Now, just to be clear, the argument that Richard Nixon should be credited with a popular vote win in 1960 doesn't rest on theories about dead people voting in Chicago or cows voting in Texas. It does rest on a fuller understanding of Southern voting history.
Before going further, credit where credit is due. This analysis isn't something I discovered on my own. Instead, it derives from a pair of articles published in PS: Political Science and Politics. The first, authored by University of Illinois at Urbana-Champaign professor Brian Gaines, appeared in the March 2001 edition of that journal. The second, by George Mason University professor Gordon Tullock, appeared in the January 2004 edition. Even back in 1960, Congressional Quarterly concluded that it was Nixon, not Kennedy, who had won the popular vote, for the reasons that follow.
If you asked your average political aficionado when the South began to leave the Democratic Party, the answer would probably be 1964. In truth, that exit has much deeper roots. A better starting date is 1938, when FDR conducted an unsuccessful purge of conservative Southerners. The Democratic share of the vote in the South steadily declined from that date forward, as the national Democratic Party fully embraced progressivism.
Things famously came to a head in 1948, when the Democratic National Convention (to Harry Truman's private consternation) adopted a pro-Civil Rights plank in the party's platform. The Southern delegation walked out of the convention and formed the Dixiecrat Party.
But -- and this is critical -- the goal of Dixiecrats was not to win the popular vote or Electoral College outright. They recognized this as impossible at a time when Reconstruction was still a living memory for many voters (in fact, the last Civil War veteran didn't die until 1956).
Rather, the Dixiecrats hoped to deny either party a majority of the electoral vote. That would throw the election to the House of Representatives, where each state is allotted one vote. The 11 states from the Old Confederacy would surely hold the balance of power in such an election, and could extract assurances on civil rights from whichever party wanted the victory the most.
It didn't come close to working (somewhat surprisingly, in retrospect), and there wouldn't be another major effort by a Southern candidate to split the Electoral College for another 20 years. But Southern states didn't give up their quest. In 1956, South Carolina, Alabama, Mississippi and Louisiana offered up "unpledged" slates of electors who would be free to vote for whomever they wished, and could make the difference in a close election.
This brings us, finally, to 1960. In that year, the canonical recitation advises us that Sen. John F. Kennedy defeated Vice President Richard Nixon in an incredibly close popular vote, 34,220,984 to 34,108,157. That's a difference of only 112,827 votes.
It's also inaccurate. Three states -- Louisiana, Mississippi and Alabama -- offered unpledged slates of electors. In Louisiana, the unpledged delegates came in third place to Kennedy and Nixon, receiving only 21 percent of the vote. In Mississippi, the unpledged electors won, edging out Kennedy by three percentage points; those electors eventually voted for Sen. Harry Byrd of Virginia.
But Alabama did something very, very different. At the time, voters there did not cast ballots for Democratic or Republicans tickets. Instead, they cast 11 votes, one for each elector from the state. Thus, it was possible to cast six votes for Republican electors, and five votes for Democratic electors, if one so chose.
Those electors had been selected by the parties in the primary. In the 1960 Alabama Democratic primary, 24 electors ran as unpledged, refusing to be bound by the decision of the Democratic convention. They faced off against 11 "loyalist" candidates, who agreed to accept the national candidate. This actually gave the loyalist forces an advantage; there were 11 slots in the Democratic slate, so the odds were greater that the unpledged electors would lose out by having their votes divided too many ways.
But after an election, an extremely close run-off, and a re-count, unpledged electors claimed six of the 11 slots for the Democrats, while loyal electors were awarded the remaining five slots.
In the fall, all 11 Democratic electors defeated the Republican electors. As promised, the five loyal electors eventually cast their ballots for Kennedy. As they suggested they would, the unpledged electors joined their Mississippi neighbors in voting for Byrd.
But the popular vote? It was a mess. After all, some people cast as many as 11 votes, and others case as few as one. We can only estimate that about 550,000 people voted overall. The end result is that the six unpledged Democratic electors each received between 320,957 and 324,050 votes, totaling 1,934,826. The five loyal Democratic electors each received between 316,934 and 318,303 votes, totaling 1,587,900. And the 11 Nixon electors each received between 230,951 and 237,981 votes, totaling 2,588,790 votes.
