Two Florida Senate polls give mixed results

Florida Everglades Public Policy Polling (PDF). 7-26/29. Likely voters. MoE ±3.3% (4.4% for Republican primary) (5/31-6/3 in parentheses, using registered voters):

Bill Nelson (D-inc): 45 (49)
Connie Mack IV (R): 43 (36)
Undecided: 13 (15)

Bill Nelson (D-inc): 45 (47)
Mike McCalister (R): 40 (33)
Undecided: 15 (20)

Bill Nelson (D-inc): 46 (47)
Dave Weldon (R): 39 (31)
Undecided: 15 (22)

Connie Mack IV (R): 47 (34)
Dave Weldon (R): 14 (6)
Mike McCallister (R): 10 (10)
"Someone Else": 6 (9)
Undecided: 23 (28)

Public Policy Polling has been consistent in finding a solid lead for incumbent Democratic Sen. Bill Nelson, even when other pollsters have been seeing a closer race or even a small lead for Republican challenger Rep. Connie Mack IV. That ends with July's sample, though, as they find a race within the margin of error for the first time. Also, for the first time, they find Nelson performing about the same as Barack Obama (who leads by 1 in the same Florida sample).

Part of that, no doubt, has to do with PPP's shift from registered voters to likely voters, which tends to shift the terrain a few points to the right, but PPP's Tom Jensen also chalks that up to the barrage of third-party negative advertising that Nelson has undergone with, at this point, little effort to retaliate. (Nelson doesn't lack for money himself, but appears to be marshalling his resources for the closing months.) That's evident in Nelson's declining approvals, down to 35/40, and a shift from a Nelson edge among independents (48-28) in the previous June poll to a 45-38 Mack margin among indies this time.

On the other hand, nobody really likes Mack either, who has 23/35 favorables. PPP's picture of this race looks a lot like the presidential race, where there's a Democratic incumbent that many people are lukewarm about but an outright-disliked GOP challenger.

Quinnipiac for CBS/NYT. 7/24-30. Likely voters. MoE ±2.9%. (6/19-25 results in parentheses, using registered voters):

Bill Nelson (D-inc): 47 (41)
Connie Mack IV (R): 40 (40)
Undecided: 12 (17)

Quinnipiac, on the other hand, has tended to show a much closer race between Nelson and Mack, with most of the races falling inside the margin of error ... and with today's sample they've bounced in the opposite direction from PPP, finding Nelson up by 7 against Mack. (You've no doubt already seen the presidential toplines for this poll, where Barack Obama leads by a similar margin of 6.) Unlike PPP, they've stopped polling the GOP primary (coming up soon, on Aug. 14) and the non-Mack head-to-heads, but that's reasonable, considering how likely Mack is to the nominee.

What's remarkable about this Quinnipiac poll is that they, too, just made the jump from registered to likely voters, and yet the needle moved slightly in the Democrats' direction. Conservative pundits have, of course, chalked that up to changes in the Quinnipiac poll's partisan composition as compared to 2008 exit polls.

As an aside here, for those of you who are blogosphere old-timers, doesn't the poll-spinning coming out of the Republican blogosphere this year remind you greatly of the graveyard-whistling that was coming out of the left blogosphere in the closing months of 2004? Oh, the pollsters are all using old assumptions that don't take into account how much people have changed over the last 4 years because they're so pissed off. Oh, all the pollsters have agendas, except for the one true pollster that we really like (except in our case, it was Zogby!). Oh, the "incumbent rule" will save us, because the other guy is polling under 50 percent and all the undecideds will break our way. And so on ...

Quinnipiac for CBS/NYT. 7/24-30. Likely voters. MoE ±2.9%. (6/19-25 results in parentheses, using registered voters):

Sherrod Brown (D-inc): 51 (50)
Josh Mandel (R): 39 (34)
Undecided: 9 (14)

Quinnipiac for CBS/NYT. 7/24-30. Likely voters. MoE ±2.9%. (6/19-25 results in parentheses, using registered voters):

Bob Casey Jr. (D-inc): 55 (49)
Tom Smith (R): 37 (32)
Undecided: 8 (17)

Quinnipiac, as they did a month ago, also polled Ohio and Pennsylvania as part of the same wave. As you've probably already seen in the presidential toplines, Ohio seemed to get a little tighter (but still in lean Democratic territory, with a +6 Obama margin) while the spread in Pennsylvania got even better (now +11 Obama). The Senate races followed the same trajectory'Ohio got a little closer, while Pennsylvania stayed about the same'but it's really academic, since in both cases, Sherrod Brown and Bob Casey are over the 50 percent mark against underwhelming opponents.

I've just gotta say, maybe it's time for Quinnipiac to think about swapping in Virginia'one of the ten or so states they have as part of their stable of states that they poll'as the third leg in their monthly troika of polls instead of Pennsylvania. It's not 2000 any more, and Pennsylvania just isn't looking like much of a swing state these days, borne out not just by polling but also by both parties' (and outside groups) lack of spending money here either at the presidential or senatorial level. Or maybe they could think about adding one more large eastern seaboard state to their portfolio, maybe one that rhymes with Borth Barolina?

Finally, here's an intellectual exercise for you: how the heck does Mitt Romney plan to win the election without winning Pennsylvania, Ohio or Florida? After some fiddling around, I've found the best path to 270 that doesn't include any of those states would be for him to flip not just Indiana and North Carolina, but also (ahem) Virginia, New Hampshire, Iowa, Colorado, Nevada, Wisconsin and Michigan. Let us know in comments if you can think of a more plausible route.