Data quality is the foundation of every credible humanitarian assessment. Yet in Somalia's complex operating environment — characterised by insecurity, population displacement, limited infrastructure, and diverse linguistic contexts — maintaining rigorous data quality is one of the most persistent challenges facing research and MEL teams.
At Himma Consultancy, we have conducted household surveys, needs assessments, and monitoring exercises across all Federal Member States and Somaliland since 2016. Over that time, we have identified the same data quality failures appearing again and again — and developed systematic protocols to prevent them.
The Most Common Failures
1. Enumerator Bias and Social Desirability
In Somalia's community settings, respondents frequently tell enumerators what they think they want to hear — particularly on sensitive topics such as food consumption, protection concerns, or programme satisfaction. When enumerators are from the same clan or community as respondents, this effect can be amplified or reversed depending on social dynamics.
The solution is not simply to use outside enumerators — it is to design instruments that minimise socially desirable responses through indirect questioning, use of vignettes, and careful piloting. Enumerators must also be trained to probe neutrally without leading respondents.
2. Translation Failures
Somali has significant dialectal variation, and technical development terminology — empowerment, governance, resilience — does not always translate cleanly. We have seen survey instruments where back-translation revealed that the Somali version of a question asked something fundamentally different from the English original.
A question about "household food security" that becomes "whether the family has food today" in Somali translation produces data that measures something completely different from what was intended.
Our protocol requires bilingual review of all instruments before piloting, followed by cognitive interviewing with community members to test comprehension of each question.
3. GPS and Location Errors
In IDP settlements and informal urban areas, GPS coordinates are often recorded incorrectly — either because enumerators record the coordinates at the start of the day rather than at each interview location, or because devices are unreliable. This makes geographic analysis meaningless and prevents verification of site visits.
4. Incomplete and Inconsistent Records
Skip logic failures are among the most common issues in digital data collection. When an enumerator selects an incorrect response to a filter question, entire sections of the questionnaire are skipped — producing records that appear complete but contain systematic gaps.
Our QA Protocol
Himma's data quality assurance framework operates at four levels:
- Pre-field: instrument piloting, enumerator training and certification, and device configuration checks
- In-field: daily supervisor review of submissions, back-checks with 10% of respondents, and real-time monitoring of completion rates and outliers
- Post-field: data cleaning protocols, duplicate detection, range checks, and logical consistency verification
- Reporting: transparent documentation of data limitations and their implications for findings
The Bottom Line
Data quality in humanitarian surveys is not achieved by using the right software or hiring experienced enumerators alone. It requires systematic attention at every stage of the research process — from instrument design to final reporting. In Somalia's environment, where decisions based on bad data can mean aid going to the wrong people or missing the most vulnerable entirely, this is not a technical nicety. It is an ethical obligation.
