I wrote this in my last post.

t wouldn't take much effort with a database or even a bit of coding in Java/Python/Perl/etc. to match early primary voters to early general election voters, since the voter registration ID is included in the data. But, you would only be able to match about 45% of general election voters, at best (4M primary voters divided by 9M primary voters).

It's actually about 22%, not 45%, if you compare early voting data. I just downloaded it from both party primaries back in March. Only about 50% of all voters in the primaries voted early. I don't have voter registration numbers for people that voted on primary election day (2020-03-03), so I can't compare them.

Otherwise, I did exactly what I described above: matched the voter registration number. This is what I found:

• 11.08% of the early voters in the general election voted early in the Republican primary.
• 10.00% of the early voters in the general election voted early in the Democrat primary
• 78.92% of the early voters in the general election did not vote early in the party primaries.

I'll rerun my analysis after the last day of in-person early voting is reported tomorrow. I'm also planning to look at the breakdown of in-person vs. mail-in voting.

A final caution: this analysis is based only on participation in a party primary. It cannot account for people that voted in one party's primary, then voted for the other party in the general election. It could occur for at least two reasons:

1. The voter doesn't like the candidate that won their party primary.
2. A voter normally affiliated with one party "crossed over" to vote in the other party's primary in order to promote a weaker candidate.