This blog has moved.

Please visit our new page here

« Q Poll: Sen. Bill Nelson tied with Connie Mack, who's killing GOP rivals. Voters back Stand Your Ground | Main | Rick Scott's 'I've-never-tried-to-shoot'-an-elephant gaffe with King of Spain »

Not so fast, GOP. Unlike Q Poll, Marist FL poll says Obama, Nelson winning FL, Rubio makes 0 difference

Well, now we have a real political tussle -- not so much in the presidential and Senate races in FL, but among dueling polls.

Yesterday and today, Quinnipiac University found that Republican Mitt Romney was beating President Obama and that Republican Rep. Connie Mack was essentially tied with Sen. Bill Nelson. Quinnipiac also found Sen. Marco Rubio could give a small, 2 percentage-point boost to Romney's ticket.

But this morning, an NBC-Marist poll found different results. Some excerpts:

And in Florida, 46 percent of registered voters prefer Democratic Sen. Bill Nelson to 42 percent who said they would vote for Republican Rep. Connie Mack.....

In Florida and Virginia, Obama leads Romney by an identical four-point margin, 48 percent to 44 percent, among registered voters, including those who are undecided but leaning toward a particular candidate....

The NBC-Marist polls also make this clear: Adding potential home-state politicians to the Romney ticket doesn't change the results much in these battleground states.

An Obama-Biden ticket vs. one featuring Romney and Florida Sen. Marco Rubio produces the same head-to-head result in Florida -- 48 percent to 44 percent. But Romney adding former Florida Gov. Jeb Bush narrows it to a two-point Democratic lead, 47 percent to 45 percent.

What gives?

First off, these fluctuations simply happen with surveys. That's why polls have error margins (3% for Martis and 2.4% for Quinnipiac).

Also, there's a chance that Quinnipiac and Marist asked a different question about identifying party ID and registration (lengthy background here). We know that Quinnipiac asks respondents about party ID and doesn't scale (or weight) its polls to make sure that the Republican, Democrat and independents are represented in proportion to how they appear in the voter-registration rolls. Marist, however, could have weighted its poll or specifically targeted voters to ask about their party registration. If Marist used the same method as Quinnipiac, then please refer to the first point: This stuff happens.