Electoral Vote Predictor 2004:   Kerry 328   Bush 210

Data in Excel format
Battleground states
Info about the states
Previous report
Next report


dark blue Strong Kerry (168)
light blue Weak Kerry (63)
white and blue Barely Kerry (97)
white Exactly tied (0)
white and red Barely Bush (11)
light red Weak Bush (65)
dark red Strong Bush (134)
Needed to win: 270

News from the Votemaster

Finally, the suspense is over. We now know whether there was a post-convention bounce for Kerry. Answer: there was. Of course, there always is, and it evens up after the other convention, but for the moment, Kerry has received a boost. He is now ahead in the electoral college 328 to 210, with all 50 states and D.C. reporting for duty (we had two ties until today). The Wall Street Journal's estimate using a different methodology is 318 to 220. Close enough for government work.

Zogby has released polls in 16 battleground states. As usual in such a tight race, there is good news and bad news for all the candidates (except Nader, for whom all the news is bad--he is dropping practically everywhere). First the good news for Kerry. He is ahead, albeit by small margins, in 13 of the key 16 states in which the election will be won or lost, including a 3% margin in Florida (which is still within the margin of error). What's more, Michigan and Pennsylvania, two states hard hit by job losses, have been solidly in his column for well over a month now. In addition, Kerry and Bush are still statistically tied in Tennessee for the third poll in a row. Making Tennessee a battleground state is good news for Kerry. Bush was supposed to win this one without a fight.

Now the good news for Bush. He has increased his shaky lead in Ohio to 5%, so we now rate the state weak Bush instead of a tossup. This is very good news indeed for Bush. For him, no Ohio, no White House. Florida flipped compared to yesterday, but the difference is within the margin of error. Also good for Bush, he is ahead again in Nevada, but don't bet on it. It's illegal. Even in Nevada. Also positive for Bush is that Kerry's 10% lead in New Mexico has vanished into the desert. It is now 1% for Kerry, a statistical tie.

To get a better idea of how much bounce there was, see the daily graph of the electoral college score over time. No doubt more polls will be flooding in this week. As an aside, Mason-Dixon had a poll in Nevada showing the same thing as Zogby, but since Zogby's was held a day later, in accordance with our policy of "latest poll wins" we only show Zogby's. Similar, Fairbank-Maslin-Maullin's poll of Washington state two days before Zogby's shows exactly the same 8% Kerry lead as Zogby. For more information on polling and our methodology, see the FAQ .

Here is a summary of the state of the states. Bush absolutely must win Florida and Ohio. It is hard to conceive of any realistic scenario in which he wins the electoral college without both of them. For Kerry, these would be big pluses, but not fatal if he loses both. His great strength lies in his consistent leads in the Midwest, from Missouri on north. This region is reeling from job losses and whatever numbers come out of D.C. in the next few months, the facts on the ground are that a lot of people have lost good jobs and are not likely to get equally good ones before November. If Kerry can maintain his lead in this region, Teresa can call the moving company. Of course, Karl Rove understands this very well, so Bush is going to spend a huge amount of time and money in the region from now until the election. Thus the states to watch are Minnesota, Wisconsin, Iowa, Missouri, Michigan, Ohio, Pennsylvania, and Florida. Forget the national polls. The popular vote doesn't matter, as we saw in 2000. It is these eight states and another half dozen or so smaller battleground states where the action will be.

Some people have asked me why I don't include the margin of error on the site. There are three reasons. First, it is virtually always the same in the state polls: 3% to 4% because the MoE depends only on the sample size, generally 500-1000 people in the state polls. Second, MoE matters a lot in a survey where the underlying mean is stable, like trying to determine the average height of the citizens of a state. Then all the variance is due to sampling error. In such a volatile election, the underlying mean is changing wildly so the sampling error is not the biggest source of noise. New Mexico just went from 'strong' (a 10% Kerry lead) to effectively tied (a 1% Kerry lead). That is probably not due to just statistical fluctuations. Some people in New Mexico changed their minds. Third, adding one more item to the charts tends to clutter them up. There is enough information already.

We are starting an RSS feed of the site today. If you know what this is and like beta testing, give it a shot and report back any problems. Remember, if at first you don't succeed, it probably involves a computer.

To get the dates and sources for the polling data, click on the Excel spreadsheet at the top right. To bookmark this page, type CTRL-D. To help publicize this website, please link to it to improve its Google PageRank and tell your friends about it.

-- The votemaster