As Expected

Chris Bowers notes an interesting finding: If you take the current Obama-McCain national polling matchup and compare it to the Kerry-Bush result in 2004, you'll see an 8.1 percent swing in favor of the Democrats in the national popular vote. And if you apply an 8.1 percent swing in favor of the Democrats in each state, you get a map that's really the same as the map produced by the current state-by-state polling:

Not the most earth-shattering result on the planet, but a useful reminder that even though in principle the Electoral College allows for crazy divergences from the popular vote result, in practice gains in national polling tend to be distributed fairly evenly. So if you want to keep tabs on the race over the summer months there's probably no need to bother with the state-by-state polls.

Comments

Well, it is earth-shattering in that it represents a data point against the prevailing theory that the hard-working Americans of the Rust Belt are super-plus Negro haters that will never vote for Obama.

Posted by Gabriel | June 18, 2008 6:47 PM

Hey Matthew. I'm a regular reader but I usually don't comment.

Have you looked at monte carlo simulations of the electoral college vote? The only sane way I have seen to predict the election. Using polling data (not great, but it's what we have) it simulates the election 10,000 times. Right now the source I've found has Obama winning 98.5% of the time. Kind of eye-popping. Kind of shows how the media is desperately going to perpetuate the myth that this is a close race.

You can find that at a blog called "Hominid Views", with which I have no affiliation. I actually discovered it about an hour ago.

Posted by akachazz | June 18, 2008 6:52 PM

Did Virginia secede again?

Posted by careful1 | June 18, 2008 6:53 PM

or you can just put fivethirtyeight.com in your feed reader.

Posted by dbt | June 18, 2008 6:53 PM

Not the most earth-shattering result on the planet, but a useful reminder that even though in principle the Electoral College allows for crazy divergences from the popular vote result, in practice gains in national polling tend to be distributed fairly evenly.

your reasoning is circular. first, you add 8.1% to all state polls, which is another way of distributing the national 8.1% evenly among all the states, and then you use that electoral map to show the the popular vote "tend[s] to be distributed fairly evenly" across the country.

Posted by upyernoz | June 18, 2008 6:53 PM

Hey Matthew. I'm a regular reader but I usually don't comment.

Have you looked at monte carlo simulations of the electoral college vote? The only sane way I have seen to predict the election. Using polling data (not great, but it's what we have) it simulates the election 10,000 times. Right now the source I've found has Obama winning 98.5% of the time. Kind of eye-popping. Kind of shows how the media is desperately going to perpetuate the myth that this is a close race.

You can find that at a blog called "Hominid Views", with which I have no affiliation. I actually discovered it about an hour ago.

Posted by akachazz | June 18, 2008 6:55 PM

whoops.. it put my comment up twice.

Posted by akachazz | June 18, 2008 6:59 PM

akachazz, I don't know what crack your simulator is on but I don't think their numbers can keep up with Nate Silver's at 538. (Nate was previously seen doing baseball stat work at BP, he invented PECOTA).

Posted by dbt | June 18, 2008 6:59 PM

upyernoz wrote:
your reasoning is circular. first, you add 8.1% to all state polls, which is another way of distributing the national 8.1% evenly among all the states, and then you use that electoral map to show the the popular vote "tend[s] to be distributed fairly evenly" across the country.

I think you missed the part of his post where he said "And if you apply an 8.1 percent swing in favor of the Democrats in each state, you get a map that's really the same as the map produced by the current state-by-state polling". The idea is to compare the electoral map you get by adding 8.1% uniformly to every state with the electoral map you get using state-by-state polling, and note that the results are the same; nothing circular about that.

Posted by Jesse M. | June 18, 2008 7:00 PM

The state-by-state polling numbers are still going to be useful in tracking where the campaigns (particularly the McCain campaign) is going to be spending its resources (both in terms of money and candidate appearances). The closer the race is in traditionally solid GOP states (VA, NC, GA, CO), the less time McCain can spend trying to win MI, OH, PA. And like Nate at 538 pointed out today - a close race in Florida absolutely destroys McCain's strategy.

Posted by south side | June 18, 2008 7:04 PM

your reasoning is circular. first, you add 8.1% to all state polls, which is another way of distributing the national 8.1% evenly among all the states, and then you use that electoral map to show the the popular vote "tend[s] to be distributed fairly evenly" across the country.

