Perhaps people confuse/conflate it with a more broader category, like https://www.nbcnews.com/feature/nbc-out/nearly-1-5-young-adults-say-they-re-not-straight-n1270003

Well, the survey does not show the distribution of answers. So my guess (it's a guess, not data) is the people living in densely populated areas answered lower numbers, and people living in rural or remote areas ridiculously overestimated the number of trans people based on how much of a big "issue" medias are making of them.

For sure. Many Americans are confused by percentages. They do not understand that 20% is equivalent to saying "in a room of 100 people, 20 of them are trans", and even if they did understand that, they wouldn't have the proactive reasoning to make sure their percentage estimates add up/overlap in a way that makes sense, e.g not implying that 20 people in the room are all Hispanic Asian atheist Catholic bisexual transgender millionaires.