Editore"s Note
Tilting at Windmills

Email Newsletter icon, E-mail Newsletter icon, Email List icon, E-mail List icon Sign up for Free News & Updates

July 21, 2008
By: Kevin Drum

WINNING....Via Dan Drezner, Clive Crook tells us that there's a simple metric to predict the winner of a presidential election:

Alan Abramowitz, a politics scholar at Emory University, has shown that summer head-to-head polls convey almost no information about the forthcoming election. (Subsequent head-to-head polls are not much better.) Instead, he has a simple "electoral barometer" that weighs together the approval rating of the incumbent president, the economy's economic growth rate and whether the president's party has controlled the White House for two terms (the "time for a change" factor). This laughably simple metric has correctly forecast the winner of the popular vote in 14 out of 15 postwar presidential elections.

Of course, it's worth pointing out that Abramowitz's "time for a change" factor alone correctly predicts 12 out of 15 postwar presidential elections, so those other two variables don't really have to do much heavy lifting.

In any case, Abramowitz's metric, which ranges from -100 to +100, gives John McCain a score of -60 this year, which means he's as doomed as any candidate ever. This suggests two thing: (a) Obama is going to win a very convincing victory, and (b) the only real way for the McCain campaign to give itself a chance is by going negative early and hard. I'll put money on both those things.

Kevin Drum 10:45 AM Permalink | Trackbacks | Comments (35)

Bookmark and Share
 
Comments

BUT, BUT, BUT...didn't McShame "pledge" not to have a negative campaign? Wouldn't that be a "flip-flop" that MIGHT get noticed...NAH, saw someone lead into a new ad for his campaign that they characterized as NEGATIVE!!! And, that's from our unbiased MSM...HA!

Posted by: Dancer on July 21, 2008 at 10:51 AM | PERMALINK

That's consistent with the model developed by Yale's Fair which uses institutional and economic factors. The GOP will just have to steal it again.

Posted by: doom on July 21, 2008 at 10:58 AM | PERMALINK

They're too inept even to go negative in a way that doesn't just invite snickering. Kiss McOld's wrinkly ass goodbye.

Posted by: Steve LaBonne on July 21, 2008 at 11:00 AM | PERMALINK

Oh, and doom- this one ain't gonna be close enough to steal.

Posted by: Steve LaBonne on July 21, 2008 at 11:02 AM | PERMALINK

Maybe I'm naive, but I'm not so sure going negative is as effective as it used to be. Lots of people are fed up with that kind of stupid approach to choosing a president. Look where it's gotten us in this century...

Posted by: cmac on July 21, 2008 at 11:17 AM | PERMALINK

You can read Abromowitz's views directly at Larry Sabato's Crystal Ball, specifically http://www.centerforpolitics.org/crystalball/article.php?id=AIA2008061901

Posted by: Walsh on July 21, 2008 at 11:24 AM | PERMALINK

the only real way for the McCain campaign to give itself a chance is by going negative early and hard. I'll put money on both those things.

The problem with going negative early is that McSame will shoot his ammo before the battle is really joined. Plus, it totally opens him up to a whithering counter attack, for his flip floppy DC politics of personal destruction.

Posted by: -ck- on July 21, 2008 at 11:30 AM | PERMALINK

The very idea that going negative can win otherwise lost elections is rebutted by the Abramowitz and Fair models, as well as Lichtman's Keys to the Presidency method -- i.e., elections that supposedly turned on effectively vicious campaigns ('88, '04) were actually environments favorable to the incumbent party...marginally in '04, solidly in '88. Conversely, a venomous campaign in '92 did nothing to stop Clinton's election. This has led too many to imagine Clinton as some political genius; in fact, it was just a perfect year to be the opposition candidate.

Though not as perfect as this year. No matter what system you use, this year ranks as one of the worst environments for an incumbent party in the past century -- 1920, '32, '52 and '80 are the only comparable years, and each of those ended in thumping out-party wins. Racism might shave a point or three off Obama's margin, but the circumstances are too starkly pro-change for that to do anything but turn a landslide into a solid victory -- whatever the polls say on any given day.

