« “I’m with the band” | Main | We might report, but please don't decide »

unfutz, ampersand, victory

Your Humble Blogger happened on Unfutz’s Electoral College Survey via Alas, a Blog and initially, it looked pretty darned cheering. After all, he shows 39 sites, of which 28 show Sen. Kerry with enough electoral votes to win, 5 show him leading but have enough undecided states that if they all go to Our Only President, Sen. Kerry would lose, 5 show Our Only President with enough electoral votes to win, and 1 shows Our Only President winning, but with enough undecided states that if all they all go to Sen. Kerry, Our Only President would lose.

In other words, 28-5, just counting winners, and 33-6 counting leaners. Good odds, no? The trends are also good, such as they are.

My first concern was that ampersand, bless him, wants Our Only President and his cronies out of the White House, which shades the link a bit, and from an initial look at Ed Fitzgerald’s unfutz, Mr. Fitzgerald wants Our Only President and his cronies out of the White House, too. As does YHB, where you are getting the link even now. So, you know, consider the source.

Still, Mr. Fitzgerald clearly wants to include, for instance, the Young Conservatives. I don’t think he’s stacking the deck, but of course, I don’t know. The numbers I’m seeing there are similar to the numbers that I’m seeing in other places, anyway. Which brings me to my more serious concern about the survey. Unlike Prof. Pollkatz, aka Stuart Eugene Thiel, who analyzes a variety of different actual polls, Mr. Fitzgerald is simply stacking up a list of websites who are making their own analyses. Undoubtedly, some of those analyses are based on the same polls, or on some of the same polls. That is, five different people may be looking at the same Ohio poll, and putting OH in the Dem column, while only one or two people look at the other poll and put it in the Rep. Is it a good idea to count each of those as a different source?

Clearly, if you want to do serious analysis, this isn’t the way to do it. It’s like averaging averages; it doesn’t work. What is on unfutz is a kind of zeitgeist meter, a sense of what people are sensing. And, of course, I’m pretty happy about it, myself. Still, don’t mistake that for analysis, and certainly not for prediction. It’s a long way until November.


Post a comment

Please join in. Comments on older posts will be held for moderation. Don't be a jerk. Eat fruit.