Articles and Analysis


Morning Status Update for Tuesday (10/28)

Topics: Status Update

Today's update is posting a little earlier and is slightly abbreviated (for reasons that will be clear in the next post).

We had what I believe is a new one-day record for new statewide poll releases (37), but if you are looking for evidence of "tightening" you won't find it here. Fifteen of today's polls represented updates from the same pollster from previous tracks conducted earlier in October: Eight (8) of the 15 showed a nominal shift in Obama's direction, 7 in McCain's direction and 1 showed no change in margin.

081028 new polls

The best evidence of stability in the race comes from four new surveys in Florida, four in Missouri, two in Ohio and five in Virginia that collectively leave the margins in each state within a few decimal points of where it was yesterday. Needless to say, the McCain campaign has little hope without a significant narrowing this week in states like Virginia, Colorado, New Mexico and Ohio.

Once again, the pattern of change is mostly random: Of 13 competitive states showing any change in margin today, seven showed slight shits to McCain and six showed slight shifts to Obama.

081028 trends

The eight daily trackers again showed mostly random movement. Of those that reported yesterday (which excluded the Battleground survey which does not interview on Friday or Saturday), three polls showed nominal shifts to McCain, two to Obama and three showed unchanged margins.

081028 trackers



"seven showed slight ****s to McCain."

An unfortunate, if funny, typo.

Good analysis, though.




I'm sure many would echo the sentiment, even if it was a typo.


Gary Kilbride:

I like to ignore state polls and look at the shift in net points in the national polls. It has been somewhat for McCain in the past few days. This time +7 for McCain compared to +2 for Obama for a net of 5.

On that National Tracking Update, at far right the table should say Change Since "Sun" instead of "Sat"



Another FL. poll up this morning. O+5.



I wonder if consistency in support for one candidate against variation in the other candidate's levels should be considered "softness" in the second candidate's levels.
For instance, all VA polls have Obama at 51-52, while McCain varies from 40-47. In AZ, there are just two polls, but McCain's at 49-51, while Obama's 41-46.

Folks can focus on the national polls all they want, but it's the states, period (see Gore, Al). I seriously doubt either candidate will win by 5% - Not.Gonna.Happen.

Finally, it's just fun to see North Carolina in the same "true toss-up" state as Missouri. Perhaps even truer (!) - in MO, Obama's a constant 48, while McCain's 46-48. But in NC, Obama's 48-50, and McCain's 46-49. McCain/GOP-boosters must feel like throwing up...

Aside - I just read that the RNC is starting to advertise in WV and MT! Not looking good... But we gotta GOTV. Polls are no substitute for what happens Nov 4.



This may not be the correct place to raise this, but two requests to Pollster:

1. Could more of an effort be made to post poll results as they are actually released.

2. The "typos", particularly in the top-line numbers, are now legend on this site. This must surely be embarrassing.

In other words, this is a plea for a professionalism in performance and presentation to match the intellectual acuity and uselfulness of this site.

Thanks, in advance.



Yes, I too make a plea for 'professionalism' and 'lack of typos', because I too am an ungrateful SOB who wants to nitpick about this wonderful thing you bring us.

(Or in other words, the complaining about typos says more about the people complaining than the people making the typos.)



It really pays to read the "about" sections of your websites when assessing them. This one appears to consist of two men with five interested friends and, like most internet groups, has nothing concrete to "sell" (and therefor i imagine a low ad-based budget). To get not only a page a day of concise analysis in addition to the constant data management and other logistical notes is surprisingly effective even with the occasional typo.

On an added note, the contributors to this site have expressed ties to the Democratic Party, New York, Massachusetts, and Washington DC, which does not mean this is a partisan site nor that the data is manipulated, but it lets you know which salt to take things with.



There's a legitimate time and place for nit-picking. In public, on the day a family member has passed away, are not legitimate places or times.



Labeling people based on where they come from is a little disingenuous.
If you look a little more closely at the Pollster.com contributors, you'll see some with GOP ties (Lombardo, Soltis) and some with Democratic ties (Omero, Blumenthal). And Mark B. has always been a champion of greater transparency from pollsters - so the reading public gets the complete picture, not just what the pollster wants us to see.



I know this sounds distasteful, but I like the idea for it "cynicism value" -

Intrade ( http://www.intrade.com/ ) should have two new markets: one to see if Obama will be assassinated before his first term, and second to predict the number of assassination attempts during his first term.

PS I am a big Obama supporter - so I am not suggesting anything - this is nothing but pure, healthy cynicism.



When you live in Mass or New York or even DC there's a culture that you're surrounded in even if you call yourself unbiased and strive for the greatest level of neutrality. What someone from this culture calls indisputable from the side of a fence another will call subjective, and vice versa.

I didn't mean it as an accusation, everyone here works very hard I'm sure to keep things fair. But even neutrality is subjective. Find me a reputed website which presents "data" (and lets remember that there is no such thing as presenting data; the moment it is read it becomes objective information) that is MORE in Obama's favor and you'll deserve a medal.



