Okay first thing's first: everyone just calm down. Sit down, have a cup of coffee, close pollster.com for the day before you have a freaking heart attack.
If you are worried about last-minute poll movements, here's a long laundry list of facts that I've been compiling over the past couple of days, related to history and psephology. I've done a bit of amateur polling in my time. Hopefully some of the following can soothe your nerves in the final week of the campaign.
First thing's first: lets deal with a couple of obvious facts.
1. Elections are about issues.
This election is about one massive issue in particular - the economy. It's bad. When the economy is bad, it rarely favours an incumbent party. Polls have shown consistently that Obama is more trusted on this particular issue.
National Security has taken a back seat to the economy, but there is anecdotal evidence in some states that Obama is not far behind McCain on the key question of leadership.
If there was a final-week poll reversal in favour of John McCain in this economic environment, it would be an unprecedented event.
2. The polls over the past few days have been narrowing slightly. This is normal.
Why? Because the last undecided voters are finally starting to make up their minds. It is not unusual for them to be breaking for McCain. In the absence of any major event or new revelation in the campaign, it's completely normal.
If you think about why people would still be undecided in this sort of toxic economic environment, the question these voters still have in their head is probably one about Barack Obama, the man. Perhaps some of these voters are racist, or nervous about Obama's inexperience, but are still weighing up whether to ignore that because the economy is so bad.
These people will not break 100% for John McCain on election day. Even if they went for McCain 75-25%, on Current Polling Obama would still be slightly ahead.
If you look at most of the tightening over the past few days, barely any of it has come off Obama's margin. It has mainly come from McCain's number edging upwards. This is also normal. Did you really expect that John McCain's final popular vote tally would be down at 42%? I certainly don't think that's going to happen.
Far more likely - in the event of an Obama victory - would be McCain getting 46-48% of the popular vote, and Obama at about 50-53%. I do have a reason for suggesting this, and I'll come back to it tomorrow.
3. Barack Obama's "national popular vote" poll numbers are still at the 50% threshold in most polls, and they have been like this for weeks.
This is important because it means Obama has already - in polling at least - reached a stable threshold he needs for victory. If Obama keeps polling between 49-52%, and McCain is still back down around 44-46%, Obama does not need to worry quite so much about the people who still haven't made up their minds. McCain's only hope rests not just on convincing all the remaining undecided voters, but also CHANGING THE MINDS of people who currently state their intention is to vote for Barack Obama. Given that Obama's base of support has been rock solid in the 49-52 band over the past few months, that's a very tall order indeed (unless people have lied to every single pollster over the past 2 months, or if there is a massive Bradley effect).
Of course, Obama shouldn't campaign on the assumption that remaining undecideds don't matter - he should try and win as many as he can. He has the ground game and the funds to do it, so why not? But from a polling perspective, it's a positive sign.
Why? Well, here's one Pollster's example.
I was digging around the Gallup Poll website yesterday when I came across a great page. In particular, I took an interest in Gallup's record at polling past presidential elections. They put up a very good graph on this page that is worth noting.
http://www.gallup.com/...
Gallup notes that in almost every single election since 1952, the person who lead in the Gallup Poll of Registered Voters (not Likely voters) one week before the election ended up winning the popular vote. The only two exceptions were 1980, where Reagan came back, and 2000, where Al Gore won the popular vote, but still lost the election.
At this point, what we are all worried about is some kind of amazing poll reversal - that's what it would take for McCain to win. So lets look at 1980, where Reagan came behind spectacularly. The first thing we should note is that although Reagan was indeed way behind in the last poll, there were quite a lot of undecideds - way more than there are in this current election. There was also a libertarian who got 6% on election day. Most critically, Carter's poll number was slightly low for an incumbent seeking re-election - only 47%. This meant that although he was theoretically ahead, he had not sealed the deal.
