Neighborhood Crime and Travel Behavior
Project Number: 2405
Principal Investigator:
Christopher E. Ferrell, Mineta RA, (Ph.D. UC Berkeley, June 2005),
Associate and Senior Transportation Planner – Dowling Associates, Oakland,
CA.
Institution:
Mineta Transportation Institute
Telephone Number:
(408) 924-7560
mti@mti.sjsu.edu
Project Objective:
This study hypothesizes that urban density and neighborhood crime have
been confused in the minds of the public as well as the conceptual and
statistical models of transportation researchers. Each having an opposite effect on mode choice, it is assumed
that to the extent that crime rates are higher in older, denser urban areas,
crime rates have masked and countered the effects of density on mode choice,
reducing our estimations of its importance.
This research proposes to study the effects of neighborhood crime on
mode choice.
Abstract:
While crime is assumed by transit agency staff and researchers to have a negative effect on transit usage, it has not been studied thoroughly using statistical modeling techniques and is rarely if ever used in mode choice models for travel demand modeling. In part, this may have been because crime data was not available except for larger jurisdictional areas. The gradual introduction of increasingly sophisticated computer database and analytical tools to local police departments has meant that crime data is being collected and made available at the neighborhood level. The City of Oakland has made crime data available online with an interactive, web-based mapping tool. Crime data from Oakland is available with geo-locational attributes, allowing individual crime events to be mapped in urban space. While Oakland’s data is unique in its detail and availability, Bay Area cities such as San Jose, San Francisco, Berkeley and Fremont all make crime data available online, often aggregated to the neighborhood or police “beat” level. Palo Alto is also on the forefront of this work in the Bay Area. Their GIS analysts are working to make crime data available to their officials in “real time.” With these new data sources, an examination of the relationship between crime and travel behavior will be more productive.
Proposed Work
scope:
This study proposes to collect crime data from individual police departments around the San Francisco Bay Area, and using Geographic Information System (GIS) tools, overlay these data with census data that describes neighborhood demographic and physical characteristics (such as urban density) and with travel diary survey data commonly used to develop mode choice models. It is expected that the appropriate crime data will be available from only a portion of these jurisdictions. Using these data, the effects of urban density and crime on mode choice can be separated and statistically controlled. Analysis can potentially be conducted at several geographical levels, with regional data being used to measure the effects of crime on a wide and varied sample of Bay Area communities, and more focused, fine-grained analysis can be conducted where the data available is of sufficient detail and quality. In such cases, individual cities such as Oakland and Palo Alto could be studied with crimes counted and indexed surrounding specific transit stations. Assuming findings that support the hypothesized relationship and confirms the importance of crime as a predictive variable for travel behavior, this roject will also document the state of current crime data availability in terms of data type, format, and availability for use by transportation researchers and practitioners. To the degree this data is useful and important as a model variable, the potential for it to be made regularly available to transportation modelers can also be explored.
This study will utilize data from three primary sources:
1) Crime report statistical data to be collected from Bay Area police departments. This data will ideally be disaggregated, but if it is unavailable in this form, at a minimum, the team will seek to obtain data aggregated by local neighborhood, census tract, police department “beat” city council district, or some other neighborhood-level aggregation.
Initial contacts with several of the larger city police departments in the Bay Area indicate that data is available from all of these cities, and several of them make their data available online. In terms of data quality, the most appealing and potentially rewarding data will be available in disaggregated form with precise location information. While not all of these cities record precise crime locations, data will often be available at the census tract, city district, or police patrol beat-levels.
2) U.S. Census of Population and Housing Data (2000). These data will be used to describe the general physical and demographic makeup of Bay Area neighborhoods to provide adequate statistical control variables for the travel models that will be developed. Census journey-to-work data may also be used as a dependent variable for developing models of mode choice at the neighborhood level.
3) Bay Area Transportation Survey (BATS) 2000 Data: Collected by the Metropolitan Transportation Commission, these data are used to calibrate the region’s travel demand model and contains detailed travel data for a sample of Bay Area residents. BATS 2000 data will provide information on individual travel behavior for statistical analysis.
By combining these datasets using computer database and Geographic Information System applications, the relationships between urban form, travel behavior and crime can be measured.
