Who will win the 2008 Presidential Election? Do the math...

[calteacherguy]

Who Will Win the 2008 U.S. Presidential Election?

Eric Schulman
Alexandria, Virginia
Bio...http://members.bellatlantic.net/%7Evze3fs8i/me/index.html

Daniel Debowy
New York University School of Medicine, New York, New York

Abstract
Our 2003 algorithm for determining the winners of United States presidential elections correctly determined the winner of each of the 55 U.S. presidential elections between 1789 and 2004. We apply the algorithm to 44 Democratic and Republican candidates for the 2008 U.S. presidential election and find that the Democrats have nine tickets with electabilities greater than 150, whereas the Republicans have five tickets with electabilities greater than 150. The most electable ticket, with an electability of 264, has Republican former Governor Tommy G. Thompson as the presidential candidate and Senator Charles T. Hagel as the vice presidential candidate. The next most electable ticket, with an electability of 260, has Democratic retired General Wesley K. Clark as the presidential candidate and former Vice President Albert A. Gore Jr. as the vice presidential candidate.

<snip>

4. Discussion
As can be seen in Table 7 above, the most electable ticket has Republican former Governor Tommy G. Thompson as the presidential candidate and Senator Charles T. Hagel as the vice presidential candidate. The next most electable ticket has Democratic retired General Wesley K. Clark as the presidential candidate and former Vice President Albert A. Gore, Jr. as the vice presidential candidate.

We assume that major party primary voters are rational and will understand the empirical power of our algorithm. Democratic primary voters will therefore nominate Wesley K. Clark, Albert A. Gore Jr., William B. Richardson, or Christopher J. Dodd as their 2008 presidential candidate. This candidate, being rational, will choose Albert A. Gore Jr., Birch E. Bayh III, or Christopher J. Dodd as their vice presidential running mate. Republican primary voters will nominate Tommy G. Thompson, George E. Pataki, Michael D. Huckabee, or James S. Gilmore III as their 2008 presidential candidate. This candidate, being rational, will choose Charles T. Hagel or Richard B. Cheney as their vice presidential running mate (whether Mr. Cheney would accept a job that, as Vice President Nelson Rockefeller noted, "has no responsibility and no power" is beyond the scope of this paper).

Not everyone believes that major party primary voters are rational. Political prediction markets such as Intrade.com allow users to place bets on which candidates they think will win elections; their users show a remarkable confidence that major party primary voters will choose candidates with low electabilities. For example, on March 20, 2007, it cost Intrade users $45.30 to buy futures that would pay $100 if Hillary D. R. Clinton became the 2008 Democratic nominee for president. On that date, Rudolph W. L. Giuliani III futures cost $42.00, Barack H. Obama futures cost $30.00, John S. McCain III futures cost $20.10, and Williard M. Romney futures cost $16.00. Those five were the only candidates with Intrade presidential futures prices above $10 and their presidential electabilities range from a low of -110 to a high of 0. In contrast, Tommy G. Thompson (presidential electability of +154) and Wesley K Clark (presidential electability of +110) futures cost only 40 cents each. Figure 1 shows the electability vs. Intrade futures prices for 18 Democratic and 21 Republican candidates for president (solid circles) and 8 Democratic and 7 Republican candidates for vice president (open circles). The linear regression line shows that, in general, the higher a candidate's electability, the lower their Intrade price.

More...http://members.verizon.net/~vze3fs8i/air/pres2008.html

[TwilightGardener]

1. WHAT???? Gore/Clark, I can understand, but Charles T. Hagel?

I honestly had to think about who that was for a second, LOL! Never think of him as "Charles". Tommy Thompson only looks formidable on paper, because of his resume.

[napi21]

2. Algorithms can do a lot of things, but I sure don't see it having a clue

THIS time.

[calteacherguy]

4. Perhaps not, but it HAS predicted the past 55 results accurately...

Of course past performance is not guarantee of future results, and it assumes people will behave rationally...quite big assumptions.

Still, it's fun to ponder. :-)

[napi21]

8. Perhapse it's too soon? I don't know what data they used

but I have to believe the results change over time. The results they stated sound a bit more like their info was about a year old when lots of people thought Clark would be an early candidate, and I just can't explain the Thompson result at all.

[TwilightGardener]

3. And CHENEY as VP???? These folks are nutty.

[calteacherguy]

5. Either that or the algorithm falls apart when neo-con VPs are involved.

I mean this folks are certainly outLIERS in the American political spectrum.

