3. INTRODUCTION
Study Area & Data Sources
This study specifically examines illicit drug activity in the city of Boston, Massachusetts
using both RTMDx Utility, and ArcMap 10.4 software. Boston is located in the Northeastern
United States and has an area of approximately 48 square miles. Boston was chosen for this study
because of its relatively high crime rate and high levels of poverty. Boston’s poverty rate is 21.9%,
nearly 10% above the national average of 13.5% (U.S. Census, 2010). According to U.S. census
projections, the city contains a population of about 667,000 people, 24% of which are African
American.
For the current study, we will be examining administrative data found on the City of
Boston Data Portal and the Massachusetts Office of Geographic Information database. The data
utilized in reference to drug activity in this analysis will be drug-related arrests in accordance to
police incident report data from November 2015. Environmental risk factors associated with drug
activity determined during the literature review process will be examined primarily using city
permit, code enforcement, and Massachusetts Bay Transportation Authority data. This data will
be used to produce a Risk Terrain Model (RTM) to forecast locations of drug-related criminal
activity in Boston.
4. METHODOLOGY
RTM Procedures
Risk Terrain Modeling is a form of hot spot mapping used to study spatial crime
vulnerability (Caplan and Kennedy, 2016). Instead of mapping the crime incidents themselves,
RTM assesses vulnerability utilizing social and physical risk factors associated with criminal
activity. Unlike other methods commonly to predict criminal activity, such as the Kernel Density
Estimation (KDE), RTM identifies “hot areas” based on environmental criminology (Drawve,
Moak, and Berthelot 2014; Drawve, 2014). According to a study comparing RTM to other hot spot
methods (including KDE), RTM was found to be far more reliable and consistent at predicting
crime (Drawve, 2014). Based on the research surrounding RTM in reference to its reliability, and
the fact that it focuses on the environment as opposed to the individual, this method was chosen
for this analysis.
A RTM is created using the RTMDx Utility software, which conducts a series of analyses
to create a model that best explains the relationship between the risk factors and the outcome
event. To use the software, various parameters must be set including the study area, block
length, cell size, outcome event, and up to 30 risk factors (Caplan et al., 2013). For the purposes
of this study the parameters were set to the following: the study area is the city of Boston; block
length is 348 feet; cell size is 174 feet; model type is “aggravating”; the outcome event is reported
5. drug crime from November 2015; and risk factors were identified based on a review of both the
literature and the data.
The RTMDx Utility software provides a variety of settings to determine if a spatial
relationship exists between the identified risk factors and the selected outcome event. When
entering a risk factor into the software, the user is provided an option to select a method of
operationalization, maximum spatial influence, and increments used in the analysis (Caplan et
al., 2013). Predictors of drug activity were identified through both a review of literature
surrounding environmental criminology and the available data. Through this review, 8 risk factors
were identified: alcohol outlets, abandoned properties, fast food establishments, public schools,
convenience stores, night clubs, property maintenance code violations, and bus stops. Each of
these risk factors was found on a variety of different sources within either the city of Boston Data
Portal, or the Massachusetts Office of Geographic Information database.
The block length parameter was determined based on the average block length of Boston.
Using both a city of Boston base map and a Massachusetts road map shapefile in ArcMap 10.4,
we calculated the average block length in the city. The cell size parameter was determined using
the block length input. Caplan et al. (2013) recommends the cell size parameter be set to half of
the block length to ensure the most realistic representation of the physical environment in the
model. Finally, the “aggravating” model type was selected to determine if a positive relationship
exists between the risk factors and the outcome event (Caplan et al., 2013). In our case, we used
an aggravating model in an effort to determine if a relationship exists between the risk factors
identified and drug crime in Boston.
6. STUDY FINDINGS
Table 1: Risk Terrain Model analyses results
Table 1 contains the risk factors that were found to be significant in relation to drug crimes
in Boston. Our analysis found 6 of the 10 risk factors to be significant. Unsafe structures
(condemned housing), 311 service requests for street light outages, and 311 service requests for
needle pick-ups were found to be the most significant of the inputted risk factors. According to
the analysis, places within 174 feet of unsafe structures (condemned housing) in Boston are
approximately 6 times more at risk of drug activity than other areas in the city.
