On the definition and interpretation of interviewer variability for a complex sampling design