Data gained by interviews and questionnaires administered to public officials and other players in the policy process are often invaluable and may not otherwise be available to researchers. Care is required, however, in using both such techniques and the data acquired. Questions must be properly framed to elicit the needed information. Questions that are "loaded" and therefore bias responses, or that are so general as to create strong doubt about their intent need to be avoided. Officials and others may not always respond fully or candidly to questions, their memories may be hazy, and they may overstate their own role in events. Data gained from these sources obviously should not be viewed as gospel. Rather, they should be checked against other sources, used with care, and regarded as representing particular viewpoints on some event. Good judgment called for.
Many studies of policymaking take the form of case studies; that is, they focus on particular programs, statutes, or areas of public policy. Case studies have been the butt of much criticism because, being narrowly based, they do not permit sound generalization. "What is a case study a case of?" is a common gibe. Preferred studies are those dealing with all the cases in a universe, such as all regulatory commissions or sunset laws, or a meaningful sample thereof, such as Supreme Court decisions on the rights of the accused or the benefit decisions made by a welfare agency. These afford a better basis for generalizations. Case studies, however, do have a variety of 48 They can be used to uses to test theories; to develop new theories; to provide detailed, contextual analysis of events; to analyze deviant cases that contradict our generalizations; and to help provide an "intuitive feel for the subtleties and nuances of the policy process and the practice of politics. There is plenty of room in the study of policy for both case studies and more general and comparative studies. To draw on a now passé Republican analogy, policy study should be viewed as a "big tent.