I’ve updated the numbers from the last post. Here are the changes in a nutshell:
- I accidentally gave Obama 3 votes from Montana in the last post, so he was actually only winning by 190 votes, not 196.
- North Dakota is now leaning Obama with 2.5%. This brings Obama to 196 if no Bradley Effect exists.
- Florida dropped from 5% to 4.6%, which means that if the Bradely Effect is 5%, Obama only wins the election by 34 votes instead of 96 (FL27 + ND3 away from O and toward M).
- If the Bradley Effect is 6%, Obama wins by 16 votes instead of 22 (due to my ND error).
- If the Bradley Effect is 7%, Obama wins by 16 votes because Virginia bumped from 6.5% to 8.1% in the past three days, thus moving 13 votes from McCain to Obama.
- If the Bradley Effect is 8%, McCain wins by 2 votes (ME4 and NM5 away from O and toward M).
So, based on current polling numbers, Obama could take a 7% dive in every state, and he’d still win the election. This adds a full percent from the previous post!
These numbers assume a lot of things (that everyone polled will vote, that all states suffer equally from the Bradley Effect, etc.), but I think this goes a long way to demonstrating the statistical tolerance that Obama can have before it costs him the election.
Here is the data file: joshandersonpollingdata1017