Methods in Medical Research

This volume consists of four sections: the largest is on "Statistics in Medical Research" and is a general account of the non-technical aspects of the subject; a section on "NMethods in Environmental Medical Research" makes up the second major portion of the book; and the volume is completed by two smaller sections, of some 30 to 40 pages each, on "Some Methods of Studying Human Genetics" and the "Design and Construction of Metabolism Cages." The variety of subject-matter is greater than in previous volumes of the series, though there is much material that will have a general appeal to workers in medical research. Donald Mainland, with the help of his assistant, Lee Herrera, has prepared the chapters on statistical methods, and the work is characterized by the clear writing for which Mainland is known in this field. The purpose of the section is not to provide details about how to perform particular statistical calculations, but rather to give a "realization of what modern biological statistics implies throughout the conduct of any type of medical investigation." It is debatable, however, whether a worker in medical research is likely to learn much about statistics by studying general principles alone. For most people, the consideration of specific problems, especially when combined witlh the working of numerical examples, is the only way to develop an understanding of the wider issues, and, in fact, the authors have recognized this by using illustrations, including in some cases numerical data drawn from medical research of current interest. Following the introductory chapter there is an account of what is meant by the effect of chance, or random, factors. An understanding of this is necessary in the analysis of results of experiments, but it is equally important in the design of experiments, for in this latter process one first excludes, or decides to make independent observations onl, important systematic factors, and then, by the act of deliberate randomization, ensures that the remaining factors will affect the results in a random manner. It is, therefore, clear that the question of what is to be treated as a chance factor is controlled in part by the investigator, and it is equally clear that the design of the experiment must be married to the analysis of the results. The two following chapters give a general account of the planning of investigations and the analysis in relation to the planning. The description is …