INDC Journal

« Bush Up in New Poll | Main | Missing Explosives Update »

October 27, 2004
"Stanford Predicts: The 2004 Presidential Election" (Actualisé)

Posted by Bill

A team made up of Stanford Statistics students and supervising professors is running a predictive election model, and Bush is currently up ... BIG:

Kerry 21.0%
Bush 76.4%

The numbers above represent the probabilities that either candidate wins enough votes on the Electoral College to be elected President, as of the latest available polls.

Why their analysis is different:

Our analysis attempts to make finer distinction from poll data, apart from declaring a state Red, Blue, or Swing. It proceeds in two steps:

Step 1: compute the probability that a given candidate wins a given state, given the latest polls for that state, rather than categorizing the state as Red, Blue, or Swing.

Step 2: roll up the probabilities thus computed for each of the 50 states plus D.C. to infer the nationwide probability that a given candidate wins the majority of the electoral colleges (270 or more votes in the 2004 election).

Interesting. Check it out.

UPDATE: No, I don't put much stock in fancy statistics, either; I rely on yard signs. This one was spotted in Powhatan County, Virginia by reader Ned May:

notjaques2.jpg

Je suis d'accord! Chirac est un ver qui mange du fromage!

UPDATE: Princeton however, says "non!"

Predicted median with undecideds: Kerry 309 EV, Bush 229 EV (probability map)

Median outcome, decided voters only: Kerry 264 EV, Bush 274 EV (probability map) (Trends to 10/12)

Zut!

But wait! Is the Princeton analysis unbiased? Check out what the good professor says on his site:

I do not take donations. If you would like to express your support, you are welcome to do so politically through ActBlue. For balance, Republicans may donate through the NRSC.

"For balance." Furthermore:

Dear fellow Kos readers,

I am undertaking a meta-analysis of state polling data to calculate a current snapshot of the probable range of election outcomes. Like most of you, I have a strong bias about how I want the presidential election to turn out. However, I wanted a measure that did not have that bias. Read my preliminary findings here.

Heh. A good scientist never tips his hand, le professeur.

Meeeerd-euh! Je n'y crois pas à ces conneries!

Si les cerveaux étaient de l'essence, ils ne seraient pas suffisants pour faire tourner un kart de fourmis à l'intérieur d'un beignet!

(Via commenter StatSam)

Posted by Bill at October 27, 2004 12:01 AM | TrackBack (5)

Comments

Well, statistics are funny things, the way I read it, someone other that Bush or Kerry has a 2.6% chance of being elected.

They have this single most glaring error. They are only as accurate as the weighting given to Democrats and Republicans are in the polls they are using.

If the internals are available for a given poll, and I agree with the weightings, I will give a small credence to a particular poll, but since they aren't even telling us which polls they are using (which some other very well known electoral map consolidators are doing), I'm afraid I have to give this one a complete and total pass.

I like their answer though :)

Posted by: Thad O at October 26, 2004 11:41 PM

I want it to be true, but like Thad I'd like to see what polling data they're using.

Posted by: Dean Esmay at October 26, 2004 11:47 PM

Love the yard sign, and I agree with it completely.

Posted by: Just Me at October 27, 2004 12:07 AM

Great yard sign.

It's odd, but around my neck of the woods (where Bush will carry 70% or more of the votes) there are very few yard signs about national candidates. There are a ton of signs for locals, particularly Sheriff, which slighty alarms me, I mean, what if you put up a sign, and the other guy wins... (just kidding)

Posted by: Thad O at October 27, 2004 12:15 AM

Actually, that's fromage mangeant le singe de reddition

Posted by: joe shropshire at October 27, 2004 12:18 AM

Google language tools says that would be: "Cheese eating the rendering monkey" is that what it was supposed to be?

Posted by: Thad O at October 27, 2004 12:21 AM

The missing 2.6% doesn't necessarily go to a 3rd party; it could be part of an error term or model uncertainty.

Posted by: Joe R. the Unabrewer at October 27, 2004 12:22 AM

Chirac is a cheese eating the worm?

Posted by: Spoons at October 27, 2004 12:32 AM

Er, doesn't that say (roughly) Chirac is a worm-eating cheese?

Chirac est un ver qui mange fromage.

(Something like that. Too lazy to babelfish.)

Posted by: Angie Schultz at October 27, 2004 12:33 AM

Is that better?

