We seem to have gotten past some point of no return where we have a half-dozen polls or so to look at almost every day: 
A lot of these polls are good news/bad news for both candidates. In Rasmussen's poll of Mississippi, for instance, Barack Obama hasn't improved his numbers from the May edition of the poll, which had also showed him trailing by a 50-44 margin. On the other hand, that result had looked to me like an outlier before, and now can probably be taken more credibly. While Mississippi remains a longshot for Obama, keep an eye on Louisiana, which has similar demographics but much bluer party identification figures.
The two polls of Texas show the race tightening, but probably not enough to make the race interesting. There's a good rundown here of the pros and cons of Obama investing resources in Texas. The only thing I'd add is that there is far more room at the margins for the Democrats to make up ground with registration among Latino voters than among African-Americans. If Tejanos vote in anything resembling the same proportion that they constitute of Texas's citizenry, the state could be quite competitive.
But it's the SurveyUSA result in Ohio that I want to focus on. Obama leads by 2 here, but had been ahead by 9 in SurveyUSA's may poll of the state. That previous poll had shown a heavily Democratic sample -- 52 percent Democrat, 28 percent Republican, 18 percent independent -- and had triggered a lot of discussion about whether pollsters should be weighting their results by party ID. SurveyUSA does not do so -- although if it had applied the May distribution of party IDs to this poll, it would have shown Obama ahead by 10-11 points rather than by 2. Conversely, if SurveyUSA had applied the June party ID distribution to its May poll, that poll would have shown a dead heat rather than Obama ahead by 9.
I do not mean to be a fair-weather fan on the idea of weighting by party ID. As I implied the other day, I suspect that pollsters are facing something of a trade-off between volatility and potentially introducing bias. Weighting by party ID will almost certainly reduce noise, and perhaps make it easier to perceive trendlines -- but if the pollster's guesses about party ID are wrong, they may be reducing the turbulence but landing at the wrong airport. I'll say this: if a pollster doesn't know what it's doing, I think it should be letting the numbers speak for themselves. On the other hand, if the pollster has a robust and thoughtful method for weighting by party ID, it might be worth the trade-off. It is interesting that, taking two of our three highest-rated pollsters, Scott Rasmussen is a firm believer that you ought to weight by party ID, and Jay Leve at SurveyUSA is a firm believer that you ought not.
Finally, there have been some further refinements to the simulation model based on everyone's feedback, which I'll get around to explaining in a bit.
Friday, June 27, 2008
Today's Polls, 6/27
-- Nate Silver at 4:33 PM
Labels: kentucky, mississippi, new jersey, ohio, party identification, rasmussen, survey usa, texas, today's polls
60 comments
Just reposting from the last thread-
Look at the huge spike on the distribution chart at about 295. I know I've seen them before, and sometimes at curious places, but I don't think I've seen one this big. Just another request to have some sample maps for those highly probable (if even 1 or 2% likely) outcomes.
Also, with the time dependence now kicking in, will we start seeing more spikes? You would think that with the passage of time, the swing states would be whittled down to two or three, causing the bell curve to look more like a castle roof. The flip side of course is that if Obama actually makes good on his effort to flip the "big sky" states and the number of swing states increases.
The big spike is at 293. That's Obama winning
Kerry's 252
+ Iowa (7) = 259
+ Colorado (9) = 268
+ New Mexico (5) = 273
+ Ohio (20) = 293
Right now, there's kind of a big gap between Colorado, which is the worst of the states listed above and where we have Obama at 62%, and then a whole big chunk of states in the 40s; Nevada at 44%, Virginia at 44%, Indiana at 43%, Missouri and Florida each at 36%. So there's a pretty big area where Obama wins exactly those 293 EV's and those only -- then he sort of gets over the hump and a whole bunch more states become available to him.
Finally, there have been some further refinements to the simulation model based on everyone's feedback, which I'll get around to explaining in a bit.
I'm hoping part of the refinement is to return another decimal of precision to the state-by-state Win % data. It was great when you had it and I miss it.
McCain is definitely going to draw less of the white vote in Texas than Bush did, and Obama is definitely going to draw more of the minority vote in Texas than Kerry did.
If the Obama people or his grass-roots activists run a breakneck party registration campaign in the South (including TX) this summer and fall, there could be a serious electoral upset. I'm just not sure the Obama team is ambitious enough to go for that just yet...although the fact that they've indicated they want to fight in Florida, North Carolina, and especially Georgia(!?) might suggest otherwise.
With so many unregistered members of target DEMographics in magenta states, I'm not sure late June polling is going to provide much of an insight into what's going to happen in the Sun Belt on Election Day. TX, LA, MS, FL, GA, and NC are big grey areas; I'd say conventional wisdom dictates that those states are more than likely to go to McCain, but it really depends on what kind of campaign the Obama people - either officially, with paid campaign staff micromanaging everything, or unofficially, with particularly enthusiastic Obama supporters working on their own initiative - decide to run.
