Articles and Analysis


50%, 8%, and October: Some Historical Context

Saturday's Gallup tracking poll revealed two big numbers for Obama. Obama hit 50% in the tracking poll and took an 8% lead over McCain. This isn't the first time that Obama has hit the 50% mark, and it isn't the first time he has held a lead of 8%; but now we are in the last month of the campaign and numbers like these in October usually mean electoral success in November.

Using Gallup's compendium of presidential trial heat polling since 1936, I counted16 candidates who received 50% support or higher in an October Gallup poll. Hitting the 50% mark was a very good predictor of victory. Of those 16 candidates, just two failed to win the general election--Al Gore and Thomas Dewey.


An 8% lead has also been difficult for trailing candidates to overcome. Only one candidate who held a lead of at least 8% in October ended up losing the election--once again, that was Al Gore in 2000. There were thirteen other occasions since 1936 where a candidate had an 8% lead or greater in at least one October poll, and in each case that candidate won.


While the history doesn't look good for Republicans, the McCain campaign can take some solace from 2000. In that campaign, Gore's support reached as high as 51% (and he had a 10% lead over Bush) early in October. However, he quickly lost that lead after the first debate and Bush actually built his own 13% lead in a late October Gallup poll. This was the only time that both candidates reached 50% and held a lead of at least 8% in October Gallup polls--yet another way in which the 2000 election was truly unique. The McCain campaign will have to hope they can duplicate Bush's 2000 comeback rather than go on to defeat like the other 13 candidates who found themselves behind by 8% in October.

NOTE: Mark Blumenthal emailed to point out that the 2000 Gallup tracking poll was highly volatile in 2000 and it may be the case that this volatility was at least partly responsible for both Gore and Bush hitting 50% (and for both taking 8% leads) during October that year.



Now we should factor in 50% +8 a week *after* the first debate...

Bush had the first debate, all the firsts are gone in this race.



It's looking very good for Obama, McCain is throwing his final desperation tactics at him now.



Of course, there are quite a number of people who believe that Gore did indeed win in 2000 - and that if Jeb and Katherine hadn't been in charge of the count in Florida, we'd probably be ending President Gore's second term about now.

But that aside, of course Gore certainly won the *popular vote* in 2000, which is after all what the national polls were measuring. So in that sense only Dewey is an outlier - and given the primitive state of polling in that time, perhaps that's going back too far for any congruent circumstances.

Of course, in this election year, *anything* is possible. I'm certain it's going to get a lot tighter before it's all over.



ThatSeattleGuy is quite right that polling was primitive for the Dewey-Truman race.

To be more specific on the polling in the 1948 race, Gallup used quota polling. This was criticized in the report on the poll failure issued by the Social Science Research Council and the method was quickly abandoned.




I agree with everything you have said, except the last thing you said. I don't really understand why you are so certain that it will get a lot tigher before it is over.

If this year is like 1980 or 1988, Obama's lead could just continue to climb or at the very least stabilize. I actually think a stabilization or slow continual climb is more likely at this point than a tightening (absent an outside event like a terrorist attack or a terrible debate performance from Obama which seems rather unlikely).

What are your thoughts that make you so sure it will tighten before the end?



I too, agree with seattleguy, that the race will most likely tighten up considerably in the next 4 weeks. For starters, I feel that the American public has a memory of about 4 days. The greatest reason for Barack's recent climb in the polls is the economy. And although the economy certainly isn't going to strengthen, it will most likely stabilize and be pushed back into the subconscious of the American people as more newsworthy stories (like pitbulls and lipstick) start to swamp the media. I feel that the people that are undecided must have to be of the lowest intelligence and will be swayed by such minutia.

Also, as hinted at before, the october surprise, is only a surprise in its quality. It will happen, and as we know, so many people are driven by fear. Any domestic terrorist attack, or (another endorsement of the democratic candidate by osama bin laden) will send the mindless hordes back into the secure, terror-ridding arms of the red. How the republicans have cornered the market on national security is beyond me, but that is a fact that we must accept.

