Pollster.com

I'm starting to think that Ras and GWU aren't really polling anymore. They're just going to keep repeating the same #s over and over and say the race is "stable."

That's 5 straight days Obama's been at 49 in the rolling average. I think we can assume that the past 4 days he's been at 49% in their daily numbers.

And with McCain being 47, 45 then three straight days at 46 - I think we can assume his number has probably been 46 for each of the past 4 days.

Gives us a pretty strong basis to work off of to track the rest of the Battleground's daily numbers.

I think undecided people are mentally inept, or just DUMB. How can you not be decided? The differences are very stark. Do you want four more years of the same, or a new direction?

New PEW Research National (Likely Voters)

Obama - 53
McCain - 38

Undecided people aren't undecided. They just aren't telling pollsters who they are going to vote for.

@MichaelJason

Don't worry, the polling on undecideds is screwy. It's either the 2% of racist people in American or the 4% of Republicans who don't want to go on the record and say they are voting Democratic.

Undecideds tend to break away from the incumbent leadership of the last failed government. The inroads that Barack Obama has been making with independent voters shows that he will fair better with the undecideds.

Pew seems to be a great poller but they seem and outlier like GWU on the other end. The real telling point was that Obama had a 19% advantage in those already voting...which is to say they are LOCKED in votes. The remaining may be split of just not vote, giving an obvious advantage to Obama......where it shoudl be.

Pew shows Obama 52-36 among RV.

Obama ahead 53-34 among early voters.

That is a huge margin for McCain to overcome on election day.

I can't imagine Obama winning by 15 points, but if he does Kohut is going to look good. Certainly a contrast with the Battleground poll, which must have a completely different LV model.

@s.b.

That's not entirely true. A good 50+% of undecideds actually never end up voting ... they aren't undecided ... they just don't care about the election.

@Thatcher

It seems reasonable, but there's no mathematical reason to assume that. For instance, assuming that a pollster using a three-day rolling average gives a result 50% for Obama for 7 straight days, during which time Obama had daily polling numbers x_1,x_2,...,x_7 (so that the rolling averages are (x_1+x_2+x_3)/3, ...., (x_5+x_6+x_7)/3) gives a two parameter family of solutions x_i (5 equations, 7 unknowns).

@griffstep

I see your point. Well made. I guess it was too good to hope for.

@s.b.
Good point -- another issue is the massive amount of "refusals," respondents who don't want to be interviewed.

Although they don't like talking abiout this, some pollsters estimate that the refusal rate is approaching 90%.

@Thatcher

PS- And anyway, the point of a rolling poll is to wash out the noise in daily polls. Who cares what the daily numbers are? (I mean, other than boomshak, who loves to play the OBAMA FALLING OFF BASED ON MY BACK OF ENVELOPE CALCULATIONS OF YESTERDAYS RASMUSSEN DAILY #S!!!!!!!!!!)

They give lousy golfers a break by subtracting strokes from their score, right? So why can't we do it for politicians?

McCain and Palin are playing miniature golf, and losing. So let's give them an extra five percentage points in the polls.

Obama leads by at least eight percentage points in states with 272 electoral votes, so he wins anyway. But it might make the losers feel better, if we pretend the contest was close.

Pew is really going out on a limb this election. If the results are more where the trackers show it -- in the 3 to 8 range -- Pew looks like idiots.

On the other hand, if the win is double-digit, Pew's a genius.

Give them credit for sticking their necks out. Time will tell if it got chopped off or not.

@GriffStep -

I agree with you - Actually, that is why I was hoping we could say what Battleground's dailies were ... so that we could counter Boom's blathering.

Sigh ...

@Thatcher

The funny thing is, this sort of analysis shows that anybody can make the daily numbers come out however they like, by setting the initial conditions appropriately. That's why I have a good chuckle every time I see someone post something having to do with the daily numbers that "must have been the cause" of some shift in a tracker.

@Thatcher:

That makes a lot of sense. I don't buy this "Bradley effect" crap where people supposedly lie to the pollsters about who they're voting for so as to not appear racist. Voters have plenty of ways to rationalize a vote for John McCain besides his race (Obama's too liberal, or he's inexperienced, or you want to see a woman in the White House, or you want offshore drilling, or you think Obama's going to raise taxes, etc.)

But it makes sense that people who don't intend to vote might lie about that to a pollster. Who wants to admit they don't care enough about the country to vote?

As for voters who are genuinely undecided right now: how many of them are going to have a strong enough opinion about one of the candidates to show up next week and wait around in line for who knows how long? Even if they do make up their minds, their support is unlikely to be very strong, and an enthusiastic voter is much more reliable than a leaner.

The election and polls come down to the likely voter models.

If you utilize the traditional models, Obama has a small lead of 2-4% points. That is a great number for Obama, but it means that he wins narrowly, like the Dems "lost" the past two elections.

If you utilize an adjusted traditional model, which includes more first time voters, more ethnic voters, and more voters in the younger age block - you have a lead for Obama that is double, in the range of 8-12% points.

It will be interesting to see how the "likely voter" model adjusts after the 2008 election.

I'm looking forward to reading the post-mortems by a number of these pollsters. As a Democrat, I hope that GWU's pollsters are wrong and that they'll have to explain themselves afterwards. But if GWU is right, then others are wrong and will face the same issues.

In the meantime, it would be nice if they responded to Nate Silver's challenge about their likely voter modeling. Because GW is quite far from the consensus of where this race stands.

Virginia Voter Suppression:

http://hamptonroads.com/2008/10/phony-flier-says-virginians-vote-different-days

Phony flier says Virginians vote on different days

I've seen this type of flier so many times - 2 times in VA and once in Orlando (in Spanish).

I wonder if it ever is really effective? Maybe in the days prior to 98% of households owning a TV set ... but I doubt anytime in recent years.

Remember, undecided can also mean they're not interested in either candidate this year. Likely voter models may not filter out the "i've voted in the past but this time I'm not going to" people.

The polls are varying largely on their internal ratios. Depending on whether they use 38% D - 32% R, or 37-33, and what ratio they give to each age bracket, etc. They are all trying to "guess" which groups are going to vote in what amount. Most of the daily changes are tinkering with the percentages. Whichever pollster matches the voting populace most closely will be seen as the genius this year.

As far as that flier, I've seen that before. Kinda funny actually.

@Thatcher

that flyer was probably posted by a democrat so that they could wail about republicans. Anybody stupid enough to fall for that probably shouldn't be voting anyways.

According to today's Pew Research results, one third of Republicans believe Obama will "do too much to help Blacks." That is all one needs to know about the composition of the McCain coalition.

I have heard that this pass Sunday was the last day of poll tracking by most of the polling groups.