Comments on “Effects of Change Interventions: What Kind of Evidence Do We Really Have?”

Abstract

This is a very important article that should start an important conversation about the state of research in the field of organization change and development. I agree completely with the premise of the article that change strategies and interventions should ideally be chosen based on evidence, not intuition or values about change. The article does an excellent job of framing the scientific principles by which research should be evaluated. The four-level classification of research based on the extent to which the study is controlled and enables conclusions about internal validity is sensible and grounded in accepted standards of normal science research. Although one might argue with the way in which the sample of studies was chosen, changing the sampling criteria would only make marginal changes in findings. Consequently, the findings that the vast majority of research in organization change does not come close to meeting standards for normal science research is defensible. Indeed, if asked in advance of the study I would have predicted these findings. The conclusion that the wide variety of variables and measures being employed is hindering the convergence of knowledge about what works and does not work is also valid and has been cited by many others as a barrier to progress. The article is, therefore, an important challenge to the field of change management with which I agree. It should cause scholar-consultants who are in a position to encourage senior managers to undertake meaningful research to evaluate their practice. Having said this, I have reservations about some of the implicit and explicit conclusions of the article. I offer these to encourage more complexity in our thinking about the reasons for the current state of intervention research and what might be done about it. Let me start with a critique of the research method the authors employ. The method is too distanced to enable us to learn more about how the poorer research designs were executed. The authors fail to examine a subsample of these studies to learn if the