level (e.g., a county or district program delivered via local agencies such as schools, clinics, and hospitals), which are then expected to affect factors at the household/individual level (e.g., lifestyle choices and eating behaviors). Different intervention components at different levels are mediated by social and human factors operating in their natural settings. All such factors call for appropriate observation, measurement, and modeling that are constrained by traditional experimental and quasi-experimental designs. Complex interventions often are constrained in simplistic experimental designs and accompanying analytic models that assume a treatment to be a single manipulated factor (similar to a medical therapy).

Third, while the level of certainty (or internal validity) of research designs continues to be salient, external validity increases greatly in importance when one is identifying and appraising research evidence on public health issues such as obesity. Effective public health interventions need to have generalizability, transferability, and sustainability beyond small-scale research studies (Green and Glasgow, 2006).

Finally, not all questions that arise in taking social action are best answered by cause–effect designs or even research-based evidence. While the issue of whether a particular treatment works to reduce obesity may be answered with experimental designs, questions such as what causative and protective factors could potentially be targeted by interventions call for different approaches and research methods. Correspondingly, different criteria must be applied to build the needed evidence base. In short, the type of evidence must be matched to the policy or practice objective at hand. Potential alternative research designs to the RCT are described in Appendix E.

Transparency in Decision Making

Transparency implies openness in decision making, effective communication, and accountability. The ability to make public health decisions as transparent as possible is likely to foster public trust, better risk communication, and financial accountability (Honore et al., 2007; Laforest and Orsini, 2005; McGloin et al., 2009). Few frameworks for evidence-based decision making have made transparency a central theme; the committee decided in its deliberations that transparency was a goal of its framework. Transparency in summarizing and communicating the evidence used in making a decision is discussed in Chapter 7.

Approaches to Measurement

Measurement of progress in preventing obesity can take several different forms. The recommended indicator of obesity for adults and children is BMI—calculated as weight in kilograms divided by the square of height measured in meters (kg/m2). Beyond measurement at the individual level, consideration of how to measure the effects of environmental and policy changes is a relatively new aspect of obesity research, and appropriate measures and evaluation designs are still being developed. Chapter 8 and Appendix E provide more detailed discussion of methods for measur-



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement