Past research has identified social and environmental correlates of behaviors thought to be associated with obesity and weight gain among children and adolescents. However, few novel, modifiable correlates are being identified, and much less research has documented the efficacy of interventions designed to manipulate those correlates. These efforts have been inhibited by limitations in study design, measurement and conceptual models. The predominant biomedical and social science research paradigms, emphasizing reductionist approaches to understanding etiological mechanisms of diseases and risk factors, may be responsible for actively slowing efforts to prevent obesity and improve population health. This approach has left many of the most important applied research questions unanswered, and the all-too-common refrain: "we don't know what works?" We propose an alternative research paradigm to overcome these barriers. The alternative research paradigm represents only a subtle conceptual shift but results in substantial implications for generating new hypotheses, research questions and research designs, to shorten the cycle from research to improved population health.