Speaking of polling bias, we usually talk about average polling bias with the election. But that obscures a ton of variation by state.
In 2020, polling underestimated Trump by a large amount. But not in every state! Polls actually underestimated Biden in Nebraska's second district and in Colorado. In Georgia, polls were close to spot on as well.
In 2016, likewise polling underestimated Trump by a large amount. But again not in every state! Polls actually underestimated Clinton in Nevada. They were close to spot on in Arizona and Georgia.
And as I've repeatedly mentioned, historical polling bias averages around zero as it has favored both parties over the years. In 2012, polls underestimated Obama. In 2000, polls underestimated Gore. By similar amounts as polls underestimated Trump in 2016 and 2020.
And in 2018 and 2022, polls underestimated Democrats. So this isn't a recent phenomenon.
I want to emphasize the importance here of this variation. Say polls underestimate Trump again. Harris could still win if Pennsylvania or North Carolina or Georgia end up with a polling bias in her favor. That's entirely possible.
Just as it is entirely possible polls systemically underestimate Harris across the board. In that case, she'd win by a landslide.