Survey sampling issues in primary care research.

Survey research methods are often used to answer questions in primary care medicine. Consequently, investigators must understand and address sampling issues. Probability sampling designs have particular utility for questions that seem most appropriate for survey research. These designs provide each member of the population with a known probability of selection, and if performed properly with an appropriate sample size, they provide a sample with essentially no selection bias. Examples of random and stratified sampling designs are presented. These two probability designs, along with nonprobability sampling techniques (convenience, purposive, quota), are discussed in terms of bias, error, and possible limitations.

Organizational effects on mentally retarded adults: a longitudinal analysis.

Normalization has gained wide acceptance as a goal that residential institutations for the mentally retarded should strive to achieve, but many organizations have been shown to have difficulty achieving the goal. Theories developed from the organizational contingency perspective suggest that organizations with bureaucratic structures will have particular difficulty accomplishing the nonroutine tasks associated with normalization. Our major purpose was to test the usefulness of such theories for the evaluation of mental retardation facilities by ascertaining whether a less bureaucratic organization for the mentally retarded would achieve greater success than a more bureaucratic organization. The closing of a large public hospital and the subsequent transfer of most of its residents to two new facilities (one of which was more bureaucratic than the other) allowed us to examine bureaucracy's effect on treatment. As predicated, the analysis showed that the less bureaucratic organization produced a greater average positive change in behavior than did the more bureaucratic organization. A number of clinical and demographic characteristics of the residents which could have influenced the observed changes in behavioral level were identified and controlled. They were not found to explain the differences between facilties. Other factors, which could not be controlled in this study, provide suggestions for future research.