How to Audit the Data Behind a Scientist Ranking

Confirm the person, identifiers and linked works. Look for name collisions and accidentally joined author records before analysing numeric indicators. A flawless formula on incorrect identities still produces unreliable results.

Record each source, the time it was queried, included document types and any transformation or normalisation. Check whether duplicate publications, historical affiliations and retracted works are treated according to a published rule.

Choose a few records and verify their counts against source-linked evidence. Examine outliers and surprising jumps. Sampling cannot prove the entire dataset is correct, but it can reveal systematic matching or counting defects.

Ask who can approve corrections, how methodology changes are versioned and whether recognition decisions follow independent eligibility rules. Transparency is a process, not simply publishing one formula.