Tuesday, November 18, 2014
The Problem With Crime Maps
I've never been completely convinced of the legitimacy of crime maps. There are obvious dangers of stigmatizing neighborhoods and areas as hot-spots of crime based on crime data which may not be entirely accurate. The accuracy of crime data depends on a number of factors, including victims reporting crime, the police accurately recording the data and the data being accurately plotted to the correct location.
This last point was apparent in this Seismograph San Francisco Bike Theft Map. While creating the map Seismograph found that there was a hot-spot of bike thefts at 850 Bryant Street. It turns out that this is the address of a police station. Stolen bikes with no location data were simply plotted by the police at the address of the reporting police station. 850 Bryant Street is therefore the bike theft equivalent of Null Island.
I'm also fairly sure that most crime heat-maps are not normalized for population density. The result is that crime heat-maps often seem to closely resemble population maps. Areas with a high population density appear to be crime hot-spots, although they really only show that where there are more people there is proportionally more crime.
Brazilian crime map Radarea has attempted to overcome some of these problem by crowd-sourcing the reporting of crime. By allowing the victims of crime to directly input the details and locations of crime they presumably hope to record some of the crimes not reported to the police and to eliminate some of the errors in the recording of crime data.
Radarea claim that the map will help tourists visiting unfamiliar areas to identify neighborhoods with high levels of crime and will also help locals discover the safest areas in their neighborhoods. These are some bold claims. I'm not convinced that crowd-sourcing crimes will provide data that is any more accurate than official crime data. It seems to me Radarea are running the same risk of providing inaccurate crime maps which may well give users an entirely false picture of the relative safety of different locations.