Though similar to unsafe structures (condemned housing), abandoned properties only
have a Relative Risk Value (RRV) of 2.33. These findings suggest that in the City of Boston, places
within 1044 feet (3 blocks) of abandoned properties are approximately 3 times less at risk of drug
activity when compared to places within 174 feet (1/2 block) of unsafe structures. After
researching abandoned properties in Boston, we found that the city secures abandoned
properties to limit access to these places. This could account for the large difference in both the
spatial influence and RRV score between abandoned properties and unsafe structures.
NAME OPERATIONALIZATION SPATIAL INFLUENCE Relative Risk Value (RRV)
Unsafe Structures (Code Violations) Proximity 174 6.26
Street Light Outages (311 Calls) Proximity 174 3.05
Needle Pick-up Requests (311 Calls) Density 1044 3.03
Take Out Restraunts Proximity 870 2.76
Bus Stops Proximity 348 2.41
Abandoned Properties Proximity 1044 2.33
7. Risk Terrain Modeling (RTM) predicts places at risk of criminal activity based on
environmental factors utilizing a series of regression analyses with each of the inputted risk
factors and the outcome event (drug activity). For the purpose of this analysis, the environmental
risk factors from Table 1 were utilized to predict the places in Boston, Massachusetts at greatest
risk of drug activity. Figure 1 is a risk terrain map illustrating these places. According to the model,
the areas colored red are approximately 32 times more at risk of drug activity in the city than the
areas colored dark blue.
Figure 2 shows significant clusters of drug crimes in Boston in November 2015 using
Kernel Density Estimation (KDE). When compared to the RTM map (Figure 1) we can see that the
much of the drug crime clusters are located in the “risky places” identified by in the RTM analysis.
Based on the commonality of these findings, Figure 1 allows us to identify the places where drug
markets are likely to displace after an intervention takes effect. Because RTM analyses rely on
environmental risk factors, RTM is much more reliable at forecasting where crime is likely to
Figure 1: Risk Terrain Map of Places at Risk of Drug Activity Figure 2: KDE Hotspot Map of Drug Crime Concentrations
8. occur when compared to KDE; however, KDE is found to be more accurate (Drawve, 2014).
Because we utilized both KDE and RTM, it can be concluded that the model is both accurate and
reliable; however, this cannot be proven without assessing the validity of the model through
proper interventions to determine if drug crimes displace in the places identified by the RTM.
RECOMMENDATIONS
According to the model unsafe structures (condemned housing), 311 service requests for
street light outages, 311 service requests for needle pick-ups, take out restaurants, bus stops,
and abandoned properties were all found to be significant predictors of drug activity in Boston,
MA. We acknowledge that addressing each of these six risk factors may be difficult due to fiscal
and resource constraints. Because of this, we have decided to address the risk factors with a
Relative Risk Value (RRV) above 3.0 (unsafe structures, 311 service requests for street light
outages, and 311 service requests for needle pick-ups). Based on our findings, the following
recommendations are proposed:
1. Prioritize addressing condemned housing notices and street light outages in risky
places
• Expedite the demolition or “boarding-up” of unsafe structures (condemned
housing)
• Utilize LED replacement bulbs to ensure longevity and reliability of street lighting
• Create a phone app to easily report streetlight outages
9. 2. Address places with high amounts of 311 service requests for needle pick-ups
• Promote social services for those addicted to drugs in these places
(advertisements, clinics, needle exchange programs, etc.)
o Expand Addicts Health Opportunity Prevention Education (AHOPE)
Program
o Create needle “drop bins” to reduce the reusing of needles
• Work with community leaders to promote services for addicts while encouraging
good police-community relations
• Prioritize foot and bike patrols
3. Routinely assess the distribution of drug activity in reference to the RTM
• Look for displacement of drug activity and tailor the distribution of resources
accordingly (ex. streetlight outage and condemned housing prioritization)
• Hold quarterly meetings to assess and redistribute resources and services
accordingly
4. Maintain data currently being collected relating to the identified risk factors to easily
reassess risky places within the city