Features
Everything DataFlow checks, and how it helps you fix it.
The checks a careful research assistant would write, set up from your form and run on every sync, with a fix ready for each problem and a record of every change.
High-frequency checks
Set up from your form when you import it, and run on every sync. Each flag says which rule fired, with what threshold, against which group.
- Duplicate IDs and interviews sent twice
- Interviews too short or too long
- Outliers by IQR or z-score, within villages or teams
- Ranges and the form's own constraints
- Missing answers, and “don't know” and refusal rates
- “Other, specify” answers that fit an existing option
- Your own rules in XLSForm syntax, and impossible combinations from a cross-tab
Back-checks
A back-check survey revisits some interviews. DataFlow links it to the main survey and compares answers by type: ones that shouldn't change, ones that test the enumerator's skill, and ones that may change, within a tolerance. Differences are counted per enumerator, per back-checker and per question.
Enumerator monitoring
Six statistical signs of made-up interviews per enumerator, against the others working in the same area. If the form records GPS, journeys no one could make between interviews. Flag rates per 100 interviews for each person.
Corrections and audit trail
Corrections sit on top of the raw data, which is never modified. Each one records who, when and why, and can be undone. The activity log is tamper-evident, and every download includes a Stata do-file that replays each correction from the raw data.
Field Report
One file, about half a megabyte, that opens offline in any browser: progress and forecast, every check and its trend, team pace and quality, and the cleaning backlog. Three sharing levels, with a preview of exactly what goes in.
Personal-data protection
Personal details found in the form and in free text, and locked in a vault that also opens in VeraCrypt. Maps show areas, not exact places, unless someone asks, which is logged.
A local AI assistant
It explains flags, prepares fixes for you to approve, matches “other” answers to existing options in any language, and writes a check from a sentence. It runs offline on the computer and never makes up numbers: DataFlow counts, the AI explains.
The Datadom
Each study is a small kingdom that grows as its data gets clean. Stars reward care, not speed, and old problems block new levels.
