On The Issue News

[cynatnite]

I received an email from them a little while ago. Here's some of what they said:

OnTheIssues predicts that Bush will win the popular vote. But Kerry will win the electoral race by several dozen electoral votes. We base our election prediction on the results of the 1998 gubernatorial election in Minnesota, where Jesse Ventura was polling down three points the weekend before the election, yet won the election by three points. A spread of six points is greater than the margin of error in any serious poll, and understanding why Ventura won -- and why the polls were wrong -- is the basis of our presidential prediction.


Pollsters don't actually report the number of people who say they're going to vote for Bush or Kerry. The pollsters adjust the raw numbers by factors that account for historical demographic trends. For example, if only 25% of people aged 18-25 voted in the previous election, then anyone in that demographic group is discounted by 75% in polls in the current election. That's called 'normalization' by the pollsters, and is built into every scientific poll. The demographic groupings are determined and applied by each pollster, which is why poll results often differ.

There's one more factor to consider. Many people don't like to tell pollsters who they will vote for, which the pollsters report as 'undecided'. Those people often don't like incumbents, for the same reasons as they won't tell pollsters their preference. What we see as healthy American iconoclasm means any incumbent has to have a strong lead among voters who call themselves undecided, because those voters tend to vote for the challenger. That factor could amount to another percentage point or two, since the 'undecided vote' is running at 3-4 points this weekend. Typical analyses give the undecided vote to Kerry by a factor of two-to-one, which means another net gain for Kerry of a couple of percentage points over what the pollsters show.

Now onto the specific numbers. OnTheIssues uses electoral-vote.com as our source of polling information, which you can see at http://www.ontheissues.org/elect_frm.htm . Based on the analysis above, unless Bush is ahead by more than 3 percentage points in a state poll this weekend, we predict that Kerry will win that state. The contentious states (with electoral vote count shown) that we predict will go to Kerry are:
- VA (13) - Bush up by 3 points.
- HI (4) - Bush up by 1 point.
- NM (5) - tied in the polls.
- NH (4) - tied in the polls.
- NV (5) - tied in the polls.
- MI (17) - Kerry up by 1 point.
- IA (7) - Kerry up by 1 point.
- FL (27) - Kerry up by 1-2 points.
- PA (21) - Kerry up by 3 points.
- MN (10) - Kerry up by 3 points.

All the pundits have been predicting Ohio as the Florida of 2004 (i.e., the most contentious state). Bush is currently up by 4-5 points in Ohio. With that size lead, we predict Bush will win Ohio despite the factors in the above analysis. The overall count is:
- 196 electoral votes in strong pro-Kerry states
- 229 electoral votes in strong pro-Bush states
- 113 electoral votes in the states listed above, which we predict go to Kerry, 96-17.

The sum of the above is that Kerry wins the electoral vote 292-246. We predict that Virginia and Hawaii will both be tighter than shown above but will both go to Bush. So we predict Kerry wins those states where he currently is tied or better -- in other words, any state in which Bush is not currently ahead, will go for Kerry.

We are not endorsing Kerry -- we don't endorse candidates but we do endorse political participation and we especially endorse voting based on the issues. We note for the record that in 2000 we similarly predicted a Bush popular vote victory and a Gore electoral vote victory and we were incorrect on both counts (although few pundits predicted a split, and we did). And viewers should note that this prediction is based on very wonkish concepts -- we wholeheartedly admit to being wonks!