Recent findings in the literature on child abuse have been inconsistent and often contradictory, perhaps in part because most researchers treat child abuse as a homogeneous psycho-social phenomenon and commit the error of group averaging. One methodology which deals directly with the problem of homogeneity error is the typology approach. This methodology attempts to develop a system of categories that are sensitive to whatever groupings exist in a population. Employing the typology approach, the present, ongoing research explores demographic, personality, emotional and stress variables and includes families from different ethnic groups as well as a broad range of abused children in a control group design. To date, data have been collected on 12 families within each of three groups: (a) abusing families involved with the Children's Aid Society of Metropolitan Toronto (CASMT), (b) non-abusing, problem families in CASMT, and (c) non-abusing, non-problem families whose children attend a day nursery. The following instruments are being used: (1) Basic Personality Inventory, (2) Personality Research Form (altered form), (3) Schedule of Recent Experience, (4) The Vanderveen Family Concept Test, (5) The Michigan Screening Profile of Parenting, and (6) a study questionnaire measuring demographic and social factors. It is hypothesized that child abuse can be understood in terms of types of abusing families and that there are identifiable differences between abusing and non-abusing, demographically similar populations. Data that have been collected are presented in tables and figures. (Author/RH)