Null Hypothesis Statistical Testing (NHST)
摘要
Null hypothesis statistical testing has been called “the backbone of psychological research” (Gerrig and Zimbardo, Psychology and life, 16th edn. Allyn and Bacon, 2002, p. 46), and it is used as the primary, if not sole, statistical testing procedure in the discipline. NHST allows a dichotomous decision (“reject the null” as “significant” or “fail to reject” as “nonsignificant”) to be made about a statistical hypothesis and is used as evidentiary support to license inferences from the sample tested in a research study to the greater population of interest. Debates pertaining to NHST have existed since the birth of the discipline of psychology, however, and major controversies, including concerning the hybridization of Fisherian and Neyman-Pearsonian logics of NHST; the mischaracterizations of the p-value, null hypothesis; and the testing of psychological theories pose challenges to the methodology. Amendments to NHST, as well as alternative methodologies, are discussed as potential remedies.