Quality control of a post-distribution monitoring database…

International NGO · Written recruitment test

Practical exercise · Assistant level

MEAL Assistant (data entry and data quality)

Duty station: Bosaso, Bari region (Somalia)
Time allowed: 60 min
Marks: 100 marks
Pass mark: 60 %
Profile: Assistant, 0 to 2 years of experience

Instructions

The test lasts 60 minutes. It is marked out of 100; the pass mark is 60.
You may use a calculator or a spreadsheet. Show your workings: a correct result without workings receives only half the marks.
Any outside help (another person, artificial intelligence) is forbidden.
The documents are fictional. Do not make unjustified assumptions; if you make one, state it.
Document 1 is a 14-row extract selected for the exercise: the percentages requested refer to this extract only.

Original practice test, built on the format of written tests used by United Nations agencies and international NGOs. The context and figures are fictitious.

Documents provided

Document 1. Extract from the post-distribution monitoring (PDM) database, October cash transfers

Row	Household ID	Enumerator	Date	Sex of head of household	Household size	Amount received (USD)	Interview duration (min)	Satisfied	Knows a complaints channel
1	BOS-0412	E1	14/10	F	6	90	32	Yes	Yes
2	BOS-0415	E1	14/10	M	5	90	28	Yes	No
3	BOS-0433	E2	15/10	F	7	90	35	No	Yes
4	BOS-0433	E2	15/10	F	7	90	35	No	Yes
5	BOS-0451	E3	15/10	M	4	900	30	Yes	Yes
6	BOS-0467	E3	16/10	F	38	90	27	Yes	No
7	BOS-0480	E4	16/10	M	6	90	6	Yes	Yes
8	BOS-0482	E4	16/10	F	5	90	7	Yes	Yes
9	BOS-0491	E4	16/10	M	8	90	6	Yes	Yes
10	BOS-0503	E2	17/10	F	6	60	33	No	Yes
11	BOS-0519	E1	21/10	M	5	90	29	Yes	Yes
12	BOS-0522	E3	17/10	F	3	90	31	Yes	No
13	BOS-0530	E1	18/10	F	6	90	30	No	No
14	BOS-0545	E2	18/10		5	90	34	Yes	Yes

Document 2. PDM quality control rules (note from the MEAL officer)

Survey period: 14/10 to 18/10. Distribution: 2 and 3 October, USD 90 per household (single amount, whatever the household size).
Minimum duration of a complete interview: 15 minutes (the questionnaire has 46 questions).
Plausible household size: 1 to 20 people.
All fields are mandatory. A household must appear only once.
No data is changed in the database without a record: each correction is entered in the cleaning log (row, original value, new value, source, date, author).

Document 3. Results of the MEAL officer's phone checks (to be used for question 2)

BOS-0451: the household confirms it received USD 90 (typing error).
BOS-0467: the household has 8 members (typing error).
BOS-0519: the interview took place on 17/10; the tablet date was set incorrectly.
BOS-0545: the head of household is a man.
Enumerator E4: of the 3 households reached, 2 state that they were never interviewed. Decision: E4's 3 interviews of 16/10 are excluded and will be redone by another enumerator.
BOS-0503: the female head of household confirms she received only USD 60; she says that "a person at the distribution point kept the rest".

Question 130 marks

Check document 1 against the rules in document 2. For each row with a problem, state the nature of the problem and the action to take (correct after verification, delete, call the household back, report). Present your answer as a table.

Your answer

0 words

Model answer

Row	Problem	Action
4	Exact duplicate of row 3 (same ID BOS-0433, same answers)	Delete row 4 and record the deletion in the cleaning log
5	Amount of USD 900 impossible (single amount of USD 90): probable typing error	Do not correct "by guesswork": check by phone, then correct with a record
6	Household size of 38 outside the plausible range (1 to 20)	Check by phone, then correct with a record
7, 8, 9	Interviews of 6 to 7 minutes, far below the 15-minute minimum, all conducted by E4 on the same day, with identical "Yes / Yes" answers: risk of questionnaires filled in without an interview	Put the 3 rows on hold, report to the MEAL officer, have the households called back by another person
10	Amount received of USD 60 instead of USD 90. This is not necessarily a data entry error: it may be a diversion	Do not correct the data. Report it the same day to the MEAL officer for referral to the complaints mechanism
11	Date of 21/10, outside the survey period (14 to 18/10)	Check with E1 and the household, correct with a record
14	Sex of head of household missing (mandatory field)	Call back the household or enumerator E2, complete with a record

Rows 1, 2, 3, 12 and 13 show no anomaly under the rules.

