Conflicting Virginia Polls

By Jim Ellis

May 22, 2017 — Early in this election cycle we’ve already seen several special and odd-numbered year campaigns produce conflicting polling data, and at the end of last week, a new example came forth. Two new polls from the Virginia governor’s race, Democratic primary, produced opposite results and both can be questioned in terms of reliability.

Earlier in the week, the Virginia Education Association released a Public Policy Polling survey (May 9-10; 745 likely Virginia Democratic primary voters), which projects Lt. Gov. Ralph Northam leading former US Rep. Tom Perriello (D-Charlottesville) by a substantial 45-35 percent margin.

Late last week, the Washington Post and the Schar School of George Mason University released their sponsored Abt Associates poll (May 9-14; 1,604 Virginia adults; 351 likely Democratic primary voters; 264 likely Republican primary voters) that produced a much different result. According to this polling sample, it is Perriello who actually holds a 40-38 percent preference lead among the most likely June 13 Democratic primary voters.

Not only do we see inconsistent conclusions from this pair of surveys, but also methodological questions arise. The Public Policy Polling survey has the stronger sampling group particulars, but may have bias problems. PPP features a robust sample of 745 Democratic primary voter respondents but the poll was conducted for an organization that is outwardly supporting Northam, and the 10-point advantage for their candidate is beyond any previously released independent figures.

The Washington Post Abt survey is methodologically flawed. The 1,604 respondent pool is obviously a good size, but the individuals are identified as “adults.” When attempting to determine registered and likely primary voters, the total sample shrinks considerably. Finding only 351 Democratic primary voters and 264 similar Republican participants shows the high number of non-voters included in the original universe.

To put the sample size in perspective, the 351 Democratic primary voters may qualify as an adequate sample for a congressional district poll, but not in a state with 11 CDs. The Republican sample is even low for a congressional contest, let alone a statewide campaign. Therefore, the Post/Schar poll has an unacceptably high error factor.

Comparing all available data, it does appear that the Northam-Perriello primary is highly competitive and likely to end in a close result. More importantly, it defines what many other Democratic primaries will look like across the nation as the Bernie Sanders activist faction continues to oppose the Democratic establishment.

Witnessing both senators Bernie Sanders and Elizabeth Warren (D-MA) publicly endorsing Perriello, while establishment officeholders such as Gov. Terry McAuliffe and senators Mark Warner and Tim Kaine back Northam along with traditional Democratic support organizations such as the Virginia Education Association, we see similar divisions to those formulated in the Clinton-Sanders 2016 presidential primary.

The remaining question is will the two candidates stretch so far to the left in attempting to clinch the nomination, and particularly so in vote-rich northern Virginia, that they cede the middle ground for the general election.

For his part, former Republican National Committee chairman and 2014 US Senate candidate Ed Gillespie, at this point the presumed GOP gubernatorial nominee, understands the dynamic and is navigating more toward the ideological center for what he anticipates being a close general election campaign. In many ways, this Virginia gubernatorial race could prove a precursor for the myriad of partisan coming campaigns soon to be contested during this active mid-term election cycle.

Leave a Reply

Your email address will not be published.