ESTATÍSTICA EXPERIMENTAL

Abstract

This text was prepared for use by students of course of Experimental Statistics discipline of graduate programs in the areas of agricultural sciences at the Federal University of Pelotas. It is available in the literature a lot of Experimental Statistics texts, the most known in the English language. However, almost all of them emphasizes the statistical analysis of experiments, making only references to the conceptual and methodological aspects that underlie the experimental research. This is a source of failure of many experiments. This approach is sometimes justified with the argument that many aspects of the experimental plan are more own of the scientific method and of the specific research area than of the statistical method. The author does not agree with this argument, since the modern methodology of experimental research arose precisely from the need of the consideration of the statistical method already in the experiment planning phase. In fact, the most significant contributions of Fisher for experimental research originated from the emphasis on experimental design methodology, as soon as he found that the amount of information generated by the inference of an experiment can not be greater than that contained in the data. Moreover, the statistical method is a component inseparable from the scientific method, particularly the experimental research. The modern conceptual and methodological basis of experimental research originated precisely in the development of statistical methods for use in the experiment. In these circumstances, we can not study experimental statistics without a solid understanding of the conceptual and methodological basis of experimental research. The initiative of this writing arose from the need to fill this gap, providing material to the student and researcher quite sparse and difficult to access in the literature. Besides, many concepts are presented in the literature in vague and inaccurate way, which does not allow the establishment of a coherent and comprehensive conceptual and methodological basis. The usual concepts are reviewed and formulated with the precision necessary for the proper understanding of its real meaning, in order to establish a rational basis for the methodology of experimental research and in particular for the Experimental Statistics. A more detailed exposition of design of experiments is the subject of a volume in preparation. The author acknowledges the contributions of any kind to better match the text to its purposes.