It was phrased a bit poorly, but I think you're probably misunderstanding the point -- it isn't circular. Obama-McCain national polling is showing an average pro-Democrat swing of 8.1% compared to Kerry-Bush national polling. If you then add 8.1% to the results from 2004 in each state, it should match up to the statewide polling we're seeing now. Only... it doesn't, since Obama isn't polling 8 points better in Michiga or New Hampshire than Kerry did, and he's doing substantially better in Alaska/Montana/North Dakota/North Carolina and a bit better in Virginia. So that makes the point wrong rather than circular.

Posted by jbryan | June 18, 2008 7:05 PM

Oh crap! Nate Silver of BP? This is like telling me the guy who invented chocolate also invented peanut butter. 538 it is then!

Posted by akachazz | June 18, 2008 7:06 PM

You may want to pass this information along to President Gore.

Posted by Jim | June 18, 2008 7:42 PM

Only... it doesn't, since Obama isn't polling 8 points better in Michiga or New Hampshire than Kerry did, and he's doing substantially better in Alaska/Montana/North Dakota/North Carolina and a bit better in Virginia.

But determining the margins state-by-state isn't relevant--it's only about who wins state-by-state. 50%+1 is just as good as 75% as far the electoral college is concerned (NE and ME excepted). MY's point (perhaps wrong) was that, for purposes of EV counting, watching national polls is close enough to state-by-state to ignore state-by-state 'til Labor Day (at least).

Posted by zaleriana | June 18, 2008 7:51 PM

Pushing the 10% of the population that typically causes problems. Thankfully, that 10% are voting for Obama, this time around.

Guess that 10% are the true center of American politics, ehh?

Posted by zic | June 18, 2008 7:52 PM

These polls are worthless because they neglect the Wilder effect. How many of those white middle aged men and women who say they are going to vote for Senator Osama are lying? I suspect that a sufficient number of them are lying such that Senator McCain will win an overwhelming electoral college victory. Remember, you hear it here first.

Posted by SLC | June 18, 2008 7:57 PM

One of the things that really surprised me in 2004 was how almost all the state polls had Kerry winning Ohio. He looked very good EV wise, given his leads in Ohio and Penn, even while losing the national polls.

The Bush comfortably won Ohio by 4. The state poll averages were way off, while the national poll averages were almost entirely accurate.

So I'm biased towards believing the info we get from national polling and ignoring 538 simulations (as smart as Nate is).

Posted by Shock Mouse | June 18, 2008 8:07 PM

I don't think it follows that just because the two maps are the same TODAY at the beginning of the campaign, they will be the same in November. The point of campaigning is to try to change the polls in some states more than others. Maybe that's misguided, but these maps don't shed any light on that.

Posted by Led | June 18, 2008 8:08 PM

It also blew my mind when I found out Silver = Poblano. Like when it was revealed that Ken Tremendous is a writer for "The Office" and sometimes plays Mose, Dwight's cousin.

Posted by 55 | June 18, 2008 8:08 PM

SLC,

I'm not really sure what you are saying makes any sense. There is no reason for a closet racist to say they will vote for Obama in a poll when they could just say they are voting for McCain. I somehow doubt that the pollster is asking the follow up question, "is it because Obama is black?" when they a respondent chooses McCain.

Posted by Ethan P | June 18, 2008 8:42 PM

How many electoral votes does Lake Michigan have? Looks like a winnable swingstate.

Posted by phil | June 18, 2008 8:50 PM

SLC, what Ethan said. And did the Wilder effect show up in any of the many polls conducted during the nomination contest?

I'm more concerned about the "black people and other Democrats show up at the polls and find that they've been mysteriously removed from the rolls or are faced with an 8-hour-long line or have their newly required ID rejected or are otherwise prevented from voting" effect.

Posted by KCinDC | June 18, 2008 9:00 PM

There was no notable Wilder Effect in the polling for the relevant races in 2006. So, it appears the Wilder Effect may be in the past at this point.

Posted by DTM | June 18, 2008 9:22 PM

Uh, no, in practice the Electoral College is an archaic system that favors the voters of one state over the voters in another state. As a Texan, my vote is more or less worthless. And so is my Californian boss's.

