Balkinization  

Saturday, November 01, 2008

The Economy, Stupid

JB

Yale Economist Ray Fair's Presidential Prediction Model offers a very simple prediction method based on real per capita GDP and measures of inflation, along with dummy variables for party incumbency and presidential incumbency. His October 30th prediction, based on the latest economic data, predicts that Obama will will 51.91 of the vote and McCain will win 48.09 of the vote, for approximately a four point spread. Fair notes that his model may actually overpredict the Republican vote because "the economy since the end of the third quarter appears to have gotten much worse. People may perceive the economy to be worse than the economic values through the third quarter indicate."

Meanwhile, Larry Bartels notes in a recent paper that the electorate tends to focus on the economic news right around the time of the election; if things are going bad, they tend to punish the incumbent party. In addition, the electorate is somewhat myopic, focusing only in the most recent economic news, and discounting what happened in the first three years of a president's term. Neither Fair's model nor Bartel's account of voter response to economic news looks good for McCain.

What is perhaps more important is that if Fair's and Bartel's accounts are largely correct, then most of the issues in the campaign-- other than the economy itself-- have been largely peripheral to the outcome. Neither Sarah Palin, nor William Ayres, nor Obama's birth certificate, nor the number of houses McCain owns, or even Joe the Plumber has made much difference to the outcome of this election.


Comments:

Jack:

The Nickelodeon vote by the kids has a better predictive track record than does Fair, unless I missed the Reagan-like Bush landslide in 2004 and Gore's comfortable majority victory in 2000.

There has never been an election like this one in the history of the Republic and most predictions are pure speculation.

Based on the current readings of the most historically reliable likely voter tracking polls, there are four possibilities:

A) Obama is winning by a couple points if the polls are correct that there will be a surge of new Dem voters for Obama AND there is NO Bradley effect.

B) Obama is leading within the margin of error if the polls are correct that there will be a surge of new Dem voters for Obama AND there is a Bradley effect similar to 2006 when the undecided broke 2:1 or more for the white candidate in biracial races.

C) The race is in a dead heat if the Dems, GOP and Indis turn out in the same proportions they have in prior Dem heavy elections like 1996 and 2000 AND there is NO Bradley effect.

D) McCain is leading within the margin of error if the Dems, GOP and Indis turn out in the same proportions they have in prior Dem heavy elections like 1996 and 2000 AND there is a Bradley effect similar to 2006 when the undecided broke 2:1 or more for the white candidate in biracial races.

Early voting has not shown anything like a Dem+7 advantage for Obama. Indeed, the much vaunted youth vote is a no show as usual while there is the expected surge in African American voting for Obama in the South.

Although we will not know the existence of extent of a Bradley Effect until the actual ballots are counted and compared to the polls, we do have evidence that there will be such an effect in the Obama v. McCain contest:

1) Four of the five 2006 statewide biracial races had the undecided break for the white candidate at a 4:1 ratio and over 2:1 when the Tennessee senate race outlier is factored in.

2) In the 2008 Dem primaries, there was a significant Bradley Effect in favor of Clinton in states with lower African American populations and a reverse Bradley effect in the South for Obama. The net was an advantage for Clinton and presumably McCain in 28 primary states.

3) The "undecided" are conservative churchgoing Reagan/Clinton Dems and Indis who disproportionately love Palin, distrust Obama and historically vote GOP for President.

4) Zogby's Friday poll has these Reagan/Clinton Dems breaking hard for McCain and giving him a small 1 point lead.

4) Obama is campaigning in states filled with these "undecided" Reagan/Clinton Dems warning his supporters not to get complacent even though the Dem media polls claim he will win in a landslide. Obama's internal polling is telling him something else.

I suspect that this race will be far closer than the Dems anticipate and the outcome may not be known until Wednesday.
 

Shorter Bart: If all the models of voter participation that all the professional pollsters are using are dramatically wrong, McCain would be within the margin of error, albeit usually still losing. Then let's hope for racism in the polling both.

A few facts. First, while Republicans did get all excited last night when Drudge leaked that a ONE DAY of a three-day Zobgy poll with a small sample had McCain up 1, let's remember taht Zogby's three-day tracker shows Obama up 5.

Further, check out two other polls so far this morning: Rasmussen, O +5, a gain of 1 point since yesterday; and Dieago-Hotline, 0 + 7, unchanged from yesterday.

Or to quote the guy at pollster.com: "no good news for those looking for a late shift to McCain." The RCP average (hardly a lefty site) has Obama up 6.4, and slated to win between 311 and 353 EVs.