Marking grid

Duplicate, out-of-range values, out-of-period date and missing field all identified (12 marks)
E4 anomaly identified with its three indicators: duration, same enumerator and same day, identical answers (8 marks)
Row 10 treated as a possible sign of diversion and not as a mere data entry error (5 marks)
Correct actions: check before correcting, keep a record, do not delete without a decision (5 marks)

Your mark for this question / 30

Question 225 marks

After the checks in document 3, establish the cleaned database (list of rows retained). Then calculate: (a) the average household size; (b) the percentage of households that received the full amount; (c) the percentage of households satisfied and the percentage knowing a complaints channel, in total and then by sex of the head of household. Comment in two sentences.

Your answer

0 words

Model answer

Rows retained (10): 1, 2, 3, 5 (corrected to USD 90), 6 (corrected to 8 people), 10, 11 (date corrected to 17/10), 12, 13, 14 (sex M). Rows excluded: 4 (duplicate) and 7, 8, 9 (E4 interviews cancelled).

(a) Average size: (6 + 5 + 7 + 4 + 8 + 6 + 5 + 3 + 6 + 5) / 10 = 55 / 10 = 5.5 people.

(b) Full amount: 9 households out of 10 (only BOS-0503 received USD 60) = 90%.

(c)

	Number	Satisfied	Know a complaints channel
Female heads of household (rows 1, 3, 6, 10, 12, 13)	6	3 (50%)	3 (50%)
Male heads of household (rows 2, 5, 11, 14)	4	4 (100%)	3 (75%)
Total	10	7 (70%)	6 (60%)

Comment: female-headed households appear less satisfied and are less aware of complaints channels. With 10 households, this is only a signal to be checked on the full sample, not a conclusion.

Marking grid

Correct list of rows retained and excluded (7 marks)
Correct average and full-amount percentage (6 marks)
Correct table by sex and total (8 marks)
Cautious comment on the small sample size (4 marks)

Your mark for this question / 25

Question 320 marks

Propose four checks to build into the data collection form (for example KoboToolbox) or into the Excel file to prevent these errors in future. For two of them, give the exact Excel formula or constraint (the household ID is in column B, household size in column F, starting from row 2).

Your answer

0 words

Question 425 marks

Write the message to the MEAL officer (120 words maximum) reporting on your check. Also state, in two lines, what you must not do regarding enumerator E4 and household BOS-0503.

Your answer

0 words

Other tests for this profession

Panel interview question for this profession

Tell us about a time when you discovered an error or an inconsistency in data you were entering or checking. What did you do?

Competency assessed : Rigour and data quality control

What the panel expects : A specific error with figures, the method used to spot it, compliance with the procedure (check before correcting, keep a record, inform the supervisor), and a measure to prevent the error from happening again.

Possible follow-up questions

How did you check the correct value before correcting it?
Who did you inform, and when?
What did you change in the tool or the procedure afterwards?

Strong answer

During a post-distribution monitoring survey, every evening I entered and checked the questionnaires from six enumerators. On the third day, I sorted the interviews by duration: 11 questionnaires from the same enumerator lasted less than 8 minutes, against 35 on average, with almost identical answers. I did not delete anything. I recorded the cases in the cleaning log and alerted the MEAL officer that same evening, with the table. He had 5 households called back: 3 had never been interviewed. The 11 questionnaires were redone by another team, before the analysis. I then proposed an automatic daily report in KoboToolbox of durations and out-of-range values by enumerator. On the next survey, 420 questionnaires, we spotted and corrected 9 anomalies on the first day, instead of discovering them at the analysis stage.

Weak answer

"I am always very careful when I enter data. If I see an error, I correct it directly so that the file is clean." This answer shows a correction made without verification or a record, with no specific example or figures, and without informing the supervisor.

← All tests

Quality control of a post-distribution monitoring database

Instructions

Documents provided

Question 130 marks

Model answer

Marking grid

Question 225 marks

Model answer

Marking grid

Question 320 marks

Model answer

Marking grid

Question 425 marks

Model answer

Marking grid

Result

Common mistakes that cost marks

Useful references