Analysis of Crime Data Near Brownfields Sites


FACT SHEET
Analysis of Crime Data Near Brownfields Sites
EPA 560-F-18-185
October 2018
Introduction
Crime and fear of crime are of great concern to community
residents and local officials and can influence public and
private investments. To examine crime as an issue for
revitalization sites, EPA's Office of Brownfields and Land
Revitalization (OBLR) analyzed spatial and temporal data
regarding crime associated with brownfield property
revitalization projects. This fact sheet outlines EPA's approach
for collecting and analyzing crime data near brownfield sites
and is intended to serve as a guide for stakeholders who may
be interested in conducting similar analyses, or building upon
this research and forging partnerships with criminology
researchers to examine issues in their community. The
objectives of EPA's analysis were to:
•	Investigate potential spatial (location-based) and temporal
(time-based) connections between brownfield sites and
crime rates within a 1-mile radius of brownfield sites, from
at least three cities within each EPA region.
•	Determine if there are differences in crime rates within a %-
mile radius and between %-mile and 1-mile radius of
brownfield properties.
•	Evaluate if there are differences in crime rates before and
after key brownfield activities (milestones), such as cleanup
and redevelopment.
This fact sheet outlines the data used for this analysis and
assumptions about that data, the methodology used,
limitations and challenges for such analyses, and
recommendations and resources for additional analyses.
Available Data
The two key data streams necessary to conduct an analysis of
crime data near brownfield sites are:
• Crime data represented both temporally and spatially by
individual crime occurrences with dates, and point spatial
data (latitude and longitude). EPA researched two sources
of crime data for this project:
o Census block group crime indices are available through
the ESRI Online platform via Applied Geographic
Solutions (AGS) CrimeRisk (AGS 2018). However, since
census block groups are much larger than the areas of
interest for this analysis (%-mile and 1-mile radius
around sites), this data could not be used to determine
the difference in crime rates at brownfield sites.
Therefore, census block crime data was not used,
o Individual crime location data is maintained by local law
enforcement agencies but the data is not always publicly
available. Individual crimes and their discrete locations
captured by latitude and longitude are required to create
"heat" maps, which illustrate the density of crime
occurrences within an area of interest. Local law
enforcement data is often stored in databases that
include crime activity [including the U.S. Department of
Justice (DOJ) Uniform Crime Reporting (UCR) program]
and latitude and longitude coordinates. Many cities and
agencies make their crime data available for free as a
download. EPA used individual crime data because
point-specific data can be used to evaluate crime rates
spatially and temporally.
• Brownfield site data, including latitude and longitude, and
start and end dates associated with key activities at those
sites. EPA maintains brownfield site data in the Assessment,
Cleanup and Redevelopment Exchange System (ACRES)
database. ACRES data fields critical to this type of analysis
include the unique property identification (ID) code,
address, latitude and longitude, and key start and end dates
for brownfield site milestones, such as: Phase I and II
environmental site assessment, cleanup, and
redevelopment.
Data Selection and Preparation
EPA used a two-step approach to select brownfield sites to
analyze, because both crime data and brownfield site data are
required. First, EPA determined which cities had available
crime data and identified the time period for which the data
was available in those cities; EPA then selected brownfield
sites in cities where revitalization milestone activities had
occurred during the time period with crime data.
Crime Data
The UCR list of agencies that report crime data varies from
year to year depending on several factors. At the time of the
analysis, 2016 was the year with the most recent publicly-
available data and was used to identify cities for the analysis.
Cities were further targeted by selecting only those with
available location-specific crime data. No types of crime were
excluded.
Brownfield Site Selection
Using the list of cities with available UCR crime data, the
ACRES database was queried to identify associated brownfield
sites. In addition to being in a city with UCR data, a
brownfields site also needed to:
1.	Include an activity end date such as Phase 1 or Phase 2
environmental site assessment, cleanup, or
redevelopment within the crime data period,
2.	Have a closed grant,
3.	Contain latitude and longitude information,
4.	Include an activity start and end date after 2005.
Sites with redevelopment end dates were preferentially
selected. EPA selected three cities in each of EPA's 10 regions
using the method outlined above. For each city, ACRES
typically contained a short list of brownfield sites that met the
four criteria, and these were selected for analysis.
For each brownfield site, EPA defined a time period for
analysis by comparing the brownfield activity end date to the
available time period of crime data. An equal number of years
was analyzed before and after the brownfield activity end
date. For example, if a site had an activity end date in 2014
and the crime data for the associated city was available from
2009 to 2016, two years before and after 2014 (the milestone
date) were analyzed (2012-2016).
1

-------
Methodology for Analysis
Analysis of Crime Data Before the Brown field Milestone Date
This section outlines EPA's methodology to conduct
geographic information system (GIS) and quantitative analysis
of the data. The applicability of this methodology to assess
additional brownfields sites will vary depending on the type
and availability of data for a particular site.
GIS Analysis
Once a brownfield site was selected and a date range for
analysis was identified, GIS was used to establish a 1-mile
radius area around each site location (latitude and longitude)
and calculate the distance between the brownfield site and
each individual crime location. For this study, the data was
plotted on two panels for each analysis: (1) before the
brownfields site milestone date and (2) after the milestone
date to produce the following maps:
Crime Incident Heat Map
This type of map depicts individual occurrences of crime at a
location within a defined time period (ArcGIS 2018). Colors are
used to show differences in the frequency of crime over time
with green representing a lower crime frequency, red and
yellow representing a higher crime frequency, and grey
representing areas where no crime events were documented.
Hot Spot Analysis
This type of map displays hot spots as orange grid cells where
crime density is higher and cold spots as areas of blue grid
cells where crime density is lower within a region compared to
the whole 1-mile radius, prior to and after the brownfield
milestone date. The hot spot analysis divides the 1-mile area
around the brownfield site into a spatial grid and then
compares crime density in clusters of grid squares to the
whole area. If a grid square's crime density is significantly
different than the whole area, then it is identified as either a
hot (higher incidence of crime) or cold spot (lower incidence of
crime). A comparison of the two maps before and after the
brownfield milestone activity date identifies the potential
migration of crime events away from, or towards, a site.
Temporal Trend Analysis
This type of map displays the significance of change in crime
within individual grid cells over the time period analyzed,
within an individual cell, dark purple depicts an increase in the
count of documented crime events over the time period and
dark green depicts a decrease in crime events over the time
period. If crime is steady before redevelopment and decreases
around the brownfield site after redevelopment, then this
analysis would not show any change in the before
redevelopment panel, but would show a decrease (a color
change to blue) in crime after redevelopment close to the site
and less of a change farther away from the site.
If all areas under analysis are increasing in crime, a trend
would not be identified because there is no statistical
significance from one grid cell to another. This analysis
provides another way of examining changes in crime over
time, rather than identifying location-specific hot spots.
[crime Incident Heat Map (January 10, 2011 • January 10, 2013) |
| Hot Spot Analysis (January 10, 2011 • January 10, 2013)|
[Trend Analysis by Grid (January 10, 2011 - January 10, 2013)|
2

-------
Analysis of Crime Data After the Brownfield Milestone Date
Quantitative Analysis
EPA also calculated the crime density per year. The crime
density per year was calculated for %-mile intervals within 1-
mile around a site to determine if the crime rate varied with
distance from a brownfield site. The following charts are
examples of what was prepared for each site to display the
crime rates around each brownfield site over time; each chart
also shows the data sets before and after redevelopment.
Chart A
Chart A shows crime density within a 1-mile radius from a
brownfield site for all years included in analysis.
Chart 6.1A Crime Density within 1 Mile - Baton Rouge, Louisiana
Before Redevelopment After Redevelopment
¦within 1/4 mite
¦	between l/
-------
Chart C
Chart C compares crime density within a %-mile radius of a site
to crime density within a %-mile and 1-mile radius of the site
for all years included in the analysis.
Chart 6.1C Average Crime Density Inside and Outside 1/4 Mile - Baton Rouge, Louisiana
Before Redevelopment
3
"Sf 2500
¦	within 1/4 mile
¦	between 1/4 and 1 mile
After Redevelopment
ll
1/10/11-1/10/12	1/10/12-1/10/13	1/10/13-1/10/14	1/10/14-1/10/15
Chart D
Chart D shows the percent change in crime density after the
milestone activity end date for a brownfield project site for
each %-mile interval.
Chart 6.1D Percent Change in Crime Density Before and After Redevelopment -
Baton Rouge, Louisiana
Limitations/Challenges
Some notable limitations and challenges of this analysis
include:
•	Brownfield sites in non-city areas were not included; the
crime activity in rural areas is typically measured on the
census block level and this analysis was restricted to point
data analysis.
•	Any apparent reductions in crime after brownfield
revitalization activities efforts cannot be definitively
attributed to the brownfield activity; this analysis does not
control for changes in other factors (economic activity,
vacancy rates, police presences, demographics) that may
influence crime rates.
•	The analysis relies on reported crime data. Not all crimes
are reported and reported crimes may be influenced by the
level of policing in an area. Changing patterns of policing
can influence the crime rate in an area. For example, if part
of the brownfield revitalization effort includes an increased
police presence after redevelopment, reported crimes
could increase simply because the same criminal activities
are being addressed and reported more often.
The precision of the spatial data related to crime data varies
based on each municipality's privacy standard. Some
municipalities track crime locations at the exact address,
while others use the nearest block.
Some types of violent crimes are excluded from public
datasets to protect the identity of the victims.
Natural or jurisdictional (city limits) boundaries around or
through a brownfield project site may impact the analysis
and influence quantitative results. For example, the
migration of crime around a natural feature, such as a river,
presents a physical barrier, while a city limit may have no
impact. However, the extent of crime data available around
a brownfield project site may be limited to less than a 1-
mile if a city limit is within that distance.
Recommendations for Further Investigation
Crime changes per capita could be analyzed instead of
crime changes per square mile.
Population changes could be included in the analysis using
census block data.
A similar analysis of 30 smaller, more rural cities could be
conducted if point spatial data becomes available for rural
areas in the future.
Partnerships could be developed with local police
departments and a brownfields redevelopment coordinator
from each city to evaluate redevelopment goals and best
management practices associated with crime mapping to
support those efforts.
References and Resources for Additional
Information
Applied Geographic Solutions (AGS).2015. "CrimeRisk".
https://www.appliedgeographic.com/crimerisk/. Accessed on
January 12, 2018.
Harwood. 2012. "Characterizing Populations NearOSWER
Sites". Final Draft with Errata Sheet. April 30.
Uniform Crime Report (UCR). 2018. Personal Communication
between Joel Peters and Texas UCR staff. January 15.
U.S. Census Bureau. 2016. "Urban and Rural".
https://www.census.gov/geo/reference/urban-rural.html.
Accessed on February 9, 2018.
U.S. Department of Justice (DOJ). 2005 "Mapping Crime:
Understanding Hot Spots". August.
https://www.ncjrs.gov/pdffilesl/nij/209393.pdf. Accessed on
January 12, 2018.
U.S. DOJ, Federal Bureau of Investigation. 2017. Crime in the
United States, 2016. September.
https://ucr.fbi.gOv/crime-in-the-u.s/2016/crime-in-the-u.s.-
2016/resource-pages/about-cius. Accessed on
January 12, 2018.
4
-------