Really, it's quite simple. In 2012, there are 928 official voters for the Trophy. 870 of them are sportswriters, 58 are living former winners, and there's one fan vote. Many of those voters will publicly declare their votes - in print, on TV, on the radio, on the net - and if we count enough of them, we'll know who the winner will be.
It takes a little math (which we're happy to do) but mostly, it takes your help - if you hear an official voter go public, let us know.
How it works:
- We count actual votes. We count only the picks from people who claim to be actual voters. (The organization doesn't release a list of voters, so we can't verify.)
- We make two assumptions. 1) That the voter turnout in each region will be equal, and 2) that the undisclosed votes in each region will mirror the publicly disclosed votes,
- Based on those assumptions, we total up the votes in each region and then extrapolate the totals.
- We continually update as more information comes in.
Dude, it looks screwed up. Sometimes, especially early on, the total actual votes don't seem to correlate to the projected totals. Not to worry - that's usually because there's a huge disparity between the regions in the number of ballots we've found. Also, if we don't know what region a voter belongs in (often, network TV guys) we assign a fraction of that vote to each region.
Why do the numbers add up to more than 100%? Every Heisman voter must list three names. Top vote gets 3 pts, second place gets 2 pts, third gets 1 point. Therefore, each ballot has a total of six points - but the most any one candidate can get is three points. A unanimous #1 selection would get three points on every single ballot. To make things as clear as possible - 100% is the maximum possible points each candidate could earn. (And yes, that means that the total projection values will add up to 200%, not 100%.)
When's the projection good enough? Well, we've consistently gotten it right with around 20-25% of the vote counted. But we'll let you know. It has a lot to do with how close the race is.
Early in the data collection process, we expect the numbers to be off somewhat - since people who are promoting underdogs often will talk about their pick early. The more data that arrives, the better the projection gets - so if we're missing a voter, let us know.