Articles and Analysis


Daily Status Update for Friday 10/31

Topics: Status Update

Is the race for president "narrowing" enough for John McCain to catch Barack Obama by Tuesday? While our national trend has closed slightly over the last two weeks and today's new polls show a whisper of a narrowing in the battleground states where the McCain campaign has increased ad spending, time is short and Obama's lead still looks daunting.

We entered 37 new polls into our database yesterday. A little less than half (16) represent updates to previous polls conducted by the same pollster in October, 11 are updates from September or earlier and 10 are first impressions. Ten of the recent tracker updates represent slight nominal improvements for Obama, 5 represent slight upticks for McCain and one showed no change in margin.

081031 new polls

The impact on our trend estimates in battleground states is also mostly random, with a slight edge to Obama: The margin on our trend estimates shifts slightly in Obama's direction in 9 states and slightly in McCain's direction in 5.

081031 trends

The trend estimates provide a small glimmer of hope for Republicans: Even though the changes in the trend estimates are generally small and possibly random in direction, they have moved slightly in McCain's direction today in four of the six states where the McCain campaign increased its media buy this week. Nielsen reports increased ad spending by the McCain campaign this week in Colorado, Florida, Georgia, Ohio, Pennsylvania, and Virginia. Of these, all but Ohio and Georgia shifted slightly red yesterday (and we had no new polls in Georgia). Still, the shift across the five states was very small: the margins narrowed by an average of just 0.3 percentage points leaving Obama still leading in these states by an average of 5.1%. At this rate of change, McCain will need more than the four remaining days to close those gaps.

Two new polls in Virginia by AllState/National Journal and Marist College both show Obama leading by 4 points, a narrower margin than the roughly eight point lead Obama has held on our trend estimate for most of this week. The new polls narrows Obama's lead on the Virginia trend estimate to 7.2 percentage points (50.9% to 43.7%), tipping the state back to the light blue "lean" Obama category. In the rank of states, Virginia is the one that looks most likely to put Obama over the total of 270 needed to win. So if McCain can somehow catch Obama nationally, Virginia remains as critical a battleground state as any in the country.

On the national trend, there has been much continuing speculation about whether the results show a "narrowing" and if so by how much. If you look at our national trend chart, we definitely show a narrower Obama margin now (5.4% as of this writing) than about a week ago. Obama's current national trend estimate now stands at 49.5%, about a point lower than his smoothed high of 50.9% a week ago, but still higher than at any point this year before October 10. McCain's current 44.1% represents a slightly less than two percentage point increase since a smoothed low of 42.2 on October 12. Also, of 7 of the 11 new national surveys yesterday show slightly narrowed margins since the previous non-overlapping sample from each pollster; only two show slight gains for Obama.

081031 new national

The obvious question is whether McCain can continue to narrow that gap enough over the next four days to catch Obama. The rate of change (roughly two points on the margin over the last two weeks) seems inadequate. Still, Obama continues to lead on every national poll and yesterdays tracking updates show no decisive shift in either direction -- 3 polls tick slightly to Obama, 2 to McCain and two show no change in margin at all.

081030b trackers



Corrected link for trackers



Argggg put the URL on the name anyway one more try




The 'slight' narrowing makes me nervous....
I still worry about an 'external event' that might shift things...

And hooray for the return of civil comments!



New polls from PPP overnight (Obama-McCain, O+ since last PPP poll):
CO 54-44 (+0)
NM 58-41 (+6)
MI 55-42 (+3)
OR 57-42 (No prev.)
WV 42-55 (-5)

I should note that I am not sure about the CO internals. CO SoS figures say among ("active" or all) RVs, Dems and Reps are ~1/3, with unaffiliated at 30%; PPP's poll has Dems 42, Reps 39, Unaff 19. PPP's poll also has "already voted" at 65%, but MIB returned and EVs are less than 50%, so either PPP oversampled MIB/EVs or people have MIBs and say they have voted, though they haven't returned the MIB.

GOTV, folks!

Didn't bin Laden's tape come out the weekend before E-day in 2004?



Its does not look like the narrowing in the national trackers has had much of an effect on the race, The probability projections at http://election-projection.net only show a barely perceptible shift toward McCain.



After noticing that comments were turned off yesterday, that a lot of work goes into this site. I thought maybe not just civil comments, but a thank you might be in order.

I just want to say that I have enjoyed turning to Pollster throughout the primaries and the presidential election. I've gotten lots of insight and I appreciate what you guys do.




