Home > politics, Statistics > Pollitricks: More Election Doodle Dandy

Pollitricks: More Election Doodle Dandy

2012 October 29

This is a time series of the probability of a win by Obama (blue) or Romney (red) based on my monte carlo of battleground state elections based on the rolling ‘current’ polls of these states posted on RealClearPolitics. (eg Colorado). The state poll means are adjusted by 0.6 pts in favor of Republicans based on some analysis by Nate Silver of the 538 blog. The chart contains daily data starting from Oct 21 and reflects the ‘current’ polls at the end of the listed date. I have one earlier data point, Oct 13, from data I manually downloaded when I started this project. Currently, I update this chart manually once a day.

A quick chart of the state-by-state probabilities used in the Monte Carlo. Updates every two hours.

The EV monte carlo also updates every two hours. For details of production, see my previous post.

For those interested in comparing Python and R charting, make your browser wide and click here. I attempted to make these two charts as similar as practical. The R chart is in the upper left hand corner. Be aware that I have more experience with R than Python. Nevertheless, it appears to me easier to get an aesthetically pleasing chart in R than Python.

  1. Ned
    2012 November 16 at 6:39 am

    Is that right-most peak on the distribution the one representing the actual result (332 EVs)?

    Which direction was your 0.6% bias adjustment applied? Did you expect the state polls were biased towards R or D?

    Just curious.

  1. No trackbacks yet.
Comments are closed.