Yale professor predicts 57.5% for Bush (NY Times Magazine)

[tomfodw]

Ray C. Fair is a professor of econometrics at Yale. He says he has come up with a model of predicting presidential elections. In an interview with him to be published in tomorrow's New York Times Magazine, he predicts Bush will receive 57.5% of the popular vote. He says his model has been accurate in the past to within 2.5%. He has a book on the subject, Predicting Presidential Elections and Other Things (Stanford University Press, 2002).

I won't recapitulate the interview, which is kind of stupid (the interviewer, someone named Deborah Solomon, keeps trying to trick him into revealing a supposed anti-Kerry bias; he keeps insisting his model is free of any bias, although he says he supports Kerry). She never goes into his actual methodology or asks him to prove his assertion that he has predicted past elections so accurately.

Anyway, since what he is predicting is so different from all the polls (which he admits and says polls are "notoriously flaky this far ahead of the election"), I'm wondering what everyone else thinks of this.

Has anyone here read his book? If so, is his methodology convincing and reliable? Is anyone worried by his prediction?

Obviously, there's a danger of it becoming self-fulfilling. There's also the possibility that it could make the Republicans complacent and overconfident while spurring Kerry and his supporters on to greater effort.

[cynatnite]

1. Why Vote?

If anyone takes this jackass seriously they may not vote.

Polls don't mean shit in this election, IMO.

Cyn:)

[Cha]

2. I think Kerry is going to win and I have no idea

what the percentage will be. We'll talk in November and see what this guy has to say for himself when his prediction doesn't come true.

[TruthIsAll]

40. My model says it's 99%+ Kerry wins...and I have the numbers to back it up.

http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=132&topic_id=619614&mesg_id=619614

[Mick Knox]

3. links are always nice nt

[tomfodw]

9. Not up yet

Since it's tomorrow's NY Times Magazine, the Times Web site has not yet posted the article. Subscribers to home delivery of the Times get the magazine on Saturday. Sorry.

It's a very short interview, in any case, and as I said above, the interviewer never gets into the specifics of Fair's methodology (other than to bring up the claim in his book that, in her words, "economic growth and inflation are the only variables that matter in a presidential race"). All he says is that Iraq and other non-economic issues won't make much difference. He says, "If the equation is correctly specified, then the chances that Bush loses are very small."

As a historian, I tend to mistrust reductionist arguments, especially one like this that appears to smack almost of economic determinism. (Especially since economic growth in this case does not appear wildly pro-Bush, considering the stalled recovery.)

[jmowreader]

54. And that's how this prediction is flawed

The only sectors that are showing economic growth are war-related industries.

[geek tragedy]

4. Because it's a fact that academics know everything about

how the real world operates.

[Champion Jack]

23. HA HA HA HA HA HA !!!!!

That's the best one I've heard tonight

[Kierkegaard]

5. Anyone who would

push out statistics but not their methods of derivation is full of bovine excrement...

[tomfodw]

16. Blame the interviewer, not the professor

It's a stupid interview because the interviewer doesn't ask about his method. That's why I asked if anyone had read the book (in which, presumably, the professor explains his methodology).

[Kierkegaard]

19. I agree. I should have been more specific

as to whom I was attributing blame.

[lancdem]

6. There's no way Bush will get even close to 50 percent of the vote

Don't worry; there are all kinds of models predicting all kinds of results.

Bush got less than 50 percent of the vote last time, and it's ridiculous to think Bush has any chance of increasing that percentage by such a huge amount.

[In Truth We Trust]

18. EXACTLY!!!!

[realFedUp]

7. I'd also predict 57%...

for Kerry/Edwards.

Course that's just one person's opinion.

[DemocratSinceBirth]

8. He Bases His Analysis On Things Like GDP, Personal Income Growth,

Et Cetera.....


These determinative models don't always work... I think they predicted a Dukakis and Gore win... Wait Gore did win....


I don't think the professor has factored an unpopular war into his model...

[lancdem]

13. Or the fact that GDP isn't all that matters

Remember, the economy was growing strongly in 1994, but the Dems still got killed in the mid-term elections because people weren't yet feeling the positive effects of the expansion.

[DemocratSinceBirth]

15. Exactly.....

And the professor doesn't factor into his analysis that Bush is the first pres since Herbert Hoover to have negative job growth and that there is a good chance the equities markets will be loer than when Bush took office....

[fujiyama]

38. Hell

I think the economy was showing some upward signs when Bush Sr was in office but people weren't really feeling it -- people need jobs. Now his son has the country stuck in a war and people aren't happy about friends and relatives being killed for a reason that seems very shaky at best to many.

