Three Roles for Statistical Significance and the Validity Frontier in Theory Testing

Abstract

This study offers a method for empirically testing theories operationalized in the form of multivariate statistical models. An innovation of the method is that it distinguishes testing into three separate forms, “effect testing,” “prediction testing,” and “theory testing,” where statistical significance plays a separate role in each one. In another innovation, the researcher specifies not only his or her desired level of statistical significance, but also his or her desired level of practical significance. Statistical significance and practical significance each serve as a dimension in a two-dimensional table that specifies the rejection region – the region where the researcher can justify the decision to reject the theory being tested. The boundary of the rejection region is the “validity frontier,” which ongoing research may advance so as to reduce the size of the rejection region.