weighed against the disadvantages of the limit on how much data can be collected using in-person or telephone data collection procedures. Although it is possible to collect contextual data on family violence using social surveys, it is much more difficult to collect these data with a survey instrument than by other means. Surveys tend to be unable to adquately collect data from members of the lowest socioeconomic groups. Because the base rate of the severest forms of violence is so low, surveys also fail to locate adequate numbers of family members experiencing the most injurious or life threatening abuse.
There is no question that if we are to improve our understanding of family violence, there are many research issues and problems that need to be overcome and solved. First and foremost, the field of family violence must continue to improve upon the definitions of abuse, violence, and the family. Until such time as the majority of investigators are employing similar definitions for the central concepts in the field, confusion and contradiction will dominate the study of family violence.
More research studies need to be based on larger and more representative samples. There is a compelling need to collect data which can be generalized to larger populations. Survey research is a means of generating a general causal model of family violence. Such a model could then be subjected to more intensive investigation using smaller samples and more intensive assessment with psychometrically sound measures.
Although the problems of fielding adequate longitudinal designs are many, there is a need to employ more of these designs. Change and time order should not exclusively be inferred from cross-sectional data. Panel and cohort studies are a necessity if knowledge of family violence is to be advanced.
Students of family violence need to attend to the major measurement issues in the study of violence and aggression. The field has been well served by the Conflict Tactics Scales, but a single scale is not a solution to measurement problems. Weis ( 1987) points out that there is not one validation study of family violence yet published. An experimental design which compares measures on the same sample is imperative.
The field must move beyond accepting conventional wisdoms and post hoc conclusions as theory. For the study of family violence to be truly advanced, programs of research must begin to test the various notions, hypotheses, and propositions which have been developed over the past 25 years.
This paper is part of the Family Violence Research Program at the University of Rhode Island. The program is funded by a grant from the National Institute of