Diary

Perrin’s Predictive Election Algorithm — the Final pre-election results

This algorithm is designed to predict the net gains or losses by the each party in both houses of the U.S. Congress.

I have reviewed most recent data which I use as the values in the algorithm. In half of the variable data inputs, the Democrats showed some modest improvement.

However, in the other half of the variable data inputs, the Democrats showed statistically significant declines. It turns out that the variable data points the Dems declined in are the most powerful within my algorithm. Meaning, the data points the Dems declined in have the largest statistically significant impact on the out come of the algorithm’s results, while the places the Dems showed modest improvement have a much smaller impact on the final predictive result.

Here are the final, pre-election results:

A net loss by the Dems of 77 seats in the U.S. House, and a net loss of 10 seats by the Dems in the U.S. Senate. The algorithm final number in the Senate was on the cusp of an 11 seat loss, but by the smallest of margins, the answer was 10.

As I mentioned the first time I published these results, I was conflicted about which data to use in the most powerful variable in the algorithm, so I used a wildly different data point for the House calculation, versus the Senate one — mainly to see which would be the most accurate.

It turns out, also, that I used a different calculation in the algorithm for the House prediction — mostly because the data point is so different from the Senate’s, it needed a different calculation. Given the difference in this calculation, and the significantly different data used in the House versus the Senate results — I am really using two similar, but definitely different algorithms.

(For those who would like to know a short history of these algorithms, you can find it here.)

Out of curiosity, I used the House algorithm to predict what the electoral college vote would be if Obama was running for re-election on Tuesday. The results are that he would win 175 electoral college votes, 95 short of the 270 he would need to win.