Read on Twitter
Random thought: a lopsided Democratic edge in early voting may be a bit of an underrated challenge for pollsters. It's a bit dangerous to have one party lock in its votes, while the other party's voters don't get that kind of boost in their turnout probability
Let's take an extreme example to illustrate: suppose a 50-50 race with *100%* turnout, and suppose 100% of Party A votes early, while no one in party B votes early.
That poll will be biased toward party A, unless it assigned 100% turnout to all of party B (which wouldn't happen)
That poll will be biased toward party A, unless it assigned 100% turnout to all of party B (which wouldn't happen)
IRL, the issue isn't underestimating Republican turnout. The real issue is that the Dems who don't vote early are less likely to vote than Republicans who don't, but they'll probably all tell you'll they'll vote. Alternately, you have too many Dem early voters
We'll handle the latter just fine. The former is tougher and it's something I'll need to think about.
In the past, we'd just clock in the 'early voters' as 100% likely to vote and slowly degrade the turnout probability of those who didn't vote early to match an overall turnout estimate. I'm not sure that'll fly if absentee voting is super lopsided.
