How Much Can a Filter Version Change a Workforce Measure?

Research question: How often does a changed inclusion filter move a synthetic measure across a declared threshold when source records stay fixed?

Method: Construct forty-eight monthly datasets, then apply paired filters that alter status, location, worker type, or effective-date treatment. Preserve each query version.

Measures: Report denominator changes, absolute and relative movement, threshold crossings, and reproduction errors. Lock the classification and calculation rules before review.

Decision boundary: Reviewers identify evidence conditions. They do not choose the correct population, interpret results, set materiality, or authorise a restatement.

Limitations: The datasets omit many joins, late records, privacy constraints, and correction histories. A live pilot needs approved definitions, privacy review, an audit trail, and a stop condition.

References: National Institute of Standards and Technology, Cybersecurity Framework 2.0 (https://doi.org/10.6028/NIST.CSWP.29); National Institute of Standards and Technology, Privacy Framework 1.0 (https://doi.org/10.6028/NIST.CSWP.01162020); U.S. Government Accountability Office, Standards for Internal Control in the Federal Government, GAO-14-704G (https://www.gao.gov/products/gao-14-704g); Philippine Statistics Authority, Annual Survey of Philippine Business and Industry (https://psa.gov.ph/statistics/annual-survey-philippine-business-and-industry). These publications provide control, privacy, evidence, and sector context. They do not evaluate Outsourced Employment or these synthetic cases.