Current Trends in Statistical Inference
Robin L. Plackett · Journal of the Royal Statistical Society Series A (General) · 1966
can be traced back for 150 or 200 years. During most of this time, there have been differences of opinion about the validity of the methods proposed, and it seems to me unlikely that there will ever be general agreement, or that arguments will be discovered which conclusively show the irrelevance of highly developed lines of thought. My view is that the differences of opinion should be patiently explored, and that the amount of agreement already in existence should be emphasized. We are dealing with ideas which go rather deeper than logic or mathematics and the important point is, not that we should start from the same assumptions, but that we should reach essentially the same conclusions on given evidence, if possible. However, we have to start somewhere and the variety of possible viewpoints immediately presents us with difficulties of classification which cannot really be discussed in a general way. The many recent developments in Bayesian analysis, and the fundamental nature of Bayes's theorem, suggest looking first at this field.