In one more week, Georgia will hold its earliest primary election ever and give a definitive answer on the race everybody is watching, the Republican primary for the U.S. Senate. Recent polls have established some clear trend lines as to which GOP candidates should advance to the runoff election. 

Businessman David Perdue has been running in first place, but his support has fluctuated below the 30 percent level.  U.S. Rep. Jack Kingston of Savannah and former secretary of state Karen Handel appear to be battling for the second spot in the runoff.

U.S. Rep. Paul Broun of Athens, who’s always had problems raising money, has settled back into fourth place, while Rep. Phil Gingrey of Cobb County has dropped to fifth place in most recent polls.

These trends suggest it will be Perdue facing either Kingston or Handel in a hellacious runoff election. But if there’s one lesson we should have learned about Georgia politics, it is that polls in a primary campaign can be very unreliable. To illustrate, we need look no farther back than the Republican primary in the governor’s race four years ago.

For months in 2010, the poll results kept telling us that John Oxendine, a highly controversial insurance commissioner for 16 years, was leading the primary field and seemed to be the most likely nominee to run against former governor Roy Barnes in the general election. I’m not talking about dubious surveys conducted by sloppy polling firms, either. These included polls commissioned by the state’s largest newspaper, the Atlanta Journal-Constitution, on behalf of a consortium of Georgia newspapers that paid big bucks for the survey numbers. Just six days prior to the primary election, the Journal-Constitution released a poll showing Oxendine in first place with 31 percent among likely Republican primary voters, with Karen Handel at 23 percent, and Nathan Deal at 18 percent.

The executives at Cox Enterprises must have had some suspicions about those numbers, because their polling firm was immediately sent back into the field to conduct a second statewide survey. Those results came out three days before the election and showed that Handel was in first place with 29 percent support, followed by Oxendine at 22 percent and Deal at 20 percent.

When the primary ballots were actually counted, Handel finished in first place with Deal making it into the runoff with her. Oxendine, the leader in so many pre-election polls, finished in fourth place and out of the running.

It’s hard to imagine how polls conducted for the state’s largest newspaper could have gotten it so wrong. These polls not only projected a candidate in the runoff who missed the runoff completely (Oxendine), they also did not include in the runoff the candidate who actually won the election for governor (Deal).

I bring this up to poke a little fun at the Journal-Constitution, which takes itself far too seriously as a purveyor of “exclusive” journalistic scoops.  But I also want to make a useful point: Primary elections are very difficult to poll accurately because voter turnout is so volatile.  

The primaries next Tuesday could attract as few as 20 percent or as many as 30 percent of the state’s registered voters. With that kind of volatility in turnout, surprises can happen—just ask John Oxendine.

It may be, as the polls are suggesting, that Perdue will actually finish first in next week’s balloting. It is also a possibility that Broun could come back from his fourth place position and make it into the runoff election.  We may even find out that there are thousands of Gingrey supporters that the polls are somehow missing. Who knows for sure? 

The actual voting on May 20 will tell us whether any of these polls were accurate. That uncertainty is what makes elections so much fun to watch.