thearchivist.funSampling bias
A dataset built from search results, surviving pages, popular accounts, or one language can overrepresent the material easiest to find. This shapes what conclusions are possible. A large number of records does not automatically remove the effects of how those records were selected. Describe the source population you care about and the actual route by which examples entered the collection. Record exclusions, inaccessible spaces, and missing periods. If you use a convenience set, present it as such and keep conclusions tied to that set. A qualitative case study can still be valuable without representing every user or community. Avoid reporting a percentage as a population estimate when the collection design does not support it. The useful question is not only how many records exist, but which relevant records had a chance to appear.
Context and provenance
A collection process that systematically favors some material over other relevant material.
Sources and evidence
Editorial review: 2026-09-23. New evidence may change this account.