If Osama is captured, then Bush may get that sort of popular vote. Otherwise, there isn't a chance in hell. Even then, I see the highest % to be around 53-54. I don't see Bush winning anything more than 55%.

[Yupster]

55. Clinton was brilliant

For the whole campaign, Clinton and Gore kept blasting the "worst economy in the last 50 years."

By the time the press realized the numbers were actually moving up pretty strongly, it was too late to make that argument. When Bush tried, he was pushed down for being out of touch.

It worked because it was hammered in for months and months. Thanks James Carville for his laser beam focus. "It's the economy stupid."

[papau]

24. Under Bush, Personal Income Growth does not equal free cash growth

And GDP is counting as domestic too Much that us spent overseas.

I suspect when the economic numbers for today get cleaned up under Kerry, the formula will again work.

[Sparkly]

10. No, that was a Yale professor's prediction of Bush's final exam average

But it was misoverestimated.

[Guaranteed]

11. LOL, trust your gut on this one.

Sounds pretty ridiculous, doesn't it? It probably is.

[WI_DEM]

12. I've heard of his model before, but

this is an election unlike other elections. Even if Bush won (which he won't) he will never receive 57.5% when he got only 48% in 2000. The country is too divided and too many people want to kick Bush's ass out of the WH.

[kikiek]

14. Only a prediction based on economic factors which have fallen out

of Bush's favor again. Many unemployed in this country, and underemployed. Also read that in previous elections factors that led to the winning candidate were not included, and his prediction matched by fluke (5 for 6). Also many polls are not able to predict accurately because of the high number of 1st time previously unregistered voters. Also if Bush doesn't win by a landslide or loses this blows the validity of his methods to predict. That blows his theories already in my book.

[frylock]

17. i'll wait and see what coach stramm has to say

[tomfodw]

20. Fair predicted 58% in May

Someone posted an article about Fair to the Bush_Be_Gone mailing list (not sure where the article originally appeared; there was no citation or link). Fair predicted pretty much the same total for Bush back in May. The economic picture looks someone different now, I would think, but he's still predicting just about the same thing. Not knowing anything about his methodology, I have to say, it still looks a bit fishy to me.

One relevant excerpt:

"Fair thinks the most relevant risk is the potential gap between perception and reality about the economy. Fair says that his model's prediction of a Bush victory in 1992 was thwarted despite a recovering economy, because public perceptions of recovery lagged the reality.

"This time around polling data suggest that the same thing might be
happening, with voters' negative perceptions of the economy strikingly at variance with its true health. But Fair notes that the current economic recovery has already been underway longer than the one in 1992, and the next election is still six months away."


Like many academics, Fair seems quite wedded to his theory and reluctant to consider anything that might make it seem less than perfect.

Also, there are economists who dispute the "true health" of the economy and the recovery. Maybe the voters are not as stupid as he thinks, in that maybe their perceptions of the economy are more accurate than what economists want them to think.

[tomfodw]

25. In replying to my own post...

...I see that Fair is blaming one of his own inputs for his failure.

"Fair says that his model's prediction of a Bush victory in 1992 was thwarted despite a recovering economy, because public perceptions of recovery lagged the reality.

"This time around polling data suggest that the same thing might be happening, with voters' negative perceptions of the economy strikingly at variance with its true health. But Fair notes that the current economic recovery has already been underway longer than the one in 1992, and the next election is still six months away."

In other words, he was right in 1992, but the voters were too stupid to get it. And they're too stupid again this year.

[JohnnyRingo]

21. I've personally led 4 new voters to register ...

They were already motivated By Buxh politics, so there wasn't much arm twisting involved.

I'm sure this has been done by hundreds of thousands of other Democrats across the country.
Meanwhile, I don't think the Repubs are having such an easy time finding new registrants.


There are vastly more Dems than there are Repubs in the country, unfortunately they don't go to the polls. I think this year will be different.

Stick THAT in his model and see how it works.

[Doctor Smith]

22. I can develop a model which is 100% accurate

when applied to past elections, and can predict 30%, 50%, 80%, or any other number you like for Bush in the upcoming election.

Anyone who learned how to solve a system of simultaneous equations in high school can do the same.

[Yupster]

56. True

always easier when looking back.

[Radical Activist]

26. Yale professor found smoking crack in office

will be the headline on this guy next week.

[devinsgram]

27. This sounds like the guy

who did the astrology charts for Gore and * in 2000 and said it would be Gore because of their charts and also Sylvia Browne predicted Gore would win in 2000 and now she says on the Montel show that it will be *.

