The Biggest Democratic Swing We Ever Recorded Was a Republican Collapse

The largest single-cycle move toward Democrats anywhere in our data is Salt Lake County, Utah, in 2016: a relative swing of 34.84 points. It is not close. Across the 18,671 county-cycles with a prior cycle to measure against, nothing else touches it. Four of the seven largest pro-Democratic swings among counties casting at least ten thousand votes are Utah counties in that one year.

Here is what happened underneath that number. Between 2012 and 2016, Democratic votes in Salt Lake County rose from 146,147 to 175,863 — a gain of 29,716. Republican votes fell from 223,811 to 138,043 — a loss of 85,768. Votes for everyone else rose from 11,841 to 104,962. Of the total movement between the two major parties, 74.3 percent is Republicans leaving. Our headline figure describes a Democratic surge that Democrats mostly did not receive.

The rule that produced it

That rule is correct and we are not proposing to abandon it. It also has a failure mode, and Salt Lake 2016 is the purest case of it in twenty-four years of returns. When a candidate takes a quarter of a county's ballots overwhelmingly from one side, removing that candidate from the denominator does not remove the distortion. It concentrates it. The party that lost those votes shrinks against a party that did not grow, and the metric books the difference as movement.

One thing the convention never does: name the wrong winner. We checked all 21,783 county-cycles in the file. In every one of them, the candidate leading on the two-party margin is also the candidate who led on all ballots cast. The distortion is not about who won. It is about how much moved, and in 2016 that error is enormous.

2020 is the control

Four years later, Salt Lake County Democrats added 114,043 votes — 64.9 percent more than 2016, and nearly four times the raw gain the 2016 swing was built on. That is what an actual Democratic surge looks like in this county. Our metric recorded 2020 as a relative swing of 2.87 points toward Republicans.

2016: Democrats gained 29,716 votes and we recorded a 34.84-point move toward them. 2020: Democrats gained 114,043 votes and we recorded a 2.87-point move away.

The clean way to see it is each party's share of all ballots cast, which no third-party rule can inflate. Measured that way, the Democratic share of Salt Lake County rose 3.71 points in 2016 and 11.05 points in 2020 — the later gain nearly three times the earlier one, and that understates it, because a third-party surge depresses every share it does not receive. The order of those two cycles is the reverse of the order our leaderboard puts them in.

Utah in 2016 is not a normal cycle

How unusual this is can be stated exactly. Of 21,783 county-cycles in the file, 17 have a third-party share at or above 20 percent. Fourteen are 2016. Twelve are in Utah. The highest we have ever recorded anywhere is Cache County, Utah, 2016, at 35.65 percent. Statewide, Utah's non-major-party vote went from 25,027 in 2012 to 305,523 in 2016 — 27.00 percent of all ballots, against a national 5.61 percent, both summed from the county file. Utah's Republican vote fell by 225,369 over the same span.

In Salt Lake, Cache, Davis and Utah County the Republican loss runs from 2.9 to 6.0 times the Democratic gain, and the third-party share sits above a quarter of all ballots in each. The bottom two rows are different, and they belong here. In Summit County, Democrats gained more votes than Republicans lost — 2,431 against 1,551 — on a third-party share of 13.59 percent, roughly half the state's; that swing is mostly a real one. In Wasatch the two are close, 872 against 1,105, and the third-party gain of 2,764 is larger than either. The rule does not fail everywhere it could.

We cannot tell you who got those votes

This is the part that should make a reader cautious about the whole exercise. Our source file for 2016 carries exactly three candidate labels nationwide: Donald Trump, Hillary Clinton, and OTHER — 3,158 rows each. Every minor candidate in the country that year is collapsed into one bucket. We can tell you that 104,962 Salt Lake County ballots went to someone other than the two major nominees. We cannot tell you who, or how they divided. We are not filling that gap from memory, and no number here depends on filling it.

What we would change

Not the metric. Reintroducing third-party votes into the denominator would corrupt every cycle to fix one. Run 2016 that way and Utah County's swing goes from 25.61 to 43.84, and Cache's from 29.21 to 42.63 — on the counties where the third-party surge was largest, the artifact gets bigger, not smaller. Salt Lake argues against the change as much as for it: under the all-ballot rule its 2020 swing is −0.41 rather than the −2.87 we publish, so the alternative would have buried the very correction this piece is about.

What should change is the labelling. A relative swing computed in a cycle where a quarter of the ballots went somewhere our data cannot see is a different kind of number from one computed where 3 percent did, and we currently present them identically — same column, same leaderboard, same flip ledger, which logs Salt Lake as flipping Democratic in 2016 on this basis. Seventeen county-cycles need a flag. That is a cheap fix and we have not made it.

The 2016 reading was not wrong about where Salt Lake County was heading. Its lean stood at D+9.84 in 2016 and D+12.05 in 2024, further from the nation on the Democratic side than the artifact year itself. The destination was real. The route we recorded was not, and 34.84 of Salt Lake's 35.60-point total shift since 2000 — 97.9 percent of it — is booked to a single cycle in which Republicans left and Democrats barely arrived. We published the four states we got wrong. This is one we are still publishing.