The need to develop methods for assuring confidentiality of social research data stems from two related problems. First, research subjects who are emabarassed or threatened by an inquiry about their private lives may refuse to respond or may distort their response, thereby assuring for themselves the confidentiality of particular information; as a consequence, the accuracy and precision of estimates of parameters in social research may be critically undermined. Second, the social researcher in Germany as in the United States, has no legal protection against judicial appropriation of the data for non-research purposes; if obtained from the researcher, the research subjects’ responses may lead to legal or social sanctions against him.

The mathematical methods described in this article have been developed to alleviate these problems and, more specifically, to increase the strength of the researcher’s promise that “confidentiality” will be maintained. The randomized response method, the unrelated question method, and the newly developed contamination method permit one to aquire information without needlessly jeopardizing or embarrassing the research subject. The methods, based on simple laws of probability, are summarized and compared in the following remarks.