mplsinneoi

Reblogged from jakke

Great analysis.

As requested, here are the election prediction maps for the 2012 US Presidential election. Click on the maps to enlarge them.

Because it’s not clear whom the Republican candidate is going to be, I’ve mapped things out for Rick Perry, Mitt Romney, Michelle Bachmann, and Ron Paul. I’m using all the two-way matchup polling data I could track down from the last two months. This is hopefully going to be a semi-regular feature around here, depending on the availability of polling data and also on how the nomination race progresses.

The model I’m using here is called national uniform swing. Basically, it works like this: I take national level polling data, and then assume that all the shifts in national numbers since 2008 correspond to uniform shifts across the entire country. (I realize this is a big and dubious assumption, but in practice a national uniform swing model does a pretty decent job of predicting changes. Once the Republican candidate is determined then it will be more worthwhile to look at state-level data.) Anyway, after that I use the margins of error on the polling data to generate win probabilities for each state, and then translate that into electoral college votes. Nebraska and Maine’s special system for allocating electoral votes is included here, but I haven’t drawn the results on the map for the sake of clarity.

Under current polling, note that Romney is the only viable candidate. The election is still a long ways away, and obviously lots of things could happen to shift the polling (e.g., a big new war, large-scale bank failures, Rapture, raptors) and also other candidates like Huckabee could conceivably enter the race - but right now, Romney is the only Republican candidate who has any chance of winning. Bachmann and especially Perry still have relatively low name recognition, so that could change as more people learn about them, but right now Romney is by far the most competitive candidate the Republicans could nominate.

Interestingly, with Romney as a candidate it’s entirely possible that the electoral college returns a tie, with 269 electors for each of Obama and Romney. You can read what ilyagerner has to say about how that would be handled. It would not be pretty.

Also, many states are entirely locked out of consideration in the presidential election, because they have a greater than 99% chance of sticking with the same party as last time. Only the following states (or districts) went with the same party as last time with probability less than 99% under any scenario:

Arizona

Florida

Georgia

Indiana

Iowa

Mississippi

Missouri

Montana

Nebraska (and its first and second districts)

Nevada

New Hampshire

North Carolina

North Dakota

Ohio

Pennsylvania

South Carolina

South Dakota

West Virginia

Don’t see your state on that list? Then there’s basically no Republican candidate who could make your state at all competitive. And many of these states only come into play if Romney is the Republican candidate. Notably missing are four of the five most populous states, which collectively account for about 100 million people.

If you have any questions or feedback (especially on the data visualization, because that’s something I’m really trying to work on), please let me know.

mplsinneoi

Notes