ABSTRACT
The representative sample fallacy refers to the assumption, implicit in most survey research, that a representative sample of the population is always superior to clinical samples. This chapter examines some of the difficulties in more detail, with particular emphasis on the role of injuries and of frequency of assault as criteria for identification of cases of child abuse and wife beating. An important potential problem with measuring child abuse and wife beating on the basis of assaults, regardless of whether an injury resulted, is the possibility that the cases with injury may differ in other ways from cases where there were assaults but no injuries. In the case of children, the parental behaviors used in the Conflict Tactics Scales to measure abuse, such as kicking, punching, or hitting a child with an object, are acts that go beyond ordinary physical punishment.
