Well , it's finally here, and not a moment too soon. I haven't posted in months. I'll blame some of the procrastination on today's election. Like many former able-bodied workers, I have spent countless hours watching the web wires for the latest info on the moose hunter of all elections.
One place where I've probably spent 538 hours or so is at fivethirtyeight.com - a meta-meta-polling site that claims to present "electoral politics done right."
Named for the total number of possible electoral votes, FiveThirtyEight.com is the creation of Nate Silver and Sean Quinn, who have developed a unique methodology for polling analysis:
- Many polls are used to produce a weighted average. The weights are based on a reliability index determined by that pollster's historical track record, the poll's sample size, and the recentness of the poll.
- A regression estimate based on the demographics in each state is used to account for outlier polls
- An inferential process allows states that have not been polled recently to have their results modified , effectively makeing them "currrent"
- The election is simulated 10,000 times "in order to provide a probabilistic assessment of electoral outcomes based on a historical analysis of polling data since 1952. The simulation further accounts for the fact that similar states are likely to move together, e.g. future polling movement in states like Michigan and Ohio, or North and South Carolina, is likely to be in the same direction."
See the site for a much more thorough description of their methodology.
And especially visit it to see the display of results, particularly of the simulated elections.
Which is why I'm posting now, at 1:20 AM EST on November 4th. The latest simulations has Obama winning in 98.1% of the model elections, which ultimately leads to a prediction of Obama getting 52% of the vote to McCain's 46.1%. The electoral breakdown is predicted to be 346.5 to 191.5 for Obama (I could not find where the 0.5 electoral vote comes from)
Naturally I am as interested in the accuracy of these predictions as I am interested in the outcome. I will post an update to this piece as soon as the final numbers are tallied, to check the reliability of FiveThirtyEight.