Marquette has one of the strictest Likely Voter models. You have to have voted in the last midterm for you to get through their screen.
I might need to confirm that with Charlie. That seems like a pretty big assumption for one pollster to hang their hat on, given what 2014 looked like.
I’m pretty sure that’s what there methodology has said in the past, but I’d have to double check.
Found it, you have to tell them that you are absolutely certain that you are going to vote. Even saying that you are very likely to vote bumps you out of the likely voter screen. Other pollsters typically will include those in that second tier after absolute certainty.