Citizen science data quality begins in the workflow
Reliable contributions do not appear because a database rejects bad values at the end. Quality is shaped by the task, guidance, instrument, context, participant feedback and validation route.
“Can we trust citizen science data?” is too broad a question. Trust depends on what the data is intended to support, how the method is performed, which contextual information is preserved and what checks happen before and after submission.
Translate the protocol into actions
A research protocol written for trained specialists cannot simply be placed in a help screen. The application must turn it into observable steps, show what a valid contribution looks like and prevent impossible or misleading states where practical.
- Acquisition guidance
Timing, position, orientation, repetition and environmental conditions. - Required context
Device, location, method version, calibration state and relevant observations. - In-app checks
Ranges, completeness, consistency and method-specific warnings. - Quality indicators
Visible confidence or validity signals without implying certainty. - Review workflows
Expert assessment, peer review, consensus or escalation where appropriate. - Traceability
Processing versions, edits, exclusions and provenance preserved for analysis.
Different tasks need different validation
A photograph of a species, a water-colour observation, a camera-derived measurement and a classification of astronomical imagery do not share one universal quality model. A project may need training examples, repeated observations, reference targets, calibration, consensus thresholds, automated anomaly detection or expert adjudication.
Do not hide limitations
A strong participant experience can make a method feel effortless. It should not make uncertainty disappear. Supported devices, environmental constraints, calibration validity and intended use should be explicit to participants and downstream users.
Relevant experience
iSPEX required a guided optical acquisition and processing chain. Black Hole Finder uses experience and agreement inside a classification workflow. Mini Secchi translates a physical field procedure into guided observations.
Design this into your project
These decisions are most effective when made with the research method and participant workflow—not added after the app has been built. Tell Pocket Science about the goal, participants and constraints.