A methodological note on the selection of friends as controls

Abstract

A number of issues inherent in the selection of friends as controls in case-control studies are illustrated in the light of a recent dietary study, based on hospital and friend controls. Preselection of certain characteristics of the controls by the interviewee is almost unavoidable. The choice of controls to be used in a case-control study must reflect the nature and type of hypothesis which is being tested. Advantages and drawbacks of potential control groups must be weighted against each other.