The RCT is generally considered as the “gold standard” in evaluating the effects of psychological and behavioral interventions. The primary goal of an RCT is to determine whether an intervention works by comparing it to a control condition, usually either no intervention or an alternative intervention. Secondary goals may include identifying the factors (e.g., age, sex, health, cognitive status) that might moderate the effects of the intervention, and understanding the processes through which the intervention influences change (i.e., mediators or change mechanisms that bring about the intervention effect), according to the theory guiding the trial (Gitlin, 2013; West & Spring, 2007).

In an RCT, individuals or groups are assigned to treatment conditions at random (i.e., they have an equal probability of being assigned to the treatment or control). That helps ensure that the effects of the intervention can be causally attributed to the differences between the intervention and control, not to some extraneous factor(s). To the extent that the investigator can rule out alternative explanations and minimize systematic error (or bias), then the causal inference about the experimental effect is said to be “internally valid.” If the results of the intervention can be generalized to a population of interest, that is, the population the intervention was designed to help, the intervention is said to be “externally valid.” However, there are many factors that can threaten the internal and external validity of a study and lower confidence in the findings.

The original conceptualization of threats to validity was articulated by Campbell and Stanley (1966) and has changed very little over time. Common threats to internal validity that RCTs address include history, maturation, selection, temporal precedence, regression to the mean, attrition, and testing and implementation (Melnyk & Morrison-Beedy, 2012; West & Spring, 2007). There are also external threats to validity that can be addressed by RCTs, including sample characteristics, setting characteristics, and effects due to testing. However, RCTs have evolved and become increasingly more complex, creating new challenges for maximizing validity and minimizing bias (Mohr et al., 2009). For example, the use of a “usual care” group as a control may not be sufficient when the intervention requires an investment of time and energy that excludes all but the most highly motivated participants. In this case, any differences that emerge may be due to the composition of the samples participating in the intervention and control conditions (i.e., selection), and not to active ingredients of the intervention (Lindquist, Wyman, Talley, Findorff, & Gross, 2007).

Randomization of individuals or groups (e.g., schools, work sites, clinics, or communities) to an intervention or control condition represents the best strategy for ruling out alternative explanations, but it may not be possible to control for every conceivable threat to the internal validity of one’s experiment (Melnyk & Morrison-Beedy, 2012). Today, behavioral interventions are conducted in the real- world of increasingly complex health care environments and diverse communities, which has increased their external or ecological validity, but which has also made it much more difficult to exert tight control over threats to internal validity. Thus, investigators are faced with inescapable trade-offs between internal and external validity, which makes it difficult or impossible to minimize threats to both at the same time (Freedland, Mohr, Davidson, & Schwartz, 2011).

Given the complexity of today’s real-world environments, researchers have to weigh a variety of scientific, practical, and ethical issues and other considerations in choosing control groups for RCTs of behavioral interventions in health care and other settings (Westmaas, Gil-Rivas, & Silver, 2007). These considerations include the costs of the control conditions, feasibility, potential contamination, ethical issues, recruitment, retention, and sample size. As mentioned earlier, it may not be possible to control for every potential threat to internal validity. Rather, it may become necessary to control for whichever threats are deemed to be most important at the expense of remaining vulnerable to the less important ones (Freedland et al., 2011). In addition, it may be unethical to use a traditional control group if that deprives participants of treatments that have already been incorporated into routine health care practice (Freedland et al., 2011; Mohr et al., 2009).

< Prev   CONTENTS   Source   Next >