As most politically aware observers know, Donald Trump lost the 2016 presidential popular vote to Hillary Clinton. In fact, Mrs. Clinton won the popular vote in 2016 by about 2.1% points, which is roughly in line with the 2.4% popular vote margin of victory that George W. Bush secured in the 2004 election. Mr. Trump won the presidency thanks to his relative strength in the Electoral College; he secured 306 pledged Electoral College votes, a majority of the 538 cast. To understand his advantage in the 2016 electoral vote, it helps to understand the concepts of the tipping point state and the Electoral College Lean.

## The Tipping Point State and Electoral College Lean

The tipping point state, a concept popularized by Nate Silver of FiveThirtyEight, is the state which yields the winning candidate their decisive 270th Electoral College vote if we order all the states by the winner’s margin of victory. Take 2016 as an example. We can line the states up by Mr. Trump’s percentage margin of victory, starting with Wyoming, a state he carried by roughly 46 percentage points. Wyoming gives Mr. Trump his first 3 electoral votes. His next largest victory was in West Virginia which netted him 5 electoral votes, bringing his total to 8. Continuing in this manner we eventually reach Wisconsin, a state that Mr. Trump only carried by 0.8%. With Wisconsin in his column, Mr. Trump reaches exactly 270 pledged Electoral College votes. Thus Wisconsin was the tipping point state in the 2016 election and the tipping point margin was 0.8%.

Comparing the tipping point margin to the national popular vote is a useful exercise. Basically, it tells us how far away the losing the candidate was from winning in the Electoral College.

Let’s measure the disparity between the popular vote and the Electoral College vote with a metric that I am calling *Electoral College Lean.* Electoral College Lean is defined as the winning candidate’s percent margin of victory in the tipping point state minus that candidate’s percent margin of victory in the national popular vote. Thus a positive Electoral College Lean means that the winning candidate did better in the Electoral College relative to the national popular vote. A negative Electoral College Lean has the opposite interpretation.

Take 2016 as an example. Mr. Trump won Wisconsin, the tipping point state, by about 0.8% while losing the national popular vote by 2.1%. So the 2016 Electoral College Lean was 2.9% in favor of the Republican candidate. By mapping out the Electoral College Lean in each election since 1976, we can see that this 2016 Lean was unusually large.

Prior to 2016, the largest Electoral College Lean since 1976 was 1.8%, which occurred in both the 1980 and 2008 elections. Interestingly, in neither of those elections did this Electoral College advantage affect the final outcome. In 1980, Jimmy Carter lost the popular vote by a whopping 9.7% points to challenger Ronald Reagan; the election was not close enough for Mr. Carter’s relative advantage in the Electoral College to matter. In 2008 Barack Obama enjoyed a 1.8% Electoral College Lean over John McCain, but he won the election by a wide margin and did not need this advantage. In contrast, Mr. Trump’s 2.9% Electoral College Lean in 2016 was decisive in his victory.

The 2.9% Electoral College Lean in 2016 dwarfed the average absolute Electoral College Lean of roughly 1.0% (and this average includes 2016). So 2016 was very much an abnormal year in terms of a popular vote Electoral College split.

I should note that that the Electoral College has *not* systematically favored the Democrats or Republicans in the elections since 1976. The average Electoral College Lean, including party in our calculus, was basically 0% over this time period (it was 0.05% in favor of Democrats if we round to two decimal places). Moreover, in the three elections preceding Mr. Trump’s 2016 win, Democrats enjoyed the relative Electoral College edge; Mr. Obama had Leans of 1.5% and 1.8% in his two victories. Of course, this may be little comfort for Democrats considering that they have been on the wrong end of an Electoral College/popular vote split in two of the last five presidential elections.

## What Does This Mean for 2020?

The natural question for the 2020 presidential election is will the Electoral College advantage that Mr. Trump received in 2016 persist? Will Joe Biden, the presumptive Democratic presidential nominee, need to win the popular vote by 3 percentage points or more to win the election?

At this early stage in the campaign, I’m hesitant to predict whether or not Mr. Trump’s Electoral College Lean will grow or shrink in 2020. Looking at history, we might suspect that 2016 was an abnormality and expect the Lean to shrink at least somewhat towards the mean. However, some early polling suggests that Mr. Trump may still enjoy a significantly larger advantage in the tipping point state than he does nationally.

Let’s take FiveThirtyEight’s newly released polling average which shows Mr. Biden currently up 9.2% nationally. Digging into their state polling, we see that Mr. Biden leads Minnesota by 6.6%; this would be the tipping point state in an election result that exactly mirrored FiveThirtyEight’s polling average. So Mr. Trump would have a 2.6% Electoral College Lean in such an election, though he would lose handily. Of course, much could change between now and the election in November; this is only a very imprecise early ballpark estimate.

Still, as a thought experiment, I think it’s useful to think about a hypothetical election where we know the tipping point is 3 percentage points to the right of the country, like in 2016. That is, suppose we know that Mr. Biden needs to win the national popular vote by at least 3 points to win the Electoral College. How big a lead would Mr. Biden need to have in the national polls to feel safe in such an election?

Well, let’s use the polling data discussed in this FiveThirtyEight piece by former FiveThirtyEight contributor Harry Enten (currently at CNN). Enten provides the final week polling average for every presidential election from 1968 to 2012, and we can calculate that the actual national popular vote is roughly normally distributed about the this polling average with standard deviation of about 2.9 points. Using this simple model of the popular vote, and the assumption that the Democrat needs to win the popular vote by at least 3 points to win the Electoral College, we can calculate how likely Mr. Biden is to win the election based on his final week polling lead.

Under the assumption that the tipping point state is 3 points to the right of the country, we see that Mr. Biden approaches a 90% chance of victory if he polls at about 6.5% or 7% above Mr. Trump nationally in the week before the election. A 9 point lead, approximately what Mr. Biden enjoys currently, would almost certainly result in his election in this scenario. A 3-5 point lead would be a much more tenuous advantage; Mr. Trump would have somewhere between a 25% and 50% chance of capturing the Electoral College in this model.

Of course, at this early stage we can only make educated guesses about each candidate’s relative strengths in the Electoral College. Still, we can see that Mr. Trump’s Electoral College Lean in 2016 was very large by historical standards. If he is able to to maintain or widen his relative advantage in this domain it would make the Democrats’ path to victory a more uphill battle.