Evaluation: Design, Indicators, and Use of Findings
What this domain tests
Evaluation answers two questions: was the program delivered as planned, and did it work? Distinguish formative evaluation (during implementation, for improvement) from summative evaluation (endpoint judgment of merit and worth), and process evaluation (fidelity, reach, dose, satisfaction) from outcome and impact evaluation (change in knowledge, behavior, health status, or system performance).
Choose indicators that are valid, reliable, feasible, and sensitive to change; specify numerator, denominator, data source, and collection frequency before data collection begins. Establish a baseline and, where feasible, a comparison group-historical, matched, or randomized-to reduce attribution error. Recognize threats: regression to the mean, secular trends, maturation, attrition, and instrumentation changes mid-stream.
Budget for evaluation, apply data quality checks, protect confidentiality, and report to multiple audiences in plain language, including unfavorable findings and limitations. Evaluation should feed a documented improvement cycle rather than sit on a shelf.
Traps
Questions reward activity counts as though they were outcomes, accept post-only satisfaction surveys as impact evidence, or overlook a missing baseline. Watch for ceiling effects at already high-performing sites and for claims that a pre-post gain "proves" the program caused the change.
Rule of thumb
State your causal claim in inverse proportion to the strength of your design.
Worked example
Satisfaction rose from 70% to 92% after training, with no comparison group and no behavior measure. Correct conclusion: participants were satisfied; effectiveness remains unproven. Recommend a comparison and a behavior indicator before making broader claims.