Posted by: demtom on July 21, 2008 at 11:30 AM | PERMALINK

Negative doesn't stick to some guys. Obama seems to be able to shrug it off, probably because most people don't take it seriously, for some reason. Reagan, unfortunately, was the same way. Maybe Obama won't be able to keep up his immunity to this stuff, but so far, so good.

Posted by: David in NY on July 21, 2008 at 11:30 AM | PERMALINK

I think this is basically correct, though there's one potential x-factor: we have a non-white candidate running. Since we've never had one as a major party nominee, there's no way to know whether or how much it'll affect the outcome. Still, if there was ever a year for the dems to try a non-white candidate, this is it.

Posted by: jimBOB on July 21, 2008 at 11:34 AM | PERMALINK

This suggests two thing: (a) Obama is going to win a very convincing victory, and (b) the only real way for the McCain campaign to give itself a chance is by going negative early and hard. I'll put money on both those things.

You forgot (c), McSame constantly lying about his positions and Obama's, and (d) the so-called "liberal media" letting him get away with it.

Of course, there's no sense in putting money on those, as they've been happening now for months.

Posted by: Gregory on July 21, 2008 at 11:36 AM | PERMALINK

These kinds of metrics are ridiculous. For any set of 15 non-random data points you'd have to be retarded to not be able to come up with a reasonable sounding metric that is in accord with 13-15 of the data points. You start with common sense and tweak it until the data fits. There's no predictive value there beyond the value of common sense--i.e. the "model" adds nothing. Abromowitz's metric is completely useless as a predictive tool for which elections in the future will depart from the results common sense would dictate.

All this really shows is that Abromowitz failed his statistics class.

Posted by: R Johnston on July 21, 2008 at 11:37 AM | PERMALINK


Ummm it's not just "The McCain Campaign" that will go negative Kevin, the entire shadow establishment/Republican/Corporatist enterprise will try to suppress voting, slant coverage, manufacture October surprise/s, and God knows what else etc.

There's a slimy hand on the scale ... so: We have to fight tooth and nail, no matter what the "simple electoral barometer" says!

Posted by: ♪ ♪ on July 21, 2008 at 11:38 AM | PERMALINK

Abramowitz's model is all well and good, but how many of the elections it predicts correctly have included a black Muslim elitist flip-flopping snob with a terrorist name and a wife who's ashamed to be an American?

Posted by: Lanco Yokel on July 21, 2008 at 11:40 AM | PERMALINK

Oh, and doom- this one ain't gonna be close enough to steal. - Steve LaBonne

Caveat: it will take a historic slaughter for this election to not be close enough to steal. I'm talking greater than 60:40. I suspect that anything less than 65:35 will still be close enough for the Republican'ts to steal enough votes here + there to throw the selection into the Supreme Court again, where they can be assured of a positive outcome.

Long, paranoid story short: it ain't over, not by a long shot.

Just because you're paranoid don't mean they're not after you,
-Z

Posted by: Zorro on July 21, 2008 at 11:45 AM | PERMALINK

Going negative is a given. Look for some wildly incendiary (yet false) accusations to be leveled by surrogates. Very personal stuff, malicious in the extreme. Pedophilia, adultry, current drug use, theft in office, domestic abuse. It won't have to bear any semblance to reality, the embarrassment of being so wrong of no concern to those putting it forth. The aim will be for just a scant few percentage points of the electorate to fall for it (and they will), the most gullible or prone to racism to change their votes. I can hear Wolf Blitzer now; "We're about to bring you a story you may not wish to hear, news that Barack Obama and a goat were sighted behind the barn........more after this break."

Posted by: steve duncan on July 21, 2008 at 11:49 AM | PERMALINK

Oh yes, a few October Surprises to expect, aside from OBL making a reappearance:

* A sudden increase in threat levels, similar to what we saw in Fall '04.

* A sudden drop in gas prices. After all, the oil industry can't afford for the best friends it's ever had to leave the White House.

* Significant numbers of US troops will return from Iraq. They'll be sent right back after the election, of course, win or lose.

Can anyone think of other dirty tricks?

-Z

Posted by: Zorro on July 21, 2008 at 11:51 AM | PERMALINK

metrics? Ain't that some kinda sissy Euro shit?

