This page is no longer active

Da Blog has moved to MorganWick.com. Please update your bookmarks, links, and RSS readers.

Wednesday, October 29, 2008

A new way to look at election results!

I know this is kind of short notice - I've been sidetracked by my ongoing platform examinations - but I'm wondering if anyone wants to help me track the results on election night? Ideally I would have some sort of PHP and MySQL system set up for inputting and tracking results, but because of Freehostia's restrictions that would mean letting people into my Sandsday MySQL account, so I may have to rig something up on Wikipedia. I had trouble keeping up with the results on Super Tuesday, which was actually about on the order of the number of presidential races on election day proper. Trying to keep up with that, plus Senate and House races, would take me forever. I'm thinking 25 races per person would be more than a workable number; that would mean two people to do the presidential races, one to do the Senate, and 17 people to do the House, for 20 total and I would be one of them. I don't think I have 20 visitors a day, let alone that many willing to volunteer, so I may have to double up some of the work on some people. Comment on this post or e-mail me at mwmailsea at yahoo dot com if you're willing to refresh pages all night.

If I can't get anyone, I'll just put all my Presidential race work on Da Blog, mostly just my projected electoral vote counts (and if I get only one or two volunteers I may just give them an Excel file and have them post work on their own blogs). The sources of semi-raw vote data with all candidates I relied on in primary season might not be available anyway, which probably means the dreaded patchwork of sources.

The reason I'm planning on doing this is to provide a demonstration of my projection system. There's more explanation here, but the idea is to take race projection at least partly out of the hands of potentially biased analysts and making it less reliant on possibly flawed exit polls. It's based on one thing and one thing only: the results themselves.

Rather than one level of projection, which favors an early rush as networks fall over themselves to project and then boredom most of the rest of the way, this creates three, two of which are based on objective mathematical formulas:

Projection: Only invoked when the race appears to be a sure thing but the mathematical formulas haven't confirmed it yet. This occurs only at my discretion, and is based on the results themselves, not invoked the instant the polls close (unless the candidate is unopposed).
Autoprojection: Invoked when %1 > %2 + (1-P), where %1 is the vote percentage of the first place candidate, %2 is the vote percentage of the second place candidate, and P is the percentage of precincts reporting. 1 represents 100%.
Confirmation: Invoked when P * %1 > P * %2 + (1-P), using the same values as above.

If I feel like it, I may have more tomorrow, but I'm already late in trying to get home and see the Obama infomercial when it airs on the West Coast!

No comments: