Research

Original analysis of public records: a dataset read end to end and reported, with the method stated and the counts reproducible from the source. This is the section where the publication does its own measurement rather than explaining someone else's. Each piece says what the record covers and what it misses, cites the snapshot it was computed from, and avoids drawing a trend from a number that reporting bias could explain on its own. The record's own labels and classifications are reproduced rather than corrected, and where a pattern could come from how incidents were recorded rather than from what happened, the piece says so.