Who shat on McCain? America?



I love this site, it's the first one I read every day. And the comments are always thought-provoking.

Can anyone recommend a good analysis of why the New Hampshire Democratic primary results were not predicted in the polls? I'm an Obama supporter that would like to be talked down from my paranoia (but only if warranted).




"What someone from this culture calls indisputable from the side of a fence another will call subjective, and vice versa."

And sometimes the person on the other side of the fence is simply demonstrably wrong. The argument for complete relativism is the only thing holding up most current anti-science arguments. Yes, it's only a theory. Theory encompasses some pretty sure things, including the force of gravity. Data does not become subjective just because you print it.

We are talking about sociology here, which is very murky water at the best of times, and is by its nature subjective. But the act of applying statistical trending to the existing data points, with full transparency of methods, is about as good as it's going to get.

If I sound a little edgy about this, it's because there I have seen (subjectively) a strong push right now to convince conservatives that the polls are all wrong and made up by liberals, culminating in the accusation that liberals are somehow stealing the election by dirty polling. This kind of relativism fuels that.




Sorry, forgot the most important bit: "Find me a reputed website which presents 'data' ... that is MORE in Obama's favor and you'll deserve a medal."

Actually, I go to this site before 538, BBC, CNN, or any of the other kajillion poll maps *because* it makes the most carefully neutral estimates. (I'd usually say conservative estimates, but that word's a little fraught in this context.)

It had my state, Washington, in light blue for months, and there's never been any doubt which way Washington's going to fall. It has states in yellow that are polling over 50% for one of the candidates. It is bending over backwards to show any wiggle room in the numbers.

I'm not sure why you see it as swinging so much farther in Obama's favor than the other aggregators. Take it up with the collectors of the original data points.



@Gement - excellent riposte.

Pollster.com just aggregates data - and the map upfront ensures folks don't have to read the analysis. Even Mark B.'s morning updates are just a read of the numbers from a statistical perspective. It just so happens that the numbers continue to favor Obama - too bad if you don't like that, but it doesn't make it partisan. [FWIW, Professor Franklin's own Big Ten poll at least in the first edition showed things way tighter for Obama than other polls.]

By the way, a new NV poll is out, Obama +10:
http://www.suffolk.edu/32717.html (via dkos: www.dailykos.com/story/2008/10/28/134349/60)


There are a few very large, unacknowledged elephants that are sitting in on the polling discussions.

1) The 'RCP average' of all polls is flawed. Obama's lead nationally is between 4 and 5 points, not 7-8.

2) All polls publicly listed now expect a higher Democratic turnout than has been since the 30's. This is a guess and, with party affiliated polled people voting 8-1 for their party's candidate, it's a guess that significantly impacts the poll result. And its not a scientific guess. Turnout projections are famously wrong. Just check the democratic primary this year. If voting this year is 2% higher Democrat this year than 2004, this is essentially an even race. Current polling projections give Dems a 6-7% advantage.

3) Obama needs to be ahead by 5-6 points in the national polls to be ahead in the battlegrounds. This is because of his intense support in the coastal blue states. By this I mean the battlegrounds as a whole. And Obama does not need to win all the battlegrounds to win the Elec college. But make no mistake, these states are very tight.

4) Obama is African American and no African American in a significant race in the past several decades has out-performed his or her final pre-election poll numbers. Republican or Democrat. Win or lose. However, this year may be different.

This race is very, very tight with Virgina being the key state unless McCain picks of Pa.



@ NDChristie

I commend to your attention the site ElectionProjection.com, which is the work of a Republican. I am no scholar of polling, but evidently it has a good track record. Today it shows Obama with 375 EVs and McCain with 163. A quote from today's "dispatch" on the site's home page:

"Eight days from Election Day, here is this blogger's conclusion: Barack Obama will win this election in a landslide. He will capture at least 350 electoral votes and win the popular vote by 7% or more. McCain may benefit from some measure of the Bradley Effect, but that advantage will be overwhelmed by African-American turnout, Palin-induced defections by moderates, and under-funded, unenthusiastic GOP get-out-the-vote efforts."

He goes on to state that "The blue wave of 2006 may be matched by the blue wave of 2008. (Note to my fellow conservatives out there: I know, I know - I hate reporting all this bad news as much as you hate reading it!)"

Also, @dave101: Please confine these musings to a non-public forum. Forgive me if I'm skeptical of your support for Obama.


Peter Løhmann:

First of all I would like to say a big thank you to everyone behind Pollster.com. I spend quite a bit of time reading this page, which I guess is a bit odd considering I'm a Danish national.

Secondly: I've have just been looking at various state regressions, and with the aid of the wonderful flash tools looked at the number of undecideds in various states. It varies quite a lot, being low in NC and high in for instance MT and ND. Is there any possibility that an analysis could be done before the election that look at the number of undecides in various states still up for grabs and conclude on the basis of those? The fewer undecides there are, the less likely a drastic change in candidate support should theoretically be.


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.