The use of statistics in behavior analysis: When to use it and when not?

Abstract

Behavior analysis originated from the tradition of the biological sciences, that favor the replication of data analyzing directly the behavior of individual subjects. Other areas of psychology that originated from the social sciences, favor the replication of data analyzing statistically grouped data, either of the behavior of the same subjects or of groups of subjects.  Although behavior analysts favor the analysis of individual data, there has been an increase in the number of articles published in specialized journals in which inferential statistics has been used to determine the reliability of a finding.  In this paper the basic premises of both strategies to infer the reliability of a finding, that derived from the biological sciences and that derived from the social sciences, are presented. The paper also includes the main criticisms and the arguments against those criticisms made by the defenders of each strategy. The conclusion is that each strategy has its own strengths and weaknesses and that in order to determine the reliability of a finding, a researcher should analyze the data following the strategy that allows him/her to advance scientific knowledge by comparing the current findings with those from previous research and thus answer unambiguously the research question.