it identifies likely voters either.
But I do have a hunch about what is going on.
In determining their likely voter models, each side will examine what the Moran campaign used as criteria: past voting history in Virginia and self-reported likelihood of voting on June 9th in the upcoming primary. Each side will then develop a model of the electorate based on how they assign weights to these different criteria.
The Moran camp believes that the turnout on June 9th is likely to be quite low. (perhaps around 150,000-250,000 voters) I would guess that a lot of voters who participated in the Presidential primary of 2008, but not in the Senate primary of 2006 or the Lieutenant Governor primary of 2005 get excluded from their model.
The McAuliffe campaign believes that turnout will be considerably higher and that they will be able to tap a number of “federal Democrats” who voted in the Obama-Clinton matchup, but did not vote in either the Senate or LG primaries in 2006 and 2005. In essence, their “screen” is likely to be less severe than the one employed by the Moran campaign.
In the final tally, the only poll that really “matters” comes on June 9th.
But given the influence that polls can have on fund raising and media coverage, it is not surprising that the Moran campaign has released internal results that can reduce the impact of the recent public polls.
It would have been campaign malpractice to do otherwise.
Leave a Reply