Trusting Citizen Science

Olu: For my ecology project I'm using bird sightings uploaded by the public through an app, and my supervisor says the data isn't reliable enough for a dissertation. I think she's being old-fashioned, because there are two million records and no research team could ever collect that many.

Ingrid: Size doesn't fix bias, though. Most of those records come from parks near cities on sunny weekends, so you'd conclude birds avoid farmland and rain when really it's people who do.

Carlos: Misidentification is the bigger issue for me. I volunteered on one of these apps, and beginners constantly log rare species that turn out to be common ones, which would wreck any analysis of range changes.

Olu: Both problems are known and there are statistical corrections for them. You can weight records by how often a location is visited, and the app flags unusual sightings for expert review before they're accepted.

Ingrid: If you actually apply those corrections and show them in the methods chapter, I'd say the data is fine for common species. I just wouldn't make claims about anything rare.

Carlos: Expert review helps, but it only catches the sightings somebody noticed were odd. I'd still want Olu to compare a sample against professional survey data from the same areas before trusting any of it.