As to early voting, Obama doesn't need the "youth vote." Among other things, he's getting the African American vote, and others. For example, today's stats from early voting in NC:
-2,350,000 voted (66% of total votes cast in 2004); AA at 26.3%(6-7 pts above models in pollsters internals); turnout 52% Dem, 30% Rep, 18% Ind.; women 56% which is 4% above pollster models. Early voting is also looking good for Obama in Colorado and other states.

Remember, this is a race in which McCain would have to sweep ALL of about 6-8 "battleground" states, several of which show Obama with significant leads. For example, at the moment, the Pollster.com average gives Obama a 6.3 percentage point lead in Ohio. Even if McCain were to gain 6.5 percentage points on Obama in every state, thus carrying Ohio, he’d still be losing in all the Kerry states plus Iowa, New Mexico, Colorado, and Virginia and thus lose the election.

McCain can, of course, still win. But he would need something to change dramatically in the next three days.

Or maybe Bart understands turnout models better than all the professional pollsters, including those with loyalties to the Republican Party. Or maybe not.
 

P.S. to Bart. Today's Gallup is not encouraging for McCain. Indeed, it shows movement in favor of Obama:

Gallup 11/01:

RV
O52
M41

LV (Extended)
O52
M41 (-1)

LV (Traditional)
O52 (+1)
M42 (-1)

Oh yeah, and I spelled "booth" and "that" incorrectly in my original post. More coffee, more polls!
 

jslater:

The polls vary between Obama +13 and Obama +4. These "professionals" are far far outside their own margins of error apart from one another. By definition, around 3/4 of them will be wrong.

This variance is largely caused by the enormous ahistorical spread in the partisan weightings used by these polls, which are between Dem+6 and Dem +12. The Dems have not voted in these numbers since LBJ. The early voting does not reflect Dem+6 and Dem +12.

If you honestly believe that Obama is winning by ten points as claimed by Gallup, I will wager $1000 here and now that that Obama will not win the vote by ten points. If you are unwilling to put your money where your mouth is, you have proven my points and please do not waste our time posting this drivel in the future.
 

P.P.S.

Also, McCain seems to be gambling everything on flipping PA, a gamble that seems like at best an extreme longshot. He has basically NEVER led in any PA polls, and many polls have had McCain down double-digits.

Even though a few recent polls have McCain within single digits (8 in this morning's Morning Call), PA is a pretty blue state. It has gone blue in the last four elections, and I believe I'm correct in saying the percentage Dems have won in PA has outpaced the percentage Dems have won nationally in ever Prez election since 1948.

So if you're into historical trends, PA looks good for Dems.
 

The default in Bart's list iist is of course A: the polls are broadly correct, nothing changes in the next two days, and Obama wins easily with around 350 electoral votes and a clear majority of the popular vote.

What Bart doesn't want to think about is outcome E: there is no Bradley effect, but there is a 2-point cellphone effect among young voters, and a further 2-point GOTV effect from Obama's very superior ground game. Why, bart, do you think there are disgracefully long queues for early voting al over the South?

The result here is a landslide: Obama carries most of the supposedly tossup states, North Carolina, Indiana, Missouri, Nevada, and gets 380+ electoral votes. The Dems near 60 senate seats. The GOP suffers a historic disaster, a Cannae, and faces a near future of internecine bloodletting as cultural and economic conservatives battle it out for a chance to be defeated in 2012.

I think D is likelier than E; but I made a bet on Intrade on Obama carrying North Carolina when the contract stood at 20 - now it's 65. Go Tarheels!
 

Bart:

"Drivel"? My, aren't you testy -- my guess is your mood darkened after seeing all the other polls today that didn't confirm the tiny morsel of encouragement Drudge spoon-fed you from Zogby.

I was just posting what the current polls actually say. Again, perhaps all the professional numbers are "drivel" and, like Rove famously but incorrectly claimed in 2006, you've "got the math." But I don't think so.

Also, I never said Obama will win by 10. I said he's ahead outside the margin of error, and there's no discernable movement towards McCain. Which means it's very likely that Obama will win. If he wins by 4, well, that's still better than any non-incumbent Prez winner has done in many decades.
 

jslater said...

Also, McCain seems to be gambling everything on flipping PA, a gamble that seems like at best an extreme longshot. He has basically NEVER led in any PA polls, and many polls have had McCain down double-digits.

