Alan Abramowitz has a good piece up at Sabato's Crystal Ball on swing voters in the 2020 election. He presents data for 2020 and earlier elections going back to 1952 using National Election Study data. Interesting stuff.
His basic argument is that swing voters have declined as a share of voters over time but still remain a significant and potentially decisive share of the electorate. He notes:
"It is clear that there are far fewer swing voters today than only 20 or 30 years ago. However, swing voters have not disappeared, and they can still play a crucial role in elections. In 2016, for example, although there weren’t many swing voters, there were about three times as many Obama-Trump voters as Romney-Clinton voters. That was probably enough to swing Michigan, Wisconsin, and Pennsylvania to Trump and hand him a victory in the Electoral College.
In 2020, there were even fewer swing voters than in 2016. However, it appears that swing voters favored Joe Biden over Donald Trump by about a two-to-one margin."
Also interesting is his breakdown of left-right issue preferences among consistent vs. swing voters.
"The contrast between the two distributions [of consistent and swing voters on left-right issue preferences] is striking. The distribution of voters who consistently supported one party is bimodal. Consistent Democratic voters are located overwhelmingly on the left side of the scale while consistent Republican voters are located overwhelmingly on the right side of the scale. Only about one-fifth of consistent voters are located near the center of the scale. In contrast, more than three-fifths of swing voters are located near the center of the scale. Only a small minority are located well to the left or right of center."
Not surprising I suppose but interesting nonetheless. Swing voters are still very much a thing.