Standards used in this guide

The recipes use published methods and measures wherever one exists. Where the program's workstreams provide a tool, the guide uses it.

Methods and tools

Resource	Use	World Bank use	In this guide
REaLTabFormer (paper)	Transformer-based synthesis of tabular and relational data with a memorization control (MIT)	The program's synthetic data workstream	Chapters 2, 4, and 5
synthpop (paper)	Sequential synthesis with CART and parametric methods; utility and disclosure measures		Chapters 2, 3, and 4
SDV and SDMetrics	Copula, network, and multi-table synthesizers (Business Source License); quality and privacy metrics (MIT)		Chapters 2 to 5
sdcMicro practice guide	Disclosure control practice and risk measures	Written at the World Bank	Chapter 4

Measures

Measure	Use	In this guide
Total variation distance; Cramér's V; Pearson correlation	Marginal and association comparisons	Chapter 3
Propensity-score utility (pMSE) and confidence interval overlap (Snoke et al., 2018)	General and analysis-specific utility	Chapter 3
Distance to closest record; attribute inference; membership inference (Shokri et al., 2017)	Disclosure risk of synthetic data	Chapter 4

Documentation and governance

Standard	Use	In this guide
World Bank microdata schema	The catalog record of the synthetic file, with links to the real survey	Chapter 6
Metadata Editor and NADA	Creating the record and publishing the file with its access type	Chapter 6
Creative Commons BY 4.0	The licence for open synthetic files	Chapter 6
UNECE, Synthetic Data for Official Statistics: A Starter Guide	The reference introduction and use cases for statistical organizations	Chapters 1 and 7
Access chapter of the microdata cookbook	Tiers, disclosure control records, and the access workflow	Chapters 1, 4, 6, and 7

Corrections

Standards change versions and URLs. If a reference here is wrong or out of date, open an issue with the row and the corrected source.