The most important sentence on this site is also the easiest to skip: dispatch data is not crime data. They are related, but they measure different things, and treating one as the other produces bad conclusions in both directions.
What dispatch data measures
A dispatch record says: someone reported something, and the agency sent a response. It captures operational activity — everything emergency services actually spend their time on. Most of it is not crime: medical emergencies, automatic alarms, hazards, collisions, wellbeing checks, disputes that end with a conversation. Even the crime-shaped categories (assault, break and enter) are initial reports: some will be unfounded, reclassified, or duplicated across services responding to the same incident.
What crime statistics measure
Official crime statistics — the kind published annually by police services and national statistics agencies — count verified, founded offences, categorized after investigation under standard reporting rules. They are slower (published with months of lag), but they are the authoritative record of how much crime actually occurred. They also miss what was never reported to police at all, which victimization surveys try to correct for.
Where each one fails
Dispatch data over-counts (unfounded and duplicate reports) and under-describes (initial labels, no outcomes). Crime statistics under-count (unreported offences) and arrive late. Dispatch data reflects reporting behaviour — areas where people call more, or where proactive policing generates more records, show more activity for reasons unrelated to underlying conditions. Crime statistics reflect recording policy, which changes over time and differs between services.
When to use which
Use dispatch data — this site — for recency and granularity: what kinds of activity an area sees, its rhythms by hour and day, how this month compares to the area's own history. Use official crime statistics for questions about actual crime levels, long-term trends, and between-city comparisons. The honest approach, and the one Explore takes, is to present dispatch data as exactly what it is: a recent, detailed, imperfect operational picture — labeled with its period, aggregated for privacy, and never dressed up as a crime rate.