A post at Reddit looks at an interesting issue: How often each of the states has voted with the winner of a presidential race. The only problem is that the chart that was shared is hard to read and, frankly, ugly.

We decided to do one better.

Yes, New Mexico is the state that's voted with the winning presidential candidate most often -- 92.3 percent of the time since 1912, when it backed winning candidate Woodrow Wilson. It's only bucked the national trend twice, voting for Gerald Ford in 1976 and Al Gore in 2000. The latter, we'll note, was pretty close.

The least predictive state? Alabama. That is, if you exclude Washington, D.C. But we'll come back to that.


You'll notice we also broke out the results by another metric: How predictive states have been recently. After all, that New Jersey voted for the Federalist candidate in 1796 doesn't tell us a whole lot about what to expect more recently. So, the best predictors over the past 60 years have been Ohio and Nevada (the former missing 1960 and the latter, 1976).

Over the past 60 years, the Deep South and D.C. have continued to be outliers in predicting winners. Why? Because they almost always vote the same way.

If you compare the accuracy of predictions with the frequency of correct picks, you get a bell curve, with lower rates of prediction in more partisan places.


Which shouldn't be a surprise! Swing states (which we wrote about earlier) are the states that pick winners, and are, necessarily, states that move back-and-forth between parties. The swingiest states end up picking winners more of the time. They "swing" the election, if you will.

If you want to pick a president, move to New Mexico or Ohio and flip a coin on how you plan to vote. It's not an elegant system, but it seems to work.