The implication by @reidepstein & the @nytimes that vote by mail has a partisan advantage, and that one election result "Upend[s] Old Theories" is just wrong, grossly irresponsible, and verging on journalistic malpractice. A correction or retraction is called for. https://twitter.com/gronke/status/1252629241469865985

One party mobilized its supporters to vote by mail, the other did not. To find more votes for the first party by mail than in person votes says *nothing* about the effects of moving to an *all* vote by mail system. Nothing. It holds zero inferential value.

All this article does is baselessly boost *false* theories about partisan advantages from all-vote-by-mail systems (used by states as politically distinct as UT and WA). In doing so it discourages moving to a system that can literally save lives.

Retract it. Now.

To state the methodological problems more clearly, consider this alternative hypothetical analysis, which uses exactly the same approach as the @nytimes piece

Imagine a state allows voters to cast a ballot using an orange-colored ballot or a purple-colored ballots. They count the same, voters can choose whichever they want.

Then one day one party starts telling its supporters to use the orange colored ballots only

While the other party doesn't tell it's supporters which color ballots to use.

When the ballots are counted, the first party has a greater advantage in the orange colored ballots than the purple colored ballots

Is this surprising? No.

Does it tell us anything at all about what would happen if the state moved to only using the orange colored ballots? Not in the least.

Would it upend research that examined shifts in other states between the color of the ballots required (and found zero partisan effects), just because it counted the orange and purple colored ballots separately? Not at all.

In saying that it is informative to look to differences in votes cast in-person and votes cast by by mail and infer anything about all-mail systems is actively misleading and misinforming the @nytimes audience. There's a reason previous research didn't do this; it's bad research.