The Political Machine's statistical model is not poll based. When we designed it, we concluded that since ultimately it's a video game, we should only concern ourselves with the macro-based statistics. That meant not worrying about day to day fluctuations in polls. Instead, the goal was to use data (political party registrations, exit polling, census data, etc.) that was publicly available to get a clear understanding of the people who live in the states.
From there, we then look at the political issues of the day and determine, to best of our ability, how Democrats, Republicans, and Independents feel about a given issue. The Political Machine, hence, has an issue-driven model rather than a poll-driven model. Depending on what issues matter at the time of the snapshot, different candidates have different advantages. The model has an error factor of around +/- 3% and it cannot take into account with great precision the effect of advertisements, endorsements, speeches, TV interviews. The Political Machine has all these things but our top priority was ensuring the game was fun and so the effect of those 3 things (ads, endorsements, interviews) has more to do on the "fun factor" than anything else. But still, it does seem to display an interesting level of accuracy.
Of course, we're only in June and the issues still need to be tweaked between now and election day. Consider below:
June 2004 model (minus ad effects)
The census data in The Political Machine doesn't change. Instead, the evaluation of different issues change. Move the economy slider down and Kerry does better. Bring up global tensions and Bush does better. Which issues will matter in November may be quite different than they are now. But according to the model -- which again is not tweaked based on polling data beyond what the % of registered Democrats and Republicans are in a given state, it seems to roughly reflect the current polls -- Kerry has a 2% national lead over Bush according to the poll. One should point out that third party candidates like Ralph Nader are not factored into here. But it is our view that Nader is not likely to have a significant impact anywhere other than perhaps Wisconsin.
If the election were held today, based on the issues as we evaluate them today, Kerry would win 305 to 233. The big state going against Bush right now that is on the edge is Ohio. Ohio could go either way and the economy has a lot to do with that along with a host of other factors that could slide one way or the other. That said, typically our models favor the Democratic opponent. Minus advertising, Republicans tend to be slightly behind. But in this election, the Republican candidate has a significant advantage in campaign funds (contrast that to Dole in 1996).
Here is one potential Bush victory path. Through effort in Wisconsin and Iowa and victory in Ohio, Bush manages to win by a mere 2 electoral votes. However, that assumes Bush wins in West Virginia also. In this scenario, the election is decided in Ohio and West Virgina with help from Wisconsin and Iowa.
Some polls shows Bush leads in Oregon and Michigan. We don't claim our model is the end all be all predictor. It is, after all, a video game for consumers. But barring a strong third party candidate, we don't see that happening. Similarly, despite Mr. Ridge's position in the administration, Bush is not likely to win Pennsylvania. Kerry should pick Gephardt as his running-mate as it gives Kerry his bet shot at picking up both Missouri and Iowa which, Bush needs to win at least one of them to win. But candidates don't always pick who they should pick. Bush, for instance, picked Dick Cheney instead of Tom Ridge as his running mate which cost him Pennsylvania and led the way to the Florida debacle.
The states to keep an eye on are these:
-
Ohio
-
Wisconsin
-
West Virginia
-
Iowa
-
Missouri
-
Arizona
National polls are kind of interesting but not terribly useful because they tend to have a +/-2% to 3% level of error which, unless a candidate is able to have a definitely 2% lead, doesn't indicate who would actually win the election. What really matters is who is winning in these 6 states. Bush must win nearly all of these.
Election 2000 results.