Survey uncertainty register
StatMedAnalysis | 7 November 2026 | Use aggregate or hypothetical figures only.

Use one row per threat to a specific result. This register is qualitative: do not add unlike uncertainties into one score or widen a confidence interval by guesswork.
1. Identify the result and its sampling uncertainty with units and confidence level.
2. For each additional threat, document the mechanism, affected result, evidence and feasible sensitivity analysis.
3. Revisit after each sensitivity analysis and record whether the stated interpretation changed.

Result / target population / period: ______________________________________
Estimate and sampling margin (level, units): _______________________________
Threat | Mechanism | Data supporting concern | Direction known? | Sensitivity analysis | Decision/owner
______ | _________ | _______________________ | ________________ | ____________________ | ______________
______ | _________ | _______________________ | ________________ | ____________________ | ______________
Residual limitation for results text: _____________________________________
Fictional example: 18% ±3 percentage points (90%). A boundary revision added a neighboring area. The sampling margin describes sampling error within the published geography; it does not quantify the revision. Repeat on stable geographic units if available. Mark the direction unknown until examined.
Completion check: each asserted direction has evidence; no unsupported numerical bias correction has been applied.