Not that I am excited to write this, but…
A very smart observer of Florida politics shared with me their notes explaining why Charlie Crist’s six point lead in Rasmussen’s first survey of Florida’s 2014 gubernatorial race is pretty far askew of reality once the survey’s crosstabs are considered.
On multiple levels, Rasmussen’s 750-voter sample fails to represent Florida as we know it.
First, take a look at the survey’s partisan makeup: 33 percent Republicans, 37 Democrats, and 30 percent “no major party.” Then look at voter turnout in 2010 and 2012, in which Republicans made up 44 percent and 39 percent of the voting population, respectively. In 2010, just 21 percent of voters were “no major party” voters, and this was true for just 16 percent in 2010. The oversampling of third party voters, and the underrepresentation of Republicans suggest an immediate disconnect from what we can expect in six months.
Then, consider that Rasmussen didn’t provide geography tabs. We all know that Florida is comprised of some pretty distinct regions — but this survey doesn’t let us see where these respondents are from. Getting the right regional balance is critical. Without it, predictive value drops off dramatically.
Finally, on age, Rasmussen’s sample is flat out too young. In 2010, just 15 percent of voters were between 18 and 39. But in this survey, about 24 percent were in this age bracket. Likewise, Rasmussen’s sample of voters over the age of 65 is seven percentage points lower than voter turnout in 2010. Doesn’t take an advanced degree to shake your head at that one, huh?
Crist’s six point lead is overcome easily by these sampling disparities, and Rick Scott knows that. Don’t be fooled by the survey’s top line results.