So yes, I think that Obama's lead should continue, however, the democrats are playing fair against a team on steroids. the prognosis is never certain.





My statement was over-simplifying what in my mind is actually complex and fluid situation. Certainly I'm not a professional pollster (nor, even less, a statistician). My belief is that with the McCain campaign signalling that they will be pulling out the stops and going full-negative, combined with the memory of this week's economic crisis fading at least somewhat in voters' minds over the 31 days until 11/4, the margin between the candidates in national polls is more likely to tighten than it is to stabilize at current levels or to widen.

To go out on a limb with specific numbers (just so someone can bring them back to my face later :) I will predict that the current Obama lead in the national polling numbers will shrink to half the current margin at some point before election day - though it could then go either direction after that.

Of course, that's in the end irrelevant; what matters is what happens in each of the contested states and their contribution to the overall EV. And there I've no clue: with the Bradley effect on one side, the seemingly all-out-of-historical proporation Obama GOTV effort on the other, plus the random X factor of potentially unpolled or under-accounted-for cell-phone-only voters, it's reasonable to say this one's going to be difficult to call accurately at that level right through Election night.

Certainly the momentum, numbers, and statistics are all solidly on Obama's side right now. But note that while a purely mathematical simulation of current polling numbers/trends puts his victory chances at 95% or better (for example, see election-projection.net), the Intrade numbers - in which traders presumably factor in a much more complex set of variables, including the possibility of an "October surprise" - currently have it at just 66%. Neither number is likely "correct", but the spread between those two is so large as to invite some thought as to what might be missing from an analysis that only looks to current polls and polling trends.


PS: Question for you - I'm going from memory, but in 1980 didn't Carter maintain a close race until near the end, when he went into free-fall? How does that inform your thought that the current Obama lead should stabilize or widen in the coming days?



No. No. No. Talking in a historical sense about candidates recovering from 8 points down at 4 weeks to go is like the gambler at the roulette wheel who bets on red because the last 10 spins have been black.

Like the roulette wheel spin, each election is statistically unique -- there is nothing statistically to be learned from preceding election data.

In other words: explain how the 1936, 1976, 2000, and 2008 elections possess the same independent variables. Answer: They don't.

It is one thing to predict the likelihood of McCain recovering from his polling deficit -- it is entirely unscientific and unreasonable to make predictions about McCain recovering from his deficit by way of comparison with that of preceding candidates, unless one is making the case that the country has not changed at all.



The roulette wheel example is inherently flawed, because the odds are 50%, so each spin is its own independent outcome. If the odds at the roulette table were not 50% red and 50% black, lets say 55% Blue and 45% red, it would be advantageous to go with the majority and the most likely outcome, which is supported through the historical data.


Matt Herlihy:

Whitetower, certainly there must be something said of the fact that only 2 candidates have overcome their opponent after hitting 50% in October. Maybe the American electorate has been different every time, and all of the characteristics of each race are unique, but certainly I don't think its a stretch to suggest that overcoming such a disadvantage so late has been, historically, difficult.

As more people make up their mind and move one way or another, and less people report that they are open to changing their mind, then naturally there is less movement possible within the polls.



It would be interesting to know how the (opinion-polled) electoral votes looked like when those candidates had 50% and/or +8%, in particular whether the successful candidates had the required electoral vote at least in their safe / likely & lean states. Currently all the Kerry states (-NH, +NM, +IA) are at least leaning to Obama, giving him 260 EV and needing only 10 EV out of the 93 EV available in the tossup states (all 2004 Bush states +NH)


Pacific moderate:

With Gore, keep in mind that he won the popular vote by half a million votes, so 2000 doesn't really undermine the trend. If not for the pecularities of the butterfly ballot design in West Palm Beach County (note the spike of several thousand Pat Buchanan votes in that liberal county), he would have won the 2000 election.


