TruthIsAll
05-25-2008, 02:52 PM
2004 Election Model: A Pre-election Chronology of DU Postings
http://www.geocities.com/electionmodel/2004ElectionModelDULinks.htm
In the four months prior to the 2004 Election, I posted Election Model electoral vote projections and probabilities on the Democratic Underground. The following links comprise an historical chronology of those posts. I have also included post-election analytical commentary. It’s interesting to review the daily DU post commentary as the projection model was updated.
The analysis was based on the latest 50 state and 18 national polls. The state model used a Monte Carlo simulation of 5000 election trials to determine the expected electoral vote. The probability of a Kerry victory varied each time the polls were updated. It was simply the percentage of the 5000 trials in which he received more than 269 Electoral votes. Of course, the projections throughout the timeline all assumed that the election would be held the following day.
Prior to the final projection, the only assumption in the model was that Kerry would win 60% of the late undecided vote. It was changed to 75% in the final, based on the opinion of professional pollsters (Zogby, Harris and Gallup) that historically, the challenger could expect to win 67-90% the undecided vote - if the incumbent was unpopular. Bush had a 48% approval rating on Election Day.
The final projection, based on the Monte Carlo state polling model, was that Kerry would win 51.8% of the two-party vote and capture 337 electoral votes. The national polling model projected that Kerry would win 51.6% of the two-party vote.
The pre-election projections were confirmed by the exit polls. Pollsters Edison-Mitofsky in Jan. 2005 reported that Kerry led the average unadjusted state aggregate exit poll: 52-47%. Bush won the vote: 50.7-48.3%.
All links to the election model show the final projection analysis. The election model graphs include the full statistical trends.
http://www.geocities.com/electionmodel/2004ElectionModelDULinks.htm
In the four months prior to the 2004 Election, I posted Election Model electoral vote projections and probabilities on the Democratic Underground. The following links comprise an historical chronology of those posts. I have also included post-election analytical commentary. It’s interesting to review the daily DU post commentary as the projection model was updated.
The analysis was based on the latest 50 state and 18 national polls. The state model used a Monte Carlo simulation of 5000 election trials to determine the expected electoral vote. The probability of a Kerry victory varied each time the polls were updated. It was simply the percentage of the 5000 trials in which he received more than 269 Electoral votes. Of course, the projections throughout the timeline all assumed that the election would be held the following day.
Prior to the final projection, the only assumption in the model was that Kerry would win 60% of the late undecided vote. It was changed to 75% in the final, based on the opinion of professional pollsters (Zogby, Harris and Gallup) that historically, the challenger could expect to win 67-90% the undecided vote - if the incumbent was unpopular. Bush had a 48% approval rating on Election Day.
The final projection, based on the Monte Carlo state polling model, was that Kerry would win 51.8% of the two-party vote and capture 337 electoral votes. The national polling model projected that Kerry would win 51.6% of the two-party vote.
The pre-election projections were confirmed by the exit polls. Pollsters Edison-Mitofsky in Jan. 2005 reported that Kerry led the average unadjusted state aggregate exit poll: 52-47%. Bush won the vote: 50.7-48.3%.
All links to the election model show the final projection analysis. The election model graphs include the full statistical trends.