For the Democrats, it must feel like 2000 all over again. A day after Quinnipiac and American Research Group showed Barack Obama with his first leads in Florida, Rasmussen finds him still 8 points behind. That's a slight improvement from Rasmussen's last poll in Florida, which had shown Obama trailing by 10 points, but obviously a long way from something that would get the job done in November.
We talked a little bit yesterday about the differences in accounting for party identification between the different pollsters. Those differences seem as though they're particularly large in Florida, where Rasmussen has consistently had Obama way down, but Quinnipiac more competitive.
At the end of the day, however, what matters is not so much the Florida result in the abstract, but where it stands relative to other states. Our model calls Florida a toss-up -- but it's also giving Obama credit for about a 5 point lead nationally. If the election tightens, is Florida still a state in which he wants to invest resources? Were I running his campaign, I'd commission some internal polling of the state, as Florida is so resource-intensive that a decision may need to be made on it relatively early.
That's about it insofar as the polling goes today. Rasmuseen's national tracker still has Obama ahead by 3 points; Gallup will not be releasing numbers today. Economist/YouGov also has some national numbers out that I'd failed to be mindful of; Obama leads by 3 points in their poll conducted this week, and led by 4 last week, after having trailed in most of The Economist's polling throughout the month of May.
Thursday, June 19, 2008
Today's Polls, 6/19
-- Nate Silver at 11:00 AM
Labels: florida, today's polls
94 comments
Nate-
Is there any place where we can see the national polls that you are using? I think it might be appropriate to include a national "state" where we can see thoes polls and their weights.
Florida is an example of why you shouldn't use national polls for states where there is data regularly coming out. The demographics of Florida are not that representative of the country as a whole in some respects. And there are polls on Florida literally every week. I trust the Florida polls as predictors better than an aggregation of national polls. The Democratic party's handling of the Florida party primary hurt them there too, and that won't be captured in national polls. You have changed your measures to help Obama in Florida by adding in the national polls, though it appears 'objective' when it really isn't.
Actually, Anon @ 12:24, there were three polls of Florida in the last two days, but no other Florida polls since May 19. It looks like everyone polls Florida only once a month, and they're all at the same time. So until yesterday, we had no idea what had been going on in Florida in the past month!
That, and the regression Nate uses is all about things like the demographic differences between states.
For one thing, both the Republicans and Democrats stripped Florida of 1/2 of its primary votes. The GOP and Dems ended up treating Florida exactly the same.
Another thing, he didn't "change his measures" to use national polls in Florida. Do you have a problem following this site? He said that the model shows Obama ahead by 5 nationally but a tie in Florida. Thus, in a tie nationally, he could easily fall behind in Florida. He wasn't saying that national polls will count in the Florida model.
@11:24: If you take the average of the three new Florida polls, Obama is ahead by 0.3 points. I have him ahead by 0.9 points. So ... I'm not sure what your point is.
If Obama is within 5 points of Florida, an advertising push there is worthwhile for him. He has a lot of avenues to a win without Florida; McCain has almost none. The major media markets in Florida are quite expensive, so Obama could put a huge drain on McCain's more limited coffers by mounting a Florida attack. Even if he still loses Florida (and I want to see the next two weeks of Florida polls, if the Republicans push the oil drilling issue hard), he gains a strategic advantage in other states like Michigan and Ohio, not to mention states like North Dakota and Alaska. Pushing at Florida is a good plan for Obama.
Are there breakdowns of Party identification in the Rasmussen poll vs. the Quinnipiac and/or ARG polls? Twould, of course, be interesting to compare.
In addition to already making a few stops in FL, the Obama campaign hired Steve Schale to head state operations. They've made it abundantly clear that they are going to fight for Florida. I have a feeling they've already done extensive internal polling and picked their battlegrounds otherwise they wouldn't be contesting Georgia either (they clearly see something the rest of us don't)
Nate, what are the percentages for both candidates on "loses Florida, wins election?"
Nate, Take a Bow man!
Obama is Reading 538, or at least his team is
http://marcambinder.theatlantic.com/archives/2008/06/obama_airs_first_general_elect.php
They've DROPPED Oregon, Washington and New Jersey in their ad buy (From the original Organizing 17) and added North Dakota, Indiana, Montana and Alaska (Redefining their idea of swing states)
Sorry, link here
The issue I have with this site is you guys putting far too much weight on national polls, when you post your state data.
McCain is winning Florida right now. Hell he is leading in every poll there except two, and that Quinnipiac surely looks to poll more Democrats if you look at their polls across many states.
I like what you are doing, but clearly questioning the way you are doing it. You are thinking too objectively with these polls.
Think November, and predict it that way. WV and Indiana are locks for the GOP, but you have them as toss up's or lean Democrats. C-mon now...
North Dakota, Montana, and Alaska are all super-cheap to advertise in, so they probably figured, why not?
I'm just waiting for someone to declare that McCain has an advantage because he has better team chemistry and intangibles. He's a Proven Veteran, after all.
MVRed,
When you say he's "leading in every poll there except two," you're missing the whole point. Previous polls in Florida (beyond the two showing Obama ahead and one behind) were taken at least a month ago, at a clip of two per month. Those polls are no longer really relevant, with Obama wrapping up the nomination and pivoting to the general election. They are rightly weighted far less than polls that were completed over the last few days.
Nate is projecting towards November. We don't have new poll data for Indiana and WV, so we can't say that they are locks for anyone. What we do have is poll data from regions that are demographically similar to those states, so we can use those to predict how they would react.
How can he be "thinking too objectively" with polls? Are you saying that the conventional wisdom that WV and IN are red states is more powerful than what the people of those states (and people that are like them) are actually saying?
To Ben: "For one thing, both the Republicans and Democrats stripped Florida of 1/2 of its primary votes. The GOP and Dems ended up treating Florida exactly the same."
The "exactly the same" part isn't true, and you know this. The GOP gave 1/2 delegates before the primary, while the Democrats waited to do that only after declaring candidates shouldn't campaign in the state (and they didn't, except for fundraisers) and only post hoc alloacted 1/2 delegates once it was moot for the final delegate totals. Anyone living in Florida probably ran into Giuliani or McCain at the grocery store before the primary while Obama and Clinton were nowhere to be found. The DNC screwed up the handling of Florida, even if the final outcomes of 1/2 delegates was the same on paper.
MVred:
The issue is that most of the florida polls are old, taken during the democratic primary. Since two of those polls show a bounce and one shows no effect, and we are seeing a bounce nationally and in dozens of other states, it makes sense to adjust the OLD data in accord with the national bounce. This is in the post "adjustment II: we know more than we think." Furthermore, new polls are not adjusted.
In Florida's case, the adjustment is small +0.6. It's really not worth nitpicking.
Beginning to think there will be a serious anti-McCain Barr Effect.
I just want to point out how ridiculous it is that all the republicans on this site are moaning about what a bad idea it is to take into account time-adjustment for momentum and to include national polls into the model.
It will all help McCain just as much once he has some momentum of his own in the polls. The knee-jerk reaction "you must be wrong about every state that Bush won because there is no way they could possibly be won by a democrat" is just uninformed and annoying.
I'm anon @11.24. Nate said "@11:24: If you take the average of the three new Florida polls, Obama is ahead by 0.3 points. I have him ahead by 0.9 points. So ... I'm not sure what your point is."
You wrote this: "Our model calls Florida a toss-up -- but it's also giving Obama credit for about a 5 point lead nationally." It sounds like you are combining recent polls in Florida with national polls to predict your Florida outcome. This seems wrong, or else I misunderstand. Either way