Instant Run-off Voting is the way to go. Really, I wish Obama would make this part of his campaign, as I think it would fit in with the "changing our politics" bit.

Posted by socctty | June 18, 2008 9:35 PM

"I'm not really sure what you are saying makes any sense. There is no reason for a closet racist to say they will vote for Obama in a poll when they could just say they are voting for McCain."

There is a well known and pervasive effect where survey results are wrong where matters of taboos are involved because respondents won't want to reveal socially unacceptable views, even anonymously.

I think there's every reason to believe that Obama will receive less support in the voting booth than is indicated by polling.

The question is: just how much less?

When I played with that electoral college map thingie the other day, I was left feeling much more confident that Obama will win than I had before. It just won't take much better performance than the Dems managed in the last election to win by a small margin. I live in New Mexico, and I'm certain it will go to Obama, for example. (I don't think it's going to go to the Republicans again for a long time, demographics are against that likelihood.)

It's easy to assume a large margin, or even a landslide, if you look at how low Bush's popularity has gone and how poor of a candidate McCain is likely to be.

But I think people have been underestimating the resistance there will actually be, on election day, to voting for a black person. I'm not happy about this, of course, but I well know that racism is alive and well in this country. It's going to hurt Obama. But I don't think, right now, that it's going to hurt him so much that he doesn't win. But it will be close.

Posted by Keith M Ellis | June 18, 2008 9:47 PM

Re DTM

I is yet to be seen if the Wilder effect will be negligible in a national race for president, which is quite different then a local or statewide race.

Posted by SLC | June 18, 2008 9:47 PM

Re: How many of those white middle aged men and women who say they are going to vote for Senator Osama are lying?

Please check your calendar. Middle aged people have all grown up post Civil Rights era, indeed, post 1960s. They don't have any connection to Jim Crow and KKK racism. The GOP habit of waving that smelly old tye-dyed hippie shirt against the cultural upheavals of the 60s is fast approaching its freshness date. I know because I have joined the ranks of the middle aged, like it or not, and though born in the 60s my first political memory is of Richard Nixon resigning, not of "Burn, Baby, Burn", George Wallace or even Vietnam. Now if you want to make a point about senior citizens being unwilling to vote for a Black man, I do, regettably know many of them who won't. But every last one of them will also not vote for McCain because they all learned to hate the GOP with a fresh passion. Maybe they'll vote for a Green, or for Bob Barr, or some other minor candidate-- or just not vote at all. But McCain is in big trouble and his groupies need to sober up and understand that fact. 2000 should have been his year. This year Jesus Christ would have trouble running as a Republican.

Posted by JonF | June 18, 2008 9:56 PM

"There is a well known and pervasive effect where survey results are wrong where matters of taboos are involved because respondents won't want to reveal socially unacceptable views, even anonymously."

Right, but saying you are going to vote for McCain isn't socially unacceptable. 40%+ of the country is going to do it. It isn't like they are being asked in the abstract "would you vote for a black man?"

At any rate, the primaries offered little or no evidence of a systematic polling bias due to race. If it didn't show up in 50-some heavily polled contests it seems unlikely that it will show up now.

Posted by Leo | June 18, 2008 9:58 PM

Keith, if half the country is voting for McCain, how can it be a socially unacceptable view?

SLC, why would a Wilder effect show up in the general election but not in the primaries?

I certainly agree that racism is alive and well. I'm just not seeing any evidence that people are lying to pollsters about voting for Obama. After all, there are plenty of nonracist reasons to vote for McCain, so what's the point of lying?

Posted by KCinDC | June 18, 2008 10:00 PM

SLCunt seems intent on slipping subliminal shit into its posts. Since SLCunt likes calling for bannings, can I suggest that repeat uses of the accidental-on-purpose Bin Laden typo earns a ban?

Posted by pseudonymous in nc | June 18, 2008 10:05 PM

SLC,

Um, obviously that cuts both ways--it has yet to be seen if there WOULD be a Wilder Effect in a national race for President, since there has never been a black versus white matchup between the major parties. So since all the evidence FOR a Wilder Effect is from local and state races, I don't see why that evidence can't be countered by more recent evidence of the same kind.

