Louisiana Data Conflict; Kasich Down in Ohio

Two days ago, a pair of polls were released into the public domain projecting that Rep. Bill Cassidy (R-LA-6) is faring well against Sen. Mary Landrieu (D). Yesterday, Public Policy Polling publicized a counter-study showing the senator to be in much better political shape, thus calling the Republican data into question. Why the stark difference? We’ll explain shortly.

The two Republican polls were conducted by OnMessage for the National Republican Senatorial Committee (NRSC) and Harper Polling for a conservative website. The OnMessage data (Aug. 12-15; 800 registered Louisiana voters) gave Sen. Landrieu only a 45-41 percent advantage over Rep. Cassidy. HP (Aug. 14-15; 596 registered Louisiana voters) found even better results for the Baton Rouge congressman, actually placing him ahead of the incumbent on a 47-45 percent count.

The PPP data (Aug. 16-19; 721 registered Louisiana voters) forecasts quite a different take. According to these results, Sen. Landrieu has a comfortable lead over Rep. Cassidy, 50-40 percent, when the two are paired in a hypothetical post-primary December 2014 run-off election.

Seeing Democratic and Republican pollsters surveying the same race at the same time but arriving at drastically different conclusions happened relatively frequently during the last election cycle. Particularly in the presidential campaign, we often saw the Republican data placing GOP nominee Mitt Romney in much better position against President Obama than was actually the case.

The chief reason for the past projection disparity was the turnout screening mechanism used in qualifying those who constituted the various sampling universes, and such is undoubtedly the case with these conflicting Louisiana numbers.

In the presidential year, the Democratic pollsters were much closer to accurately forecasting the participation model in what will prove to be the higher turnout year of 2012. Now heading into the mid-term cycle, where voting participation is always lower than in presidential elections, it may be the GOP numbers that yield the more accurate prediction.

As we know, who turns out always determines an election winner, and no one suggests that election year 2014 will be any exception to that rule. Defining the most accurate turnout model, however, will clearly be subject to further scrutiny and deliberation.


Public Policy Polling (Aug. 16-19; 551 registered Ohio voters) provides a new and much different perspective about the Ohio governor’s race, too. In June, Quinnipiac University released its Buckeye State data that projected Gov. John Kasich (R) to be commanding strong favorability ratings (54:32 percent positive to negative) and showed him holding a sizable 47-33 percent advantage over Cuyahoga County Executive Ed FitzGerald (D). Yesterday, PPP released job approval data that puts him in an upside down sphere (42:47 percent) and actually behind FitzGerald 35-38 percent on the ballot test.

It appears there are several reasons for the disparity. First, the two polls were not taken during the same time period, so detecting slippage for Kasich could be a legitimate finding. Second, the PPP sample of just 551 registered voters for a large state like Ohio is likely too small, meaning that the error factor is higher and reliability not as strong. Third, the polling sample appears to skew at least slightly toward the Democrats, again cutting against Gov. Kasich.

While the Q-Poll may have been an anomaly because it projected the incumbent to be in stronger position than he actually was, this latest PPP survey could also conceivably be in the same category, but this time the abnormality would favor the Democrats.

Considering the Ohio electorate’s swing nature and some of the controversial stands the governor has taken during his first term in office, a close 2014 battleground election will soon be upon us. Therefore, it isn’t particularly surprising that we are already seeing conflicting polling. Expect more of the same throughout the entire campaign.

Leave a Reply

Your email address will not be published. Required fields are marked *