I am a novice to the deep dive into the numbers that you guys take and I have a question; I keep hearing the candidates speak about something they call their own "internal polling". What does this mean? Is there some super seceret poll that they take that is more accurate then the number crunching done here, or is it just BS spin to put a happy face on reality>



@ CAR54

Agree. Thank you to the Pollster.com crew, and particularly to Mark in this difficult time for him and his family.


Still, your post leaves me wondering: in the midst of the heat that led to the comments being turned off, CAR 54 where were you??? :)



I didn't get a chance the other day to let you know how much I appreciated your tribute to your father in law. He must have a very resiliant man.

I thought you might be intested; a colleague of mine was the head of an interesting 10 year quantitative study of resiliance in holocaust survivors and the impact of tbis resiliance on their children.

It is called "the transcending trauma project" done by council for relationships in philadelphia.



Question: Even if McCain were to "catch up" to Obama in seven or eigth battleground states and the national trackers by Tuesday morning, wouldn't he still be behind because of the millions of ballots already cast (and millions more that will be cast today)?



Question: Even if McCain were to "catch up" to Obama in seven or eigth battleground states and the national trackers by Tuesday morning, wouldn't he still be behind because of the millions of ballots already cast (and millions more that will be cast today)?



I'd like to second CAR54'ssentiments. I'm confident that, no matter what the outcome is on Tuesday, the vast majority will accept that life goes on, and only an unfortunate few will choose to continue to wallow in hate and fear.



Not only are there still the "undecideds", but the fact that Obama was overpolled in the primaries, which peple seem to be ignoring. States with large black populations are the only exception and they are all Red states with serious Republican machines. There are enough states within the "overpolling" margin seen in the primaries for McCain to take. I would also say that the social pressure to be seen as supporting obama or remain quiet ie not even tell pollsters how you are voting has increased significantly since the primaries. California polls were wrong by what 10%, New Hampshire 8%, how many states are within that margin or much less??



It is significant that the latest poll actually in the field in Virginia -- the CNN/Time poll, 10/23-28 -- shows a 9-point margin: the Fox poll showing a 4-point race was taken on the 26th and the Marist poll with the same margin on the 26th and 27th. Also, Obama supporters can take heart from the fact that he's over 50% in all three. It will probably get tight by Tuesday, but I live in Virginia and it very much feels like Obama is strong here.


Vicente Duque:

Mark :

Very Sorry for the bad comments in your site. I have been insulted and harassed by scoundrels everywhere in my life, also in the internet in English and Spanish, etc ... and in several countries. I hope I am not a bad weed.

Today's bets at INTRADE :

NewMexico (5EVs) ...Obama 87.5, McCain 12.5
Colorado (9EVs) ..... Obama 83.5, McCain 16.5
Nevada (5EVs) ....... Obama 81.0, McCain 21.0
Florida (27EVs) ...... Obama 72.5 McCain 31.9
Ohio (20EVs) ........ Obama 83.3, McCain 25.0

Iowa (7EVs) .............. Obama 93.3, McCain 7.0
New Hampshire (4EVs) . Obama 87.0, McCain 15.0
Pennsylvania (21EVs) ... Obama 85.5, McCain 17.5
Virginia (13EVs) ......... Obama 84.8, McCain 19.0

How can a man lose with such wonderful bettors and casinos behind him ???



Vicente Duque



Looks to me that PA is tightening but not at a rate that could actually put it in play. If Obama holds on to PA and takes VA, it is all over. He could lose every remaining battleground state, including NV, CO, and NM where he is leading outside of the margin of error, and still get to 272. He must feel comfortable enough about PA not to go back there in the final three days and to concentrate on red states that he thinks he can turn blue. Watch out, though, if he adds a stop in PA at the last minute. Based on the polls, where the candidates and their surrogates are going to be appearing, it seems to me that Obama has settled on a multi-pronged approach to getting to 270: (1) hold on to PA, take VA; (2) hold PA, take FL; (3)hold PA, take OH; (4) hold PA, take NM, CO, and NV; (3) lose PA, but win VA, NM, NV, and CO (gets him to exactly 270). A lose PA, win VA scenario, however, seems a bit unlikely, although given the amount of time Sarah Palin is spending in PA, who knows.



