01
A validation report
Agreement between the reconstruction and the collected values, at national, regional and category level. This is the section that tells you whether to take the rest of the report seriously.
A retrospective quality review of a round you have already published
Most offices start here. Take a closed, published round — of any programme — and run an independent reconstruction against it. Because the collected answers are already known, the exercise validates itself: you can see exactly how well the reconstruction reproduces your figures before you rely on it for anything.
What a retrospective review produces
Four deliverables, all reviewable by your own staff
02
A ranked list of flagged records
Every flag carries the reason it was raised and the explanation behind the estimate that raised it. Ranked, so a team with limited review capacity can start at the top and stop when it chooses.
03
Divergence clustered by pattern
Collection-unit effects, geographic patterns, category-specific reporting problems, and artifacts introduced in processing or arriving from administrative feeds. A pattern is a finding; a scattered outlier usually is not.
04
A written note on each pattern
In a form suitable for internal methodological discussion — the kind of note that can be tabled at a methods meeting rather than translated first.
Why offices start here
There is nothing at risk in a closed round
- The round is closed and published. No released figure can be altered by anything in the report.
- Accuracy is checkable. You verify the reconstruction against data you already hold, before it informs any decision.
- No change to fieldwork. Nothing about how your enumerators, supervisors or instruments work is touched.
- The findings stay internal. The report goes to your methodologists. What, if anything, is done with it is entirely your office's decision.
What usually follows
The review decides the next step, not the vendor
If the reconstruction reproduces your figures closely and the flagged patterns turn out to be real, offices typically extend the same approach to an active round, so divergence surfaces while fieldwork can still act on it. If it does not, you have learned that cheaply, on a round where nothing was at stake. Both outcomes are useful, and we would rather you find out on a closed round than a live one.
Scope of a first review
- One programme, one round. A single completed round of a single programme — deliberately narrow.
- Your infrastructure. Executed inside your own deployment. No microdata crosses a border.
- Your registers, your call. Which administrative registers are included, if any, is decided by your office.
- Your methodologists in the loop. Quality rules are configured with your staff, not imposed by us.
Start with a round where nothing is at stake
A technical briefing identifies a suitable closed round and sets out exactly what a review of it would produce.