Analyzing Crime at Light Rail Stations

Virginia Lenvik
American River College, Geography 350
Data Acquisition in GIS, Spring 2010
virginia_lenvik@yahoo.com
Abstract
The types of crimes occuring around light rail stations in Sacramento County are examined for possible patterns.
Introduction
In the Sacramento region, the Sacramento Regional Transit District (RT) operates bus routes and a light rail system. RT’s light rail has 2 lines and 47 stations. A variety of transportation alternatives is generally thought of as a benefit to the community. Some residents, however, object to having a light rail station located in their neighborhood because they believe it would draw crime into the area. Do crime statistics support that belief? Many things influence the locations of crimes, such as population density and if the area is commercial or residential. Is public transportation another factor? Do light rail nodes influence where crimes occur? A study of the relationship between crimes and light rail stations might reveal a pattern. A strong correlation could suggest that this type of transit does draw crime to an area. And, the types of crimes near light rail nodes could give us more information about the effect of light rail on a neighborhood.
Background
Many things affect crime and some studies suggest that criminals do not travel far to commit most crimes. This is called distance decay. Criminal spatial behavior theories are examined in Mapping Crime, Principle and Practice (Keith Harries Ph. D., 1999). Identification of hotspot areas can be useful for resource allocation.

Mapping crime data has become much easier recently because there is very detailled, geocoded data available. In the paper (found on the Geography 350 site) Crime Analysis: A Look Into Residential Burglaries in the City of Sacramento, (Auturo Smith, 2004) many useful sources for data can be found

Crime statistics can be difficult to present when looking at a large area like a county. The Manual of Crime Analysis Map Production (Velasco, Mary and Rachel Boba, PhD, 2000) suggests various ways to work with point data.
Methods

The crime data for 2009 from the Sacramento police department contained 69,178 records (offenses and count shown at right), but many of those would not be of interest for this study. Vehicular crimes and DUIs would obviously need to be excluded. It seems unlikely that bank robberies would be committed by someone who would use the train as "getaway" transportation. White-collar crimes were also excluded from consideration. Traffic offenses had been one of the largest groups (behind burglary and "crimes against person"). The process of excluding crimes that were not of concern to residential neighborhoods reduced the number to 35,209 crimes. The types excluded are shown here. Some data was not usable and had to be excluded. The crimes were then grouped into categories called violent, vice, property and nuisance. In this scheme, vice crimes include prostitution, drugs, etc. Violent crimes are weapons, homicide, assault etc. Vandalism was included in nuisance crimes rather than property crime. The crimes were then grouped into categories called violent, vice, property and nuisance. In this scheme, vice crimes include prostitution, drugs, etc. Violent crimes are weapons, homicide, assault etc. Vandalism was included in nuisance crimes rather than property crime.

To study crime in neighborhoods, land use information was obtained to determine residential areas. The land use information was much too detailled (as the crime data had been). Parcels were aggregated and dissolved over several iterations. The initial shapefile contained 4,693 land use types. Those became 12 types, shown below.

Finally, only three types of land use are shown here (with the light rail station locations).



Results

Parcel data from Sacramento County provided land use information for the study area. This data was used to determine which rail stations were in more residential areas. After some study, it was discovered that by calculating the first charater of the LANDUSE field to a new field, 12 types could be symbolized. Twelve colors are difficult to differentiate and the detail was not necessary, so the final grouping above was used.

Light rail stop locations were found at the Sacramento Area Council of Governments (SACOG) website. The site has a GIS clearinghouse with data and metadata. Light rail stations were buffered to create areas possibly influenced by the points. The buffer was assigned the name of the station it represented using a spatial join.


Geocoded crime data for 2009 was obtained from the Sacramento police department. Crime in this database is described in detail in one field and in a more generalized way in another field which uses coding called Uniform Crime Reporting (UCR) which was developed by the FBI. The police department provides a document to decode the crime types. The UCR program is summarized here and in documents from the U.S. Department of Justice. The method for data aggregation is described above. By intersecting the categorized crime data with buffers of stations, a subset of data was created that was associated with the stops. Charts and reports created from that subset identify hotspots and show which nodes have the most crimes.
Analysis

While the land use data was more easily generalized, quite a lot of study of crime types needed to be done. The Sacpd.org database stucture seems to be well thought-out and extensible. However, one problem encountered was the crime report type called "crimes against person". The records in that group/classification were a mix of every category of crime. It was unfortunate that so much data was unuseable. Other data were hard to classify as well. "Public peace" included begging and mooching and one disturbing incidence of lynching. Since lynching is a homicide and a felony we know that the data is not perfect.

Conclusion


The goal of the analysis had been to identify high-crime areas, to see if there was more crime near light rail nodes and determine which nodes were hotspots. It was impossible to say if light rail draws criminals to a place with this type of analysis, but it appears that there is no correlation. The theories presented in the background research may be true. Factors like population density and finding a suitable target probably affect criminal behavior more than something like the ability to access a place with public transportation. This study did suggest that light rail stations in downtown had more crime than those in residential areas. So, while it did not show that light rail is a conduit for crime into neighborhoods, it did identify which stations had the most crime. This type of hotspot identification information is useful for allocating security resources.


References