Protecting data quality in surveys
Checklist and tips for collecting clean, analyzable field data
Why data quality matters
Even the most advanced analysis is useless if the raw survey data is biased, inconsistent, or incomplete. Data quality issues multiply in field surveys where multiple enumerators, devices, and environments are involved. Protecting quality means ensuring responses are reliable, consistent, and ready for statistical analysis without excessive cleaning.
Core principles
- Clarity: Questions must be unambiguous and tested before launch.
- Consistency: Enumerators should follow identical protocols.
- Validation: Each response should be checked against logic and ranges.
- Documentation: Keep clear metadata, survey versions, and logs.
Checklist for survey design
- Keep questions short, direct, and free from jargon.
- Avoid double-barrelled questions (two ideas in one).
- Use skip logic to avoid irrelevant questions.
- Balance open vs closed questions: open for depth, closed for analysis.
- Translate and back-translate questionnaires for multilingual studies.
Enumerator training & monitoring
Field workers are the guardians of data quality. Invest in:
- Training sessions with mock interviews.
- Clear manuals and FAQs accessible on devices.
- Spot checks and shadowing during fieldwork.
- Daily debriefs to resolve issues early.
Technical validation in digital tools
Use mobile survey platforms or Excel forms with built-in checks:
- Range checks: e.g., Age must be 18–99.
- Mandatory fields: Prevent submission with blanks.
- Skip logic: Hide irrelevant sections automatically.
- GPS & timestamps: Verify location and timing of responses.
- Device ID: Track which enumerator submitted each form.
Data cleaning & monitoring
Even with preventive measures, cleaning is still required:
- Check for duplicate records.
- Run outlier detection (e.g., income 100× higher than median).
- Cross-validate related questions (household size vs. household members listed).
- Audit time taken per interview — extremely short durations often mean poor quality.
Audit-ready practices
For projects funded by clients or governments, audits are common. Keep:
- A frozen copy of raw data.
- A processing log (what cleaning and transformations were applied).
- Version history of questionnaires.
- Enumerator performance reports.
Handover checklist
Final tips
- Pilot surveys are essential — 20 test cases reveal most problems.
- Never rely solely on enumerator honesty — add system checks.
- Incentivize accuracy, not speed.
- Document every decision — assumptions today become audit issues tomorrow.
Data quality is the foundation of trust in survey research. Protecting it ensures your analysis is credible and decisions are based on reliable evidence.