Lesson 8 of 8
Unit · What the monitoring did not see
Eleven numbers and six denominators
Everything this course computed, on one page, with the table that says which figures may be compared with which. Six different denominators across eleven indicators, and no layout will warn a reader about that on its own.
Count the denominators before writing anything
The course has produced eleven indicators. They come from two files, and they run on six different denominators.
| Indicator | Denominator | n |
|---|---|---|
| At least basic drinking water | All households | 2,403 |
| At least basic sanitation | All households | 2,403 |
| Basic hygiene service | All households | 2,403 |
| Open defecation | All households | 2,403 |
| Below 15 L/person/day | All households | 2,403 |
| Round trip over 30 minutes | All households | 2,403 |
| E. coli risk class | Households tested | 802 |
| Chlorine at the household | Households tested | 961 |
| Water point functionality | Monitoring visits made | 2,629 |
| Points working at every visit | Water points | 242 |
| Population-weighted functionality | People served | ~96,700 |
| Chlorine at the water point | Visits tested | 811 |
Six denominators, and only the first six rows share one. Every horizontal comparison across a boundary in that table needs a sentence, and every one made without a sentence will be made wrongly.
import pandas as pd
denominators = pd.DataFrame({
"indicator": ["basic water", "basic sanitation", "basic hygiene",
"E. coli safe", "point functionality", "reliable points"],
"denominator": ["households", "households", "households",
"households tested", "visits made", "water points"],
"n": [2403, 2403, 2403, 802, 2629, 242],
})
print(denominators)# Build this table first, in code, and print it above the results.Write it as data, not as prose. A denominator table generated by the same script that computed the numbers cannot drift from them; a paragraph describing the denominators can and will.
The report
WASH situation, three districts, 2024
ACCESS n = 2,403 households
At least basic drinking water 56.3% (82.3% on an improved
source; 25.0 points sit on
limited service for
collection time alone)
At least basic sanitation 43.2% (62.0% have an improved
facility; 18.8 points are
shared)
Basic hygiene service 34.1% (67.8% have a facility;
soap observed at half)
Open defecation 13.3%
Safely managed not computable: availability was not asked and quality was
tested on a subsample.
QUANTITY AND TIME n = 2,403 households
Median 23.7 L/person/day
Below Sphere minimum of 15 L 13.4% 323 households
Round trip over 30 minutes 39.3% 944 households
Queue time not separated from walking time. Distance not recorded.
WATER QUALITY n = 802 tested (33.4%)
Free from E. coli 53.6%
High risk (>100 CFU/100 mL) 7.5%
Improved sources 61.1% safe n = 625
Unimproved and surface 27.1% safe n = 177
Tested subsample over-represents untreated water: 533 households reporting
home treatment have no chlorine reading.
WATER POINT FUNCTIONALITY 242 points, 2,904 visits due
Visits finding a working point 74.4% 2,629 visits made (90.5%)
Points working at every visit 33.9% 82 of 242
Population-weighted 83.3% ~96,700 users
Range if every missed visit found a failure: 67.4% - 74.4%
Failure shapes: 82 never failed, 80 intermittent, 46 seasonal, 34 down at
year end. Seasonality is confined to protected wells and springs.
Median days down at the visit: 68 under community committees, 29 under
private operators.
NOT COMPARABLE
Water quality figures are on a tested subsample and do not share the access
denominator. Functionality figures count visits, points and people and are
three different quantities. Districts are not ranked: Nord-Ouest visited
82.7% of its due rounds against Sud-Est's 96.3%.Four rules the layout has to obey
Group by denominator, not by theme. The block headings above are denominators wearing topic names. A reader who takes one number from each block has taken four numbers with four bases, and the blank line between blocks is what tells them so.
Put n in the heading, not the footnote. A footnote is read after the comparison has already been made.
State what is not computable. Safely managed drinking water, queue time and distance are all absent, and each is something a reader will assume is in a WASH report. An absent indicator that is not named reads as an indicator at zero or as an oversight.
Put the non-comparability in its own block at the end. Not as a caveat under each number, where it becomes noise, and not in a methods annex, where it is not read.
What each number is for
A report that only classifies is an assessment. A report that also says what to do is a deliverable, and the WASH figures divide cleanly by the decision they inform.
| Finding | Decision it informs |
|---|---|
| 25 points on limited water for collection time | Where to site new points |
| 18.8 points on shared sanitation | Whether the programme is building or subsidising |
| Soap at half the facilities that exist | Hygiene promotion against hardware |
| 46 seasonal points | Dry-season alternative, not rehabilitation |
| 34 down at year end | Spare parts and management, not new construction |
| 68 days median downtime under committees | The maintenance model itself |
| Nord-Ouest at 82.7% coverage | The monitoring plan before the next report |
The last row is the one most reports omit and the one most likely to change next year’s numbers. A monitoring system that cannot reach a district in the rains produces a report that cannot describe it, and fixing that is cheaper than any of the others on the list.
What this course did not cover
Named plainly, because a WASH analyst will be asked for all of them.
- Menstrual hygiene management needs questions this survey does not ask — private space, materials, disposal — and the MHM indicators are not derivable from what is here.
- School and healthcare facility WASH are separate instruments with their own ladders, and the household ladder does not transfer to an institution.
- Faecal sludge management is the second half of safely managed sanitation and needs a containment and emptying survey.
- Cost per beneficiary and tariff analysis need financial records; the
fee_collectedfield says whether a fee was taken, not how much or where it went.
A report that names its gaps is more credible than one that appears complete, and a gap named is the first line of the next terms of reference.
What comes next
Module 4 continues in other sectors — food security and the IPC, protection case management, education attendance — and each brings the same three questions in new vocabulary. You have now met them three times: what did I count, over what, and what should change because of it.