So a quarter of Americans would like to see some legal recognition for gay relationships, but would also prefer -- again, if given the option -- that such recognition not be described as a marriage. Judging by the data in other surveys, when you narrow the choice to a yes/no for same-sex marriage, a significant segment of that in-between group may end up saying yes. But one can safely assume that it's at least a somewhat reluctant yes, given that the people offering it would prefer another option. And this would seem to be exactly the population that might say one thing to a pollster and end up voting another way in a referendum later on.