Like just about everyone else, Im more than a little confused about the Louisiana Senate race.
[IMGCAP(1)]According to OnMessage Inc., which conducts polling for Republican challenger John Kennedy in the Louisiana race, the two-term Democrat-turned-Republican state treasurer trails incumbent Sen. Mary Landrieu (D) by the mid-single digits, and the race has been closing. The pollsters most recent three-day roll is just 5 points 47 percent to 42 percent.
That assessment of the race is light years from where Landrieus campaign, national Democratic insiders and even plenty of veteran Louisiana campaign watchers see the contest, and a Mellman Group poll (also based on tracking) for the Democrat shows Landrieu holding a huge advantage 52 percent to 34 percent over Kennedy.
The difference between a 5-point race and an almost 20-point contest isnt merely one of degree. While both polls show Landrieu ahead and more likely to win, the GOP survey shows a competitive contest, while the Democratic poll shows a race that already is over.
The National Republican Senatorial Committee decided to pull out of Louisiana until it saw recent OnMessage tracking and decided that Kennedy still had a chance to overtake the Democrat. The NRSC bought another week of TV in the state on the basis of the OnMessage track.
Last week, both the Democratic Senatorial Campaign Committee and the NRSC had TV buys up in the state. Apparently Democratic strategists werent entirely confident that the race was over, even with a 15- to 20-point lead.
What else do we know about the polls and the race?
While Democratic insiders argue that the Louisiana race has moved little in weeks, OnMessage tracking found Landrieu holding solid double-digit leads from Sept. 23 to Oct. 2, after which the race closed noticeably. Indeed, on Sept. 23, the GOP poll found Landrieu ahead by 20 points, 53 percent to 33 percent, over Kennedy.
On Sept. 27, Rasmussen Reports, an automated poll about which I have often expressed great reservations, showed Landrieu ahead of Kennedy by 13 points among likely voters, 54 percent to 41 percent. On Sept. 28, OnMessages three-day roll found virtually the same margin, with Landrieu holding a 14-point lead, 52 percent to 38 percent.
What else do we know?
We also know that the Democratic and Republican pollsters disagree about what percentage of the total electorate will be black. Blacks constitute a few more percentage points of the total sample in the Mellman Group poll than in the OnMessage surveys. This alone would explain some of the difference in the ballot test, since race is a strong predictor of the vote in Louisiana.
We know that the Mellman survey shows Landrieu with a much higher favorable/unfavorable net rating than the OnMessage survey, and that Kennedys favorable/unfavorable rating is good in the GOP poll and terrible in the Democratic survey.
What we have, then, is a consistent set of differences in the two surveys. Democratic tracking has McCains lead more narrow, Landrieus lead much bigger, Landrieus identification ratio much more positive and Kennedys ID ratio much more negative than does the Republican track.
If experience counts and this year, at least in the presidential campaigns, it hasnt then we have to consider the pollsters.
Landrieu is notoriously fickle when it comes to consultants, so it should come as no surprise that she has switched pollsters. After using Al Quinlan last cycle, she has switched to Mark Mellman, a highly regarded political veteran whom I met many years ago when he was still a graduate student at Yale. In 1996, Mellman worked in Louisiana for an outside group that ran an anti-David Duke campaign when Duke was one of the Republican candidates running against Landrieu in an open-seat Senate contest.
Kennedys media and polling are being done by OnMessage, a relatively new combination of Brad Todd, Curt Anderson and Wes Anderson, whom I have also known for many years and hold in high regard. Whats most notable about the team is that it handled polling for both of Louisiana Gov. Bobby Jindals (R) races, including his win last November.
The 2007 gubernatorial election was the first statewide contest in Louisiana after Hurricane Katrina, and there is no doubt that the storm caused some residents to leave the state, changing Louisianas demographics and altering its electoral arithmetic. OnMessages involvement in that race and the firms accurate polling for Jindal cannot be ignored.
Im not certain where the Senate race really is now, but Im pretty certain that Landrieu is not going to win by 20 or 18 or even 15 points. Id be surprised if she won by 10 points.
There is no doubt that she is currently ahead (both sides agree about that), and even some Democrats believe that the Senator will not get a lot of the undecided vote. My guess is that Landrieu will win with a final margin that is much closer to the margin now found in OnMessage tracking than in the Democratic polling.
As I write this, the NRSC is advertising in the state, while the DSCC has not extended its TV buy. Of course, it could decide to do so at any time.
Jindal, a wildly popular political figure in the state, so far hasnt put his full weight behind Kennedys effort, and he has been politically averse in partisan races so far. But if the governor comes out strongly for Kennedy soon, he could affect the races dynamic.
Some observers no longer think this race is even worth watching. Im not so sure. On election night, this contest may be closer than a number of races that those same people are watching breathlessly right now. And unlike many others, I have not ruled out a surprise.
Stuart Rothenberg is editor of the Rothenberg Political Report.