And that doesn't make theoretical sense anyway. The Presidential election is actually just a bunch of statewide elections, and I see no particular reason to assume that somehow the same white people in the same states would be telling the truth about Senatorial candidates but not Presidential candidates.

Posted by DTM | June 18, 2008 10:07 PM

"I certainly agree that racism is alive and well. I'm just not seeing any evidence that people are lying to pollsters about voting for Obama. After all, there are plenty of nonracist reasons to vote for McCain, so what's the point of lying?"

I think you guys are looking at this the wrong way. The racists who won't vote for Obama because they are racists know that they won't vote for him because they are racists. They don't want other people to know.

This isn't going to matter much to those who are also die-hard Republicans who no one would have expected to vote for a Democrat, anyway.

But this will matter to independents and Democrats who are racist. To those, to varying degrees, they will feel that they are revealing their racism by saying that they will vote for McCain.

A lot depends upon what other questions they are asked in the poll. If they are asked to identify how they are registered, or have voted in the past, or will vote for other offices on the ballot, then the effect will be large.

But I suspect that even when there's no other information asked for in a poll, there's still going to be a certain number of racists who will feel guilty knowing that their choices are motivated by racism and will lie about it because of their guilt, even if there's absolutely no way the poller could tell that they weren't someone who is a strict Republican or just likes McCain. Racism is a guilty secret now in the US, known racism is the contemporary scarlet letter, and I think people will lie to conceal it, even when they don't need to.

Posted by Keith M Ellis | June 18, 2008 10:25 PM

Keith: If that's so, why didn't the effect appear in the primaries? It's hard to imagine a more comprehensive experiment on the subject of racism among Democratic and independent voters.

Posted by Leo | June 18, 2008 10:35 PM

Keith,

I'm sure there is some number of people like that still around. But again, data from 2006 (and the 2008 primaries) suggests there aren't enough of them around anymore to make a significant difference in the polls.

Posted by DTM | June 18, 2008 10:40 PM

I've always thought that it is quite difficult for the popular vote and the electoral vote to yield wildly different outcomes. For example, if candidate X wins the popular vote by 3%, its hard to create a scenario where he loses.

Since 3% in the popular votes doesn't seem like all that much, this would argue that candidates should campaign like its a popular election and forget the electoral system.

Posted by cube | June 18, 2008 10:45 PM

"It's hard to imagine a more comprehensive experiment on the subject of racism among Democratic and independent voters."

But it's not the same situation. A racist who is a lifelong Democrat or an independent who voted for Clinton and Kerry, who won't vote for Obama is strongly impying their racism. The same who voted for Clinton in the primaries is not.

Even if the pollster doesn't ask about past voting patterns, pretty much every Dem racist who is embarrassed about his/her racism could easily and plausibly convince himself that Clinton was the better choice and say so with much of that tinge of guilt. But saying they are voting for McCain, if they are a racist, embarrassed that they are, and know that this is by far the larger portion of why they are voting for McCain...well, that's not guilt they can hide from. They'll vote that way, but they're not going to admit it to anyone if they can help it.

Posted by Keith M Ellis | June 18, 2008 11:03 PM

cube - That's an interesting idea, but this map ( http://en.wikipedia.org/wiki/Image:2004CampaignAttention.png ) says that isn't the case.

Of the 50 or so presidential elections we've had, I think 4 of them have gone to a person who didn't win the popular vote. That's a pretty shitty record, and recent history doesn't make it seem any better.

And any way, if we all agree that more often than not the electoral college system returns the same result as the control - benchmark - popular vote, then why don't we just go by a popular vote?

Perhaps I'm going off topic with this, as MY probably intended to focus on pollsters predicting stuff and not how elections should actually be conducted.

Posted by socctty | June 18, 2008 11:03 PM

The numbers do look good for Obama...so far. It is still too early in the game for me to put too much stock in the poll numbers though. VERY cautiously optimistic right now...

Posted by Porn Store Owner Dude... | June 18, 2008 11:37 PM

*sigh*

So many red states. What's wrong with people that they'd willingly vote for more of this crap?

Posted by Anthony Damiani | June 18, 2008 11:59 PM

It also blew my mind when I found out Silver = Poblano. Like when it was revealed that Ken Tremendous is a writer for "The Office" and sometimes plays Mose, Dwight's cousin.

