Election Predictions methods

[Varaj]

I don't want just polls but more complex prediction methods. Meta poll analysis, historical methods, etc.
http://en.wikipedia.org/wiki/PollyVote
http://www.forecastingprinciples.com/PollyVote/

I'm going to dig up some of the forecasting methods that use historical studies as well.

[Varaj]

http://en.wikipedia.org/wiki/The_Keys_to_the_White_House

The Keys are statements that favor the re election of the incumbent party. When five or fewer statements are false, the incumbent party wins. When six or more are false, the challenging party wins.

1. Party Mandate: After the midterm elections, the incumbent party holds more seats in the U.S. House of Representatives than after the previous midterm elections. (False)
2. Contest: There is no serious contest for the incumbent party nomination. (True)
3. Incumbency: The incumbent party candidate is the sitting president. (False)
4. Third party: There is no significant third party or independent campaign. (True)
5. Short term economy: The economy is not in recession during the election campaign. (Uncertain)
6. Long term economy: Real per capita economic growth during the term equals or exceeds mean growth during the previous two terms. (False)
7. Policy change: The incumbent administration effects major changes in national policy. (False)
8. Social unrest: There is no sustained social unrest during the term. (True)
9. Scandal: The incumbent administration is untainted by major scandal. (True)
10. Foreign/military failure: The incumbent administration suffers no major failure in foreign or military affairs. (False)
11. Foreign/military success: The incumbent administration achieves a major success in foreign or military affairs. (False)
12. Incumbent charisma: The incumbent party candidate is charismatic or a national hero. (False)
13. Challenger charisma: The challenging party candidate is not charismatic or a national hero. (Uncertain)

KEYS TRUE: 4
KEYS FALSE: 7
KEYS UNCERTAIN: 2
KEYS PREDICTION: INCUMBENT REPUBLICANS LOSE

[Varaj]

Need to track down all 6 models mentioned here.
http://www.eurekalert.org/pub_releases/2008-06/uab-efp062008.php

[Varaj]

Interested in seeing if James Campbell has released his yet.

http://www.britannica.com/blogs/2008/08/barack-obama-will-win-the-presidency-political-scientists-forecast/

In a lively debate at the American Political Science Association convention in Boston today, an esteemed group of political scientists gathered to forecast the 2008 presidential election. The consensus: advantage Barack Obama.

Using various forecasting models with variables that included GDP growth, presidential approval, fiscal policy, the leading economic indicators (LEI), and primary results in New Hampshire as predictors of the two-party vote share, the various political scientists agreed that Obama was highly likely to win the presidency–with the exception of James Campbell, who blogs for Britannica; Campbell, whose prediction won’t be released until next week, since his model relies on the Gallup Labor Day polls (and it’s not Labor Day yet), believes it will be an extremely close election, with a slight advantage to John McCain.

The run-down of the predictions and models:

* Tom Holbrook (University of Wisconsin-Milwaukee): Model focuses on personal economic conditions and presidential approval. Two-party share of the vote: Obama 55.5%. Obama has a 92% chance of winning.
* Chris Wlezien (Temple University): Model variables include leading economic indicators and polls. Two-party share of the vote: Obama 52.2% based on August polls. Obama has a 72% chance of winning.
* Brad Lockerbie (East Carolina University): model looks at consumer attitudes and behavior and polls. He notes that economic expectations are the second worst since polls have been able to answer this question (in 1980 voters were more pessimistic). Two-party share of the vote: Obama 58%. (He also predicted a 25-seat loss in the House for Republicans.)
* Helmut Norpoth (SUNY, Stony Brook): Using a New Hampshire primary model and an electoral cycle variable (taking account as to whether the incumbent party had held the presidency for two terms or longer), his talk was entitled “On the Razor’s Edge.” Two-party share of the vote: Obama in a nailbiter with 50.1%.
* Alfred Cuzan (University of West Florida): In a talk presented by Randy Jones, Cuzan’s model is based on fiscal policy–that is, that if the share of government spending of GDP increases the incumbent party is penalized by voters. Their model also uses economic growth and length of term of the president’s party. Two-party share of the vote: Obama 51.9%.
* Michael Lewis-Beck (University of Iowa): Lewis-Beck’s model looks at economic growth, presidential popularity, jobs creation, and incumbency. His initial model predicts Obama to carry 56.58% of the two-party vote share. BUT, he notes that one cannot extrapolate beyond their data and that Obama’s candidacy, as the first African American candidate. So, attempting to tease out the race factor, he eventually concludes that 11.5% of voters would never vote for an African American candidate and that there will be a 6.51% penalty for Obama. Two-party share of the vote: Obama 50.07%. Calling this a potential Bermuda Triangle election, he notes that there could be an inversion between the popular and electoral vote a la 2000.
* Alan Abramowitz (Emory University): Boasting he has never been wrong in 20 years, Abramowitz’s model looks at GDP growth, presidential approval, and a “time for change” factor. Two-party share of the vote: Obama 54.3%
* James Campbell (University at Buffalo, SUNY): His model takes into account polls at Labor Day and the economy (GDP). As noted above, he can’t release his final prediction for a few days, but if McCain is at 46% of the two-party poll on Labor Day, he predicts McCain would get about 50% of the vote, meaning that it’s likely to be a close election, with a slight lead for McCain–thus making Campbell the odd political scientist out in Boston on the panel. Perhaps an enviable position to be in.

It was a really intriguing panel as the speakers went through their various methodologies, and voters will have the chance to prove these projections right—or wrong.

And, as James Campbell noted, predictions by political scientists are always wrong.

Of course, it doesn’t stop them from making them.

[FeatsofClay]

I wonder if the people predicting the % of votes are accounting for the location of those votes and the oddities of the electoral system. Having a candidate leading 52%-48% can be a loss for the 52%.

I also could see a lot of people answering differently than whoever answered the True/False one.

I wold like to know more. I would also like to revisit this AFTER the election and see who was close. :)

[Varaj]

I wonder if the people predicting the % of votes are accounting for the location of those votes and the oddities of the electoral system. Having a candidate leading 52%-48% can be a loss for the 52%.


The % of votes is popular vote and yes it takes in account the electoral system.


I also could see a lot of people answering differently than whoever answered the True/False one.

The Keys system has the best track record of all the methods I listed. These forecasting systems are all peer reviewed forecasting systems, the best of the best as it were. Does that guarantee the are correct? Heck no as Campbell says himself, "predictions by political scientists are always wrong." :)


I wold like to know more. I would also like to revisit this AFTER the election and see who was close. :)

It will be interesting but it is worth noting that all the models except Cambell's predict Obama as the winner.

[AZRogue]

It will be interesting but it is worth noting that all the models except Cambell's predict Obama as the winner.

That's what I felt, going into the race. The climate was just really in his favor, which isn't a bad thing, just what I perceived. It's been muddled, lately, and it would be interesting to see if that's still the case. I think those models are still correct, though.

[Varaj]

Polly on the electoral vote.
http://www.forecastingprinciples.com/PollyVote/index.php/pollyelectoral.html

That is a pretty good loss for the reps forecast.

[Hatter]

I tend to look at the current scenario and make best effort predictions based on most likely actions. Looking at the current electoral map you can see that McCain needs to win every current toss up state in order to win. Possible, but unlikely. The most likely tactic is for him to go negative and try to drive down voter turnout while mobilizing his base. Given that he is not favored by the republican base this is unlikely to be a successful tactic.