methods : : : data collection : : : Providence crime data

Acquiring the Providence crime database was one of the most challenging aspects of my study. The sensitivity of the information as well as the large task of compiling it for an undergraduate thesis were two obstacles that I faced as I worked with the Police Department.

My partnership with the Providence Police has been a valuable component both for my study and their development of GIS for the entire department. Along with other students working on urban theses at Brown, I made presentations to several municipal departments about the importance of GIS at the city level. We worked closely with the Department of Administration, Housing Codes and Inspections, and the Department of Public Works to coordinate data, assemble a databank, and stress the importance of a central lookup table for platlot numbers.

As the Police Department became a critical partner in this collaboration, they agreed to give me the dataset with every crime report written in Providence from 1987 - 2000. The database, which has about 800,000 entries, is a complete set and needed to be coded for privacy and security concerns. After coding, I worked with the database in Microsoft Access for about a month to "massage" the data into a format that ArcView can read.

For each case study, I followed a set of steps to gain a full set of those crimes committed in each case study area:

  1. Correct misspellings/typos in the dataset
  2. Query out all of the reports written in the case study area
  3. Match a platlot number to each report, either by the partial lookup table that we have compiled or by hand, studying paper platlot maps from the Tax Assessor's office.
  4. Eliminate reports written at an intersection (since I would not know at which of four possible parcels the report might have been written)
  5. Eliminate reports that did not contain a date or address
  6. Separate the table into two parts: "before" and "after" - usually I picked case study areas that had at least 3 years of crime data before and after, to make more statistically significant calculations.
  7. I next did a query that counts violations at each property, since I was interested in a tally of crimes per property, not the individual narrative of every crime.
  8. These two queries - before and after - were then exported as .dbf files and imported to ArcView, where they were matched by platlot with the RIGIS layer "provparcels."

[Click here to go to these maps and see the results of this method]

christine coletta
center for environmental studies, brown university
about this project
last updated 2/6/03