‘Significant’ and ‘Highly Significant’

Abstract

FREQUENTLY in scientific papers one finds the terms ‘insignificant’, ‘significant’ and ‘highly significant’ used in discussing the difference between control and experimental, with their ranges usually established by P = 0.05 and P = 0.01. Statistics text-books provide little or no guidance as to their proper usage because statisticians apparently disapprove of the convention. Yet the universality of the convention attests to the need for clearly defined, commonly accepted terms which properly weight the results of the experiment. The present convention is commonly accepted among non-statisticians, but, as will be shown here, is not clearly defined. Therefore an improved convention is proposed for the consideration of those who would benefit from it—the non-statisticians in research. Perhaps by improving the convention, it can be made more palatable to statisticians, as well as more useful to non-statisticians.