Saturday, September 20, 2008

Charts

I've been watching six polls over the past two weeks. I picked the polls pretty much at random. I'm doing it for two reasons - one, to see how my candidate is doing, and two, because I hear conflicting things from polls, and I wanted to try to get a sense of what's really happening.
These are end of day (or pretty close) values of projected Obama electoral votes. The entries are FiveThirtyEight.Com, Electoral-Vote.Com, Presidential Election Forecasts, Real Clear Politics, CNN Electoral Map, and CQ Politics. That double-arrowed line across the middle is at 270 -- the number of electoral votes needed to win.

The first thing that leaps out at me is that these guys sometimes seem to be off in their own little orbit -- particularly the CNN and CQ entries (which is why I moved them to the bottom of the legend; I'm thinking of dropping them). CQ hasn't budged since I began looking (wondering if somehow I'm telling it not to show current values), while CNN goes down, down down. Real Clear Politics has at least moved - once.

If you ignore those three, the remainders -- 538, EV, and PEF tracked roughly the same until the 16th -- which is when the financial market went ker-blooey. At that point, they all started showing gains, but they couldn't figure how much -- 538 jumped right up there to about 303, while EV and PEF jumped to about 273. I know this isn't a science, but that seems like a big spread to me -- even given that 538 makes a fetish of looking at lots of polls, and averaging/weighting them. So I'm guessing that the 'real' number is probably closer to 275, 280. Of course, I'll take anything over 270.

My conclusion, so far -- don't just believe one poll, but track it with others, and see when they vary. Undoubtedly, some of them will have a good reason, while others might just be sampling people in the grocery store parking lot.

No comments: