Statistical Inference

Statistical and, subsequently, econometric inferences have not undergone a cumulative, progressive process. We have seen instead the emergence of a number of different views, which have often been confused with each other in textbook literature on the subject. It therefore makes sense to approach the issue from a historical-scientific angle rather than a systematic one. We intend, using the extraordinarily complex development as a basis, to give a historical overview of the emergence of concepts that are of particular importance from the point of view of cliometrics. We shall start by describing the beginnings of modern probability theory, along with its connection with other statistical approaches. The following overview covers the basic principles of the current concepts of inference developed by R. A. Fisher on one hand and by J. Neyman and E. S. Pearson on the other. Neo-Bayesian approaches have meanwhile been developed in parallel, although they were not taken into account during the initial founding phase of econometrics. A “classic” approach was instead adopted in this respect, albeit with an additional difficulty: the taking into account of time. Cliometrics initially followed a Bayesian approach, but this did not finally prevail. Following on from econometrics, a correspondingly classic, inference-based position was adopted. This chapter concludes with a reference to a fundamental critique of the classic position by Rudolf Kalman, which we also find very promising as an inference-related concept for cliometrics. We often quote authors directly, in an effort to portray developments more vividly.