"how many students in pool?" I tick 1, "did you know any of the others?" well, I guess I know myself, but I'm not really someone else? So depending how you get the scores (ie each answer "sheet" separetly or all the answers to the same answer together), you'll get significant bias...
Good catch. And even though I hadn't thought of that until just now, it's not a problem.
Data is collected as a relational database in a pretty robust analytics platform. Easy to sort that out. (Easier than programming the survey around that issue in the first place, in fact.) Ultimately I'll simply not include "1 student in class" respondents in any analysis relative to whether you knew anyone else in the class.