For us poll newbies:
When the news says: This poll has a margin of error of +/-X%,
is there a standard confidence interval that is usually used (say 95% or two standard deviations)?
Or does it vary from pollster to pollster, and they really ought to tell you what confidence level is being used?
Ken Gallant



The MOE is usually calculated from the sample size, as 2 SD:
MOE = 0.98/sqrt(N) IIRC
Look up Wiki.
Of course, this tells you nothing about the demographic make-up uncertainties, turnout model used etc... So Gallup's LV-I and LV-II models are a step in the right direction.



@s.b. re: overpolling

As has been well reported on this site, the polls for the Democratic primary were lousy this year. But they were not consistently off in one direction as your post suggests. For example, according to
, Obama outperformed the Wisconsin polls by 13%. I don't think this can be attributed to Wisconsin's African American population.

There have been some really excellent articles here on why the polls in the Democratic primary were so bad this year. The problems were not nearly so simple or one-sided as you suggest. If you are pinning all your hopes on a massive Bradley Effect, you might be sorely disappointed come Tuesday night.




I've been flogging this horse for a while now, but Gallup has a shiny new release on early voters that makes it newly relevant.

early voters are more likely to say they have voted for Barack Obama than for John McCain, by a 55% to 40% margin. Among those who plan to vote on Election Day, the spread is much closer -- only a 48% to 45% Obama advantage. Despite some perceptions that there may be disproportionate early voting among blacks, Gallup's data show that black voters and white voters are roughly proportionate in their reports of already having voted. There does, however, continue to be a significant age skew in the early voting patterns, such that older voters are significantly more likely than those who are younger to report having already voted.

Gallup is embarassing itself here. These claims are made based upon an aggregate sample that now spans fourteen days. For the first few days they asked the question, in person early voting hadn't yet begun, so all of those who responded that they'd voted were absentee voters (6-8% of LV Expanded). By aggregating the sample, they've been overweighting that initial group of voters - they comprise roughly 60% of Gallup's sample, but by Gallup's own estimate, just 40% of the voters they're talking about.

it's leading Gallup to make junk claims. We have very good state-level data on the composition of early voters. And it's not poll-generated, it's being reported by secretaries of state. So when Gallup announces to the world that black voters aren't disproportionately represented among early voters, isn't that a little strange? At some point when polls fail to accord with reality, aren't you supposed to go back and take a hard look at your data?

I suspect Gallup's underlyign data is also off, because its sampling and screening are coping poorly with this precedent-shattering cycle. (It would explain why they predicted early voters splitting just like the rest of voters, and now show a gaping 15-point Obama lead.) But the biggest problem here is the aggregation. And I just can't understand why they're doing it.



Thanks for this great site.

I'm sure everyone else knows the answer to this question, but I am curious how exit polls will reflect early voting, if at all?

Won't they be misleading if many of Obama's voters have already voted?

I guess the $64,000 question is whether or not Obama brought new voters into the election or whether the early vote totals come out of what he otherwise wouuld have gotten on election day.




At the very bottom of the Trend Estimates for Battleground States, in the National Trend, was is the number for 10/30? You list +5.5 Obama, but it says that McCain only had a 0.1 uptick from the previous day (which was +6.7 Obama). So is 10/30 6.6 Obama, or was the uptick actually 1.2 for McCain?



To Loyal

Re your comment about where I was - I was looking at the numbers and the polls that day - not the comments. It wasn't till I noticed they were turned off that I realized something was up.

Also Mark - I loved what you wrote about your father-in-law. Sorry for your family's loss.

It's going to be really interested to see how all this conversation about undecided voters plays out when we have the facts.

No matter what you think, this has been an election for the records.


green baby green:


We appreciate the use of this site to post comments. And now that that "fellow" has been booted, let's all keep things civil - there is no need to get nasty even if you disagree...it's the American way to dissent!



If you factor todays trackers the trend is reversed.

Gallup Traditional 3 point change
Gallup Expanded 2 point change
Zogby 1 point change
Kos 1 point change

All to Obama.



As for the overpolling, I think that some may be assuming that pollsters have learnt nothing from the primaries. Not to mention that, in primaries, it's much harder to predict who will vote than in a general election. We all know people who say they vote every time. However, to them, "every time" means once every two or every four years. Also, Gallup has given us traditional and expanded numbers which I think tells us where things will go in the events of lower than expected or higher than expected turn-outs. When Pew reports that their survey of those who've already voted shows Obama ahead by close to 20 points, when Republicans are, historically, more likely to vote early, I'm not sure that many of the doubts expressed here about polling data exaggerating Obama's support can be substantiated.


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.