Replication data supports a 2026 framework by Brian Peters for diagnosing how empirical results change based on which countries are included in a sample. The analysis applies leave-one-country-out, leave-one-region-out, and permutation tests across all papers in a series. It finds core demographic results remain robust, but their magnitude can vary by 30-50% depending on the inclusion of CCA countries and resource exporters.
Use Cases
- Assess result sensitivity using leave-one-country-out tests on country-level data.
- Evaluate regional influence on model coefficients via leave-one-region-out analysis.
- Perform permutation tests to quantify the fragility of demographic findings.
- Analyze how inclusion of CCA countries and resource exporters affects effect size magnitude.
Strengths
- Framework developed and tested across all papers in a specific research series.
- Diagnostics quantify magnitude variation of 30-50% for key findings.
Limitations
- Unknown sample size and specific column structure limit direct analysis.
- Geographic coverage and temporal range of the underlying studies are unspecified.
Provenance
- Source
- Brian Peters, Demographics and Global Capital Allocation.
- Collection Method
- Replication package containing analysis code and data for diagnostic framework.
- Time Range
- null
- Freshness
- Data package last updated in April 2026.
- Geography
- Cross-country, includes CCA countries and resource exporters.