designs that would withhold services from a control group, including the establishment of "wait list controls" for overcrowded services such as battered wife shelters.
The explosion of interest in the topic of domestic violence has been accompanied by tremendous growth in research and publication on the various aspects of violence in the home. Unfortunately, although the amount of data collected in the past 25 years is impressive--at least compared to the dearth of data collected prior to 1962--the knowledge base is still far from solid. By and large, the reason for the lack of a well-developed knowledge base is the rather brief period of time the topic of family violence has been studied. In addition, the knowledge base has been built through exploratory and descriptive research. Such research can is useful for generating hypotheses, but cannot test those hypotheses. Furthermore, even well designed exploratory and descriptive research is subject to innumerable plausible rival explanations for the findings.
Although few researchers, clinicians, or policy makers held out much hope that useful data on family violence could be collected using survey research techniques, the skepticism has turned out to be unwarranted. Survey research on violence between family members has yielded estimates of the incidence and prevalence of violence in the home not confounded by the biases of official reports. Furthermore, survey data that examines associations between individual and social factors and family violence is also not confounded by the biases inherent in using clinical or official report data. The large samples used in survey research have provided opportunities to test hypotheses about various aspects of violence in the home. When the surveys employ representative samples, investigators have been able to break the confines of small, limited samples, and generalize their results to larger populations.
An unintended benefit of survey research on family violence is that those using the survey method have been forced to rigorously develop replicable nominal and operational definitions. Whereas studies based on clinical or official report data can simply define violence and abuse as those cases in the clinical or official records, survey researchers were forced to develop reliable and valid measurement technique to tap violence between family members.
Obviously, survey research is not the panacea for the problems and constraints inherent in studying, understanding, and explaining family violence. The biases of clinical and official report data have been replaced by the biases that are part of any project which depends on self-report and recall to measure human behavior. The advantages of large sample sizes and resulting statistical power are