The Texas Lyceum poll was of adults. Do you count them the same as polls of registered voters?
On the Win Percentage meter, are you tracking the percentage of electoral votes you project each candidate to win, or are you tracking each candidate's percentage chance to win?
Meaning, are you projecting that Obama will win 65.5% of the electoral votes, or are you projecting a 65.5% chance that Obama will be elected?
Hey, can we get a link or something to the current "snapshot," the results if the election were held today?
Projecting to November is admirable and all, but I always liked to look at the current electoral polling map to consider strategic possibilities. They may have just been polling averages, but they were the best damn polling averages on the internet.
Why is Virginia suddenly less than 50% for Obama? The raw data does not show any recent polls where he is behind. This one difference seems to have shifted the entire distribution significantly.
Nate, what do you think of SUSA polling running mate combinations?
Their latest Ohio poll shows 14possible combinations, and the commond thread in all of them is that Obama's percentage of Democrats goes down from 76% to
as low as 59%. Perhaps a Case for Clinton or Edwards as Obama's VP choice?
Hi Nate,
I noticed that states that haven't been polled recently aren't trend adjusted, why not?
Nate,
It would be really useful to have graphs of Quinniapac, SUSA, Rasmussen, Mason-Dixon and Research 2000 and how each performed in each of the 13-15 battleground states.
That would show us where each pollster got it wrong in these critical states and how we should interpert each results in each state.
I.e. You should have a "POLLSTER PERFORMANCE COEFFICIENT" that would measure Obama bias (for or against)
It seemed to me that SUSA underestimated Obama's performance in close or lost primaries. Quinniapac OTOH slightly favored him IIRC.
I also think that these COEFFICIENTS should factor in whether states are west or east of the Mississippi.
Now Iowa is on the Mississippi quite literally so I'll leave the difficult math to you.
I don't like SUSA's polling method because it is so different than everyone else's and I think it muddles the sample. Honestly, sitting through a poll, as I did once, for SUSA, is a taxing thing, often taking nearly 15 minutes or more. It usually doesn't even take that long to vote, at least where I live.
Many people are busy, especially younger people with shorter attention spans, and if a poll is hung up on, it's incomplete and thrown away.
Having a poll that takes that long to run with that many choices has got to weed the sample of some number of voters who may be lower information younger voters, also likely potential Obama supporters.
I wish SUSA would drop the Veep bullshit because it really tells us nothing at all at this point. Only the campaigns need to be polling for Veep picks as of now, SUSA's insistance on continuing this, as well as their inexplicable reliance on Likely voters that seems to screen out Obama support, is making for very bad sets of data. The MN poll was total junk compared to others and now this Ohio poll is the same.
You'll recall the May poll of Ohio didn't have that Veep business. It's possible then that the difference in Party ID was the loss of a certain type of Obama voter who simply didn't have the attention span or patience to sit through that entire poll. I've yet to see one of these polls conducted in this method mesh with what Q and Rass are finding.
I'd like to add, besides just MN, the recent MO poll by SUSA also was questionably whacky.
I'd like to see what the Models look like if all these Veep Selection Polls are simply tossed out of the data sets. Most state polls start with Voter Reg then go right to candidates, and at this point, with Obama's massive new GOTV operation and the GOP's proven GOTV operation, that's the better number than Likely voters.
The Obama effect is highly unpredictable and I just saw Chuck Todd yesterday on Hardball suggest that the Likely Voter polls are really a bad idea because with such vigerous GOTV operations on both sides, you can't figure it out at this point.
Could you run a side analysis, just for your purposes maybe Nate, to see if my complaint is possibly on to something? I think the SUSA polls are just whacky once too often for my taste.
On TX: These numbers "feel" about right. I don't think it's worth sinking a bunch of money into TV advertising. But I do think it's worth considerable party-building: voter registration, list building, some joint appearances with down-ticket candidates, etc. I'd probably also throw some low-dollar media into it: radio, print, direct mail.
On MS: This is one of the states where the 538 regression differs rather substantially from the trend-adjusted polling. Any thoughts which demographic factors are over-favoring McCain? (The counterpart is NV, where the demographics favor Obama but the polls haven't backed that up. A neighboring state effect?).
And, the Super Tracker turns down. Let's hope it straightens out soon....
BTW, regarding some off the new details (looks)
The super tracker projection line: thumbs up!
showing the third party share in the pie chart (+)
I am not completely clear on the snapshot value that has been there since yesterday, is this the trend-adjusted value without the "correction towards the mean" (the "election-day-projection-correction"; ugly word, i admit)??