Whaaaat? I never knew this. Now that I think about it, it's not particularly surprising since FireJoeMorgan is one of the few websites that can make me literally laugh out loud every 20 seconds, and The Office is the only sitcom on network TV that makes me laugh. But still...

Posted by Andrew | June 19, 2008 12:19 AM

So if you want to keep tabs on the race over the summer months there's probably no need to bother with the state-by-state polls.

???

Don't campaigns usually focus their forces on key (swing) states?

Isn't it the case now as always that a few states can make the difference even if a candidate has big leads in other states?

Posted by Swan | June 19, 2008 1:41 AM

http://www.youtube.com/watch?v=2HbJUE-G9pY

Posted by CRYS | June 19, 2008 2:44 AM

I don't see the relevance of constantly harping on Obama's current advantage over McCain.

First of all, Clinton had an advantage over Obama when this thing started. Look how that ended up.

Second, if, as I expect, Bush starts an Iran war, and Obama is unable to distance himself from it because of his current on-the-record policy statements about Iran, this could seriously change the percentages. You can believe that or not, but it has to be taken into account as a possibility.

And again, I have not heard one single Democrat discuss that possibility and how to counter it if it occurs.

In my view, the only way to counter it is for Obama NOW to distance himself from his previous comments on Iran and his groveling in front of AIPAC. Of course, that will leave him open to charges of "flip-flopping" - but compared to McCain record on that score, that shouldn't be a problem.

The other problem, of course, is if an Iran war starts, will enough of the population get suckered into approving it that no matter what Obama does, he loses.

If he distances himself now, I don't think it will affect his polling much. He may lose a few people who consider his current stance on Iran as a major favorable item, but over the course that shouldn't matter since most of the electorate does not approve of an attack on Iran - at least while Iraq is still going on. And it will help him when the war starts - he can go on the offensive against it while McCain backs it, and be able to differentiate himself and place himself in accordance with the majority of the electorate.

But if he doesn't do that, my prediction is he will lose if an Iran war starts before election day.

It is possible that Daniel Pipes is correct and Bush will not start the war unless Obama is elected. That would fit my prediction that one of Bush and Cheney's goals is to tie the hands of the incoming administration. But I think Bush wants McCain in and believes that starting an Iran war before the election will enable that.

Posted by Richard Steven Hack | June 19, 2008 3:00 AM

"even though in principle the Electoral College allows for crazy divergences from the popular vote result, in practice"

This is the concept of "swing", which has been a staple of political analysis outside the US for fifty years, but which American political commentators are just grasping. To be far, until recently partisan coalitions in the US have not been stable enough for "swing" to be very relevant.

Swing are the two changes in the vote for two parties, divided by two. The swing in 2004 was about 1.5% in favor of the Republicans. If swing is evenly distributed, and normally it is, the gaining party will win every state that it had not won last time, but where the margin was smaller than the swing.

Also, in principal it is possible for a candidate to be beaten 74% to 26% in the popular vote, and win in the electoral college by taking a bare majority in a bare majority of the states, and getting no votes elsewhere. The only presidential election where something like this happened was in 1860, when Lincoln got narrow majorities in each of the northern states and almost no votes in the South. Among other things, this led to a civil war after the election. Normally you don't get these sorts of extreme results, and the "electoral vote winner and popular vote loser" scenario will only happen in extremely close elections, with fraud coming into play. Fraud produced this result in 1876, the closeness of the result and a third party candidate in 1888, and all three factors came into play in 1960 and 2000.

In 2004, the Democrats gained a state they had not won in the last cycle, New Hampshire, despite a swing against them. The last time this happened was in 1960, when two Deep South states voted for an independent slate of electors, and the last time in a two party contest was in 1956, when Stevenson picked up Missouri despite something like a 4% swing to Eisenhower. The point is that this is very rare. If Obama does better than Kerry in the popular vote, he will carry all but one or two of Kerry's states. The likeliest state to defect is Minnesota, if Pawlenty is the Republican VP nominee, then New Hampshire. If he does worse, he won't be winning any states that Kerry couldn't win.

