Romney's sagging swing states

Political Wire's last roundup:

Colo: Obama 52%, Romney 46% (PPP)
Flor: Romney 52%, Obama 47% (IA)
Flor: Obama 50%, Romney 49% (PPP)
Flor: Romney 50%, Obama 48% (Pulse)
Flor: Obama 50%, Romney 45% (Zogby)
NH: Obama 51%, Romney 48% (WMUR)
NC: Obama 49%, Romney 49% (PPP)
Ohio: Obama 48%, Romney 46% (Pulse)
Ohio: Obama 50%, Romney 44% (Zogby)
Ohio: Obama 50%, Romney 49% (UofCinc.)
Ohio: Obama 49%, Romney 48% (Rasmussen)
Penn: Obama 49%, Romney 46% (Pulse)
Virg: Obama 48%, Romney 47% (NBC/WSJ)
Virg: Obama 49%, Romney 48% (Pulse)
Virg: Romney 50%, Obama 48% (Rasmussen)
Virg: Obama 52%, Romney 44% (Zogby)
Wisc: Obama 49%, Romney 48% (Pulse)

So Florida looks at least iffy, which is actually good, subsequent to some rather forbidding averages there, although North Carolina is likely gone (PPP can muster only a tie). But by the same measure Virginia is definitely in play: subtract a couple points for Rasmussen's Republican bias and we have a tie, plus two other slight leads for Obama and a quite questionable sizable one.

Comments

[AnneJ]

Well no matter how you slice it, with Ohio in the Obama column, it still comes up victory for the president, yes?

[Robert Lipscomb]

The latest from people who put their money where their mouth is.

be sure to check the state tabs to other pages at the top of the page.

http://www.oddschecker.com/specials/politics-and-election/us-presidential-election/winner

[nepat]

Robert -

For those of us who don't gamble (never have, never will), could you please take a minute to explain how to interpret the odds.

[Robert Lipscomb]

@nepat: First of all I am not a gambler either. I just find this a streamlined means of sorting throgh the polls as rorschach tests.

A "1/5" odds as currently for Obama means they will pay you $1 for every $5 you bet. If you win, you get $1 plus the $5 you put up. If you lose, they keep the $5.

This roughly (very roughly) translates ttheir general belief that Obama has a 5 /(1 + 5) chance of winning. That translates to 5/6 or 83%.

More precisely, the betting houses think they have a 17% chance of losing that bet - more less.

So, they think Obama is very likely the winner.

I like to look at these early on, say beginning last winter, because these guys use models based on things other than pure polling data (e.g., incumbant or not, improvig economy and so forth) which are good predictors. This is predictive modeling. (Based on historical modeling, I predict next August will be hot as hell in Nashville.) Of course, these guys have metrics they will never share. As polling data becomes available, they begin incorporating that. At that point, their modeling becomes a predictive/tracking model.

Nate Silver, who I like very much, is limited to crunching polling data from other sources. So his numbers track polling - thus tyracking modeling. By limiting his modeling to measured numbers (rather than predictive metrics), he can rightfully claim complete objectivity. Until recently, this has protected him from charges of bias.

The guys in the betting houses don't give a fiddler's damn if Fox News acuses them of bias or not. If the guys at Fox think they are so off base, the betting houses are open to take their bets.

During the October debate season, the polls went wild. the gambling odds held pretty steady with a slight downward drift for a while. It helped me sleep better at night.

I hope that was not TMI.

[nepat]

Thanks, Robert. Interesting indeed. And helpful. I'm thinking about placing a bet. ;-)