I've recently received several emails about the computer analysis President Elect is doing on the 2000 electoral college PE 2000 Electoral College computer analysis. As of the writing of this article, the computer has Bush leading Gore 288-250. Readers from both sides of the political aisle have asked whether I'm biased toward Bush and want to know more about what the analysis is based on. Here are a couple of those emails:

"I was surprised to see your analysis of the Electoral College as of 9/22. Do you think you might have a slight bias towards Bush? Certainly your analysis is not agreed to by most news organizations or the latest polls....Just what is the basis for your prediction?"

"Just saw your web site the first time - very interesting. Of course, your numbers don't exactly mirror what we've seen from the media the past 45 days or so. Why do your numbers differ, and how can we feel that your numbers are more accurate than the major media?...I hope you're right!"

The analysis is based on a formula I created this past summer well before most people were doing electoral college analysis. The results of each election from 1960-1996 are used to determine a state's voting tendency. Then every week I input the latest state by state polling results. Gore has been leading in national polling for about 5 weeks now and that's why most people are surprised to see my analysis. However, as I'm sure most of you know, only state by state voting elects a president, not national voting. It's only been the past few weeks that Gore has started to take the lead in state by state polls. And in most of those polls he leads in, there are enough toss-up votes that the whole election is still up for grabs. For example, American Research Group says Gore is leading Bush 204-132 with 202 votes still too close to call. Since my analysis assumes the election is today and there can be no toss-ups, it projects a winner in close states by looking at exactly how close the current poll is and then factoring in that states voting tendency. Therefore, with so many states still up for grabs, my analysis is still in line with American Research Group's and most of the others that Gore leads in.

That being said, my analysis does show Gore gaining on Bush substantially. On June 30, the computer had Bush up 366-172. By September 8, Bush was only up 303-235. The current analysis has Bush up 288-250. And with state by state polls released since September 22 (which will be posted as usual this Friday), unless the Republicans can turn things around quickly, Gore could be within 8 electoral votes of Bush.

In conclusion, I guess the main difference in my analysis and others is that mine takes more into account than just polls. If Bush is leading in a traditionally Democratic state by less than the poll's margin of error, the computer does not automatically give Bush the win in that state. If Gore is leading in a traditionally Republican state by less than the poll's margin of error, the computer does not automatically give Gore the win in that state. In states that others say are too close to call, the computer tries to predict what the undecided voters in that state will do based on how the state has voted in the past. Maybe most importantly, keep in mind that I'm not predicting a big Bush win right now. Everyone is saying this will be a close race and my prediction is in line with that. A 288-250 electoral count would be the closest presidential election since 1916!

Once again thanks for your questions, comments and interest in President Elect. This is the first election I've used this computer model with and after each election I hope to perfect it a little more, so feedback is always appreciated.