ABSTRACT
A test of significance is a formal procedure for making a decision between two hypotheses about some characteristic of a population on the basis of knowledge obtained from a sample of that population. These hypotheses are: null hypothesis; and alternative hypothesis. Testing for significance involves a comparison of the difference between the sample statistic and the parameter specified by the null hypothesis with a theoretically determined sampling distribution. The level of significance of a given sample statistic under the typical null hypothesis is a mono-tonic increasing function of the sample size. In the introduction to his article, H. C. Selvin dismissed concern with “technical errors” in the use of significance tests as well as concern with describing particular populations on the basis of a sample. The error of confusing statistical significance with substantive significance, although widely recognized, remains commonplace. Selvin clearly was technically wrong in asserting that tests of significance are illegitimate in research designs unless control is maximum.
