In a blogpost yesterday on the Electoral College, NY Times columnist Ross Douthat reported that “if you believe Sean Trende’s fascinating analysis,” Richard Nixon won the popular vote in the 1960 presidential election. I hadn’t seen this claim before, and I’m a dork, so I popped over to take a look.

Trende’s piece, published over the weekend, notes that in 1960 Alabama voters didn’t vote for presidential candidates, but for individual electors, and that they we allowed to split tickets. There were eleven Democratic electors and eleven Republicans, and voters could choose anywhere from one to eleven candidates from the two columns.

Five of the Democratic electors were “loyal,” or pledged to Kennedy, while six were unpledged and eventually wound up voting for Senator Harry Byrd. Those eleven electors all won, and the Republican electors all lost.

The question Trende asks is how we should count these votes. (Since far more votes were cast than there were voters, we can’t just tally up the scores for each slate and do it that way.) The traditional approach has been to credit Kennedy with the number equal to his highest scoring elector, and Nixon with the corresponding total for his, but as Trende notes, that results in Byrd getting no votes in a state where he obviously had substantial support.

Instead, Trende suggests, it makes more sense to give all the Democratic votes to Byrd instead of Kennedy, or split them on the basis of how many electoral votes each received. In either case, he notes, the result is a Nixon victory in the overall popular vote. Nixon won!

But no, he didn’t. Here’s why:

Nationally, according to the source Trende himself relied on, Nixon received 34,108,157 votes in 1960, and Kennedy received 34,220,984. If we set Alabama aside — take all their votes off the table, and tally up the rest of the country without them, here’s what we’re left with:

Kennedy: 33,902,681

Nixon: 33,870,176

That’s a Kennedy victory of more than 32,000 votes. So in order to claim that Nixon beat Kennedy nationally, we have to argue that Nixon beat Kennedy by nearly 32,000 votes in Alabama. And that’s not what happened — Kennedy’s poorest-performing elector in Alabama received 316,934 votes. Nixon’s best-performing elector received 237,981.

It’s not easy to say how the votes should be carved up, since Kennedy had fewer electors on the ballot than Nixon. But the evidence suggests that the vast majority of those who went to the polls voted for all the electors they could. In a senate race on the same ballot, the two candidates received a total of 554,064 votes. If you divide the total number of votes cast in the presidential race by 11 — the maximum number of electors a single voter could support — you get 555,592.3 people voting, almost exactly the same.

The vast majority of Alabama voters in 1960 voted for a full slate of electors, and the strong majority of those chose the Democratic slate. Perhaps ten thousand or so voted only for Harry Byrd’s electors, and not for Kennedy’s, but the overwhelming number took what they were given from the party they supported.

Still not convinced? Let’s look at it another way. Figure that 600,000 Alabama voters went to the polls in 1960. That’s almost certainly too high, but it’s within the realm of plausibility, and approximates the best-case scenario for the ticket-splitting hypothesis. In that scenario, John Kennedy’s worst-performing elector received the support of 53% of the state’s voters, and Nixon’s best-performing elector received the votes of 40% of them.

Kennedy beat Nixon in Alabama. Kennedy beat Nixon outside of Alabama. Kennedy beat Nixon.

There’s no other way to spin it. Sorry.