Likely voter models-- Boolean or weighted?

Suppose a political pollster calls me up. In addition to the usual questions about which candidate I prefer, etc., they’ll also ask questions like how likely I consider myself to vote, whether I’ve voted in past elections, whether I know where my polling place is, etc. Based on the answers to those questions, they’ll decide whether or not I’m a likely voter.

But is it an all-or-nothing thing, or is it weighted on a sliding scale? Do they say “If you meet 5 of these 7 criteria, you’re a likely voter, and will be included in our likely voter numbers, and if you meet less than that, you won’t be”? Or do they say “Based on our research, people who give that set of answers are 73% likely to vote”, and then add 73% of my responses to the likely voter totals? It seems like the latter would be more accurate, since it’s taking more information into account, and some likely voters don’t vote while some unlikely ones do. But all the discussions I’ve seen of the topic seem to assume there are exactly two categories.

Most have a threshold - you’re either in or your out.

The one exception I know of is the new RAND tracking poll, which (among other unique features) weights responses based on how likely they are to vote (a 70% likely voter who supports Obama counts as 0.7 votes for Obama).

A link to the survey: https://mmicdata.rand.org/alp/?page=election

And a caveat - this is a new and rather unorthodox method that has not been proven. I wouldn’t put too much faith in it until it can be rated.