The untapped potential of Bayesian region of practical equivalence for assessing null effects in multi-domain research

• Null effects are inherent in multi-modal research • Knowledge of null effects is scientifically and practically meaningful • Null hypothesis significance testing is inadequate for dealing with null effects • Bayesian region of practical equivalence as alternative approach • Promoting cumulative multi-modal research Multi-domain research has become a popular topic in building science. However, studies in this area have not yet paid sufficient attention to the inadequacy of the dominant research practices entrenched in the conventional null hypothesis significance testing (NHST) when dealing with null effects. To address this problem, we first explain why null effects inherently exist in multi-domain research and argue for the importance of null effects. We then highlight numerous limitations of NHST that are particularly relevant to multi-domain research, for example the inability to accept null hypothesis or to refute false theories, the untested and thereby unfalsifiable alternative hypotheses, the underpowered results that mislead cumulative research and aggravate publication bias, the often ignored testing intentions, the fallacious black-and-white thinking, unintuitive interpretation and common misunderstanding, and the asymptotic behavior. Inspired by advances in methodology literature, we introduce Bayesian region of practical equivalence (ROPE) as an alternative approach that allows researchers to decide for, against, or remain undecided regarding whether an effect of research target is practically equivalent to a value of interest (e.g., the null value). The Bayesian ROPE approach has several advantages over NHST, namely the ability to establish practical absence of an effect and to refute false theories, better scalability, no requirement on minimum sample size, intuitive interpretations, and independence from testing intentions. We contend that this alternative approach can promote cumulative multi-domain research and mitigate publication bias.