Electoral Vote Predictor 2004

Friday Dec 31: Kerry 252, Bush 286

31 Dec 2004 06:00:00 EDT

I am still waiting for information about the recount in Ohio and how the overseas vote went. If anyone knows, please let me know.

A reader asked an interesting question: What is the smallest number of states you have to win to be president? By sorting the spreadsheet on EVs, you get the answer easily: 10. A candidate who wins the following 10 states gets 271 electoral votes and thus wins: New Jersey, North Carolina, Michigan, Ohio, Illinois, Pennsylvania, Florida, New York, Texas, and California

Tuesday Dec 14: Kerry 252, Bush 286

14 Dec 2004 06:00:00 EDT

Today's map

One of the Democratic electors in Minnesota cast his or her electoral vote for John Edwards according to an AP story. The name of the faithless elector is not known.

A recount of all of Ohio's votes will begin this week, but few people expect it to change the election results.

Tuesday Dec 14: Kerry 252, Bush 286

14 Dec 2004 06:00:00 EDT

Today's map

Slate has a good article about which pollsters did well and which didn't, and about where the conventional wisdom about polling missed the mark. They confirm my earlier observation that Scott Rasmussen did the best job, despite his use of robodialers and talking computers instead of people to conduct the polls.

Wednesday Nov 24: Kerry 252, Bush 286

24 Nov 2004 06:00:00 EDT

Today's map

Here is some information about the exit polls.

Friday Nov 19: Kerry 252, Bush 286

19 Nov 2004 06:00:00 EDT

Today's map

The first scientific analysis of voting machine irregularities has now been published by the University of California.

Tuesday Nov 16: Kerry 252, Bush 286

16 Nov 2004 06:00:00 EDT

Today's map

The Libertarian and Green parties have collected the needed $113,600, so there will be a recount in Ohio. Ralph Nader is also asking for a recount in New Hampshire. If nothing else, these recounts will give us two data points for the popular vote in each state so we will be able to compute the standard deviations. We are all used to the concept of margin of error caused by the statistical sampling used in polls, but it may take some getting used to if there is a substantial statistical error in counting the actual votes.

Exit polls showed that 7% of the voters have a cell phone but no landline. Among 18-29 year olds, the figure is 20% and growing. Furthermore, this group favored Kerry over Bush by 56% to 41%. Since pollsters are forbidden by law from calling cell phones, this age group was undersampled in the polls, but the pollsters were able to correct for the omission statistically. If the number of cell-only voters continues to grow rapidly, by 2008, the pollsters are going to have a big problem on their hands.

Survey USA has made a comparison of the various pollsters on a variety of metrics. They also measured their own partisan bias by comparing their published polls with the final results. Conclusion: they were completed unbiased. I agree with their assessment.

Is predicting the electoral college easy? Apparently not. ARG had a contest to see if people could accurately predict the results. Not a single one of the 16,269 entries got it right. Even when they eliminated Iowa, New Mexico, and Ohio, nobody got it right.

Monday Nov 15: Kerry 252, Bush 286

15 Nov 2004 06:00:00 EDT

Today's map

The election may not be quite as finished as most people think. Take a look at this article about recounts and strategy. It may explain why Kerry has been so silent of late--he is letting others do the heavy lifting. If the expected recount of New Hampshire turns up serious problems, especially with voting machines, there will be demands for recounts elsewhere. Even if it doesn't change the result, this exercise could convince people that maybe now it is time to get serious about devising an electoral system in which every eligible voter can vote and every vote counts. We are still very far from that goal and it is a national disgrace.

In addition to the recount in New Hampshire (which Kerry won), the Libertarian and Green parties intend to file for recounts, starting in Ohio. There are two reasons to ask for recounts. First, the exit polls, which historically have been quite accurate, differ from the tallied results by more than could be expected by chance alone. Second, there are issues about the invalidated, provisional, and absentee ballots. Once the results have been certified, which could be as early as tomorrow in some states, the paper ballots are destroyed and the computer memories are cleared unless a recount has been formally requested.

In Ohio, requesting a recount there costs $10 per precinct or about $113,600 for the whole state. If you want to contribute, click here for the announcement. I was unable to get through to the actual contribution site at 5 a.m. EST Sunday morning. I don't know why, but I doubt it was due to their being swamped with traffic. You might have to call the Libertarian or Green Parties if you can't get through.

It would nice if the mainstream media were more interested in getting a complete and accurate count of all the votes, but they don't seem to be.

An analysis of the pollsters can now be found on the Compare the pollsters page. Take a look to see who gets an A+ and who gets an F.

The LA Times has an article comparing the election results to those of previous ones in which an incumbent president ran. By historical standards, this was an incredibly close contest.