More bogus numbers for another “poll”.
But it’s technically a very good poll.
Errors are “likely voter = past regular voter”: assume Trump pulls disaffected voters;
and in turnout assumption.
Neither of those errors are clearly wrong at this point.