Articles and Analysis


Morning Status Update for 11/04

This update is going to be a little bit abbreviated, as I'm off to C-SPAN for an interview that will air between 8:45 and 9:30 eastern time. While a few more final polls will probably straggle in later this morning, we have seen what should be 99% of the final round of polls and our statewide estimates are not likely to change appreciably.

In the last 24 hours 23 logged a record 90 new statewide surveys, including the 50 statewide internet panel surveys released last night from (interests disclosed) our sponsor and parent company, YouGovPolimetrix. For the sake of a reasonably sized table, if nothing else, I have omitted the YouGov/Polimetrix state surveys from the table below, but they are now included in all of our charts and statewide estimates.

Thirty-one (31) of the new statewide surveys were tracking polls that updated results released previously by the same pollster since October 15. These recent trackers demonstrated no consistent trend: 11 showed nominal improvement for Obama, 14 showed nominal improvement for McCain and 6 showed no change in the margin separating the candidates.

081004 new.png

Not suprisingly, the last round of new surveys were heavily concentrated in battleground states, including six new polls each in Florida and Ohio, four each in Georgia, Pennsylvania, Missouri and North Carolina and three new surveys in Virginia.

As is usually the case, the new polls had relatively little impact on individual trend estimates, although the nominal changes favored McCain in 15 states compared to Obama in 7.

081004 trends

The new surveys helped change our map classifications in two states. The six new surveys in Ohio had quite a spread -- from 7 point Obama lead on the Quinnipiac survey to 2 point McCain lead by Republican affiliated Strategic Vision -- but the four of the six showed margins of less than 2 points. Thus, the new surveys narrowed Obama's lead on our Ohio trend estimate to just over three points (49.4% to 46.3%), enough to shift the state from "lean" Obama to our toss-up classification.

The "narrowing" in Ohio looks a lot like the pattern discussed yesterday in Pennsylvania. Virtually all of the change on our trend estimate has been an increase in McCain's percentage, but Obama's numbers have remained flat and just shy of 50%.

Regular readers know that dividing line between "lean" and "toss-up" on are map is relatively arbitrary and based on a statistical formula that takes into account both the the margin and the average sample size in each state. It is worth considering that of the 13 final Ohio polls by each organization released in the last week, 10 showed Obama leading (by 2 to 9 points), 2 showed McCain leading (by 2 points on each) and 1 had a tied margin. So "toss-up" is probably too weak a characterization. Were I making a subjective assessment, I would consider Ohio leaning Obama.

Colorado also changed colors on our map this morning, from light blue "lean" to dark blue "strong" Obama. Even though the new Rasmussen poll shows Obama leading by just four points (51% to 47%), the new poll extends a trend that had been favoring Obama, and nudges Obama's Colorado margin up to to 7.6 points (51.9% to 44.3%) -- enough to shift to the strong category.

As for the national polls, most showed no evidence of a "narrowing" in McCain's favor. As compared to their most recent sample with non-overlapping field dates, 7 showed slight, nominal movement to Obama, three to McCain and two had unchanged margins. Our trend estimate ticks up to a 7.7 point advantage for Obama (51.9% to 44.2% as of this writing).

081104 national
That's all for now. I'm hoping to have a more in-depth look at the state trend estimates later this morning. For now I'm off to C-SPAN -- look for me between 8:45 and 9:30. And please stay tuned to Pollster all day. We will be live blogging the results and have prepared a special election night map to display network projections and results.



PREDICTION: the polls and actual votes will vary more in this election than any in recent memory. Why? Because at some level pollsters factor in expected turnout and the turnout this year will be far from historic patterns. In some cases this will benefit Obama and some cases it will benefit McCain. But the old adage that its all about who shows up is particularly true this year.


Voltron Defender of the Universe:

Mark, thank you for this site and the updates much more reliable then the RCP average. Along with 538.com you guys are my go to for polls as a self professed poll junkie.



First, let me take this opportunity to thank everyone at Pollster for putting together such a valuable resource. So, thanks!

Second, I'd love to see Pollster--and the industry in general--adopt a new classification scheme down the road. Basically, as I see it we really need to find a way to incorporate a clear signal of not just the estimated margin but also the amount of data supporting the estimate. That would help distinguish cases like Ohio (where the estimated margin is small, but there is a lot of data behind that conclusion) from Montana (where the estimated margin is small, but there is very little data behind that conclusion).

To put the same point another way, I am reasonably confident that Ohio is close enough that it is within what I think of as "GOTV range", meaning the effectiveness of the campaigns in getting their more marginal supporters to the polls could make the difference. With Montana, I am less confident that is true (I'm hopeful, but less certain).

Anyway, just a suggestion.


David in NM:

Are the numbers on the Mitchell poll in Michigan inverted? I'm finding a reference to an Obama plus 14 for Mitchell on various sites including RCP, but you're showing McCain plus 14 on this posting?


Mark, I'm watching you on CSPAN now. Thank you for your work. Sites like yours and electoral vote dot com and others make the election interesting. I voted for Obama at 6:30am in NY.

Finally, after 2 stolen elections by the Republicans, we will get a legitimate President.

Please answer this opinion poll about single payer universal health care at




Congratulations to the whole pollster.com team for the site, the analysis and the news.

Yours look a little bit conservative in your State trend estimates but I think that is a good thing.

I would like to see all 50 on the board ie more strong Repub states like Texas in case there are any huge surprises from States not polled so much where heavy turnout on enthusiasm and hopes trumps disaffection and lack of. Also at a glance if there is a reverse Bradley affect or large Ageist culture.


Mark Grebner:

The Mitchell poll in Michigan is reversed - he actually shows Obama ahead by 14 points, not McCain.


To all on Pollster.com, thanks! This site has become a daily visit for me. I've interviewed Charles Franklin for articles on BlogCritics in the past & I'm thrilled to see him a part of this site since I think he's one of the best in the business.

One question I plan to raise when the dust has settled involves my concern about traditional polling that assumes that people know what motivates their behavior. There's certainly enough new scientific evidence to destroy that hypothesis. While traditional polling may work at simple binary questions, it fails miserably at understanding complex mental constructions.

I did note that one poll used regressions to understand voter motivation, but even that has it's weaknesses.

Oops. I'm going into more detail than I intended.

So...thanks again for the best political team in...well, you get the idea.



surveyUSA polled early voters showing obama ahead by 15% in north carolina and 18% in florida. i believe half the votes in these two states have already been cast which means that mccain would have to win 60% of today's votes to carry those states, neither of which he can afford to lose.

sorry, i lost the link.



Thanks to Pollster! I'm crazy nervous/excited, can't wait till it's about 6:30 CST so I can start seeing how it all really turns out.

Glad to see the MI data was an entry error--it seemed likely!



I want to thank Mark Blumenthal and everyone connected with this site for your daily hard work, analysis, and insight. I've visited here every day since late August, but this is my first comment. Happy Election Day, everyone! God bless us all as our country steps forward into a brave, bright, and blue future!


Post a comment

Please be patient while your comment posts - sometimes it takes a minute or two. To check your comment, please wait 60 seconds and click your browser's refresh button. Note that comments with three or more hyperlinks will be held for approval.