Adam Liptak at the New York Times highlights the request for the Supreme Court to consider “faithless” presidential electors from the 2016 presidential election. The headline (not written by Mr. Liptak): “‘Faithless Electors’ Could Tip the 2020 Election. Will the Supreme Court Stop Them?”

It’s not a great headline. Could faithless electors sow chaos and discord into the 2020 presidential election? Certainly. Could they alter the outcome? That’s another matter entirely….

Below is a chart of “faithless” presidential electors since 1900. (This excludes faithless vice-presidential votes.) It’s tough to make apples to apples comparisons much earlier (or, indeed, even before World War I) because states often printed ballots with individual electors, to the extent voter expectation or reliance is a factor. But this nicely covers recent history.

It’s worth noting that there have been more “faithless” votes cast from losing candidates (11) than winning candidates (5). (If we included the 3 electors who actively attempted to vote for another candidate in 2016 but were replaced or revoted for the pledged candidate, the margin would rise to 14 to 5.) Even excluding 2016, the margin is 6 to 3. (That said, it’s hard to count the likely mistaken vote of a 2004 Minnesota presidential elector for “John Ewards [sic]” as truly “faithless” for the loser. UPDATE: An elector in 1948 was on two separate slates, one for the Democratic candidate and one for the Dixiecrat candidate, and even though the Democratic candidate won he cast hist vote for the Dixiecrat candidate. One can question whether this is “faithless,” too.) Of course, there are perils in such a small sample size. But it reflects that it’s easier for losing slates of presidential electors to protest or make a “statement” with their faithless vote. That is, there’s essentially no cost for electors of losing candidates to behave faithlessly—their candidate was already going to lose. Winning candidates, however, have much more to lose; we might expect electors to take their role more seriously (and faithfully).

Note, too, that the margin of victory can matter. In each race, I list the winning candidate’s presumptive margin of victory (i.e., how many votes the winner could spare to ensure he received a majority of the vote, and presumptive assuming no electors were faithless). Note that a faithless vote in 1972 was comically inconsequential: Nixon had a 251-vote margin of victory. Of course, 2000 is the opposite: George W. Bush could afford just one defection to retain his majority, as he had just 271 electoral votes and needed 270 to win the election. (Professor Robert Alexander has examined efforts from 2000 and beyond to court faithless electors, particularly in close matchups.) The higher the leverage of the situation, the less likely it’d be that an elector would behave faithlessly.

Win. That’s another caveat. If no candidate secures a majority, the top three receiving vote-getters proceed to the House, where the House votes and each state’s delegation receives one vote. That includes a 269-269 tie, which means, with no majority, those two candidates would go to the House.

Another way, then, to think about the faithless electors is to look at who the faithless electors cast their votes for. Did they try to make the second-place vote-getter (the runner-up) win the election? Or did they cast their vote for someone else?

Since 1900, exactly zero faithless electors have cast a vote for the runner-up. Faithless electors have cast a vote for a third-party candidate who placed third, or for a marginal candidate who’d otherwise receive zero electoral votes. They have never in recent history attempted to help the runner-up win the election.

Note how this second examination works. Faithless electors could deprive the winner of a majority. But to deprive the winner of the majority, and to give another candidate the majority (i.e., the power to change the winner) is something else altogether. And we have zero instances of any faithless elector ever attempting to do so.

Of course, past performance is no indication of future performance. But it’s another piece in our examination of presidential electors. Faithless electors would’ve needed astonishing coordination to pull off the feat, and they’re unlikely to be so malleable in the future.

It’s entirely possible that very aggressive courting of presidential electors can deny a candidate a majority. But to court them to change the winner? We have no evidence of that from extensive past practices.

Denying the majority to the winner sends the election to the House. In the 1824 election, the only post-12th Amendment election sent to the House, we saw the House choose the second-place vote-getter (John Quincy Adams) when no candidate received a majority of the electoral vote and Andrew Jackson received a plurality of the vote. It’s not clear how the House might handle an election where a candidate had the presumptive edge in the Electoral College and was denied the majority simply by faithless electors—it’s possible the House would play “constitutional hardball” and exercise its independent judgment; but, more likely, I think, is a vote for the presumptive winner to nullify the effect of the faithless electors. I could be wrong.

But, at best, faithless electors have historically sought chaos, not a different winner. I’ve explained some reason why they’ve done so. And I think it’s a reason we wouldn’t expect faithless elector to “tip” a presidential election.