Method policy
How Skolkoll documents calculation methods, handles changes and works with transparency.
Documentation principles
- All calculations are documented in
metric-definitions.json— a machine-readable data catalogue with 80+ metrics. - Each metric has: name, unit, range, data source and calculation method.
- Definitions follow Skolverket's and SCB's official definitions by default.
- Deviations from official definitions are explicitly marked in the variable dictionary.
See the variable dictionary for a complete listing.
Methodology change process
When a calculation method changes or a new data source is added, we follow these steps:
- Identification — A new data source, changed calculation logic or correction to an existing method is identified.
- Documentation —
metric-definitions.jsonis updated with new or changed definitions. - Communication — An entry is added to the changelog describing what changed and why.
- Verification — Results are compared with the previous version for a reasonableness check. Large deviations are investigated before publication.
- Publication — Deployment includes updated documentation and data simultaneously.
Transparency
- Methodology changes are published in the changelog before or at the same time as new data is published.
- Backwards-incompatible changes are announced at least one version in advance.
- Calculation methods and data sources are documented openly on the method page.
Review process
- Current state: Skolkoll is a one-person project. Methods and calculations are reviewed internally.
- Goal: External review by statisticians or researchers with relevant expertise.
- Next steps: We welcome reviewers — contact us at info@skolkoll.se or via the support page.
Limitations
- Skolkoll is not an official statistics producer — we compile and make public data more accessible.
- Data may differ from Skolverket's publications due to timing or aggregation level.
- Model calculations (e.g. SALSA) follow the published methodology but may have implementation differences.
- Comparisons between schools should always consider data quality and socioeconomic factors.