EVALUATE — Confidence Scoring

Package: tools-impact-engine-evaluate

Purpose

Assign a confidence score based on methodology reliability.

Inputs (from MEASURE)

• Effect estimate + CI bounds

• Model type

• Diagnostics

Produces

• Confidence score (0-1)

• Scenario returns (R_best, R_med, R_worst from CI bounds)

Confidence by Methodology

Model Type	Confidence Range	Rationale
Experiment (RCT)	0.85 - 1.0	Gold standard causal inference
Quasi-experiment	0.60 - 0.84	Strong but with assumptions
Time-series	0.40 - 0.59	Trend-based, confounding risk
Observational	0.20 - 0.39	Correlation, high bias risk