Gary Kilbride:

Your 2000 example is an October outlier due to the dates of the Summer Olympics in Sydney. Unlike this year, when the Beijing Games were held in mid to late August, similar to 2004 in Athens, the 2000 Sydney games were more than a month later on the calendar, ending on October 1. That naturally pushed the debate timetable back, and the masochistic sigh fest by Gore was on October 3, the first debate.

The debate timetable allowed Gore to take full advantage of Bush's horrid September to build the 8 point Gallup lead (although I remember it as closer to 5 or 6 consensus) heading into that first debate. Then he plummeted after the sigh debate and the twiddly thumbs "I agree" performance in debate #2, Bush's foreign policy recital of names he recently memorized.

Although the Olympic timetable allowed 2000 to barely fit within your October guideline, I completely agree with the premise of this analysis. Comebacks are for fools. I learned that in Las Vegas in the '80s. Sharp guys I met in the sportsbooks could instantly estimate likelihood of victory at any point of a game, based on score, situation and pre game odds. I was amazed. Slowly I caught on. Now I can't imagine being without it. The overriding fundamental is the team in the lead is generally much better off than fans want to estimate. Comebacks get the glory and we remember them, while ignoring or downplaying all the similar instances when the lead pony held on, if not widened advantage.

Give me the lead. Be my guest to try and run me down. In politics the frustrating exception, at least for me, is primaries. Likeminded people can suddenly switch on a dime late in a race and ruin long held indications, even if the lead has been substantial and logical for months. I don't pretend to understand it, and that's why I'm careful when wagering. But in the fall the dynamic is completely different, a polarized 40-45% on each side and low number of undecideds. It's hellish to overcome a late deficit.

The principal variable that keeps Obama at 67% Intrade range right now and not higher, like other estimates, is time and opportunity. It's the difference between wagering and tinkering with digits. You can't assume complete normalcy over the next 30 days, an election within a vacuum. Speculators have to allow play in the number to account for unexpected circumstance, particularly when dealing with huge outlays like 1/5 or 1/6 or higher. No one wants to be holding a ticket like that and your horse never makes it to the starting gate. Obama could be polling at 2-3 points lower than this on election eve and yet his theoretical win percentage would jump from 67% to above 90%, simply because virtually every hurdle has been cleared.




The roulette table example *is* flawed, but not for the reason you state. Regardless of the odds (50/50 or 55/45) each spin of the wheel is a unique instance and history of prior spins has no bearing under *any* circumstances.

The polling patterns being discussed are not the same as a roulette wheel; the result of this week's polls are partly related to the history of prior polls; the current poll results reflect changes to the prior results after the addition of new information. Polls are so sensitive to their own history that the publishing of a poll is enough new data to influence the next poll, even if no other data is added - that is why some countries prohibit the publishing of poll data in the last 48/72 hours before a vote.

In fact, if there was no relationship between current polls and their historical precedents then this site would be almost useless.

Comparing entire election cycles to each other (as the author did) to show what the definition of an insurmountable lead might be can be useful. The key assumption that may cause this multi-cycle comparison to be incorrect is the assumption that the maximum rate-of-change in poll results is constant, which would mean a 8% lead is as hard to overcome today as it was in the 1960s. **That** assertion may no longer be true, given the changes in media/medium (Internet, Cable News cycle times), but we will not know that until we have some data in hand to shown the assertion is incorrect.



To ThatSeattleGuy ...

Your memory of the 1980 campaign is correct.

Reagan and Carter were fairly close in the polls (much less than 10 percent) throughout most of the campaign. I believe Reagan maintained a small lead.

Then, on the weekend before the election, there was major movement away from Carter and toward Reagan.




Just a thought..

October Surprise?

Given the pop culture driven ways of much of our society, any possibility the October 17th release of the "W" movie would have any effect on the election, one way or the other?



It seems to me obvious that the polls will tighten in October:

First, McCain has turned completely negative, trying to turn the attention from the economy to Obama. Negative campaigning should bring down Obama's numbers.

Second, there is "front-runner-itis" -- whoever the front runner, the tendency is to revert if for nothing else than people get tend to sick of the same tunes.

HOWEVER, this does NOT mean that McCain will improve his chances of winning.

