Null Hypothesis Significance Testing in Behavior Analysis: Problems and Recommendations

Abstract

Widely adopted in Psychology in general, inferential statistics is also prevalent in Behavior Analysis (BA), an approach that historically favored single-case experimental studies. The dissemination of research with groups using null hypothesis significance testing (NHST) in data analysis has grown in this area. It brings with it related problems (intrinsic and due to misuse). Such problems often go unnoticed in the current peer review system, compromising the reliability of some conclusions available in the scientific literature. In this article, we explain the problems related to the misuse and misinterpretation of the NHST and compile guidance for editors, reviewers, and authors that can be adopted to minimize the problems mentioned.