Methodology
(Click on maps for larger view)
Data Acquisition and Processing: Census Block Groups
Along with our base maps, we obtained a shapefile of census block groups from ESRI Data & Maps, 2005. The projection type was a NAD 83-based GCS. Because this shapefile only contained population information, we procured census statistic tables from factfinder.census.gov, showing the 2000 median household income for every census block group within Los Angeles County.
Defining "median household income"
Median Income: Divides the income distribution into two equal groups, one having incomes above the median, and the other having incomes below.
Household: Includes all the people who occupy a housing unit as their usual place of residence.
Housing unit: A house, an apartment, a mobile home or trailer, a group of rooms, or a single room occupied as separate living quarters, or if vacant, intended for occupancy as separate living quarters. Separate living quarters are those in which the occupants live separately from any other individuals in the building and which have direct access from outside the building or through a common hall. For vacant units, the criteria of separateness and direct access are applied to the intended occupants whenever possible.
Source: Glossary, factfinder.census.gov
To join the Excel table containing census data to our spatial layer, we created a new field that would act as the "primary key" for the join, and used the formula [= TRACT & BLKGRP] to let the primary key be the tract number followed by the block group number. We did the same in ArcMap, and then joined the two datasets using this unique ID. This appended median household income information to our census block group layer.
Data Acquisition and Processing: Drug Treatment Layers
First, we created a layer detailing the location of treatment facilities by first obtaining a list of Treatment Providers from the LA Department of Health Services Alcohol and Drug Program Administration Annual Report 2002-03. We found street addresses and zipcodes through GoogleMaps, Superbook (online Yellow Pages), and the Department of Health Services' online directory. We converted these street addresses to latitude / longitude coordinates through GPS Visualizer (www.gpsvisualizer.com), and then converted those results into UTM coordinates via a conversion formula written for Excel (www.uwgb.edu/dutchs/UsefulData/UTMFormulas.HTM). The calculations for UTM coordinates were based on the GCS 80 ellipsoid, making them compatible with our pre-existing maps and layers. We inputted these UTM coordinates into ArcMap, generating Treatment Facilities layer.
Secondly, we obtained a shapefile of the Los Angeles county health divisions (Service Planning Areas, or SPAs) from the Los Angeles County Children's Planning Council (www.childrensplanningcouncil.org). There was no metadata with this shapefile, but our contact at the Children's Planning Council informed us that it was projected to a NAD 27-based GCS. Therefore, we defined the shapefile's projection to this standard.
We augmented the SPA layer attribute table with Los Angeles Public Health statistics on drug use, including the percentage of participants in each SPA primarily using each drug type.
Data Acquisition and Processing: Drug-Related Crime Layers
Because the Los Angeles Police Department could not provide us with GIS-compatible shapefiles of their law-enforcement regions, we chose to digitize these regions manually. We first added .GIF images of the LAPD police communities from the LAPD's website (www.lapdonline.org). We georeferenced these images to our base maps, to act as visual guides in the digitizing process.
The boundaries between police communities almost always aligned with census block group boundaries, so instead of attempting to draw manually the convoluted borders between police communities, we defined them by aggregating census block group borders. In a few locations, these edges of police communities did not align precisely with census divisions. In these cases, we chose to draw the boundaries along the nearest census block group boundary. This meant a slight areal misrepresentation of the 'real' police communities, but we deemed this to be preferable to the ecological fallacy of dividing a census block group between two police communities, or losing that census block group's data altogether.
Finally, we appended statistics published by the LAPD's Narcotics Division (Narcotics Arrests and Seizure Statistics, December 2002) to the shapefile of police communities. These statistics, organized by community, include the number of arrests by drug type and the total number of drug-related arrests.
Interim Assessment
After creating both the SPA (drug treatment) and LAPD (drug-related arrest) layers, we noted the large discrepancies in their coverage and divisions (see images below). The LAPD jurisdiction is limited to the city of Los Angeles and divided into relatively small communities (left). Health data, conversely, covers the entire county of Los Angeles; its SPAs are large regional units often encompassing several municipalities (right).
Because of the major discontinuities between these two areal units, we decided that we would not be able to directly compare police data to health data. Instead, we chose to compare each variable, in isolation, to socioeconomic status.
Aggregation: Creating Composite Layers
We proceeded by aggregating the census block groups within each police community, and each SPA, and dissolving them based on these larger regional units. This effectively removed all internal divisions, leaving us with SPA and police community shapefiles that retained census information. We made sure to retain aggregated census statistics during this dissolve, including summing population counts and averaging median household incomes.
As in the drawing of police boundaries, the aggregation of census block groups within each SPA had a few alignment problems. SPA boundaries did not always line up with census boundaries (see map below). In these cases, as before, we chose to align SPA boundaries to the nearest census block group boundary.
To adjust for population differences between police communities (and between SPAs), we calculated per-capita attributes (e.g., by dividing the number of participants by the total population in each SPA).
Maps
Our final step was to create maps for analysis. We chose to concentrate on two main dimensions: drug-related crime (at the LAPD police community level) and drug treatment provision (at the SPA level). We created two series of maps. The maps compare arrest or treatment statistics to another variable, usually income, to determine correlations.