Evaluate uptake and consequences of research recommendations.
Inputs, example and sources
World Bank · Impact Evaluation in Practice Second Edition
Institution publication · Chapter 7, table 7.2, printed page 140, PDF page 170
The formula is verified. Causal interpretation still requires a defensible study design, comparison and assumptions.
Keep the same place, observation period, definitions, units and source versions.
Analysis inputs. Treatment Verified recommendation uptake before [%]; Treatment Verified recommendation uptake after [%]; Comparison Verified recommendation uptake before [%]; Comparison Verified recommendation uptake after [%]; Study design and comparability evidence; Outcome definition, method and reference basis
Other required evidence. Recommendation cohort; institutional decisions; implementation; outcomes; alternative explanations.
Fictional worked example · 2025
Laggala Pallegama — research translation · Laggala Pallegama
| Input or result | Value | Unit |
|---|
| Treatment Verified recommendation uptake before | 42 | % |
| Treatment Verified recommendation uptake after | 59 | % |
| Comparison Verified recommendation uptake before | 39 | % |
| Comparison Verified recommendation uptake after | 46 | % |
| Study design and comparability evidence | Missing | |
| Outcome definition, method and reference basis | Verified recommendation uptake /example matching definition1 | |
| Difference in changes | 10 | % |
|---|
The arithmetic alone does not establish causation. Design, parallel trends, confounding, attrition, uncertainty and adverse effects need appropriate evidence.
Proposed checking role. Relevant research lead, statistician and data steward.
After the action. Verified outcomes, costs, distributional effects and evidence strength.
- World Bank Impact Evaluation in Practice Second Edition — Institution publication. Chapter7 difference-in-differences. Published difference-in-differences estimator and study assumptions; aggregate examples do not establish causation.
- SHARED domain decision contracts — Validation required. Domain decision rules and matching fictional examples. Local comparisons, constraints, source matching, sorting and follow-up proposals; developer checks do not establish scientific or client approval.
Institution and data context
Confirm local scope, data rights, parameters and responsible roles in this Institution review. Published components retain their source; they do not approve the local application.