So now I know it will definetly be Kerry! There is never a sure thing, too many variables.

[rumguy]

28. He can't factor in *'s tarnished credibility - the prison scandals

the lack of WMD - continued violence in Iraq....

His electoral model doesn't mean anything this time around...

[tomfodw]

29. Look...

...has anybody here actually read his book? I'm trying to understand his methodology, not his motives. I understand that his model appears to be severely reductionist - econometric models basically have to be be. Rather than get snarky about the guy, how much validity is there to his assertion about Bush? Not what we want/hope it to be, but how much could there really be?

[rumguy]

30. All I know is that it is based solely on economic numbers

But does it factor in all those that have lost health insurance in the last three years?

Does it factor in the outsourcing of jobs?

Does it factor in war?

It doesn't do any of that...so I would say that of all years, this would be the year a model such as this, based entirely upon economic numbers, is moot...

This is an election year unlike any election year in recent memory. How could this idiot's model factor in something like the Abu 'Grub' prison scandal? It can't, it has no way of measuring the outrage people felt about that. And remember, there are still awful, fucked up shit regarding that prison scandal that we haven't even seen yet.

Iraq is still violent and perhaps getting worse....

If I were this guy I would shut my fucking piehole. His stupid egghead model is worthless this year.

[DemocratSinceBirth]

31. It Really Is Hard To See * Winning In A Landslide....

for all the reasons you cited but this race is going to be close.....


I will work and pray for the best (a Kerry victory) but prepare emotionally for the worst(a Bush victory.)

[Yupster]

57. However, when presidents get elected,

they usually win in landslides.

In all of US history, only two presidents who were reelected won fewer states the second time than the first.

The two were Woodrow Wilson and Bill Clinton.

Most reelections are more like Nixon, Reagan FDR, Lincoln.

My own baseless opinion is that Kerry wins a race similar to 1996, by about 8 %. Not a landslide, but comfortable.

[tomfodw]

32. This is the kind of emotional response...

...I'm trying to avoid in this discussion.

The guy says he factors in economic growth and inflation, and that's basically enough. Not having read his book, I can't figure out how he comes up with his figures. I was hoping someone who had read his book would be able to tell us more details.

Snarking at the guy or even at Bush does not tell us anything we don't already know. I want a serious discussion here.

Gasoline prices, for example, are part of inflation. They've gone up significantly this year, but the guy has not changed his tune since May.

There's a housing price bubble that may burst soon. Jobs are still lagging, and consumer confidence is waning. And the guy admits that his model cannot account for stupid voters stubbornly refusing to go along with what the professional economists are telling them about how wonderful the economy is (although plenty of economists are saying it's not really so wonderful at the moment - and voters can read, after all {well, some of them}, so maybe the voters aren't so stupid after all).

Those of us who want Bush out are in a kind of sticky position - do we really want the economy to tank just so we can get Dubya out of the White House? Do we really want the war in Iraq to go badly so that Kerry's chances will improve? I don't, and I hope nobody else here does. (Although, I have to admit that when I hear bad news, I do think, well, at least something good may eventually come from this.)

[DemocratSinceBirth]

33. It's A Backwards Loooking Model....

For instance he looks at indices that have worked in the past and presumes they will work in the future....


And it has failed in the past...

[rumguy]

34. The Prof. needs to shut his fuckin' cakehole

His stupid economic models are irrelevant this year.

[Zynx]

35. They also said that Gore would get over 54% last time.

[CaTeacher]

45. he probably would have--

if all the votes had of been counted!

[Toots]

36. Maybe he is the one progamming the voting machines...

I sure hope he is wrong..

[ItsMyParty]

37. I believe I heard this guy interviewed a month or so ago

or some "professor" with same type of predictions BUT there was a big BUT at the end. He shot off his mouth and then sort of dribbled at the end that it all depends though on the economy come November, and the Iraq War and the oil situtation, blah blah (you get the picture) and that stuff could change his model. I think it's one of those classic "opions are like assholes" situations......

[sundancekid]

39. if you look at his formula variables, his Nov 2002 assumptions DO NOT HOLD

principally because when he did his update in Nov. 2002, he had PRESET the variable "war" to zero (war=0); while the variable war=0 was true in 2002, the formula should be revised to show war=1 in 2004 ... HELLO! accurate use of variables is pretty basic to research formulae!! does not seem worth buying his "predictions" at all at this point given that a KEY variable (war)has a totally erroneous assumption and value in the equation. Perhaps he has updated his formula in the NYT article, but I doubt it, because he would have otherwise published such revision on the Yale site -- he usually does, and such revision is NOT there. The site allows you to "play predictor" BUT STILL with the "war" variable set to zero and not to 1.

