Trump is right to be furious: a federal error could cost him the election

Donald Trump
Donald Trump: the polls point to an extremely tight race - Scott Olson/Getty Images North America

Democracy is a strange sport; 156 million Americans vote, and in the end the result is potentially decided by a few hundred thousand people miscounted four years ago.

A series of errors in the 2020 US census has effectively taken electoral college votes from states likely to back Donald Trump, and apportioned them to states likely to vote for Kamala Harris instead. With the polling margins razor thin, this could be enough to tip the balance from the Republicans to the Democrats.

States receive electoral college votes based on the number of Senators and Representatives they send to Congress, with the number of Representatives in turn being based roughly on a state’s population. If the census is wrong, so is the resulting distribution of seats.

The 2020 census – disrupted by the pandemic – is estimated by the Census Bureau itself to have undercounted the population of Arkansas, Florida, Illinois, Mississippi, Tennessee, and Texas, while overcounting the populations of Delaware, Hawaii, Massachusetts, Minnesota, New York, Ohio, Rhode Island and Utah.

Donald Trump and his supporters should be seething at this. All but one of the undercounted states has voted Republican in recent elections: Florida since 2016, Arkansas and Tennessee since 2000, Mississippi and Texas since 1980. The only exception is Illinois, which has been blue since 1992.

Meanwhile, the overcounted states lean largely blue. Delaware – home to President Joe Biden – has voted Democrat since 1992, Hawaii, Massachusetts, New York and Rhode Island since 1988, and Minnesota since 1976. The exceptions are Ohio (red since 2016) and Utah (1968).

Feed these over- and undercounts through the system used to apportion electoral college votes, and analysis from the Heritage Foundation suggests that the results look something like this: Colorado was given one elector more than it should have received, Florida received two too few, Texas one too few, while Minnesota and Rhode Island each cling on to a vote they should have lost.

Had the Census found just 26 fewer people in Minnesota, then the state would have missed out on an elector; it’s now believed that the population was overcounted by about 217,000. Similarly, Florida and Texas needed about 172,000 and 189,000 more residents respectively to each get an additional vote; they were undercounted by approximately 761,000 and 560,000 respectively.

With polling for the election on a knife-edge, it’s not hard to draw out a scenario where this is enough to sway the overall result. Texas and Florida are expected to vote Republican; Minnesota, Colorado and Rhode Island Democrat, depriving Trump of three electoral college votes while handing three to Harris.

If Trump wins in the swing states of Arizona, Georgia, North Carolina and Nevada, while Wisconsin, Michigan and Pennsylvania break to Harris – a result that would only take a fractional polling shift in a single state – then the Democrats will win the election with 270 electoral college votes to 268, and Donald Trump would have been deprived of victory by the errors of the federal bureaucracy.

Hopefully, this won’t matter. As the pollster Nate Silver has pointed out, it’s more likely that on the night we will find out the polls were slightly off in one direction or another, and that the vote will turn from a neck and neck race into a one-sided parade. But the idea that an election could be determined by simple errors in the census rather than by the actual votes of American citizens is dispiriting.