Some Empirical Aspects of the Sequential Analysis Technique as Applied to an Achievement Examination

Abstract

With the presentation of the early paper 1 on sequential analysis by Dr. Wald, a new frontier in modern statistics was achieved. Prior to those writings it had been shown by Professor R. A. Fisher that the proper place for determ ining the size of sample needed in an experiment was in the planning of the experiment itself. That is, before the experiment was executed. It is well known that previous to Fisher's work, most thought about the proper sample size took place after the experiment was run. We note to day the metamorphosis of the determination of sample size, for with Wald's writing we see that under certain circumstances n is no longer to be regarded as a fixed quantity but as a ran dom variable, and that the designation of the proper sample size is made by the experiment itself. It is in the sense that the statistical cog itation about n becomes a part of the experi ment rather than a prelude or post-mortem of it, that the marvel of the sequential procedure is noted. The use of sequential analysis in educational problems has been undertaken at least by two authors. 2, 3 Dr. H. M. Walker has made a con tribution in the field of item analysis, and Dr. Cowden used sequential analysis in administer ing an examination.