By Dr. Charles Bullock
Richard B. Russell Professor of Political Science, University of Georgia
Survey USA has released the first estimates of what may happen in the Republican Senate runoff in July. This poll, sponsored by WXIA-TV, has brought joy to the Jack Kingston camp since it shows their man leading businessman David Perdue by 11 points 52 – 41 percent.
These figures mark a dramatic improvement for the Savannah congressman who placed second in the May 20 primary, five points behind Perdue’s 30 percent. The surge tapped by Survey USA seemingly reflects the recent support coming to the legislator’s corner. Kingston has garnered endorsements from Karen Handel, who finished third in the primary, and fellow member of Congress Phil Gingrey who came in fourth. Endorsements by former candidates cannot guarantee that those who backed them in the primary will gravitate to the endorsee but Perdue’s most prominent television ad may have helped stampede Handel and Gingrey supporters to Kingston. Remember that Perdue characterized his leading opponents as cookie-cutter career politicians. Other than the usual give and take in the debates, Kingston avoided attacking Handel, Gingrey or Rep. Paul Broun who finished fifth in the primary. It would not be surprising if the supporters of the candidates eliminated in the first round find it easier to get behind Kingston than Perdue who characterized their first choice as a crying baby.
The results of the Survey USA poll may or may not prove accurate when the votes are counted on the night of July 22. As any pollster will acknowledge a survey done weeks before an election is nothing more than a snapshot of preferences at that moment and those preferences can change as voters get new information.
As with photography, the picture should be in focus – it should accurately record the targeted subjects. How good a job has this survey completed on June 3 – 5 of 419 individuals likely to vote in the runoff done at capturing reality? Obviously we do not know how GOP voters will act in July; that is why we turn to survey research. An aspect that I really like about Survey USA polls is that they report detailed marginals so that we can look at a number of breakouts for how subcomponents of the electorate are leaning. Having those data allow a comparison of the characteristics of those surveyed with what the July electorate may look like.
When registering Georgia voters indicate their age, gender, and race along with an address. Since 1996 Georgia’s Secretary of State has done audits of participation in primaries and general elections. Unfortunately with one exception post-election auditsof runoff participation have not been reported. Nonetheless, based on past primary turnout, we get an idea of what the electorate in the July runoff may look like.
In 2010, the most recent mid-term primary (it takes months to complete the audit so figures for this May’s primary are unavailable), 49.8 percent of the votes came from women. Thisfigure is in line with the past when men and women each cast about half the votes in a GOP primary. In light of that history, it is surprising, even stunning, that Survey USA assumes that men will cast 63 percent of the vote in this year’s runoff.
If the preferences of men and women for Kingston and Perdue distribute in the same fashion then the heavy weighting given to the male vote would affect the results. Since Perdue attracted the support of 41 percent of men and women, the gender weighting does not impact the estimate for him. Kingston, however, does slightly better with men (53 percent) than women (49 percent) so a weighting of the sample that better reflects the recent past would lower the estimate of his overall support slightly.
Age provides another dimension on which to compare the Survey USA sample with the 2010 GOP primary participation.
As figures in the table make clear, Survey USA expects young voters to constitute twice as large a share of the vote in the July runoff as they did in the 2010 primary and for voters aged 35 – 49 to cast about a third more of the vote this July than in 2010. Survey USA expects lower participation among voters 50 and older.
Over inclusion of voters younger than 35 boosts Kingston’s support while underweighting those aged 50-64 hurts Kingston. A similar assessment can be made for Perdue. Overweighting voters 35-49 helps his numbers while underweighting the oldest part of the sample hurts him. Although it might be more accurate, adjusting the weighting to bring it in line with the 2010 turnout would not change the bottom line; as of early June this survey shows Kingston ahead of Perdue.
For those of us who sit on the sidelines, polling data are like the results of a single game – they do not indicate who will win the pennant, make it to the tournament or play in the post-season. The polls are just markers along the way. But for candidates survey numbers take on greater significance. The results can hearten or discourage volunteers and evidence of a likely poor performance may even affect the candidate’s morale. Perhaps most important, survey data pointing to a likely defeat can cause contributions to dry up which may send a campaign into a deathspiral. This last possibility is not a problem for Perdue who can self-fund but it has proven fatal to others in the past.
