Nationwide, the US election was primarily a story of Democratic underperformance rather than huge Republican gains compared to 2020 – but in the swing states that ultimately decided the victor, it was the opposite story, with Trump’s gains far outstripping Harris’s losses.
Across the US, Democrats lost more total votes overall compared with 2020 than Republicans gained: Harris attracted 1.4m fewer votes than her Democratic predecessor, Joe Biden, did, while Trump attracted 1.1m more than he did in the previous election.
The figures were calculated by looking only at counties that have 100% of their precincts reporting and at least 95% of their estimated ballots counted, and comparing the vote in those areas to 2020.
Another way of looking at the numbers is that for every 78 votes Donald Trump gained nationally compared to 2020, Kamala Harris lost 100.
But in the seven swing states – Arizona, Georgia, Michigan, Nevada, North Carolina, Pennsylvania and Wisconsin – there was an inverse trend: the Democratic vote dropped very slightly but held up quite well compared to 2020, but Trump made enough gains to give him the White House. A large part of Democratic campaign spending was focused on the swing states, suggesting that this helped buoy up Democratic support – but not enough to overcome a wave of additional Trump voters.
At least 24 states also saw a larger drop in Democratic votes than any movement in Republican votes compared with 2020 (looking only at areas where counting was almost complete).
These included Ohio and Illinois. The Republicans lost more than 30,000 votes in both of these states, but this drop was dwarfed by the Democratic underperformance: Harris fell 200,000 votes in Ohio and nearly 150,000 votes in Illinois. In Virginia, the GOP gained one vote for every five lost by the Democrats.
Some states bucked the trend. Texas, for example, saw enthusiasm for Donald Trump outstrip the Democratic decline in support, with Trump gaining seven votes for every five lost by Harris. Michigan and South Carolina also saw larger Republican vote gains than Democratic losses.
But Democratic support dropped further than any change in Republican vote in at least 1,034 out of 2,480 counties analysed. (There were still 664 counties with more than 5% of votes yet to be counted as of 3:30am ET.)
Harris recorded a fall in 81% of counties (2,011 of 2,478). The party gained votes in just 19% of them.
Trump, meanwhile, saw a decrease in his vote total in 40% of the counties, while gaining in 60%.
The trend continued in the (relatively small) number of counties that flipped to Trump: of the 47 counties Trump won that Biden previously held, 31 were because of the drop in Democratic vote, rather than any significant change in Republican vote.
The significance of the Democratic collapse was especially apparent in Harris county, Texas. There, Democratic votes looked to be down by more than 100,000 while Republican votes increased by only about 20,000 – nowhere near enough to win the county had the Democrats not lost so much ground.
Similarly, Starr county flipped from the Democrats to Republicans, but not through a huge increase in the Republican vote (which rose from 8,247 to 9,443) but because a sharp fall in the Democrat vote (from 9,123 down to 6,845).
In total, there were 1,045 counties that recorded a drop in Democratic votes and an increase in Republican votes.
In the swing states – Michigan, Wisconsin, Pennsylvania, Arizona, Nevada, Georgia and North Carolina – the story was one of increasing Republican votes. About 303 counties had seen larger increases in the Trump vote than any major change in the numbers voting Democrat. Meanwhile, 66 counties saw outsized shrinkage of Democratic votes.
Nationally, turnout is expected to be down slightly on the record-breaking turnout recorded in 2020. But turnout in the swing states of Georgia, Michigan, Wisconsin, Nevada and Arizona looks likely to surpass that level – mainly caused by an increase in Trump voters.
Votes are still being counted and verified in many counties, meaning the figures in this analysis may change slightly.
