Statistics in the Design and Analysis of Physiology Experiments.

Statistics should be considered as a tool by the animal experimenter in much the same way that a chemical analysis or a radiation counter is used as a tool. Too often statistics is considered as something magical which can restore order out of chaos and perhaps absolve the experimenter from mistakes in logic and procedure. It is true of course that the statistician through his knowledge of statistical procedures can sometimes help salvage some results from an otherwise hopelessly muddled experiment. One point must be made clear at the outset—the statistician is not a tool of the experimenter although statistics is such a tool. What then is the proper role of the statistician in biological experimentation? In this time of specialization, most professional men and women are highly trained in one aspect of one field of interest. The old cliche that continued education is simply “Learning more and more about less and less” is nevertheless cogently descriptive of our age. Copyright © 1965 American Society of Animal Science. Used by permission. S T A T I S T I C S I N T H E D E S I G N A N D A N A L Y S I S O F P H Y S I O L O G Y E X P E R I M E N T S 1 L. D. VAN VLECK AND C. R. HENDERSON Cornell University, Ithaca, New York S TATISTICS should be considered as a tool by the animal experimenter in much the same way that a chemical analysis or a radiation counter is used as a tool. Too often statistics is considered as something magical which can restore order out of chaos and perhaps absolve the experimenter from mistakes in logic and procedure. I t is true of course that the statistician through his knowledge of statistical procedures can sometimes help salvage some results from an otherwise hopelessly muddled experiment. One point must be made clear at the outset t h e statistician is not a tool of the experimenter although statistics is such a tool. What then is the proper role of the statistician in biological experimentation? In this time of specialization, most professional men and women are highly trained in one aspect of one field of interest. The old clich6 that continued education is simply "Learning more and more about less and less" is nevertheless cogently descriptive of our age. Many biological problems, however, overlap various areas of specialization. A team con. sisting of many specialists may be needed to solve such problems. A not unusual team in the area of hormone secretion and function may include an endocrinologist, a biochemist, a physiologist, a physicist, and should we not include, also, either a biomathematician or a statistician? An example of such a team is the group which postulated the structure of genetic material. In this sense the statistician should be considered as a partner of the team, if not a full partner, at least a junior partner. He should not be in the position of a lawyer who. wants in at the beginning of a case, not just after the client has been convicted. This does not imply that the biochemist or physiologist needs no training in mathematics or statistics. An appreciation of the problems and techniques of statistics is important for all experimenters. This knowledge may be no 1 Invitational paper presented at the 56th Annual Meeting of the American Society o.f Animal Science, Knoxville, Tennessee. deeper than the survey courses biologists take in the humanities or social sciences. An expert knowledge is not required. In order to talk intelligently to the statistician, however, at least a knowledge of the rudiments of statistics is necessary. A reasonable statistical requirement for research biologists would be the completion of a 1-year conrse at the level of Snedecor (1956) or Steel and Torrie (1960). Many of the underlying principles and most of the mathematical derivations obviously would not be learned. Yet the experimenter would become more aware of the limitations as well as the potentials of statistics as a tool. Most biologists who have not had such formal training can acquire an equivalent understanding through independent study. I t is desirable for the statistician on the other hand, as a member of the team, to have some acquaintance with the field of the biologist. Many biometricians who have come up through biologically oriented curricula, with later specialization in statistics, will have this acquaintanceship. Statisticians with formal mathematical training, on the contrary, do not often have a biological background. A general observation, however, is that it is much easier for a mathematician to gain a general knowledge of biology than for a biologist to gain a general knowledge of mathematics. What then can the statistician offer as a member of the team? All statisticians will not agree, since Fisher has remarked that variance is a term describing the attitude of one statistician to another. Practicing biometricians, however, find the following statements by Cochran and Cox (1957) to be descriptive of the problems they face everyday: "Statisticians are often asked for advice in making inferences from the results of experiments. Since the inferences that can be made depend on the way in which the experiment was carried out, the statistician should request a detailed description of the experiment and its objectives. I t may then become evident