ABSTRACT
Moreover, this attempt to do the impossible, to make a single unified theory out of the immiscible approaches within statistics, seemed to offer a great, though illusory benefit. Such forced unifications necessarily introduced confusions and outright errors into the textbooks and into the interpretations, but these confusions and errors all promised what was most wanted: a way of rigorously inferring the validity of research hypotheses from the data. Conversely, the more accurate the textbook from a statistical point of view, the more useless it would have appeared to practicing experimenters. Perhaps this is one reason why repeated attempts since 1960 to expose these errors and confusions in the hybrid statistics have had almost no impact on practice. It was precisely those errors and confusions-that rejecting the null hypothesis informs us about the validity of the research hypothesis, about its probability, about the size of the effect, about the replicability of the effect, or about the quality of the experiment-that had made the hybrid statistics indispensable.
