Home National Politics Perriello Campaign Rips SurveyUSA Poll

Perriello Campaign Rips SurveyUSA Poll


The following press release by the Perriello campaign is in response to this SurveyUSA poll, which shows State Senator Robert Hurt (R) leading Rep. Perriello 58%-35%. In my opinion, SurveyUSA has been pretty good in the past (Nate Silver rates them third, behind Field Poll and ABC/Washington Post), the big question is whether or not they have a handle on the 5th CD of Virginia. They certainly didn’t seem to in 2008, but of course, Tom Perriello wasn’t an incumbent back then. We’ll see…

Even Robert Hurt’s Campaign Manager Agrees: SurveyUSA Polls “Not Reputable”

July 20, 1010–Ivy, VA–Lise Clavel, campaign manager for the Perriello campaign, released the following statement about the SurveyUSA poll conducted for WDBJ7 that was released today.

“Everyone knows that SurveyUSA polls are useless; even Sen. Hurt’s campaign manager has admitted that. This is the same poll that had us down 34 points in 2008, and we all know how that story turned out. The fact is that every credible poll this year has had Tom Perriello ahead or tied. We know this will be a tight race, and we know Tom’s hard work and commitment to the hardworking families he represents will lead him to re-election this fall.”

More after the “fold,” including a quote by Robert Hurt’s campaign manager saying that SurveyUSA is not “reputable for statistical purposes.” Ha.


*Sean Harrison, campaign manager for Robert Hurt’s campaign, said in an email on July 13, 2010, to NBC29 regarding the negotiation of a polling standard in the televised debate: “We would agree to a ‘polling standard’ of 10% for the Independent candidate of a poll conducted by a reputable polling company that does not use automated dialers and push button polling – Mason Dixon research would be reputable for statistical purposes, SURVEY USA would not.” [emphasis added]

*The SurveyUSA poll conducted on August 13, 2008, had Tom Perriello trailing Virgil Goode by 34 points. (Link)

*The poll methodology had no information about geography. The poll may have excluded areas of the district with more Democratic voters.

*The poll shows 3% undecided voters, highly unlikely for a race that is three months away when the GOP candidate has just been confirmed. Push button polls routinely have almost no undecided voters, which is part of why they are unreliable.

*The poll’s projection of the partisanship mix of likely voters is highly unlikely. Survey USA projects Dem: 27%, GOP: 42%, and Independents: 29%. However, if we define likely voter as voted in 2008 + [either 2006 or 2007 or 2009/ and define “Democrats” as voters in that group who are likely Democrats according to DPVA Model (strong + lean) and define “Republicans” as voters in that group who are likely Republicans according to DPVA Model (strong + lean) and Independents as everyone else, you get the following composition for November:

Dem: 41%

GOP: 29%

Indie: 30%

  • DCCyclone

    I know there are a lot of conservaDem voters in the South, and rural Virginia is culturally the South, but with no party-based voter registration in Virginia the party self-identification by polling respondents is going to line up according to actual voting behavior much more so than in places like Kentucky or North Carolina where people are likely to answer according to what box they checked on a form, and not necessarily by how they usually vote.

    That being the case, a turnout model of 41D-29R-30I is absurd.  SUSA accurately polled VA-05 on election eve 2008 and had Perriello down just 3 points at 50-47, confirming his stunning momentum then.  And that poll had party ID of 40D-38R-20I.

    For this poll that looks bad for Tom, I can actually believe 42% of voters being Republican.  I think the D and I numbers look wrong, with Ds undercounted and Is overcounted.  It actually doesn’t make sense that indies would be hypermotivated to vote in a midterm with nothing else on the ballot, as is the case in Virginia this fall that distinguishes the state from all but a couple others.

    But the problem is there’s no way to fix the turnout model to something reasonably accurate that doesn’t still show Tom down significantly to Hurt.

    I hope Perriello’s and Hurt’s private polling show a tight race as the Perriello camp claims.  This would be a good time to release their private numbers, even if they’re down 5 or so.