Research quality checks
Backcheck an independent survey subset
- Randomly select completed surveys for backchecks
- Independent staff repeat key questions
- Compare records; re-interview or retrain if thresholds are exceeded
Illustrative summary of the source-described audit of a subset of 576 endline respondents in 76 villages. The error thresholds and subset size are not supplied. No accuracy result is inferred.
See source →How
When to use
Field data collection in rural areas faces risks of enumerator error, fabrication, or recording mistakes during household visits. Without independent verification, measurement errors can distort key trial findings or baseline balance estimates. This method re-interviews a subset of respondents to audit survey accuracy and maintain high data quality.
Re-interview a randomized subset of study participants using independent backcheckers to audit survey response accuracy.
- Select a random subset of completed household surveys from each field enumerator team for quality auditing.
- Assign independent backcheckers who did not conduct the initial household interviews to resurvey selected participants.
- Re-administer key demographic, agricultural, and security survey questions using a condensed audit instrument.
- Compare backcheck responses against initial survey records to flag discrepancies in reported land area, crop choices, or security perceptions.
- Re-interview respondents or retrain enumerators whenever error rates exceed pre-specified research team thresholds.
Why it works
Systematic cross-verification ensures that analyzed datasets accurately reflect real household conditions.
Example
Research teams implemented backchecks on a subset of the 576 endline household surveys across 76 villages in Migori County, Kenya. The quality audit verified response consistency across enumerator teams, ensuring accurate measurement of farm security and crop yields.
About the study
Participants
A subset of 576 smallholder respondents across 76 villages in Migori County, Kenya.
Study Design
Independent backcheck auditing of tablet survey responses following household-visit endline data collection in a cluster-randomized trial.
Results
Quality backchecks confirmed high accuracy across enumerator entries, supporting data validity across 1,153 total household-round survey records.