Exercise · Intermediate
Two numbers, one file
The cluster reports 76% water coverage and the district reports 82% from the same survey. Both are defended, both are computed correctly, and the meeting is in an hour. Work out which is right, and write the paragraph that stops it happening again.
Two documents crossed your desk this morning, both drawing on the same household survey.
The cluster situation report says water coverage is 76%.
The district water authority’s brief says 82%.
Neither author has made an arithmetic error. Both will be in the meeting at four, and the coordinator has asked you which number to use.
The file
wash-household-survey-2024.v1.csv — 2,403 households across three districts.
What to do
One: reproduce both. Compute water coverage under each of these definitions and find which produces 76% and which produces 82%:
- households on an improved source, JMP list including delivered water;
- households on an improved source, excluding delivered water;
- households with at least basic service — improved and within 30 minutes;
- households on an improved source, restricted to the two districts the cluster covers.
Two: decide which is right. One of these is the JMP definition of the indicator both documents claim to report. Say which, and cite what settles it.
Three: write the definition entry. Six fields — numerator, denominator, the level it is valid at, the standard, the decision it informs, and the value.
Four: write the paragraph. One paragraph for the coordinator, before the meeting, that lets them chair it. It has to say which number, why, what the other one is, and what stops the disagreement recurring.
Check yourself
| Expected | |
|---|---|
| Households | 2,403 |
| Improved source, JMP list | 82.3% |
| Improved source, excluding delivered water | 79.7% |
| At least basic service | 56.3% |
| The gap between improved and basic | 26.1 points |
| District values in the raw column | 6, for three districts |
Neither document’s number is exactly one of the first two, and that is part of the finding. Work out what else could account for the remainder before assuming someone rounded.
Three questions
Two sentences each.
1. The two documents differ by six points and both authors computed correctly. Name the single artefact that would have prevented the disagreement, and say where it has to live for either author to have consulted it.
2. The basic-service figure is 56.3% — twenty-six points below improved source and twenty points below either number in the meeting. Say why neither document is reporting it, and whether that is defensible.
3. Your paragraph tells the coordinator which number to use. Write the sentence in it that the losing author can accept, and say what makes it acceptable.
The point
This is the meeting the definition panel exists to prevent, and it recurs quarterly in every cluster in this sector.
It is not a data problem. The file is the same, the arithmetic is the same, and the survey is fine. Two people implemented the same indicator name as two different indicators, and nothing in either document made that visible.
It happened on this platform. The WASH course taught the improved-source figure without delivered water and the project’s own indicator documentation had it right; they disagreed for months and it surfaced only when a dashboard computed the same number a third time. The correction ships with a change-log entry and an assertion in the pipeline — because the fix for this is never “be more careful”, it is an artefact that fails when two things disagree.