If a person at a bar knows they are being observed as part of a research study, they are likely to change their behavior, which can compromise the validity of the observational data collected.