SHOULD WE MEASURE AMBIVALENCE TO CHANGE?Substantial doubts exist about the added value of measuring ambivalence to change in motivational interviewing (MI) efficacy research. It is extremely challenging to assess ambivalence to change accurately, and its relative importance as a mechanism of MI efficacy is probably very marginal compared to the contribution of relational factors.In this issue of Addiction, Sarah Feldstein Ewing and colleagues [1] address the importance of ambivalence as a determinant of motivational interviewing (MI) efficacy in adolescents. They underline the need to deconstruct the concept of ambivalence to improve its measurement. Should we measure ambivalence more accurately when exploring MI efficacy?Ambivalence to change is a simple and didactic concept that changed the paradigm of substance abuse treatment, helping clinicians to accept patients' hesitations and relapses, and acknowledge that behaviour change takes time. The resolution of ambivalence is a central concept in MI: 'MI helps people to keep moving forward through the natural process of resolving ambivalence' ([2], p. 167). This hypothesis is based on the fairly straightforward observation that: 'most people who smoke, drink too much, or exercise too little are aware of the downside of their behavior' ([2], p. 6). Another general observation is that most individuals with a history of substance use disorder are no longer ambivalent to change once they are stabilized in abstinence. Despite data supporting the efficacy of MI in reducing the consumption of alcohol and other drugs, including among adolescents [3], there is no evidence that the resolution of ambivalence is a causal pathway in MI efficacy. There are data suggesting that change discussion, particularly the production of change talk, is associated with outcome [4], but this does not imply that the resolution of ambivalence is a mediator in the MI change process.Another concern is that the measurement of ambivalence to change is challenging, as it results from the encounter of several layers of subjects' dynamic and fluid internal, external, conscious and unconscious factors. Finally, even though the resolution of ambivalence is a critical ingredient of MI, its association with MI efficacy might, in the best case, be marginal. A recent meta-analysis of 59 psychotherapy efficacy studies [5] found minimal differences between the total treatment with and without one or more critical ingredients (effect size, d = 0.01), while relational factors such as empathy had a much larger effect size (d = 0.63). In addiction treatment, a large body of literature indicates that patients' outcomes are influenced mainly by relational factors rather than by critical ingredients, even when counsellors apply standardized treatment accurately [6]. An alcoholism treatment study found positive behaviour outcomes varying from 25 to 100%, depending on the counsellors' interpersonal functioning [7]. Recently Gaume and colleagues [8] found similar variability across 18 counsellors in a standardized ...