Use miles and tons, like a real American.

As in: we got to beat them by miles, because they got tons of dirty tricks loaded and ready to go.

Posted by: thersites on July 21, 2008 at 12:03 PM | PERMALINK

*

Posted by: mhr on July 21, 2008 at 12:25 PM | PERMALINK

If the X-factor of race were going to overthrow the historical trends, I'd expect it to show up in the polling on attitudes about whether the nation is on the wrong track on the theory that voters don't do cognitive dissonance. At a minimum, you'd expect Obama's favorable/unfavorable to be negative as a way of justifying not voting for the change everyone so obviously wants. But we aren't seeing either.

Crook is right that 1980 is the best comparison.
The race looks close now for the same reason it looked close in 1980 (until after the first debate): the challenger had to reassure the voters that he was sufficiently presidential. All Obama has to do is hold up in the debates.

Posted by: Maggie on July 21, 2008 at 12:48 PM | PERMALINK

Abramowitz's model is similar to every other forecasting model, with subtle differences. The fact is that the entire discipline of political science has been saying all year that the Democrats are going to win easily. But does the press ever listen to us? Of course not.

Posted by: arbitrista on July 21, 2008 at 1:01 PM | PERMALINK

The problem with every last one of these forecasting models is that all of them have been developed pretty recently. The number of Presidential elections any of them has predicted is in the single digits.

(The elections the models fit because they were the basis for the model clearly don't count.)

Small sample space, anyone?

Posted by: low-tech cyclist on July 21, 2008 at 1:14 PM | PERMALINK

I heard, forget where, an estimate that the Dems need a 10% margin to make it unstealable.

Posted by: doom on July 21, 2008 at 1:30 PM | PERMALINK

In any of those elections, those 14-15 elections that the Abramowitz's metric predicted, was there a black candidate you smoked pot and did a little blow, "maybe?" Just asking.

Posted by: The fake fake Al on July 21, 2008 at 1:41 PM | PERMALINK

The number of Presidential elections any of them has predicted is in the single digits.

No, not true. The Lichtman Keys have predicted not only all elections since they were created in 1980, but also retrospectively all elections going back to 1860. Really rather remarkable. Even more remarkable, the Keys system uses NO polling data, and makes its prediction up to a year in advance of the voting.

Posted by: jimBOB on July 21, 2008 at 1:43 PM | PERMALINK

Retrospectively doesn't mean anything; all it means is that they looked at past elections and tweaked the model until it fit them all. So as the other poster said, only single figures -- and given that at least one of those elections was stolen, and the actual *votes* would have gone the other way, it's clearly not exactly a perfect model.

Posted by: tavella on July 21, 2008 at 2:05 PM | PERMALINK

Tavella, Lichtman's system specifies it predicts the popular vote winner -- which is to say, it correctly had Gore in 2000 and Bush in '04, both by thin margins (5 negative keys, one short of defeat for the incumbents). So that system has been correct no matter what you think was stolen.

And 2008 is of a different magnitude. I count at least 9 negative keys (the intra-party contest key is cloudy; that one falling in addition would raise negatives to an unprecedented 10). I can agree that, given the small number of elections observed after the system was developed, you wouldn't want to bet the rent on an election with a closer break of factors. But this one is a laydown. Only Democratic paranoia or DC conventional wisdom is arguing otherwise.

Posted by: demtom on July 21, 2008 at 2:18 PM | PERMALINK

All Obama has to do is hold up in the debates.

Posted by: Maggie

I disagree. People will expect McCain to blunder. People also have over the top high expectations for Obama. Remember Obama's performances against Hilary weren't all that spectacular. So he is going to be declared the loser.

Remember both Gore and Kerry whumped Shrub in the debates - from a technical stand point. But Shrub was declared the winner because he didn't do as bad as people thought he would.

Posted by: optical weenie on July 21, 2008 at 2:58 PM | PERMALINK

OW's right; Shrub 'won' the debates against Gore + Kerry largely b/c people had such low expectations of him. In effect, their reaction was along the lines of "wow- he didn't drool on himself! Let's declare him the winner!"

Don't forget the 'with whom would you rather have a beer with test,' either. Of course, if you did have a beer with GWB, he'd probably wind up going on a bender + invade the wrong country.

