But how would you propose to change that, when looking at registration metrics? You can cite polls (which often vary, and are usually questionable) about which way independents are breaking. But the polls rely on modeling independent voter turnout. And independents are notoriously late to make up their minds. Rs and Ds are easier. Just assume that 95% of each party votes for their nominated candidate. But even that doesn’t capture turnout. However, it’s a better indicator than polls.
This can only mean that a great many Democrats voted for Trump.
This is even more pronounced in West Virginia, where Trump beat Hillary by an incredible 68-26 margin - almost 2.5 to one, despite the fact that Democrats had a huge advantage over Republicans in voter registration (that also has closed significantly since then).