Okay, nobody would dream of calling me a numbers-cruncher. However, ever since I read the Wisdom of Crowds I’ve been more and more intrigued with the power of statistical averaging. The basic rules, as I understand them, the more diverse-but-informed a group of people is, the more likely the average of their opinions will be a better guess than that of any individual expert picked at random. A group – i.e. their consensus or average — will almost always guess the number of jelly beans in a jar better than any single person over time, for example. Markets get right what individual buyers, sellers and analysts often get wrong. Think of Ramesh’s nearly fanatical faith in the Iowa political market.
So, the first grouping: Washington Post “Crystal Ball” predictions. I tallied them all up, and here’s what I found. Bush wins 49.384% to 48.969% for Kerry. If you drop the highest and the lowest, Bush’s score improves slightly to 49.54% to 48.91% for Kerry. The electoral vote is similar. Kerry gets 263 electoral votes to Bush’s 279. If you take out the highest and the lowest Bush’s margin slips but he still wins with 276 electoral votes.
On Capital Gang last night, the Gang (including our own beloved Kate O’Beirne) made their predictions. They only talked electoral votes, but again Bush won with an average of 275.6 and 262.4 for Kerry.
Now, dropping the highest and lowest is problematic because the producers of Capital Gang in their wisdom have chosen to have only two conservatives – Kate and Bob Novak — but three liberals – Al Hunt, Mark Shields and Margaret Carlson. But if you take Novak’s and Mark Shields predictions out, you get Bush 273 and 265 for Kerry. Interestingly, if you just yank out Novak, you get 268.75 for Kerry and 269.25. Rounding them to the nearest whole number, you get a tie.
Now a couple points. First, it’s always likely I’ve made some mathematical errors. Second, I have no idea how predictive any of this really is. But I do think it definitely captures one aspect of reality very well. The fact that Kerry’s average predicted numbers are lower than Bush’s reflects the fact that Bush has more routes to victory than Kerry does. I haven’t seen any predictions from unaffiliated Kerry supporters which put his margin over 300 electoral votes for the simple reason it’s next to impossible to imagine a scenario where that would happen. It’s quite easy to see how Bush could win 300 votes, even if it’s not obviously probable.