The point is that unless the popular vote margin is razor thin, the popular vote winner will get the electoral vote majority.

Posted by Ed | June 19, 2008 3:23 AM

I look at this map and if Virginia is really in play it tells me is that Obama is going to pick Webb or Keane to be his running mate.

Posted by Shrike58 | June 19, 2008 8:26 AM

The problem is not "crazy divergences". The problem is that when every election is decided by a whisker -- by differences that are significant but not substantial -- within a system that APPEARS to have more than one set of rules, the outcomes do not confer legitimacy.

Posted by Frank Wilhoit | June 19, 2008 8:28 AM

Shrike, Virginia's in play this year, whether or not he has a Virginian on the ticket with him. Webb's already said about fifty different ways that he's not going to be VP and won't accept it if offered, and Kaine seems like he's an outside shot for it. Still, if Obama takes all the Kerry states plus Virginia, that puts him at 265 (out of 269 needed). He'd really only need to pick up one other state to win the election if he takes Virginia.

Posted by Tel | June 19, 2008 9:06 AM

If Virginia is still really in play on 11/1, we're looking at a landslide and both the vagaries of the EC and the Wilder effect will be irrelevant.

If Bush starts a war with Iran and it's unpopular (you'd hope so, but the "rally 'round the flag" effect is strong), I don't see how it cuts against Obama. No matter what anti-Iranian rhetoric Obama felt obliged to offer at AIPAC, McCain is much more identified with the "Bomb Iran, no problem a war won't solve" approach.

Posted by stuck in 200 | June 19, 2008 10:33 AM

...So if you want to keep tabs on the race over the summer months there's probably no need to bother with the state-by-state polls.

No way -- Back Off Man -- The only way you'll get me to give up my InfoPorn is when you pry it from my cold, dead fingers.

Or, because a hyperextenuated need for details about the minutae of human enterprise can only provide an illusion of control in a universe where, individually, we understand very little about anything real.

Want proof? TIME recently claimed that McCain is somehow "even" with Obama, possibly in a Salvia-induced fantasy dimension.

Posted by Jemand von Niemand | June 19, 2008 10:44 AM

If Virginia is still really in play on 11/1, we're looking at a landslide and both the vagaries of the EC and the Wilder effect will be irrelevant.

If Bush starts a war with Iran and it's unpopular (you'd hope so, but the "rally 'round the flag" effect is strong), I don't see how it cuts against Obama. No matter what anti-Iranian rhetoric Obama felt obliged to offer at AIPAC, McCain is much more identified with the "Bomb Iran, no problem a war won't solve" approach.

Posted by stuck in 200 | June 19, 2008 10:47 AM

Clinton had an advantage over Obama when this thing started. Look how that ended up.

Obama beat Clinton by outsmarting and outhustling her. He ran a shaper, smarter, more agile campaign. We aren't really seeing that from McCain.

Posted by croatoan | June 19, 2008 11:13 AM

We've seen a serious Obama bounce since he became Clinton-free, and there will be a significant lift from the Denver convention, as speeches count for a lot and the guy is an awesome speaker (probably some pundits will note that the '04 speech was better, nobody else will care).
A significant question is how much bounce can McCain get from his own convention. This could be really important. The party is not in a good mood, the down-ticket people are in a funk, and the guy himself has some liabilities as a speaker (hopefully for him there will be some great speechwriting and he'll be coached on effective delivery).
Some of the operatives seem to be in the just-add-more-balloons mood, which is not conducive to tapping everything that's new in this campaign.
Also, humor me here and allow me to insert some macabre speculation: did anybody take into account that McCain's support among the older demographics could mean less in November because this is the demographic most likely to have less headcount (actuary tables et al)? And same goes for younger people who may register and vote because of the fad effect Obama is generating? Isn't the combo of these two worth a few percentage points, say, in Florida?
(Sorry for this last part, did not mean to disrespect or to wish stuff upon people, just asking)

Posted by Bouncyman | June 19, 2008 11:43 AM

Check out the Election Model Monte Carlo simulation in which Obama wins 352 EV based on the latest state polls.

http://www.progressiveindependent.com/dc/dcboard.php?az=show_mesg&forum=120&topic_id=3913&mesg_id=3913

Posted by Richard | June 23, 2008 3:12 AM