One needs to decide whether the questionnaire will be administered by an interviewer or self-administered. An advantage of a self-administered questionnaire is that it can be sent by post and returned in a prepaid envelope. The survey can therefore cover a wide geographical area. The fact that interviewers are not required reduces the need for training and resources and minimises observer bias. However, it is important to check the questionnaires on return as some answers may be missing or ambiguous.

If the questionnaire is to be interviewer-administered, then there will be a need for training. This is necessary to minimise both intra- and inter-interviewer variation. A common fault in interviewer-administered questionnaires is that of suggesting answers to the subjects. It is very important that the questionnaire is administered in the same way by all interviewers if there is more than one interviewer. An additional problem is that subjects may tell the interviewer what they think he/she wishes to know rather than the true answer. Intervention studies where compliance with dietary advice is being assessed may be particularly prone to this, but the extent of the problem will depend upon the rapport between the subject and the interviewer. Use of an interviewer-administered questionnaire requires more resources than a self-administered questionnaire, but it will reduce the likelihood of incomplete or ambiguous answers.