HIS problem is HIS low polling numbers. He may be able to pull Obama down a few notches, but it isn't likely improve his own numbers, and in fact, may make them worse.

Finally, people may get sick of Obama's "tunes", but they are ALREADY sick to death of McCain/Bush.

And Obama will NEVER let people forget it.



@ Gary,

Excellent post...



By 10/1 and following the Pres and vp debates support is nearly cemented. Negative ads help the base - what they do not do is persuade. IN fact all studies show among decided voters, they get defensive and dig in their heals.

Sydney's camp is resting their hopes on the following:

1. Polls are wrong and in the states that matter, they are within reach.

2. They can rile their base up better than ever before.

3. They will get an assist from state officials and Rep. lawyers at keeping Dem. numbers way down (surpressing the vote).

Undecides will not be paying much attention from here on out. The race is over, minus the counting. All Sydney is going to do - from here on out - - is to once again do all he can to rip this country apart into 2 highly partisan and hated camps. Nice work Sydney - that's putting America First. You and your political whore Caribou Barbi could not give a rat's ass about the American people and it clearly shows.



I'm going from memory, but in 1980 didn't Carter maintain a close race until near the end, when he went into free-fall?

Carter and Reagan did not debate until late October. The voters had been ready to boot Carter for over a year, but wouldn't commit to Reagan until they saw him looking relaxed, personable, and "presidential" in the debate.

With Gore, keep in mind that he won the popular vote by half a million votes, so 2000 doesn't really undermine the trend.

Bush had a 10+ point lead in October too, so 2000 undermines the trend regardless of who won.




any possibility the October 17th release of the "W" movie would have any effect on the election, one way or the other?

Doubtful. A large percentage of minds are going to be made up by then, and if a documentary like "Fahrenheit 911" didn't do it for Kerry in '04 (especially with a much longer pre-election lead time, and the subject matter that _directly_ involved one of the candidates), then "W." is highly unlikely to sway any remaining undecideds in '08.

Anyone who already links the outgoing Bush administration and a potential incoming McCain one in their minds has already done so; the film's not going to change that much. Even if it does affect some votes, I think its influence will be to push equal numbers of fence-sitters into the Red camp as it draws into the Blue one.



One key factor suggesting that Obama will retain his lead this year is that Obama has a significant resource advantage over McCain. For the first time in memory, the Democrats have more money than the Republicans, even counting the RNC vs. the DNC. It is unknown how much more. But given the numbers I've seen, my guess is it will be substantial.

Just to show my work, there are essentially three sources of resources for the candidates:

a) The campaigns themselves.
b) The national parties (RNC/DNC)
c) 527 groups.

McCain is limited to his $84 million in public financing for his campaign. For Obama, the sky is the limit. It is unknown at this time exactly how much he'll raise. But if his August figures are any indication, chances are it will be substantial. He reported $66 million in August. I believe his September total will at least match, and may even exceed that number.

Last I checked, the DNC had maintained a fundraising advantage over the RNC. I don't have exact figures on that offhand, but I haven't seen any reports suggesting that's changed.

The only wholly unknown variable at this point is contributions by 527s. But conservative 527s would have to *dramatically* outspend progressive 527s to even bring Obama's resource advantage to even.

I can see the Obama campaign's dramatic resource advantage helping him maintain or even extend his lead in October. I wouldn't rule out the possibility of Obama pulling away from McCain, and leaving him behind.




In regards to "front runner-itis": I don't think this argument really holds right now. It's very prescient in the primaries, and early in the campaign as voters get to know candidates. But, by October, a great majority of the electorate has made up their minds. In fact, Rasmussen today states that 45% of voters are absolutely postitively voting for Obama and 38% for McCain. Those are voters that will NOT change under any circumstance.

