Mark Blumenthal | September 22, 2009
What's up with the polls in the Virginia governor's race? Although four new surveys last week left at least one of our readers unsure "what to believe," they were mostly consistent within the usual margin of random variation. Overall, Republican Bob McDonnell still leads Democrat Creigh Deeds, but by a narrower single-digit margin than before. The more intriguing aspect of the new results is what they may imply about a new wave of television advertising now blanketing Northern Virginia.
The Northern Virginia region -- or "NoVA" as politicos call it -- plays something of a wild-card in Virginia election campaigns. It makes things interesting both because the fast growing, suburban counties that ring Washington, D.C. have trended more Democratic in recent years and because its voters live in one of the most expensive and politically inefficient television advertising markets in the country.
Definitions of "Northern Virginia" vary,** but I prefer the broader cut that includes all of the counties in the Washington, D.C. media market as defined by A.C. Nielsen (the highlighting of the nationalatlas.gov map above is mine). The D.C. market includes a huge chunk of Virginia's voters (36% in the 2008 presidential election) and has recently gone for Democrats by wider margins than the rest of the state. They gave Barack Obama a 16-point margin over John McCain (57% to 42%) and Tim Kaine a 14-point margin for Governor in 2005 over Republican Jerry Kilgore (56% to 42%). Interestingly, Obama and Kaine still carried the rest of Virginia, albeit by much narrower margins (Obama by one point, 51% to 49% and Kaine by two, 50% to 48%).
The reason I prefer to look at the larger media market, which includes a handful of rural counties beyond the DC suburbs, is that these voters get their broadcast television advertising through Washington DC stations. The DC market, 8th largest in the nation four years ago, is a wildly inefficient way to communicate with voters in Northern Virginia. Yet TV ads remain the most powerful tool available to reach the less politically engaged and therefore more persuadable voters who typically decide elections.
Since television advertising is so expensive and inefficient in NoVA, campaigns typically air their broadcast ads there later and at less volume than in Virginia's other big markets. As such, Northern Virginia voters tend to engage later in the campaign, causing some interesting late shifts especially in races with lesser known candidates. Deeds' surprise win in the Democratic primary for Governor this year was one example. Jim Webb's late surge in 2006 was especially instructive.
That year, news media stories had pummeled incumbent Senator George Allen over his now infamous "Macaca" comments in late August, yet many political observers wondered why he remained ahead throughout much of September. Webb's surge did not occur until his campaign was able to purchase significant time in the Washington DC market in October.
And that history brings us to the current race and the four new polls completed and released last week. In late August, a front page Washington Post story highlighted McDonnell's very conservative twenty-year-old graduate school thesis. The paper followed up with more front-section coverage on recriminations that followed (thus infuriating conservatives). Two automated surveys conducted soon thereafter, by Rasmussen Reports and SurveyUSA, showed no discernible change in voter preferences.
Those surveys came before both the McDonnell and Deeds campaigns launched their Northern Virginia television advertising just after Labor Day on September 8. The first Deeds spot was positive -- a version of an introductory advertisement featuring their candidate speaking to the camera that had run elsewhere during the primaries but not in the DC market.
The Deeds campaign did not start airing a negative ad on the Post "Culture Warrior" story until September 9, and one source that has been following the advertising closely tells me that the Deeds negative ad did not start airing heavily in the DC market until early last week. Either way, the four new surveys -- from the Washington Post, Rasmussen, DailyKos/Research 2000 and Clarus -- were the first to capture the effects, if any, from the initial wave of advertising. All were completed by last Thursday.
Clarus showed 20% of the voters as undecided, obviously a bit of an outlier compared to the other surveys. That result is most likely the result of interviewers pushing less for an answer and is essentially consistent with a large number (31%) who told the Washington Post's pollsters that they are either undecided or might still change their minds.
Nevertheless, the three other surveys produced reasonably consistent horse race results: All three showed results within the margin of sampling error of a 50% to 45% McDonnell lead, which is also exactly the same five-point margin as on the Clarus survey. So all four show a closer contest than the eight-point or better McDonnell leads we have been seeing on other surveys over the summer. As such, I think it's reasonable to look at our more sensitive trend lines (reproduced below), which capture the closer result shown in the more recent surveys, rather than our standard estimator. The difference is a roughly four-point McDonnell lead below versus a seven point lead using our standard estimate.
Is the change bigger in Northern Virginia? That is what I would expect given both the "Culture Warrior" story and the recent debut of television advertising in the region. Unfortunately, some of the data on that question is incomplete. The Washington Post survey shows a much bigger shift in it's Northern Virginia region (from "about even" in August to a 57% to 40% lead for Deeds now) than in other regions. The Kos/Research2000 poll shows only a not-significant single point drop for McDonnell overall, although that change appears to occur entirely in their Northern Virginia region (Deeds up 3, McDonnell down 3). And Rasmussen did not report on any regional crosstabs for their most recent survey.
So will these modest Deeds gains persist? Can Deeds continue to close the gap? Obviously, the race still has a long way to go, and McDonnell has a favorable political environment plus a lot of history on his side. The Deeds campaign stepped up the tone and volume of its attacks this week, but then, so did McDonnell. If you're a Democrat, the bad news is that Deeds is still behind on all surveys and McDonnell still holds the votes he needs to win. The good news -- perhaps -- is that these new polls may have caught just the first few days of negative television advertising. If the Deeds message works, and if it is not neutralized by McDonnell's counterpunch, then we may still see further gains in the next round of polls.
**More on regions: Most public polls have been defining "Northern" Virginia using the 703 and 571 area codes or a set of counties covering similar geography. Either way, they define "NoVA" more narrowly (26-27% of all voters) than the full DC media market.