Sam Wang doubles down on the PEC model

[mark_twain]

Today's article at PEC:

http://election.princeton.edu/2016/11/06/is-99-a-reasonable-probability/


In this article, Sam Wang mentions that the 99% win chance of Clinton for today forecast by PEC may be a bit on the high side, due to his assumed allowance for polling error, which he defines as having SD of 0.8%.

But in a more recent analysis, he assumes that the SD for polling is 1.1%, which results in a Clinton win chance of 95%, instead of 99%.

In any case, the reasonable range of polling error, according to Sam Wang, will be at most 1.5%, which maps to a Clinton win chance of 91%, still far higher than 538's forecast, and even NYT's forecast, at the moment.

One thing I like about the PEC model is that the certainty of the prediction grows as we near Election Day, which is correct by common sense.

[mark_twain]

Cmon, a 99% chance?
We've had a miss of 9% in one of the past 20 elections (Dewey vs Truman).
To say a 4% lead has a 99% chance of holding is just silly.

hate to break it to ya, but things have changed since 1948

This week, Wang said that minor adjustments to the model can make Clinton's chance to win range anywhere from 91% to 99%.

So we are still at risk of a Trump victory, but according to the PEC model, Clinton is a safe bet against typical betting odds for today.