In 2016, Donald Trump became the fourth person in United States history to become president while losing the popular vote, possible because the U.S President is selected by the Electoral College, which awards electors to each state, rather than by popular vote. This system leads to smaller states being dramatically over-represented in the process of choosing the next president and larger states being very under-represented. In fact, while Joe Biden won the presidency this year while winning the popular vote, the gap between the electoral college and the popular vote only grew.
Since 1828, the first election in which the popular vote was accurately tabulated—at least mostly, as there were some issues with Delaware—the gap between the popular vote and the electoral college has always been obvious.
This graph shows the percent of the popular vote won in each election since 1832 by the winning candidate along the horizontal axis and their respective share of the electoral vote along the vertical axis. As you can see, the percent of the popular vote won by a candidate is only moderately correlated with the percent of the electoral vote they win. In fact, variation within the national popular vote only accounts for around 46.7% of the variation in the electoral college vote, a shockingly low number. There is one particularly notable outlier that displays just how bad the electoral college could be—in 1980, Ronald Reagen won a mere 50.75% of the popular vote but captured a staggering 90.89% of the electoral vote, a gap of over 40%. A similar gap appeared in 1912, but that was complicated by the fact that three competitive candidates were vying for president that year.
This chart shows the difference between the percent of the electoral vote won by the winning candidate and the percent of the popular vote they won. As this graph shows, the gap is rarely below 5%, and 74% of the time, the gap is over 10%. Fully 42% of the time, the gap is over 20%.
This is a result of both the bias of the electoral college towards small states and the winner-take-all-method utilized by 48 states and D.C, as well as the individual congressional districts in Nebraska and Maine, which divide their votes up to the district level and don’t go by state. Because of this, winning 50.1% of the vote in a state gives you 100% of the electors in that state, meaning a victory of 1 vote in California gives you 55 electoral votes while receiving 1 less vote would give you none.
However, this can make the correlation look artificially low, because elections, where one candidate wins by a large majority will inevitably look even less close than they actually were. One way to adjust for this is to use the metric of tipping point states.
The tipping point state is the pivotal state in an election. In other words, if the winning candidate lost every state they won by a smaller margin than the tipping-point state, and won every state they won by a larger margin, and the tipping point state itself, they would win the election, but if they lost all of the states they won by a smaller margin, plus the tipping-point state, they would lose. This, then, works both ways, as the losing candidate would win if they won every state they lost by a smaller margin, plus the tipping point.
The graph above represents the winning candidate’s margin in the tipping point state and the popular vote. This is obviously a much stronger correlation, but it still displays a fundamentally flawed democracy. Around 79.8% of the variation in the way the tipping point state votes is explained by the national popular vote, which definitely shows a connection, but is also somewhat low in the scheme of things.
Two groups of dots in this chart stand out. First, there are the presidents who lost the popular vote and still became president. There are actually a significant number of winning candidates who have failed to reach 50% of the vote due to small amounts of people voting for third party candidates, beyond just the four who finished behind their closest competitor in total votes. Because of this, we will literally never know if the people wanted these candidates elected more than their competitors, one of the fundamental arguments for ranked-choice voting, which I wrote about last month.
The lowest percent of the vote any candidate has won while still becoming president since 1832 is Woodrow Wilson’s 41.84% showing in 1912. Based on how politically similar his opponents were, he probably would have lost if only one of them ran against him. For what it’s worth, Wilson also failed to secure a majority of votes in his second bid for the presidency, and won again.
The second group are the dots particularly far from the trendline. While these outliers might bias the overall trend, this is actually somewhat less of a big deal in this particular instance than it usually is in statistics, because the potential for those outlier events is probably more important than the overall correlation. For example, if the tipping point state was always within 3% of the popular vote, but the variation in the popular vote only explained 70% of the variation in the tipping-point state, that would actually probably be more comforting to know than seeing a graph where the tipping point state was occasionally 15% away from the popular vote, even if the variation in the popular vote explained 80% of the variation in the tipping point state.
The table above displays the difference between the percent of the popular vote won by the winning candidate and the percent of the vote they won in the tipping point state. Positive numbers indicate that the winning candidate did better in the popular vote than the tipping-point state, and negative numbers indicate that they did worse. The darker green the box is, the greater the disparity between the two values.
The highest disparity is a stunning -11.8% in 1836, when Martin Van Buren won the national popular vote by a commanding 14.2%, but only won the tipping point state of Pennsylvania by 2.4%. For context, that big of a gap working against the popular vote winner would have been enough to sink every president since Reagan ran for reelection and all but two campaigns that won the popular vote in the last 50 years.
While the gap has generally been smaller in the recent past due to high levels of partisanship and more intelligently run campaigns that seek to minimize the difference by campaigning more, spending more money in, and tailoring policies to the states more likely to be the tipping point, it still persists. In fact, the -3.9% gap Biden faced this year would have been enough to ensure the candidate that received the higher share of the popular vote would end up losing in four of the last six elections. Hillary Clinton in 2016 and Al Gore in 2000 lost anyway, but Barack Obama in 2012 and George W. Bush in 2004 also would not have been able to withstand that size of a gap.
The main focus of this article was on the presidency, but that doesn’t mean the House of Representatives and the Senate are better. The House is badly gerrymandered, with Republicans currently only needing to win around 48% of the vote to win 50% of the seats, with the gap expected to grow after redistricting in 2021. The shapes of the districts are absurd enough that there is even a font made entirely out of real House districts.
And the Senate is the worst of all. In fact, the Senate was intentionally designed by the founders to not reflect the will of the people, which may have seemed like a good idea when democracy was untested, but now is archaic, backward, and anti-democratic. No country in the world has a system like here, where arbitrary units of land are given equal representation instead of people.
As this graph shows, California’s Senators represent 1.5 times more people than Texas’ Senators, and 66 times more than Wyoming’s. Sixty-six! The bar for Wyoming is so small on that graph you can hardly see it.
This means that every piece of legislation ever proposed, every Supreme Court Justice ever confirmed, every tax ever passed, goes through a legislative body where people from Wyoming are considered to be worth a little over 66 times what people from California are worth.
This also means what the general public wants doesn’t matter that much to the Senate.
Right now, Democrats represent 56% of Americans in the Senate, but Republicans are likely to control 52% of the seats, which in a system as partisan as ours means they get 100% of the control.
44% of America, with Wyoming represented 66 times more than California, decides 100% of the legislative agenda of the next Congress. This is completely absurd.
America needs a massive, unparalleled overhaul to its political system. We need ranked-choice voting, an end to gerrymandering, and the abolition of the electoral college. Something drastic needs to be done about the Senate. And the thing is?
We can try to vote on it.
But even if a clear majority of people want it, that doesn’t mean it will happen. Not in America.
CORRECTION: an earlier version of this article misspelled the name of former president Ronald Reagan.