Date of Award

Fall 2023

Document Type

Open Access Dissertation

Department

Exercise Science

First Advisor

Michael Beets

Abstract

Preliminary testing of health behavior interventions (e.g., pilot, feasibility studies) are used to evaluate intervention viability prior to additional testing and resource investment (e.g., larger trials). This initial testing provides valuable information, but promising estimates of effectiveness produced during early testing are rarely reproduced in larger studies, stalling the development of effective, scalable health interventions. In the obesity intervention literature, external validity biases, features of the intervention that are not (or cannot) transfer to the larger study, are associated with these diminished effects. These study features, such who delivers the intervention, the characteristics of the population receiving the intervention or the duration and intensity of the intervention are nearly ubiquitous features of behavioral interventions leading us to believe that they exist, and negatively impact interventions beyond the field of obesity research. Establishing the construct validity of external validity bias across multiple disciplines of health behavior could inform health behavior intervention design, delivery, and evaluation. Equipped with the necessary guidance to identify external validity in early-stage studies, researchers and funders could proactively make informed decisions regarding the design and evaluation of preliminary studies, choosing to avoid these pitfalls and stabilize intervention effects across studies. However, identification of the biases alone is unlikely to reduce their inclusion in intervention development. Just as intervention features are ubiquitous across fields, so too is the research enterprise in which they are developed. Early-stage studies are considered the primary source of data to support larger-scale grant applications and must present especially compelling results to be considered competitive. Delivery of an intervention by a highly skilled principal investigator is likely to increase the effectiveness of an intervention but is also a response to the natural constraints of preliminary-study budgets. Identifying these system-level drivers of bias inclusion is necessary to mitigate their use. The purpose of this dissertation is to identify the prevalence and impact of external validity biases in multiple fields of health behavior research and explore the contextual factors surrounding the development of interventions that may lead to the inclusion of these biases. By providing a foundational understanding of the prevalence of external validity biases, as well as the complex contextual factors that may drive their inclusion in preliminary studies, these findings can inform efforts to amend the research enterprise in favor of scalable interventions development that can positively impact population health.

Rights

© 2024, Lauren von Klinggraeff

Share

COinS