SELECTING CONTROL GROUPS: TO WHAT SHOULD WE COMPARE BEHAVIORAL INTERVENTIONS?
GEORGE W. REBOK
In the universe of effect sizes that make up our RCT evidence base for psychological interventions, control conditions remain dark matter, exerting effects that are unseen, ill-defined, and for the most part unquantified.
While there have been disputes over which control conditions should be used, such debates have produced more heat than light.
—Mohr et al. (2009, pp. 282-283)
Selecting an appropriate control group is an essential component in the design of behavioral intervention studies. However, when designing protocols for these studies, often researchers do not pay sufficient attention to the control condition(s). Although the results of intervention studies typically are discussed in terms of the effects of a new treatment or therapy, any differences between the experimental and control conditions depend as much on the latter as they do on the former (Tansella, Thornicroft, Barbui, Cipriani, & Saraceno, 2006). Because control conditions can have variable effects, the choice of a control group can exert a substantial impact on research studies using the randomized controlled trial (RCT) design and other experimental and quasi-experimental designs. By definition, RCTs are comparative studies that evaluate differences in experience between two or more groups, and control conditions are the main method for removing unwanted sources of variation in accounting for those differences. Of the various control conditions, the condition where no alternative treatment is provided, also known as the “no-treatment control,” often produces the largest effect size for behavioral interventions because it is least likely to positively affect the outcome (Mohr et al., 2009). However, even in this case, differences between the experimental and control conditions could be attenuated if nontreated control participants benefitted from repeated exposure to the outcome assessments, as in a longitudinal design.
Developing and testing behavioral interventions involve an iterative and incremental process of building an evidence base and then translating and implementing the intervention, disseminating it, and then scaling it up for sustainability within particular practice or community settings (Gitlin, 2013). During the different phases of the development, testing, and implementation of behavioral interventions, control groups may serve different purposes. They are most important during testing the efficacy (Phase III) and effectiveness (Phase IV) of an intervention through an RCT, but they also may play an important role in other phases such as testing for feasibility and preliminary effects (Phase II). The purpose of this chapter is to (a) describe the role of control conditions across different phases of intervention development; (b) identify key methodological and ethical challenges in the use of control groups; (c) provide empirical examples of control group selection and design in large-scale, community-based intervention trials; and (d) make recommendations about how to select and use the most appropriate control conditions for randomized behavioral intervention trials. Given the author’s research interests and background, examples will focus on intervention trials to improve the cognitive health and well-being of older adults in order to illustrate the unique challenges that researchers face in control group selection in behavioral intervention studies. Many of the methodological and ethical issues discussed are relevant to different types of behavioral interventions with older adults as well as other age groups, such as those involving mental health, exercise, nutrition, and medication use.