ON THE USE AND INTERPRETATION OF CERTAIN TEST CRITERIA FOR PURPOSES OF STATISTICAL INFERENCE PART I

摘要: In an earlier paper* we have endeavoured to emphasise the importance of placing in a logical sequence stages reasoning adopted solution certain statistical problems, which may be termed problems inference. testing whether given sample, Z, is likely been drawn from population H, started simple principle that appears used judgments ordinary life-that degree confidence placed hypothesis depends upon relative probability or improbability alternative hypotheses. From this point view any criterion assist scaling with accept reject X has randomly HI should one decreases as (defined some definite manner) hypotheses becomes relatively greater. Now it course impossible practice scale form judgment single numerical criterion, partly because there will nearly always present priori conditions and limitations cannot expressed exact terms. Yet though irmpossible bring ideal situation into agreement real, measure essential guide control. our previous paper made use likelihood. That other forms criteria can interpreted different manner very possible, but object find connecting various sampling tests already use, could extended new problems.