Secondly, think back to the issues of the 1980 election. The economy was a mess and the Iran Hostage scandal was in the news constantly, including a significant stuff-up in the last few days. These circumstances were extremely tough for any incumbent attempting to seek re-election. In addition, there was one Presidential debate in the final week. It was the only one held. Reagan had been painted as an unknowledgeable fool on foreign policy and economics by the media and the Carter campaign, but in the debate he performed way beyond expectations. His confident performance reassured voters that he was ready to be President. Sound familiar?
The other - and far more critical point to make is this - since 1952, in Gallup's final Registered Voter Poll before the election no candidate who polled at 49% or above in a 2-way race ended up losing the election.
Where was Obama yesterday in the Registered Vote Tally? 52-42. Where is he today? 50-43. As long as Obama stays at 49% or above, the chance of a nasty surprise is quite small.
4. Comparable Popular Vote Polling Scanarios
Today's polls are all roughly around about 50% for Obama, compared with 44% for McCain. The range over the past few weeks in all the polls has been around 49-52% for Obama, and 41-46% for McCain. Obama is ahead by any measure - on the side he wants to be on.
Lets look back again through the history of Gallup Polls. I know, it's the same polling organisation, but it's also the one that's been around for the longest and has the most historical data we can tap into. This is not supposed to be a defence of the accuracy of Gallup, merely one example. You can look through the history of other polls and find your own examples, if you like.
Going back to the "last poll one week before the election", there are a couple of polling scenarios that seem to be similar to current polling. For our sake, lets consider two-way elections (no 3rd parties, so 1968 and 1992 are out) where one candidate was polling between 49-52%, and the other candidate was polling somewhere between 41-46%.
In those parameters, we get the following elections:
1952 -
Final Poll: Eisenhower 49%, Stevenson 43%
RESULT: Eisenhower WON
1956
Final Poll: Eisenhower 51%, Stevenson 42%
Result: Eisenhower WON
1960
Final Poll: Kennedy 49%, Nixon 45%
Result: Kennedy WON
1976
Final Poll: Carter 49%, Ford 44%
Result: Carter WON
1988
Final Poll: Bush 52%, Dukakis 41%
Result: Bush WON
2004
Final Poll: Bush 49%, Kerry 47%
Result: Bush WON
I threw in 2004 despite it being slightly outside the parameter - just so you can get an idea. I think it's worth noting that John Kerry was performing significantly better than John McCain is now. [update - a few polls including battleground and Rasmussen today have McCain at 46-47%, but Obama is still rock solid at 50% - at worst Obama is doing as well as Bush did in 2004.]
Two things strike me about this data. Firstly, in every case, the person leading won the popular vote and the election. On only one occasion was there a very tight result - in 1960. There were two moderately tight results in 1976 and 2004.
I am particularly interested by the data from 1952. That seems to match the closest in polling, and there are a number of historical similarities as well. First, there was no incumbent president or vice president running. Second, the incumbent was unpopular. Third, there was a foreign crisis (the Korean War). And finally, there was a real "time for change" factor after 20 years of the democrats in the white house. The republican challenger was very popular and had a good ground game, and a good slogan. Just something to think about - Eisenhower won that election in an enormous popular vote landslide.
5. For all our talk about national polls narrowing, the popular vote doesn't matter. State polls do.
And have the state polls narrowed at all? Well... not at all. In fact, Obama is already above the 50% threshold in the pollster.com average of battleground state polling in:
New Hampshire, Pennsylvania, Virginia, Michigan, Wisconsin, Minnesota, Iowa, Colorado, New Mexico - which would alone give him the win.
Obama is also at 49.7% in Nevada and Ohio.
Warning Signs to Look for
With all of this said, change is still absolutely possible in the last week.
If Obama's poll numbers in the national poll and in all the major battleground states go below 49% and stay there until the end of the week, that might be a worry. This would indicate that people who previously supported Obama have changed their mind.
If McCain was to reach a tie with Obama, at around 47% in the national poll, things could get a little bit close for comfort.
On the upside, Obama already has millions of votes in the bank from early voting. And always keep in mind the overarching issues - the economy and the war. These favour Obama. Keep them in mind.
Finally, 7 out of 10 voters believe Obama's going to win.
That's all for today, I'll be back tomorrow.