How come the poll results don’t match?

Posted May 24, 2012 by William March

Updated May 24, 2012 at 11:33 PM

Florida political junkies could be a little confused by poll results coming out yesterday and today.

The Quinnipiac University poll released yesterday shows Mitt Romney with a significant lead in Florida over President Barack Obama, 47 percent to 41 percent.

More results released today show Republican Connie Mack IV moving into a tie with Democratic incumbent Sen. Bill Nelson, with 42 percent to Nelson’s 41 percent.

But an NBC/Marist poll released today shows very different outcomes: Obama ahead 45-40 percent among those who said their choice was definite, or 48 percent to 44 percent including those who said they were undecided but leaning toward one candidate. In the Senate race, it showed Nelson with a significant lead, 43-38 percent, or 46-42 percent counting leaners.

Huh?

A possible explanation for the difference could be the partisan makeup of the respondents in the two polls.

The Marist poll sample included 43 percent who said they were registered Democrats, 35 percent Republicans and 21 percent others.

In the Quinnipiac poll, about 34 percent of respondents identified themselves as Republicans, compared to 31 percent Democrats and 29 percent minor- or no-party voters.

Democrats contended the Quinnipiac poll included too many Republicans to be representative of Florida’s voters, who are registered as 40 percent Democrats, 26 percent Republicans and 24 percent others.

“This poll is flawed” and “does not accurately reflect Florida’s electorate,” said state Democratic Party spokeswoman Brannon Jordan.

That complaint has been made in the past about Quinnipiac polls, but the pollsters respond that they don’t seek out a certain number of Republicans and Democrats in their polls.

Instead, they seek a proportionally accurate mix of respondents by race, age, gender and other demographic characteristics.

The poll then reports what those respondents say about their party identification—and that party “self-identification” can vary according to how people feel about recent political events.

“Some people register with one party and even though their political attitudes have changed, they don’t bother to change their party registration,” said Quinnipiac pollster Douglas Schwartz, when Democrats made the same complaint about an earlier poll.

FSP has contacted the Marist pollsters to ask about their sampling techniques, and will update with the responses.

Reader Comments

Post a comment

Members:

(Requires free registration.)




Auto-login on future visits

Show my name in the online users list

Forgot your password?


Commenting is not available in this weblog entry.