Description and Project dates:
Task One:
Literature review: Collect, review and synthesize past research on the subjects of mode choice modeling, the use of measures of urban form in these models, the effects of crime and perception of crime on travel behavior, methods of measuring and indexing crime statistics for urban analysis, and potential techniques for modeling the influence of neighborhood crime on travel behavior.
July 2005 – August 2005
Task Two:
Collect Crime, Travel and Census Data: Supervised student researchers will contact local
police/sheriff departments in the nine-county Bay Area to request
disaggregate or neighborhood-level aggregated crime data as well as any GIS
or non-GIS maps showing neighborhood or jurisdictional boundaries that are
used to aggregate the requested crime data.
Throughout this process, the availability of data from and contact
information for each police/sheriff department will be documented to provide
an overall view of the process by which the data was obtained and the
relative ease (or difficulty) with which researchers and practitioners can
access and utilize data in the future.
Census data will be obtained from the Association of Bay Area
Governments as well as the Census Bureau’s website. The research team already possesses the BATS 2000 data.
August 2005 – September
2005(Caution: approval for human
subject research may delay start of this phase.)
Task Three:
Process Crime, Travel and Census Data: Using
database and GIS computer applications, the spatial distribution of reported
crimes will be analyzed and indexed for inclusion in the project’s final
modeling exercises. It will then be
combined with demographic and travel data from the U.S. Census and BATS 2000
datasets.
October 2005 – February 2006
Task Four:
Develop
Travel Behavior Models: Using established
techniques for modeling travel behavior, a set of models will be constructed
using Census, BATS 2000, and synthesized crime index data to determine the
statistical power and significance of crime and urban density on travel
behavior. A measure of urban
accessibility for each travel analysis zone will be constructed for each
model application, providing a relative measure of transportation and land
use opportunities for survey respondents.
March 2006 – April 2006
Task Five:
Draft and submit
final report to MTI:
A final report for submittal to MTI will be written summarizing the results of the modeling activities, the availability of crime data for future use by urban and transportation researchers, as well as the conclusions of the researchers.
May 2006 – July 2006
Following submission of the draft final report, the following actions
will occur:
Copyedit and preparation of Peer Review Draft
Peer Review and Author’s Response
Final Editing and Pre-Publication
Printer’s Blue line Proof and Final Print
The estimated time for these to occur will be no less than two months. Final
publication and Web posting: October 2006
Total
Budget:
Yearly: $50,691
Principal
Investigator:
Christopher E. Ferrell, Mineta RA, (Ph.D. UC Berkeley, June 2005),
Associate and Senior Transportation Planner – Dowling Associates, Oakland,
CA.
Team Members:
Earl G. Bossard, Ph.D., Mineta
RA, Professor of Urban and Regional Planning, SJSU
Steven B. Colman, AICP, Mineta RA, Principal, Dowling Associates,
Oakland CA.
Students:
TBD
Technology
Transfer Activities:
Upon publication, pdf and html versions will be available on the Mineta
Transportation Institute web site. The project experience and data will be
available for community meetings. Authors are encouraged to submit articles
based on the research to relevant journals and to present the information to
end-users at conferences,
Potential Benefits of the Project:
This study will be of interest to three categories of urban and transportation planning professionals: urban transportation demand modelers, urban transportation planning researchers, and transit agency professionals. Travel demand modelers may be provided with a new source of data that could significantly improve their modeling techniques. By providing a clearer empirical picture of the effects of urban form and crime on travel decision-making, policy-level efforts to jointly plan transportation and land use to increase the use of transit and non-motorized modes may also be improved. If researchers and policy-makers can be shown a clearer picture of the effects of urban density on travel behavior while controlling for the effects of crime, a clearer and more substantial case can be made by backers of smart growth and transit-oriented development. Transit agency planners that are seeking to enhance transit ridership may also benefit from this research. While it may be assumed that increasing transit services to a neighborhood is the most effective way of increasing transit ridership, these efforts may be thwarted by high levels of neighborhood crime. In fact, in some cases a more intensive and community-based policing program for a neighborhood may be the most cost-effective means to increase ridership and neighborhood residents’ mobility using existing transit routes..
Key Words: Security, Transit Agency, Urban Transportation,
Transportation planning