Mumme Poll

Honestly? I’ve never understood the point of it. Usually ends up looking a lot like the any other poll’s top ten, despite the somewhat novel voting method, and I certainly haven’t seen it improve on the other polls in terms of predictive accuracy.

It’s an experiment to see how differently approval voting would result. It’s difficult to say that it doesn’t improve on them when a) there is only one result set as the poll is only a year old, and b) the notion of accuracy is purely conjecture on the part of the person offering a singular opinion on how the teams should have been ranked. The concept of predictive accuracy is meaningless for any poll, especially when considering it’s a league of 120 members who only play approximately 10% of all potential opponents. Seeing as that size of a pool mandates a poll, the point isn’t to find something that can accurately determine in week six who the best teams will be at the end of the year, but to remove as much of the bias as possible that is reflected in a fully weighted poll (hence the reference to Hal Mumme and his vote of Hawaii at number one).

Define accuracy in a college football poll.

You’re in the business of criticizing a poll for not being more accurate but you lack the ability to define exactly what accuracy in a poll would be?
I could use the blogpoll data which would involve re-entering a hundred or more top 12s every week, or I could set up the Mumme Poll which opens the voting to a significantly larger base of voters, collects a more diverse amount of data, and aggregates the data with little room for my personal error.

Why not? I get to participate instead of just engaging in the typical gnashing of teeth, with the typical whining, and the typical posturing. Will it affect the outcome? Probably not, but fortunately, I don’t care. Thanks for setting it up TFiT.