:rofl:

[TwilightGardener]

6. Go over to Freeperville, they'd LOVE him as prez! I'm curious as

to why they divide these candidates up into Pres and VP categories, instead of just ranking them all together and putting the most likely winners together as tickets.

[calteacherguy]

7. "Go over to Freeperville."

No thanks!

[Czolgosz]

9. You know this is a joke, right?

[calteacherguy]

14. Do you disagree the algorithm has been accurate the past 55 elections? nt

[Czolgosz]

17. Are you not familiar with the Annals of Improbable Research? Even if you're unfamiliar, doesn't this

excerpt tip you off to the joke?

1. Introduction

Unlike the Redskins Algorithm (the incumbent political party wins the presidency of the United States if the Washington Redskins American football team wins the last home game before the election) or the Red Sox Algorithm (if the Boston Red Sox baseball team wins the World Series in a U.S. presidential election year then Woodrow Wilson is elected president), the Annals of Improbable Research U.S. Presidential Election Algorithm (Debowy and Schulman 2003) correctly predicted the outcome of the 2004 United States presidential election. In this paper we apply our proven algorithm to 44 Democratic and Republican candidates for the 2008 U.S. presidential election.

[calteacherguy]

21. Look...the algorithm has been accurate for all Presidential elections to date.

That's all I'm sayin'...

[Czolgosz]

25. It didn't "predict" those results as you suggested upthread. It is a comical illustration of how you

can misuse data ex post facto to pretend to show a trend.

I can offer you the following algorithm which also "predicts" that Obama, Hillary, Edwards, Dodd, Kucinich, Richardson, and Biden cannot win based on all past elections:

Electability score:

1. Plus 100 for being white and of non-Hispanic ethnic origin,
2. Plus 100 for being Protestant Christian or, if Catholic, from Massachusetts,
3. Plus 100 for being male,
4. Plus 100 for having a last name that begins either with (1) a consonant other than D, Q, S, X, Y, Z or (2) with an A

Everyone who has ever been elected president has scored 400. Hillary, Edwards, Dodd, Richardson, and Biden only score 300. Obama and Kucinich only score 200. So who's gonna win? GRAVEL!!! I could add various "minus" factors which would not penalize Gravel. That's what this study is doing. It's showing how data can be manipulated to reach nonsensical results by selectively picking the algorithmic factors:

Presidential Electability = 5*(years as President) + years as U.S. Representative + 11*(years as Governor),
+110 if the candidate has been a four-star general officer in the United States Armed Forces,
+110 if the candidate has been a college or university president or chancellor,
+110 if the candidate is the child of a U.S. Senator,
–110 if the candidate has been divorced,
–110 if the candidate has been a special prosecutor,
–110 if the candidate was the first adherent of a particular religion (e.g., Church of Jesus Christ of Latter-day Saints) to be a major-party candidate for President,
–110 if the candidate was an officer of a lobbying organization at the time of the election.

It's a joke.

[avrdream]

27. Hey, thanks for that.

My brain was telling me that this didn't make sense but I couldn't figure out a way to explain it and you just did it so simply. Are you a statistician?

[karynnj]

28. This is great

and is exactly what you say it is a joke, they overfit a small number of data points. It says something when you get no points for being a senator or VP and points for being the son of a senator.

Wouldn't Gore have beaten Bush though under their model. I think the only positive he has was his 7 or so years as Governor which would make him about 77. Gore has 110 as son of a Sentaor and however many years he was in the house. It shouldn't have been close!

[TwilightGardener]

10. Except for T. Thompson, though, Wes Clark, Gore, and Hagel

would be probably the most qualified and electable of their parties, in terms of experience and military service. Why ain't they runnin'?

[Orangepeel]

11. Tommy G. Thompson, George E. Pataki, Michael D. Huckabee, or James S. Gilmore III

:rofl:

Even less likely than Wesley K. Clark, Albert A. Gore Jr., William B. Richardson, or Christopher J. Dodd

[keopeli]

12. Correctly predicted Bush in 00 and 04? So it can predict that a fraud will occur and change results

This is non-sense. An accurate algorithm would have to have predicted a Gore win in 00 and a Kerry win in 04 to be valid, unless they factor in the criminal mind of the candidate. How do you say, 'our algorithm compiles all necessary data and variables to predict who will win the election?' What past data can be used to predict Bush would sue the U.S. and steal the presidency in 00? What math equation accounts for how Bush would challenge over 3 million under-votes in 04, manipulate the election machine software and that Kerry would not fight back?

Whatever math this is, it seems to have a little hocus-pocus in it also.

PS - Long, wordy explanations with lots of citations doesn't make a fact out of an assumption.

Peace

[TwilightGardener]

13. They clearly favor governors--that's why it's Bush over Gore and Kerry.

Edit to add--then why not Romney? Too Mormon? Makes you wonder what they consider electable.

[calteacherguy]

15. Methodology.

2. Methods
We determined the electability for president and vice president of the Democratic and Republican candidates for the 2008 U.S. presidential election using the following formulas:
Presidential Electability = 5*(years as President) + years as U.S. Representative + 11*(years as Governor),
+110 if the candidate has been a four-star general officer in the United States Armed Forces,
+110 if the candidate has been a college or university president or chancellor,
+110 if the candidate is the child of a U.S. Senator,
–110 if the candidate has been divorced,
–110 if the candidate has been a special prosecutor,
–110 if the candidate was the first adherent of a particular religion (e.g., Church of Jesus Christ of Latter-day Saints) to be a major-party candidate for President,
–110 if the candidate was an officer of a lobbying organization at the time of the election.

Vice Presidential Electability = 4*(years as Vice President) + years as U.S. Representative + years as Governor,
+110 if the candidate has been a corporate banker,
+110 if the candidate has been a college or university president or chancellor,
+110 if the candidate is the child of a U.S. Senator,
–110 if the candidate was the first adherent of a particular religion (e.g., Church of Jesus Christ of Latter-day Saints) to be a major-party candidate for Vice President,
–110 if the candidate was an officer of a lobbying organization at the time of the election.
Total Electability = Presidential Electability + Vice Presidential Electability.
Years in office is equal to the number of years the candidate served in a particular office, rounded up as long as the partial year service was one month or more, unless the candidate moved directly from one public office to another, in which case the office in which the candidate spent a larger fraction of their time during that year receives credit for the year. Years of service for offices were verified using the standard repository of all knowledge and wisdom (Wikipedia). Although the electorate doesn’t care one way or the other how long a candidate has served as a U.S. Senator, we included this information for completeness. In every presidential election between 1789 and 2004, the major party ticket with the highest electability was declared the winner and took office.

[sweetladybug]

16. This is bullcrap.

Surely no one believe this!

[calteacherguy]

18. I believe that the algorithm has been accurate for all Presidential elections to date.

Note that it is only a measure of "electibility," not who primary voters will necessarily nominate (because they don't all accept the algorithm and don't all behave rationally). And of course, 55 elections or not, past performance in no guarantee of future results....necesssarily.

:hi:

[Czolgosz]

19. It's not bullcrap. It's a joke. It's sorta like "The Onion" but for science majors.

[TwilightGardener]

20. It's hard to believe, and may be a joke, but they have interesting

methodology, and considering how much people seem to LOVE Clark and Gore on DU, I'd think it might actually make some sense to some people--at least on the Dem side. Tommy Thompson makes NO sense, but I could easily see Hagel as someone's VP.

[calteacherguy]

22. What's hard to believe?

That it's been accurate for all Presidential elections to date? It has...do the math.

:hi:

[TwilightGardener]

23. Hard to believe because none of the top four mentioned either

stand a chance (Thompson) or are even running. I guess they don't take that into account, they just go with the possibilities. Maybe Gore, Clark and Hagel will see this and jump in, LOL!

[tkmorris]

24. Fun with statistics, but basically meaningless

All they have done is start with the results and then search for a mathematical scoring system that correctly predicts the results. This is a surprisingly easy, if monotonous, thing to do. You start with rational concepts like a big bonus for an incumbent president, former governor, or 4-star generals. I think we can all agree that those things are advantages, and by themselves probably yield correct results in 2/3 of the elections. Now you start adding in minor, and less rational, rules that will reverse the result of the increasingly fewer elections your system got wrong. Bonus points for the son of a Senator, but none for an actual Senator. Representatives get a bonus, but college presidents or chancellors get a bigger one by far. And so on.

Believe it or not I am certain I could produce a mathematical system that would "predict" the results of all these presidential elections based merely on the letters in their names. Or their birthplaces. Or whatever. This one actually starts with some logic, but at a certain point gets pretty arbitrary and can't be trusted to predict FUTURE results very reliably.

[Alexander]

26. It will be the reverse, if anything - Hagel/Thompson and Gore/Clark.

Gore and Hagel wouldn't play second fiddle.

[JNelson6563]

29. OMG, beyond stupid

Clark/Gore??? :rofl: Yeah, I'm sure the guy who has already won the presidency would run in the second spot under a guy who's never won an election.

Talk about never-gonna-happen. I'll toss this in the especially-useless file.

Oy.

Julie