TPM Track Composite: Obama Maintains Big Lead
Here's our daily composite of the six major national tracking polls. Barack Obama is essentially maintaining his big lead in the polls:
• Gallup: Obama 51%, McCain 44%, with a ±2% margin of error, compared to a 51%-45% Obama lead yesterday.
• Rasmussen: Obama 52%, McCain 45%, with a ±2% margin of error, unchanged from yesterday.
• ABC/Washington Post: Obama 53%, McCain 44%, with a ±2.5% margin of error, compared to a 54%-43% Obama lead from yesterday.
• Hotline/Diageo: Obama 50%, McCain 43%, with a ±3.4% margin of error, compared to a 48%-43% Obama lead from yesterday.
• Research 2000: Obama 52%, McCain 40%, with a ±3% margin of error, compared to a 51%-41% Obama lead from yesterday.
• Zogby: Obama 51%, McCain 41%, with a ±2.9% margin of error, compared to a 52%-40% Obama lead yesterday.
Adding these polls together and weighting them by the square roots of their sample sizes, Obama is ahead 51.6%-43.1%, a lead of 8.5 points, compared to the 51.5%-43.1% Obama lead from yesterday.
These continuing poll numbers that favor Obama have got to be extremely demoralizing to McCain supporters -- maybe even enough to keep them from bothering to vote.
October 24, 2008 5:10 PM | Reply | Permalink
It's fun to read the rantings about the polls on right-winger sites.
October 24, 2008 5:12 PM | Reply | Permalink
Well these polls look nice, but I will like it when the announcers on election night report that Obama has gotten elected, and the Senate Republicans number 39 and cannot block Democratic bills getting introduced and voted on in the Senate.
Please answer a poll on health care at http://poll.democratz.org
You can watch Countdown with Keith Olbermann at http://liberal.democratz.org
October 24, 2008 5:20 PM | Reply | Permalink
In every poll, Obama has at least 50% and McCain can get no higher than 45%. And it is showing a incrediable stability over time. It says that McCain not only has to reach out to undecideds, but is going to have to convince those who have gone over to Obama's side to switch sides.
October 24, 2008 5:12 PM | Reply | Permalink
The interesting thing is that, for the entire general election campaign, except for the time right after the RNC, McCain has struggled to break 45%. I've even seen some articles suggesting that when you step back and look at the overall trends of this campaign, this was always going to be an uphill battle for McCain. But of course, the media has pushed the horse-race narrative.
I earnestly believe that McCain's age has always been a drag for him, and, at least partially, kept him around that 45% ceiling. But the media hasn't given age the enormous attention that it's given (and is still giving) race. However, it seems that the electorate thinks age is much more of an issue. The Palin pick has only exacerbated the age issue, since most people see her as unprepared to step in if necessary. [McCain's temperament hasn't done him any favors either.]
October 24, 2008 5:40 PM | Reply | Permalink
I agree that McCain's age is a big negative for him. And the bad news for him is, he ain't gettin' any younger.
That said, I don't think that his advanced age in and of itself is the kicker here. It's the way he wears it - combined with his temper, erratic behavior, and his mannerisms, conveys the idea that he really could go at any time.
October 25, 2008 5:13 PM | Reply | Permalink
Still no Nickelodeon poll?
I'm disappointed.
Why does EC hate children?
October 24, 2008 5:15 PM | Reply | Permalink
Tick tock, tick tock, tick tock...
October 24, 2008 5:18 PM | Reply | Permalink
THIS
October 24, 2008 5:19 PM | Reply | Permalink
CAN'T
October 24, 2008 5:29 PM | Reply | Permalink
STOP
October 24, 2008 5:34 PM | Reply | Permalink
Just read that the St. Petersburg Times endorsed Obama.
October 24, 2008 5:19 PM | Reply | Permalink
And then we have former McCain advisor Fried endorsing Obama.
October 24, 2008 5:31 PM | Reply | Permalink
"As if you didn't know that it would sting... kissing the beehive..."
-Wolf Parade
October 24, 2008 5:20 PM | Reply | Permalink
Fat lady's on in 5.
October 24, 2008 5:20 PM | Reply | Permalink
It does not appear over until the fat lady votes.
October 24, 2008 5:23 PM | Reply | Permalink
She voted early.
October 24, 2008 6:13 PM | Reply | Permalink
Gimme a great big HoDean Yeeeehaaa! Awright, time to take the kids to the Halloween Parade. Wonder if they'll have McScary monsters?
October 24, 2008 5:23 PM | Reply | Permalink
The days just keep on ticking for Johnny Mac, and O's support only get stronger as his wave of historic momentum solidifies.
October 24, 2008 5:25 PM | Reply | Permalink
This is great news for John McCain!
Meet...
The honest republican
October 24, 2008 5:27 PM | Reply | Permalink
Everyone knows that polls, like facts, are biased towards Democrats
October 24, 2008 6:14 PM | Reply | Permalink
If I recall correctly, you used to weight by sample size and not by the square root of sample size. In any case, you should definitely weight by sample size and not by the square root of sample size.
While it is not necessarily true that polls should be weighted by sample size, it is definitely unreasonable and sub-optimal to weight by the square root of sample size.
First some intuition. The sample size weighted average of support for Obama is the fraction of people who were polled in any of the polls who say they support Obama. That clearly makes sense as an overall measure of support. In plain ascii equations it is (the sum over polls of (number polled)(number of Obama supporters)/(number people) ) divided by the sum over polls of number polled = overall number of Obama supporters divided by overall number polled.
Another intuitive argument is this. Imagine taking the sample of one poll and dividing it into two subsamples one of one third of the original sample and one of two thirds of the original sample. For each sub-sample you can calculate a fraction which supports Obama. Now if you weight these fractions by sub-sample sizes, the resulting number will be the fraction who support Obama in the original poll. Not so if you weight by the square root of subsample size. You can check this using say the research2000/Daily Kos poll which reports a 3 day average and averages for each day. Average 2 days (sample sizes about the same so weighting matters very little). Now do it.
or try redoing your average with not the 3 day average reported by Daily Kos but treating it as 3 separate polls for each day. If you calculate your tracker average with 3 little polls from DailyKos and use the square root of the sample size you will get a different result that the one you report. Why is this worse than what you did ? The problem is weighting by the square root of the sample size.
Finally, if each poll results differ from the true population averages only due to random sampling error then the sample size weighted average would be the minimum variance unbiased estimate of true support for the candidates in the population.
I have never seen any hint of a rational for weighting by the square root of sample size and I don't think there is any reasonable argument for doing so.
October 24, 2008 6:56 PM | Reply | Permalink
I'm curious about this square root issue also. I'm not opposed necessarily, but I'd like to see Josh's reasoning.
In some academic studies, they take the natural log of numbers rather than the numbers themselves.
October 25, 2008 12:43 AM | Reply | Permalink
I won't opine on the "optimum" weighting method, but using square root of sample size simply normalizes based MoE -- i.e., a 2% MoE is weighted twice as heavily as a 4% MoE.
The proof of this net.fact is left as an exercise for the student.
LK
October 25, 2008 4:02 AM | Reply | Permalink
A few more percent and the election will be outside the margin of Diebold error.
October 25, 2008 9:06 AM | Reply | Permalink
Rasmussen is 52%-44%
The Rasmussen Reports daily Presidential Tracking Poll for Saturday shows Barack Obama attracting 52% of the vote while John McCain earns 44%. This eight point advantage matches Obama’s biggest lead of the year in a race that has been remarkably stable down the stretch
October 25, 2008 1:41 PM | Reply | Permalink