Posted by: Bill from INDC at October 27, 2004 12:34 AM

Not "qui mange fromage" but "qui mange du fromage".

And "je consens" is not really at his place here (not false either).
"Je suis daccord" (= "i agree") is much much better.

"Je suis daccord! Chirac est un ver qui mange du fromage"
ou
"Chirac est un ver à fromage"

;-)

Posted by: frT at October 27, 2004 12:50 AM

Hi all,

This is off topic, but I figured you were the people to turn to for information. Anyone know anything about PIPA (program on international policy attitudes). A liberal friend of mine sent me a link to their homepage which has a unique election article www.pipa.org. I'm tryin to dig up some dirt on PIPA--just the name sounds like it might be some LLL group. Thanks!

Posted by: LuvyaDubya at October 27, 2004 12:58 AM

Digging up dirt? What do we look like? CBS News?

Posted by: Textbook Stupidity at October 27, 2004 01:36 AM

The proper spelling of the French term for agreement is d'accord. It usually stands by itself when used as an interjection.

Posted by: Matthew at October 27, 2004 01:59 AM

Text...

Excuse me. I should have rephrased "digging for dirt." I meant to say exposing the truth, in an un-See B S way.

Posted by: LuvyaDubya at October 27, 2004 02:06 AM

Bill...

We need to jump on this Hassan Nemazee story. The public must know.


http://www.worldnetdaily.com/news/article.asp?ARTICLE_ID=41077

Posted by: october at October 27, 2004 02:30 AM

Let's not forget the Weekly Reader: Bush-66% and Kerry-33%. Although not scientific at all, it has predicted every winner since 1956! I like the numbers. Two other nationwide school polls resulted in a Bush win.

Posted by: Pat at October 27, 2004 02:30 AM

The weekly reader one doesn't surprise me.

When I recently subbed for 4th grade, we read an article about campaigns and the election proccess. More than half of the class amused themselves by coloring mustaches and horns on the picture of Kerry.

Posted by: Just Me at October 27, 2004 07:33 AM

Yes Matthew you of course are right, it's "d'accord".

thanks

Posted by: FrT at October 27, 2004 08:12 AM

Just what are they smokin' at Stanford these days? I'll see their absurd "76.4%" chance of winning and raise them the following four sites. Read 'em and weep:

http://www.geocities.com/electionmodel/

http://synapse.princeton.edu/~sam/pollcalc.html

http://www.spacerad.com/electoral/

http://www.dalythoughts.com/ecb.htm

Posted by: StatSam at October 27, 2004 08:32 AM

Like Thad O I find the absence of yard signs interesting. As I posted here in my nominally solidly Democratic neighborhood in Chicago there are no yard signs and no bumper stickers. It wasn't like this four years ago.

Posted by: Dave Schuler at October 27, 2004 09:27 AM

Dave, around these parts, Philly, where Bush will get whomped, I have noticed that he is pulling far ahead in the bumper sticker race. Last night when I went to Bed, Bath, & Beyond (laugh if you must--I bought a blender, but it was Black & Decker, so I retain my masculinity--and, I needed it to make beer) and saw about four or five Bush bumper stickers and zero for Kerry.

As for the yard sign race, that's very different for some reason. Perhaps it has something to do with the Kerry campaign handing them out all over town lately.

Posted by: Nathan Hamm at October 27, 2004 09:32 AM

The Princeton Study links to the Daily Kos where the author states:

Mon Jul 19th, 2004 at 21:37:26 GMT

Dear fellow Kos readers,
I am undertaking a meta-analysis of state polling data to calculate a current snapshot of the probable range of election outcomes. Like most of you, I have a strong bias about how I want the presidential election to turn out. However, I wanted a measure that did not have that bias. Read my preliminary findings here.

Posted by: The Drill SGT at October 27, 2004 09:59 AM

"When I recently subbed for 4th grade, we read an article about campaigns and the election proccess. More than half of the class amused themselves by coloring mustaches and horns on the picture of Kerry."

Hahahahaha. Awesome, best laugh I've had all morning. For what it's worth, I remember doing that to Dukakis in the 3rd grade.

Yes, I am a young'un.

Posted by: Textbook Stupidity at October 27, 2004 11:33 AM

All this French is giving me a headache. I don't need a whole language to surrender should I ever lose my spine. Voting for the democrats will be sufficient enough.

Posted by: KBiel at October 27, 2004 01:31 PM

Posted by: pupputz at October 30, 2004 02:00 AM