Thursday, September 20, 2012

Election Model: Methodology Update

I've made a small improvement to the precision with which variance in the national simulation is calculated. It  now more accurately reflects the sample size of the RCP national polling average. As a result variance has slightly increased, which is good for the underdog i.e. Romney.

In spite of this change, President Obama's win probability continued to increase thanks strong polling the last few days, particularly in Wisconsin (now 93% to go for Obama), Michigan (92%), Ohio (90%), and Virginia (87%).

538s Senate forecast is up here, and I just learned today about a cool looking model for forecasting the House here.

No comments:

Post a Comment