MOST can be used to address a number of challenges facing behavioral intervention science:

Translational research.  MOST provides a clear conceptual framework for translational research, both for translating theory into efficacious interventions, and for translating efficacious interventions into effective interventions. The MOST approach can be used to optimize intervention effectiveness by examining crucial issues such as which characteristics of the program must remain constant across delivery sites so as not to compromise effectiveness; which must be tailored to individual delivery sites so as to maintain effectiveness; and which may be altered at the discretion of local delivery personnel with no ill effects.                    

Understanding the parts making up the whole. Many of today's behavioral interventions are "black boxes," that is, they have been evaluated only as a whole (e.g. Flay, 1986; Flay & Collins, in press).  MOST can provide valuable information about which components are active and which are inactive in a behavioral intervention.  This look inside the "black box" will increase our understanding of the details of effective prevention and treatment programs, and build a foundation of knowledge that can be drawn upon by the entire drug abuse field. 

Improved causal inference. Secondary analyses involving nonexperimental comparisons frequently yield data that inform intervention design in valuable ways.  However, any inferences based on such data are weaker, that is, subject to many more alternative explanations, than those based on randomized experiments (Cook & Campbell, 1979; Shadish, Cook, & Campbell, 2002).  Relying too much on the results of nonexperimental analyses can lead the investigator down a blind alley, or even lead to an intervention that may be harmful (for example, considerable nonexperimental evidence supported the idea that hormone replacement therapy for postmenopausal women was a nearly unqualified health benefit, but a randomized experimental study contradicted these findings (e.g. Cushman et al., 2004)).  In contrast, rather than relying on observational or post-hoc analyses to build interventions, MOST relies on carefully randomized and controlled experiments, including stratifying a priori on important variables that cannot be randomized (e.g. gender). This greatly reduces dependence on nonexperimental data.   Nonexperimental research and secondary analyses will always have a place in intervention science; our point is simply that it is feasible for intervention science to increase its reliance on experimental research, and the field would benefit by doing so.

Fostering innovation in intervention development.Creativity and freshness of perspective are important in behavioral intervention research, but the necessity of mounting a full confirmatory trial to test an innovation that is high-risk can discourage this way of thinking, even if the payoff is potentially great.  MOST facilitates intervention creativity by providing a way for investigators to try out innovative ideas without the necessity of investing in a full randomized confirmatory trial.

Saving time and other resources. MOST conserves program implementation resources in the screening and refining phases by employing methodological tools such as highly efficient fractional factorial designs, which are used strategically to keep the number of conditions that must be run to a minimum.  Moreover, the investigator using MOST is highly selective about mounting intervention trials, thus conserving program evaluation resources by devoting them only to those interventions that have demonstrated a high probability of success.  Therefore, although it is expected that usually the screening and refining phases of MOST will take more time to implement than is typically devoted to pre-intervention-trial research currently, MOST has the potential to shorten the net amount of time from theory to deliverable intervention in the long run, because there is less need to cycle through repeated confirmatory trials followed by secondary analyses.