A description of obstacles to proceeding from official police records to approaching complainants for an interview focuses on the confidentiality issue. One alternative, asking the police to make the initial approach, could prejudice response rates and bias the answers given. Benefits and problems found in the following alternative strategies are discussed: restricting research to the analysis of official records; sample surveys of the entire populations; and appealing for participants through advertisements or letters. The author's experiences in conducting a survey illustrate pitfalls in the ways of qualitative and quantitative research in this area. Two police forces agreed to allow a sample of their complainants be drawn from their records, provided a procedure could be worked out which avoided breaches of confidentiality. The resulting survey produced a low response rate of 30 percent. Reasons underlying the poor response and sources of possible bias are explored. Nine references.