That said, online surveys often fail at this, for instance having "gender" questions with just the 2 options (they should at least have an "other") or only accepting as "valid" answers that do not fit the entire population (For example, a survey for doctors with no explicit age cutoff limited ages to <99; at the time, there was one practicing doctor older than that - he would just have been given an error message that his age was "invalid".)