Rasmussen reports:

The latest Rasmussen Reports telephone survey of voters in the state shows Obama with 51% of the vote while McCain picks up 47%. That four-point advantage for Obama is down from a seven-point margin earlier in the week [four days ago] and a 13-point advantage for Obama earlier in the month.

Just 75% of Pennsylvania Democrats now support their party’s nominee, down from 86% in the previous survey. Obama is doing a bit better among unaffiliated voters while Republican support for McCain remains steady.


The Reagan/Clinton Dems are breaking for McCain here. At this rate of collapse, the race in PA will be tied on Tuesday.

If the undecided keep breaking to McCain at this rate, he wins PA.

If McCain wins PA. he wins the election.

This is why both campaigns are campaigning in PA until the end.
 

Bart's delusional, we all know that. But Fair's model doesn't look so good. In this year, it's underpredicting Obama's margin. And about this:

Neither Sarah Palin, nor William Ayres, nor Obama's birth certificate, nor the number of houses McCain owns, or even Joe the Plumber has made much difference to the outcome of this election.

It's true that without Palin, McCain would still be losing. But he might still have an outside shot. She's clearly set him back a few points. In a closer election, she would've been decisive.
 

Instead of Bradley, how about we consider The Cheney Effect? Republicans so cowed by terrorist talk and the authoritarianism of the past eight years that they will align themselves publicly with the Republicans' nativist strategy and poll straight along the GOP card, but actually vote for Obama because they know he's the better choice (and/or that McCain/Palin are the wrong choice). I'm pretty sure my mother is one of these.
 

My wife works in the business, and I trust she is smarter than Zogby, Fox and the like. And she thinks Obama's actual lead is roughly four points. So there are a couple conclusions:
- Obama is oretty certain to win. Four points is a solid margin.
- The Troll is smarter than he seems: his bet is *not* for a McCain win, but an Obama win under four points -- a fifty-fifty bet

To the troll: I'll give even money on a McCain win. But maker it $100,000 -- and email soon. We only have a couple days to set up the formalities.

To the rest: a four-point Obama win remains impressive. A black man, in this country? Four points might as well be forty.
 

wcw: As Chris Rock said, "George W. Bush is so bad he's made it hard for a white man to get elected president."

And Bart's just talking about the spread. He's not capable of making a real bet here.
 

Bart: Re Pa, first, Obama remains over 50. It's a big mistake to think that McCain going from 8-4 in one poll (and this is his best number) -- say, 51-43 to 51-47 -- is anywhere near as easy as McCain getting a majority (or plurality). McCain pretty clearly has a ceiling, nationally and in key states, of around 46-47.

Second, actually, Obama has several paths of winning while losing PA. I don't think he'll need them, but look at the states he's currently leading and you can see how it's done. It starts with Kerry states, NH, NM, CO, and VA. . . well, you can do the math beyond that.

Finally, here's another poll from today for ya: ABC tracking, Obama 53 McCain 44 -- same as yesterday.

Basically, no movement for McCain today. And he needs a lot of movement.

Of course there's always Sunday, Monday, and maybe Tuesday morning.

Except for the large number of early voters who in many key states are skewing Dem.
 

One ought to be ridiculed for offering to wager money to prove a point on a blog discussion: there is no reasonable way of taking the bet.

However, you can make bets legally (not for $1000, unfortunately, only up to $500, I think) on the Iowa prediction market.

So, anyone who really is confident that his or her candidate will win can place bets there. If you actually think McCain has a 50% chance of winning and want to buy some bets, there will be some very happy traders willing to take you up on the offer, since McCain is trading at around 16%. Which is probably too high . . .
 

"Obama will will 51.91 of the vote and McCain will win 48.09 of the vote"

FiveThrityEight.com at the moment is predicting 52.2 for Obama and 46.6 for McCain. Remarkable close to Fair's prediction.
 

"Bart"'s drunk the Kool-Aid:

If McCain wins PA. he wins the election.

ROFLMAO. Aside from not being true, McInsane is not going to win PA. Instead he'll lose a bunch of states that Dubya won.

And talk about moving the goal posts: "Bart"'s betting that McInsane is going to lose by less that 10%....

Now that's a sentiment sure to get the party faithful out in droves to participate in the carnage. After all, everyone likes to be a winner, and there's a strong disincentive to vote for a lost cause; no one likes to be part of a losing team; better to say "that's not my party; they deserted me when they picked those eedjits"....

Suck it up, "Bart".

Cheers,
 

Post a Comment

Older Posts
Newer Posts
Home