Oh, never mind...

-Z

Posted by: Zorro on July 21, 2008 at 3:06 PM | PERMALINK

I think Maggie is arguing from a different perspective. Kerry WAS clearly more effective in the debates in '04 -- all polls showed that. But the metrics of the election argued (barely) for retention of Bush as president, so it wasn't the dispotive factor.

This time around it's more like 1980, with the public very much inclined to fire the incumbent, and the only question remaining, does the challenger meet a minimum standard? Reagan only had to not breathe fire or keel over to win that year; Obama -- far less scary than and as gifted rhetorically as Reagan -- will easily cross the bar this year.

What those of us arguing for these systems are saying, basically, is presidential elections are not jump balls based on electoral whim. They are considered judgments based on the preponderance of the evidence. But if you want to keep believing they're, instead, determined by "read my lips" or a DUI or "There you go again", I don't suppose threre's any way we can persuade you.

Posted by: demtom on July 21, 2008 at 3:42 PM | PERMALINK

Michael Barone makes a very plausible case that this is like Gerald Ford v Carter -- Carter at this point had a monster lead, yet nearly pissed it away. If Ford hadn't made some really large gaffes - Poland not ruled by Soviet Union, etc. - he would have pulled it out. Americans like to think they like change a lot more than they really like change - when the election comes around their basic conservative instincts kick in and the momentum shifts. That's why I keep saying we haven't heard the last of Jeremiah Wright - if Obama loses this election the images of him and his church will be the reason.

Posted by: loki the michief maker on July 21, 2008 at 4:10 PM | PERMALINK

Previous elections throw no light on race. Is America ready for a black president?

My money says is that Obama can win, with great fundamentals overwhelming the millions of voters who will finally hesitate and either pull the lever for McCain or stay home. On reflection they will, with some embarrassment, find thay just are not ready for a black president. This will be an especial problems with older voters who try but do not completely shed the attitudes of their upbringing.

This will leave Obama with a much smaller victory margin than would be enjoyed by a white man with similar talents. In the worst case it could tip the race to McCain.

If Obama does win, and his presidency is largely successful and popular, the barrier to another black candidate will largely disappear.

I think Obama's best path is clear:
1. Put unpresendented resources into getting blacks registered and to the polls.
2. Come out for class-based affirmative action, officially saying race-baseed preferences should be phased out.
3. (I know it seems off-subject) Go to the center on abortion, emphasizing abortion reduction and compromise on 3rd-term abortions.

The last is important because many of the same people who have race issues are socially conservative and being 'reasonable' on abortion is a good way to prove independence from NARAL and other new-fangled pressure groups who ain't got no morals.

Posted by: tomtom on July 21, 2008 at 4:28 PM | PERMALINK

All somebody has to do is speak the truth about BHO and his past speeches and associations. Rev. Wright, Ailes and Rezco are pretty good charecter references. Then his campaign partner has problem being proud of America. Throw in some of his total reversals like FISA. Mention that religious folk and 2nd Amendment advocates are bitter people. It ought to make for a pretty good show. I love it!

Posted by: Fat White Guy on July 21, 2008 at 6:14 PM | PERMALINK

McCain started going negative some time ago.....

Posted by: Diane on July 21, 2008 at 6:48 PM | PERMALINK

"The problem with every last one of these forecasting models is that all of them have been developed pretty recently. The number of Presidential elections any of them has predicted is in the single digits.

(The elections the models fit because they were the basis for the model clearly don't count.)

It's an entirely legitimate technique to develop a predictive model based on how well it would have predicted past events. The law of gravity will predict future events even though it was developed inductively from past experience.

The core insight in both the Abramovitz and Lichtman models is that presidential elections are fundamentally referenda on the incumbent administration, and not really contests between candidates.

Posted by: Robert Levine on July 21, 2008 at 8:21 PM | PERMALINK
Post a comment









Remember personal info?










 

 

Read Jonathan Rowe remembrance and articles
Email Newsletter icon, E-mail Newsletter icon, Email List icon, E-mail List icon Sign up for Free News & Updates

Advertise in WM



buy from Amazon and
support the Monthly