False analogy. NYT/Marist aren't making assumptions about the electorate like Loras is.
Of course they are. Every pollster who polls voters is making some assumption. Even in a registered voter poll, pollsters are assuming that voters who are not registered today won't register or vote tomorrow and that voters who are registered will turn out and vote. Every likely voter poll has some sort of likely voter screen based on assumptions about the upcoming electorate, often at least partially based on voter history.