So how do you count this up? The method most frequently used is to award Kennedy 318,303 votes, representing the highest number of votes cast for a Kennedy elector. Nixon is awarded 237,981 votes, representing the highest number of votes received by a Nixon elector. Others award Kennedy 324,050 votes, representing the highest number of votes cast for a Democratic elector.
This first way is certainly defensible -- after all, a Kennedy elector did receive 318,303 votes in the state, and from a national perspective, it was an election between Kennedy and Nixon.
But was it the best way to do this? For starters, we end up with the rather absurd result that Harry Byrd received a majority of the electoral votes from the state, but is credited with zero popular votes.
In fact, if we are going to insist on awarding the state's popular vote to one Democrat or the other, it probably makes more sense to award it to Byrd and not to Kennedy. After all, his electors received the most votes.
Moreover, awarding the Democratic popular vote from Alabama entirely to Kennedy ignores the relevant electoral history: Electors who were understood to support a "Dixiecratic-ish" candidate won a majority of the slots in the state's Democratic primary, and probably would have swept the ticket had their votes not been split 24 ways to the loyal electors' 11.
The fact that many supporters of the unpledged delegates clearly preferred Kennedy over Nixon as a second choice, when forced to make such a pick, really doesn't change another fact, which is that unpledged delegates were, overall, the most popular choice in the state that year -- not once, but twice.
If you award the Democratic popular vote in the state wholly to Byrd rather than to Kennedy -- again, probably more defensible than awarding the popular vote in the state wholly to JFK -- Nixon wins the popular vote by 205,476 votes. (However, even with Nixon gaining Alabama's 11 electoral votes, Kennedy's election would have stood: His EV margin would have shrunk only from 303-219 to 292-230.)
But the bottom line in Alabama is that there really were Democrats who supported the national ticket, and there really were Democrats who supported the Dixiecrats. Had there been options for 11 Kennedy electors and 11 free electors, thousands of votes would have been cast for both, as was the case in Mississippi (although Nixon might well have won the state in that event). Allocating all of the popular votes to Kennedy or Byrd ignores this reality.
You could also award Byrd 324,000 votes in addition to Kennedy's 318,000 votes (and Nixon's 237,000 votes), but then you are allocating hundreds of thousands more votes than there were voters.
Probably the fairest way to allocate the votes -- a method proposed by Gaines -- is to add up the ballots cast for the 11 Democratic electors, and then allocate six elevenths of the total to Byrd and five elevenths to Kennedy. This reflects the reality of the state's Democratic Party: It was split between national party loyalists and Dixiecrats.
Adopting this approach results in a Nixon victory of around 60,000 votes, which is how Congressional Quarterly originally calculated the results.
In the end, there are three ways to count the popular vote in Alabama: Allocate all Democratic votes to Kennedy, allocate all Democratic votes to Byrd, or allocate the Democratic votes proportionally between the two candidates.
Two of those three methods result in a Nixon victory in the national popular vote. Historians choose the one that results in a Kennedy win. I don't think this is because of any conspiracy, nor is it due to bias. At the same time, though, I don't think it's because awarding Kennedy all of those votes is the best method either. Rather I think it's just due to a lazy counting of votes for Kennedy electors, combined with inertia. It's probably time for electoral historians to revisit that.
Of course, the most important thing to remember is that we don't award victories by popular vote, and that campaigns structure their strategies accordingly. Absent an Electoral College, Kennedy probably wouldn't have selected LBJ as his running mate and instead would have made a play in the vote-rich Midwest. But as a matter of historical accuracy, there are almost certainly five instances where the candidate won the popular vote, but lost the Electoral College.
Right now the RCP Averages are showing an odd situation. Mitt Romney leads nationally by one point, but trails in the Electoral College by a 294-244 count. Moreover, electoral vote number 270 (right now, Wisconsin) favors President Obama by a two-point margin.
While I believe that an electoral vote/popular vote disconnect of this magnitude is unlikely, it certainly is possible that we'll see another split between the two, especially if the popular vote is decided by less than a point. If that happens, Americans will once again receive a civics lesson in how presidents are really chosen.
In particular, we'll be reminded of the four canonical instances where the electoral vote and popular vote went to different candidates: 1824, 1876, 1888, and 2000. These are fairly well known to political junkies.
Far less well-known is that we should probably include a fifth such split: 1960.