The issue with The Democracy Institute is they, along with a lot of these articles, provide zero reference. So these "polls and polling metrics" I guess I just take this guy's word because he won't disclose what polls, or what metrics were used.
As for undervoting, yes the exact same thing happened. It was literally the exact same thing: People who vote only for President. There is no other factor than a ballot that is only indicating a mark for a President and no other entry is made.
That happened in Trump's favor, but now when it's Biden's it is "proof" of fraud. Why is undervoting not "proof" of fraud in areas where Trump has majority undervotes? If undervoting is what the "proof" is then all undervoting is an issue. If in context the undervoting is skewed at a large margin then fine, but conveniently the other side of the coin is left out of the challenge.
All one has to do is go read articles from 2016 addressing this same issue, the exact same issue. Or look at basic numbers such as what can be found on the consistently audited Congressional Votes Database:
2000: Republicans received 3.46 fewer votes in House races than they did for President.
Democrats received 4.41 million fewer votes.
2004: Republicans received 6.08 million fewer votes in House races than they did for President.
Democrats received 6.05 million fewer votes.
2008: Republicans received 7.69 million fewer votes for House than they did for President.
Democrats received 4.26 fewer votes.
2012: Republicans received 2.70 million fewer votes for House than they did for President.
Democrats received 6.27 million fewer votes.
2016: Republicans received 188,987 MORE votes for House than they did for President.
Democrats received 4.07 million fewer votes.
https://www.govtrack.us/congress/votes