Health behavior change: a field just picking up speed. A comment on Ogden (2016).

A central tension described in Ogden’s (2016) editorial is that between the variability (the ‘mess’) that appears inherent to human behavior – in this case in response to health-related interventions – and the determinacy science seeks. For those studying and modeling human behavior, it is hard not to be dumfounded by its ‘ill-fitting, multiple-influenced, volatile, situational and porous nature’ (Rowson, 2015, p. 318). Yet, individual behavior is far too important an affair today not to be approached scientifically, thus seeking some level of predictability. Ogden’s argument is that as this process is advancing, specifically through the effort of systematization illustrated by the work of Michie et al. (e.g., Michie, Atkins, & West, 2014), so is a profession, ultimately an entire field, at risk of perishing. My own view is that, as a scientific field, health behavior change is in its early stages and it is far too early for major paradigm shifts. Instead, I foresee that progress in this field will contribute to improve interventions for all, but where complexity and insufficient evidence will regularly force interventionists to rely on other sources of information to make decisions. Indeed, this is precisely where I would relocate the central tension of the field: to the fact that while health psychologists are in demand to build interventions that work today, currently these have to be based on rather limited data about what works, why, for whom, and under which circumstances. For example, I question the value of looking post hoc at published interventions – most of which have been poorly designed and/or reported (Prestwich et al., 2014) – and coding them based on the Behavior Change Techniques Taxonomy to arrive at ‘effective behavior change techniques (BCTs)’ through meta-analysis (e.g., Michie, Abraham, Whittington, McAteer, & Gupta, 2009). Aiming to design evidence-based interventions, researchers have often referred to ‘effective BCTs’ (i.e., BCTs known to change behavior), apparently unaware of two key considerations. First, that the effectiveness of any BCT is dependent on a number of parameters, including not only the target behavior but also the population, setting, mode of delivery, and the presence of and interaction with other operating BCTs and related mechanisms of action (c.f., Kok et al., 2015). Second that, unfortunately, we still do not know which combinations of BCTs and related mechanisms work best in which circumstances. However, to get there I do not see a good substitute for a path that resembles this: explicitly deciding which theory or theories to use; formally identifying the theory-derived mechanisms of action and the techniques/methods to target them; effectively training interventionists to deliver them and/or adequately translating them to other modes of delivery (c.f., West & Michie, 2016); ensuring validity and fidelity at each step of the way; and finally evaluating for whom and under what circumstances, and/or modes of delivery, results are most convincing. The complexity inherent to this process is undeniably great. And further complicated by the long timeframes associated with changes in the so-called lifestyle behaviors, and time is something that theoretical models often struggle to accommodate. Therefore, significant progress will require an equally great effort from the scientific community, including the capacity to produce, organize,