WHAS11/Courier-Journal Bluegrass Poll (Survey USA): 15 point lead for Paul
Rand Paul (R) 55%
Jack Conway (D) 40%
Margin of error +/- 4.2%
Rasmussen Reports (Pulse Opinion Research): 15 point lead for Paul
Rand Paul (R) 54%
Jack Conway (D) 39%
Undecided/Another candidate 6%
Margin of error +/- 4.5%
CNN/Time/Opinion Research Corporation: Conway and Paul tied
Rand Paul (R) 46%
Jack Conway (D) 46%
Neither/No opinion 9%
Margin of error +/- 3.5%
The obvious difference between CNN/Time and the other two polls is that CNN/Time is asking "registered voters" while SurveyUSA and Pulse are asking "likely voters."
The Jack Conway campaign says another difference is the proportions of Democrats vs. Republicans in each poll. Saying the WHAS11/Courier-Journal Bluegrass Poll is "dead wrong," Conway alleged that pollster Survey USA was applying a flawed methodology and an expectation of higher Republican voter turnout than had ever been seen before in Kentucky.
Yet SurveyUSA says there is no formula applied.
"It's a direct reporting," responded Survey USA's Ken Alper; "We don't touch it."
Alper explained that SurveyUSA is not asking how voters are registered to vote, but does ask if they are a Republican, Democrat or Independent. In that respect, the WHAS11/Courier-Journal Bluegrass Poll sheds even more light on how the Kentucky electorate is identifying itself:
Margin of error +/- 4.2%
These numbers are vastly different from the 20 point advantage Democrats hold in party registration:
1,625,380 Democrats = 56%
1,060,838 Republicans = 37%
191,577 Other = 6%
Total Voters: 2,877,795
"In many cases," Alper continued, "those who don't vote in partisan primaries do not actually know how they're registered, and/or may not realize they've been officially recorded as a party member for the past 20 years because they happened to vote in a single primary as a newly registered voter. Adjusting our party preference numbers to match the historical turnout based on registration data would be an apples and oranges exercise."
Yet, Conway pollster Pete Brodnitz of the Benenson Strategy Group disputes SurveyUSA's methodology and suggests the SurveyUSA numbers do not reflect political reality.
"Registered Democrats have consistently made up at least 20% more of the Kentucky general electorate than Registered Republicans," the Brodnitz memo continued, "In fact, even in two nominally Republican years (1994 and 2002), Registered
Democrats had a more than 25% advantage over Registered Republicans. Further, voters registered to neither the Republican nor Democratic parties have never been more than 5.2% of the electorate."
"SurveyUSA’s party distribution of 47% Democratic, 42% Republican, 10% Other Parties in this poll flies in the face of 20 years of Kentucky precedent," Brodnitz said, "In fact, when we adjust SurveyUSA’s party registration to the average of the last three statewide elections, the ballot changes significantly."
Alper responded to Brodnitz's critique:
The Conway email says the registered voter turnout in the 2008 election was :
Mark Blumenthal of the Huffington Post writes that the divergence in poll results is also tied to whether the poll is conducted by a live interviewer or if the poll is an automated (touch tone) call, (in WHAS11 polls, the recognizable and trusted voice of veteran news anchor Gary Roedemeier asks the recorded questions).
As the Blue in the Bluegrass blog (hat tip Page One) laments, Democrats simply do not appear anywhere near as motivated to vote for Jack Conway as Republicans are for Rand Paul.