The Constitution provides that when it comes to presidential elections, "[e]ach state shall appoint, in such manner as the Legislature thereof may direct, a number of electors . . . ." State legislatures have great discretion in deciding how to choose electors.

Forty-eight states and the District of Columbia choose the same process: the candidate who receives the plurality of votes in a statewide election wins all of that state's electors. (That is, all the state's pledged electors; electors, of course, can be faithless, as some were this election.) In Maine and Nebraska, the system is slightly different--two electors are award on this basis, and the remaining electors (two in Maine, three in Nebraska) are awarded to the winner of each congressional district.

In the 2016 presidential election, this system yielded 306 pledged electors for Donald Trump and 232 for Hillary Clinton. (The numbers changed slightly because of faithless electors.)

But there are many ways that electors might be awarded. What would the 2016 presidential election have looked like under other systems? Of course, the candidates and voters would have behaved differently, but it's worth exploring how even some seemingly small changes to the rules could dramatically alter the outcome. I created a few scenarios and assumed that all 50 states and the District of Columbia followed them. (Again, these all projected a pledged elector total.)

First, a purely proportional system. In each state, I divided the total number of electoral votes by the percentage of votes received. I used ordinary rounding rules. In a couple of states, that yielded one too many votes, and I deducted it from the candidate with the smaller fraction that was rounded; in others, it was a vote short, and I gave it to the candidate that was the statewide winner. There was no floor, so it made it much easier for third-party candidates to secure a vote.

Clinton 265
Trump 261
Johnson 10
Stein 1
McMullin 1

Because no candidate received 270 electoral votes, the election would have been sent to the House, which would have chosen among Clinton, Trump, and Johnson. The Senate would have chosen between Kaine and Pence.

Johnson would receive two electors from California, and a scattered number of votes in other states. Stein's single electoral vote would have come from California; McMullin's from Utah.

Second, a proportional system among candidates who received at least 15% of the vote. That ensured that only sufficiently "serious" candidates received the vote. This is not unlike many state presidential primaries.

Clinton 270
Trump 267
McMullin 1

This scenario would have been a nailbiter to the end, where rounding errors in some states (or, perhaps, in my own spreadsheet!) might have swung the election into a tie or to Mr. Trump. Assuredly, recounts would abound for those fractions that would have swung an elector from Mrs. Clinton to Mr. Trump.

Third, a winner-take-all if a candidate exceeds 50% of the vote, otherwise proportional among candidates who received at least 15% of the vote. Currently, states award their electoral votes to the winner of a mere plurality. But perhaps a state would choose to award its electoral votes on a winner-take-all basis only if that candidate secured a majority of the vote, not merely a plurality. Otherwise, electors would be awarded proportionally (here, using the 15% floor). This is also similar to how many states award delegates in presidential primaries.

Trump 277
Clinton 260
McMullin 1

Fourth, an elector per congressional district and two electors to the statewide winner. This would mirror the Maine and Nebraska method, but instituted nationwide. Of course, in states with a single at-large congressional representative (or the District of Columbia), three votes would go to the statewide winner. The heavy lifting has been done elsewhere to calculate these totals, leaving the simpler task to this compilation....

Trump 290
Clinton 248

Fifth, state legislatures select the presidential electors themselves. This was a common way of selecting electors through the 1824 election. I looked at the partisan composition of all state legislatures in November 2016 and assumed they assembled in a joint session to choose electors. I then assumed partisan allegiances would stand (no small assumption this election!): Democratic members would vote for Mrs. Clinton, Republican members would vote for Mr. Trump. (I assumed Congress would empower the District of Columbia's Council to choose its electors.)

Trump 338
Clinton 200

Of note, the outcome looks almost no different than the billions spent on a popular election. The Minnesota, New Hampshire, Nevada, and Virginia legislatures had Republican control, and their electoral votes would have been given to Mr. Trump (under the assumed scenario). Apart from Maine's district system, that's the total difference in the outcome. The slimmest margin came from Washington state, where the Senate was controlled by Republicans by a single vote, and the House by Democrats by two votes, giving Mrs. Clinton all 12 electors by one vote.

You can examine the actual results of the 2016 election on Wikipedia. Now, below, are the detailed state-by-state totals for each of these five scenarios.