At this point in the campaign its actually a net POSITIVE to be the frontrunner, because Americans like to settle into mob mentality. "He's doing well in the polls, lots of people like him, he must be good. Let me take a look at why everyone is supporting him". It's like a ball rolling down a hill. The stronger the narrative is that Obama is winning, the more likely he is to consolidate support and build his lead. This is EXACTLY why the media refuses to report this race as anything but "very close". It's just downright false, but the reason they are doing it is because if the Obama landslide meme takes hold, they know that John McCain is absolutely, positively guaranteed to come back from the dead.

That said, we are past the first debate, past the VP debate, into the second week of October, and have only two more debates left (usually watched MUCH less than the first one). Opinions are solidifying.

Indeed, if Rasmussen is correct and 45% of voters are solid behind Obama, and another 6% are leaning, with 5% undecided, Obama only needs 50% of a combination of his already leaning voters and the pure undecideds. Even if John McCain were to take the lion's share of undecideds, and Obama retained all of his leaners, Obama would win with 52-53% of the vote. Even if Obama LOST 15% of his current leaners and picked up none of the undecideds, he would reach 50%. A daunting task for McCain indeed.



A large percentage of minds are going to be made up by then, and if a documentary like "Fahrenheit 911" didn't do it for Kerry in '04 (especially with a much longer pre-election lead time, and the subject matter that _directly_ involved one of the candidates), then "W." is highly unlikely to sway any remaining undecideds in '08.

The release of Fahrenheit 911 did coincide with a pretty big shift in the polls, though. You could make a plausible argument that it might actually have helped Kerry more if it had been released on Labor Day weekend instead of earlier in the summer.


Claire Voyant:

Historically the race for President tightens in the final 2-weeks before the election.

I believe the same will happen this year.

The media will play the story as the "McCain Comeback"

Mr. Obama will win with just under 50% of the vote
McCain will tally about 48%

Mr. Obama will win at least 325-electoral votes and will be credited giving the democrats a filibuster proof Senate.



Until I see a poll with "weighting" for race (people lying about voting for a black I.E. Los Angeles and Tennessee) I don't believe the Obama lead is real.



West Palm Beach has already screwed-up an August election with "magic" votes, and their ballot this November instead of butterflies has arrows! Proven to be 33% less accurate than fill-in ovals. And Lord only knows what magic Diebolts black box electronic "voting" machines have in store for us.



Interesting... Another reminder that Al Gore's loss in 2000 probably wasn't really a loss IF the votes had just been counted fairly. Water under the bridge to nowhere.

Also, I would suggest the whole racial question in these polls can be explained by the fact that, in the RCP Poll averages, people indicate the generic congressional vote goes for Democrats 48.6% to 38.3% While the specific presidential vote goes to the (black) Democrat candidate over the (white) Republican candidate 48.3% to 43.3%. That increase in 5% for McCain is possibly indicative of those people who can't vote for the black guy. Just a thought.


Paul 66:

The one thing people seem to be discounting is the long lasting effect of the economic news.
I understand that starting today a lot of people will be getting their 401K statements. And this will, for a lot of the ones who have tried to keep it vague, crystallize what the downturn in the markets means to them.
If you figure that will last over the next week or so, you should see the lead climbing throughout that time.
Also when one is looking at what (nefarious) effect local officials may have, and accepting that the Republicans may have been more inclined to use their position to influence things, one has to remember that in a number of states and in particular in Ohio what was a solid Republican administration (governor, Lt Governor and Secretary of State) in 2004 is now solidly Democratic.



AAAARGH!!! The lead is back to only SIX points?!?!?

I guess the Palin effect is back again!!!??? I'm scared that Obama will fcuk up the 2nd debate, and that Bristol Palin will have "a tragic miscarriage" one week before Nov. 4th.

Just watch mccain paint this as a comeback kid moment.

Un effing believable. This race is not about Obama or Mccain, anymore, it's just about how well Palin does. Biggest mistake was not to let Biden pound her senselessly.
Ah, hindsight .....



Not to relitigate the past, but Gore may have won the 2000 election if the SCOTUS decided differently in Bush v. Gore, or the situation in Florida was handled differently. Additionally, the electoral college is an variable unaccounted for by the chart.

Just a thought.


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.