I don't understand how a "likely voter", who first answers that he's likely to vote, then says "wouldn't vote" in the GE question ...
84% said that they are registered to vote.
67% said that they are likely to vote.
62% said that they are registered AND likely to vote.
So 62/84 of registered voters said that they are likely to vote.
But 5/16 of non-registered voters said that they are likely to vote.
So Clem says: I meant to register after I moved here.
Interviewer: When was that?
Clem: Back in '78. I really liked Reagan and would have voted for him, but I somehow didn't get around to it. Then Bush came along and I kind of lost interest in politics, but with Trump I'm kind of getting excited (scratches himself).
Among the 67% likely voters, 3.8% said they "Wouldn't Vote". Among the 62% who were registered and likely, 1.9% said they wouldn't vote. So among the 5% Likely, but Unregistered, 27% said they "Wouldn't Vote".
I could not find the text of the questions asked, but they presumably asked whether someone was likely to vote and registered to vote before they started asking other questions.
Are you likely to attempt to lose weight in the next six months?
Do you currently exercise most days?
If you were to attempt to lose weight, would you walk, swim, jog, go to a gym, do yoga, reduce your food consumption, some other exercise, don't know, or wouldn't lose weight?
Answers of Yes, No, and Wouldn't are totally consistent.