The use of Statistics in Brazilian Journals of Behavior Analysis

Abstract

The use of Null Hypothesis Significance Tests (NHST) has been a topic of debate in science, psychology, and behavior analysis due to difficulties in interpreting them correctly, their tendency towards dichotomous thinking, and the frequent associated questionable practices. The behavior analysis literature has a tradition of criticizing the use of NHST for studying individual behavior, given its preference for single-case research designs (SCRD). Despite this, there has been an increase in the use of NHST in international behavior analysis journals over the last six decades. However, there has been a lack of evaluations of Brazilian behavior analysis literature. In this study, we examined the reports of inferential statistics in all empirical articles published in the Revista Brasileira de Análise do Comportamento is (Brazilian Journal of Behavior Analysis), the Revista Brasileira de Terapia Comportamental e Cognitiva (Brazilian Journal of Behavioral and Cognitive Therapy), and Perspectivas em Análise do Comportamento (Perspectives in Behavior Analysis) until April 2021. Our findings reveal that almost half of the papers published in the mentioned journals were empirical. Among these, 76.6% were experimental studies, and approximately a quarter (25.2%) used NHST for statistical inferences, with a noticeable upward trend in its use. However, only 3.2% included inferential statistics with estimation metrics such as confidence intervals and effect sizes. Importantly, we found that 69.5% of the studies using inferential statistics presented their results incorrectly or incompletely according to the APA standards in force at the time of their publication, hindering the verification and revaluation of the statistical calculations presented. Our study reveals that Brazilian behavior analysis journals show similar trends to international journals in the use of NHST, and the incomplete reporting of associated statistics is damaging to the field. This emphasizes the pressing need for improvement in the editorial policies of behavior analysis journals regarding the reporting of inferential statistics.