Watch on YouTube

6:14

From conceptual model to measurement

A model you drew for another purpose will tell you what to measure. A causal diagram, drawn to work out which variables to adjust for, is already a data collection checklist — every variable on a backdoor path has to be measured or the adjustment cannot happen. A logic model does the parallel job for a programme, naming inputs, activities, outputs and outcomes. Treatment fidelity is what separates implementation failure, where the programme was never delivered as designed, from theory failure, where it was and still did not work.

Also covered: logic models · DAGs as a measurement checklist · treatment fidelity · implementation vs theory failure