Get me, I follow everything. Hell, again I find myself in a position where I’m paid to follow everything and write about it. It’s a curse.
But RCP has Pennsylvania and Michigan as toss-ups in the presidential race? What the hell?
Seriously, maybe I’m missing something but I don’t see their methodology anywhere. Can someone point me there?
Hell, even there own numbers for PA show Obama with a 4.7% lead in polling and the D candidate with a significant win in the last three elections. In what way does that count as ‘toss up’?
I noticed that too, when a right-winger on another message board pointed out when Romney was ‘leading’ in the EV map there (for one week, because Romney’s lead in NC was 5.1%, but Michigan, Pennsylvania and Wisconsin were all ‘ties’ at around 4.5% each).
It’s a lazy map that doesn’t make any prediction except that Romney will get at least 191 EV’s (duh), and Obama will get at least 201 (double duh).
Well sure, but it’s certainly easier to just put a number threshold on it without bringing cat-eating into it.
If they put it at 3% instead, then North Carolina is no longer a toss-up but is now Lean Romney. That’s probably more accurate (move PA and MI to Obama’s side and NC to Romney’s). But as long as we know the threshold I don’t really see a problem.
Oh, and Minnesota is getting close to RCP “toss-up” territory too. One more poll showing Romney down by less than 5 will probably get it there.
That’s now at Obama 290, Romney 248. The only change in the last week has been moving Colorado from Romney to Omama. Both Michigan and Pennsylvania are given to Obama.
I’ve been monitoring it and I believe that the no toss ups map has never had Romney ahead. The closest was two weeks ago, when it was Obama 277, Romney 261. Then they shifted New Hampshire and then Colorado.
That seems about right. Virginia is listed as a dead heat but my guess is that Obama takes it as he did in 2008. That would be 303 - 235, which is no longer in the squeaker category. Even if not, then the loss of one of Ohio or Michigan or Pennsylvania still gives Obama the win.
I agree that 5% is a bit large for inclusion in toss up, but none of those three states are at 5% right now. So I don’t understand the problem.
We’re all mad that the front page map doesn’t say what we want, and the “no tossups” map is hidden.
OK, seriously, the front-page, default map is so conservative (in the sense of ‘unwilling to take risks,’ not in the modern political sense) that it’s useless. It’s a waste of a map that implies the race is close when it’s really saying they’re not willing to say anything but that the 100% likelihood states are going the way they’re going.
That’s important because people look at pictures but don’t read words. It implies a story that’s not there.
I’d say my problem is their utter irrationality concerning their definition of toss-up.
I would, seriously, bet them any amount of money they care to lose that PA and MI will be in Obama’s column come a week from now. Anything else is just, as said above, laziness or the need to generate page views.
“Toss up” is perhaps not the optimal descriptor. It implies that both candidates have an equal chance of winning. Whereas in reality, at this late stage in the campaign, a candidate ahead by 4.9% has a much greater than 50% chance of winning.
But, there are only so many shades of blue and pink that RCP can show on their map, and they have to set the cutoff points somewhere. Maybe 5% is too high for this late in the campaign. But then, both campaigns are making ad buys in Pennsylvania, so they apparently don’t regard it as bolted shut, signed, sealed, and delivered.
At any rate, it isn’t a predictive map. There is no analysis behind it. It’s a mechanical, color coded representation of recent poll results. The data behind it are transparent. You can easily choose your own threshold and make your own map. I don’t see the problem.
It wasn’t just RCP. TalkingPointsMemo’s PollTracker had Pennsylvania as a tossup for pretty much all of the week before last. Then when it stopped being a tossup, Michigan was a tossup.
The really irritating thing about when they had Pennsylvania as a tossup was that they showed Obama favored by only 2.0%, which apparently was the tossup threshold. But with one category of exception that I’ll get to, they didn’t have any polls that showed Obama leading by less than 3.0%. And if all your polls show Obama leading by 3% or more, how the fuck does your model show him up only by 2% overall?
(The exception, fwiw, was that a couple of oldish polls that showed Obama leading by only 2% had been more recently superseded by polls by the same pollsters showing him up by 3% or more. In that situation, it’s standard to either (a) discount the earlier poll entirely, or (b) take into account which way the momentum’s going, according to that pollster.)
I asked them that and suggested they change their model. They wrote back and told me it would be too much trouble to do that at that point in the game.
I stopped paying attention to TPM’s PollTracker after that. Other than to raise my eyebrows when it showed Ohio as a tossup a day or two before the election. I hope they had better reasons than Chris Cillizza did, but I doubt it.
And I should second what Jas09 said about their ridiculously wide ‘tossup’ category. They’ve changed it, but for several weeks before the election, right up to Sunday if not Monday, they had 11 Senate races in the ‘tossup’ category.