Currently, I’m compiling crime rates for cities in a particular county. Although not a perfect solution, I plan to assign the crime rates measured at jurisdictional level (often city/town/village level) to the residential properties within that jurisdiction. Also, it turns out that the data available from FBI’s Uniform Crime Reports (UCR) is based on voluntary reporting by these jurisdictions and not all cities in the county have reported their numbers. Roughly 30-40% of jurisdictions have reported their numbers.
Additionally, the crimes occurring in unincorporated cities are responded to by the county police and are not supplied with no spatial context. Crimes reported by county police are in aggregate figures and include figures from different corners of their jurisdiction which often are spatially separated by dozens of miles thus assigning crime rates from a country police department to all residential properties within the county might be erroneous.
My solution – assigning crime rates to all properties within jurisdictions whose crime rates are available through FBI UCR with their corresponding figures. For properties that lie in jurisdictions that haven’t contributed to FBI’s UCR, I relate them to the nearest jurisdiction that has reported their numbers and use those crime numbers after weighting them by population of the jurisdiction they lie in.
Thoughts? Opinions? Alternate solutions