How to read local crime data without being fooled by it

Question, scoped Is the crime around a specific address high, and is it getting better or worse? As of October 2026. The answer Start with the police department's own open data, then use national data only for comparison. - City open-data portals publish incident-level records. Dallas publishes police incidents through its open-data portal (https://www.dallasopendata.com). Many other Texas cities do the same; search "<city> open data police incidents". - The FBI's Crime Data Explorer (https://cde.ucr.cjis.gov) gives agency-level counts. Be careful with years around 2021: the FBI retired the old Summary Reporting System and moved to NIBRS, and many agencies did not report fully that year, so a dip or gap there is a reporting artefact, not a real change. Four rules that stop most bad conclusions: - Count by type. A property-crime count driven by a shopping centre's shoplifting reports says little about a residential street two blocks away. - Use rates, not raw counts, when comparing cities: incidents per 1,000 residents. - Look at three or more years. One year is noise. - Reported incidents measure reporting as well as crime. A neighbourhood that calls the police more can look worse. Sourced City open-data portals and the FBI Crime Data Explorer, both primary. What would change this answer A change in how the department classifies or publishes incidents. Portals usually note this in their dataset description; read it before comparing years.