Diffusion and Adoption of Good Science: Overcoming the Dominant Logic of NHST and the Reporting of Rubbish

ABSTRACT Purpose: This article describes and explains the arrival of the current tipping-point in shifting from bad-to-good science. This article identifies Hubbard (2016a) as a major troublemaker in attacking the current dominant logic supporting bad science, that is, null hypothesis statistical testing (NHST) in management science.

Focus: Crossing the tipping-point is occurring now (2016–2017) from bad-to-good in the behavioral and management sciences. Because of the occurrence of supporting conditions, continuing to ignore the vast amount of criticism of bad science is no longer possible. Bad science includes the pervasive use of directional predictions (i.e., “as X increases, Y increases”) and NHST. The time is at hand to replace both with a good science: computing with words (CWW) to predict outcomes accurately and consistently, constructing models on a foundation of complexity theory, and testing models using somewhat precise outcome testing.

Recommendations: Achieving parsimony does not equate with achieving simplicity and the shallow analysis of only predicting directional relationships. Researchers need to build the requisite variety into models and predict precise outcomes. Doing so requires recognizing the relevancy of equifinality (alternative configurations occur that indicate the same outcome), sign reversals, and causal asymmetry tenets of complexity theory. Testing for predictive validity of models using additional samples is a must.

Managerial implications: Executives benefit from learning that CWW accurately is possible; do not accept models of only directional predictions; demand substance in research findings; look at the XY plots of CWW complex antecedent statements and both simple and complex outcomes that you need to know about (e.g., purchase behavior).