Model Utility
Model Utility asks how useful a theoretical model is for guiding mechanism-based inference across contexts. A useful external validity model provides a shared foundation—formal or informal—that prevents broad or underspecified claims risking false positive or false negative inferences. We organize Model Utility into three components.
Components of Model Utility
MU1: Target & Estimand
A theoretical model must define the target STOUT population and estimand: the descriptive, predictive, or causal quantity of interest. This anchors the external validity question in a specific population and clarifies what quantity researchers seek to generalize or transport.
MU2: Mechanistic Structure
A model must specify the mechanism: the structure or chain by which an input leads to an output. This includes identifying modifying factors and the functional form of M-STOUT interactions that determine when and how effects vary across contexts.
MU3: Query & Identification
A model must express the external validity query and determine how far it can be answered. When answerable fully, the theoretical quantity yields an identification formula in terms of observable data. When not, only informative bounds are possible without stronger assumptions.
Mechanism-Based Inference
These three components provide a shared foundation across modeling approaches—formal and informal alike. By requiring explicit targets, mechanisms, and queries, Model Utility helps researchers avoid overgeneralization while still making meaningful claims about where and when effects apply. The goal is not to eliminate uncertainty but to be precise about what can and cannot be inferred from available theory and evidence.