It's pretty straight forward, you can buy into or debunk for yourself.


http://fairmodel.econ.yale.edu/RAYFAIR/PDF/2002DHTM.HTM

[tomfodw]

41. Thank you

This is the kind of response I was hoping for: someone who could look at Fair's model and see if it made sense. If he hasn't updated the inputs since 2002, then it's hard to see how he can expect people to take his prediction seriously.

The interviewer in the Times makes it sound as if there are only two variables: economic growth and inflation, and Fair's response does not indicate otherwise. He makes it sound as if he's trying to pare away emotion and other noise and reduce the complexity to a very simple (simplistic, in my mind) set of factors that, despite what people may think, actually do lead to accurate predictions.

Is he actually using the model on the Web site you link to? Or is he using some other model he hasn't bothered to put on on the Internet.

[oasis]

42. Teresa has a suggestion for Prof. Fair on what he can do with his "model"

:evilgrin:

[liberalpragmatist]

43. The Same Models Predicted a Massive Al Gore Win

I remember reading an article where a bunch of political scientists and academics used complex models to predict the outcome.

All said Gore. And all said Gore in a landslide - one by over 60%.

I'm not going to put to much faith into these. They can be fun. But the only results that matter are the ones on Nov. 2.

[CaTeacher]

44. I have heard this professor speak before

on statistical models--I remember mention of Bayessian Updating, Monte Carlo, and boot-strapping.

I don't put lots of faith in any of these methods....but I would like to flip through the book--statistical modeling is quite fascinating.

[Julien Sorel]

49. No they didn't.

Fair's model predicted Gore would get 50.8% of the popular vote, and he actually got 50.4%.

[liberalpragmatist]

53. Maybe not Fair's Model

But I do recall there being a study that showed Gore receiving something like 58% of the two-party vote. I'll try to find the article.

My point is these kinds of models treat history like a science. While there can be trends, each election is the product of it's own unique circumstances.

[enigmatic]

46. There's NO WAY W gets 57 percent in November..

No way; maybe if the Democrats had nominated William Hung...

[ProfLefty]

47. Lets Make Sure...

That we prove this nimrod wrong.

[CaTeacher]

48. I agree with you, Professor!

[Tom Kitten]

50. ???????

Bush came out of Yale, what a miserable failure he turned out to be! Methinks they be busy covering their sore asses!

What exactly is the reason ballyhooing Ivy League colleges as being "elite" anyways?

My sister, conservative Christian Republican as she is (and I love her very much even though we are 180 degrees apart politically, isn't that just awful) had a fellowship at Yale and said that was the most decayed college she had been to...

I call these peoples bluffs...they are convenient for the press, mostly East Coast based who have few clues about real research so they depend on tradition not thinking that 80% of the nation lies west of their time zone...where there happens to bea few thinking people also!

[tomfodw]

51. Umm...

...Kerry went to Yale, too.

Why the assumption that Ray Fair is some kind of stooge or idiot or evil person with a hidden agenda? Just because he says something we don't like and don't want to believe. If he has a workable model, we need to know why it is predicting what it predicts.

I didn't want to turn DU into a slanging match (which of course it never is anyway <g>), I wanted to discuss this guy's premises and conclusions and see if they had any validity. His 2000 prediction appears to be quite accurate (at least with regard to the raw vote totals - no way his model could have predicted the Supreme Court's criminality).

I can understand why he's doing what he's doing. The purpose of science is not to make predictions; it's to understand what is happening in the world by modeling factors that represent the world. Predictions are used to validate (or invalidate) the model, which itself is used to figure out which factors are important and which aren't. (I know most of you reading this don't need my little explanation.) Fair is saying, from what little I understand of his model, that only economic growth and inflation make a real difference and that everything else that people think matters really doesn't. That's the very purpose of science, to cut through the conventional wisdom and deliver real, usable knowledge. I'm not saying he's right (I'm not saying he's wrong, either).

I think he's wrong, but that's more a way of saying I hope he's wrong. In any case, making ad hominem attacks on him or ridiculing his motivations is completely irrelevant to the main question, the reason I started this thread in the first place: is he right? Does his model accurately reduce the complexity of the real world to an understandable simplicity? If he's right, all the snarking in the world won't change a thing.

[tomfodw]

52. By the way, here's a link to the article

http://www.nytimes.com/2004/08/15/magazine/15QUESTIONS.html


(It requires a free registration to the NY Times site to read articles there. Worth it in my opinion.)