I'm having a hard time imagining how you'd accurately measure who a non-voter would have voted for. Like how would you verify if someone is even eligible to vote? Lots of Americans (Eg green card holders, kids, felons) can't vote. Also 2% is way smaller than the Lizardman’s Constant: https://slatestarcodex.com/2013/04/12/noisy-poll-results-and...
It isn't pointless pedantry. The claim was that "a bit over half of all American voters" voted for Trump. This is simply factually untrue, as nowhere near all American voters even voted.
>I'm having a hard time imagining how you'd accurately measure who a non-voter would have voted for.
No claims are being made about the hypothetical votes by non-voters, so whether and how that metric can be measured isn't relevant to the claim being made, which is about people who actually voted. This is measured through a count of the actual votes tallied by state.
>Like how would you verify if someone is even eligible to vote?
Every state has their own methods to determine voting eligibility. This isn't relevant to the claim being made, as the set of "all American voters" is presumed to be equal to the set of "all eligible American voters."
>Lots of Americans (Eg green card holders, kids, felons) can't vote.
Individuals who cannot vote are also not relevant to the claim made, which is about the set of people who actually did vote, and what fraction of the entire electorate they represented.
It seems pointless pedantry.