WelfareWeights Engine

An open implementation of the estimation pipeline behind the Global Burden of Disease disability weights: the numbers between 0 (full health) and 1 (as bad as dead) underneath every DALY figure. Every link on this page opens in a new tab.

Read

• How disability weights are estimated (PDF) — the full guide: the survey questions, the three-stage estimator, what must be true for the weights to mean anything, and how to audit it; validation evidence and reproduction guide as appendices. Written for people who did econometrics years ago, or who know math and code but never took it.
• The same guide as markdown (browsable on GitHub).
• Validation evidence — nine lines of defense, every number reproducible with recorded seeds.
• Roadmap — what is done, what is next, and what we need (the original survey microdata) to finish the job.

Run

• The repository — Apache-2.0 code, CC-BY-4.0 documentation.
• quickstart.py — the full pipeline with uncertainty intervals on synthetic data with known true weights, in about fifteen seconds.
• The studies — misspecification battery, power curves, bootstrap coverage, and recovery at the published GBD 2010 design scale.

Context

• Reference Case Guidelines for Benefit-Cost Analysis in Global Health and Development (Robinson et al. 2019) and the HHS Guidelines for Regulatory Impact Analysis — the formal frameworks in which these weights get used.
• Salomon et al. 2012 and Salomon et al. 2015 — the methodology this engine replicates.

Part of the WelfareWeights project.