Interestingly it looks like it didn't make a difference; adding the first two groups, or the second two groups, gives you numbers not that different from the 'total'.
(If only people who voted in 2018 were permitted to vote, Biden would've won by a little over 8 points, actually almost exactly replicating the 2018 performance. If only people who did not vote in 2018 were permitted to vote, Trump would've won the NPV outright by 2 points and won the election very comfortably. But 2016 actually isn't a very important dividing line.)
Hmmm... this kind of goes against the CW that Trump brings a bunch of supporters out who otherwise wouldn't vote.
