This interchange of leading statistical authorities and endocrine investigators emphasizes the creative possibilities of statistics in endocrine research. The goal is to enhance the economy, validity, and elegance of experiments. Detailed examples using endocrine data guide the investigator through modern statistical techniques for experimental design, analysis, and interpretation. The agenda, which was prepared after consultation with experiments, includes such topics as nonparametric and sequential techniques, the use and misuse of multiple comparisons, the role of computers and biomathematics, the detection and management of invalidity, and the extraction of maximum information from bioassays. An integral part of the treatment of each topic is a discussion section in which experiments and statisticians join in an exploration of specific difficulties and obscurities. Among these are: randomization, transformations, the rejection of observations, the meaning of nonparallelism in bioassay, and determination of the confidence limits of a ratio of relative potencies.
The important new experimental techniques of competitive protein-binding analysis and radioimmunoassay receive extensive consideration. Their underlying mathematical principles are elaborated, together with the controversial issues of sensitivity and precision. Three possible approaches to the statistical analysis of radioimmunoassay data are proposed.
This volume comprises the presentations made at a Workshop Conference on Statistics in Endocrinology held at Endicott House in Dedham, Massachusetts, December 4-7, 1967.