Premium
Null Hypothesis Significance Testing Does Not Show Equivalence
Author(s) -
Barchard Kimberly A.
Publication year - 2015
Publication title -
analyses of social issues and public policy
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.479
H-Index - 31
eISSN - 1530-2415
pISSN - 1529-7489
DOI - 10.1111/asap.12095
Subject(s) - null hypothesis , equivalence (formal languages) , significance testing , alternative hypothesis , statistical hypothesis testing , psychology , bayesian probability , econometrics , statistics , mathematics , pure mathematics
Null hypothesis significance testing (NHST) allows researchers to find differences between groups, such as differences between men and women, Blacks and Whites, or heterosexual and same‐sex couples. In contrast, this article will show that NHST is mathematically incapable of allowing researchers to conclude that two groups are the same. Because of this, NHST leads to a research literature that emphasizes differences between groups and ignores similarities. To examine similarities between groups, either Bayesian statistics or confidence intervals should be used. Thus, Frost and Gola (2015) should have used one of these approaches when examining similarities between heterosexual and same‐sex couples.