4110: GEPL 5110 Geographic Information Systems
ANALYZING THE BENEFITS OF CITY PARKS Can Natural Surrounding Influence Violent Crime? Sarah Lyon-Hill Graduate Many factors exist when analyzing the presence of crime in certain geographic areas: population density, education, poverty, race, and more. This study looks specifically at the possible correlation between green space, i.e. parks, and violent crime. Reasons for this correlation include decreased mental fatigue and increased surveillance. According to Kuo and Sullivan (2001), "a considerable body of studies indicates that vegetation aids in the recovery from mental fatigue" (347) Symptoms of mental fatigue, such as irritability and decreased control over impulses, are proven psychological precursors to violence. In addition, the presence of parks provides outdoor areas for community interaction, and thus increased surveillance by community members (Kuo and Sullivan 2001). Increased surveillance or environmental cues that indicate surveillance have been shown to deter criminal violence (Newman 1972). Because most of this research deals with violence, I will focus my study on violent crime (as opposed to crime in general), which the FBI defines as murder, rape, robbery, and aggravated assault (Dept. of Justice). While considering other social factors that may contribute to violent crime, I will look at the city parks and metroparks in Toledo, Ohio, their locations within census block groups, and the violent crime that occurred during 2005 in those block groups. I hypothesize that, when holding several social factors constant, less violent crime will have occurred in census block groups with parks. Methodology and Data Sources Because of easy access to crime data and city park locations, I chose Toledo as my case study. The Toledo Police Department reported 7185 crimes in a 2005 excel file. I limited those crimes to those defined by the FBI as violent crime and geocoded 1141 crimes. For easier analysis, I spatially joined that data to a census block group shapefile. I then took a Toledo Parks shapefile provided by Dr. Iseki, compared it with pdf data sheets on the City of Toledo website, and added parks to the file using ArcGIS Editor. Overlapping both an aggregated spatially joined crime-block group shapefile and the edited parks file provides a visual analysis of whether block groups with parks contain less incidence of crime. In addition, I added a 0.1 and 0.25 mile multiple ring buffer to analyze proximity to parks. The social factors considered were poverty, race, education, and population. I downloaded these tables from American FactFinder, manipulating some data after download. For instance, I was only able to download poverty by age; therefore, I added all columns together and divided by total population, finding percent poverty within each block group. In order to find percent population who had attained an education of high school or below, I downloaded education attained by gender, added gender groups and all grade levels through high school, and divided by the total population. Using aggregated colors and shapes, I varied these social factors and examined amount of crime within each block group with parks. Afterward, I held all social factors constant, exported then into a new shapefiles, and compared them with the number of crime incidence in each block group. Finally, I looked at accessibility to well-frequented city parks and metroparks via TARTA bus routes. I downloaded census information on vehicle availability. I exported census blocks with higher rates of households without cars - areas that would be most in need of TARTA. I then selected routes that intersected both selected parks and block groups. For information pertaining to specific techniques, see Appendix 1. For more information about data and sources, see Appendix 2.
1
GEPL 4110: 5110 Geographic Information Systems
Layouts and Findings
Layout 1. This layout is an introduction to the overall area inset map of Lucas County, Toledo border, streets, parks, Maumee River. I also present pictures, which serve as examples of good parks in the region those that are well-maintained and preserve visibility by using tall trees, shrubs, and open space.
Layout 2. The top map shows all incidences of violent crime that have been geocoded. To the right of the map is a pie chart, which displays the number of violent crime by category. The bottom map is the spatially joined crime-block group shapefile that I aggregated. By examining park locations and the buffers around the parks, there does not seem to be any pattern between crime and parks. 2
GEPL 4110: 5110 Geographic Information Systems
Layout 3. Taking a closer look at high crime block groups (I exported the data into new shapefiles), it seems that much of the area is not easily accessible to parks by foot. Many of these specific parks are also not very well maintained. The lower left block group, for instance, is a part of Swan Creek that has been left natural with tall trees and very little visibility.
Layout 4. This layout presents percent poverty of each block group. Using Select by Location, I selected all block groups that intersect parks. From that selection, I used Select by Attributes to select three different ranges of poverty and then exported those data sets into new shapefiles. Looking at those data sets in terms of violent crime, one sees that most of the high crime is located in block groups with mid 3
GEPL 4110: 5110 Geographic Information Systems
poverty.
Layout 5. Here I used aggregated colors for race and symbols for crime. There does not appear to be any relationship between the two. The race data, however, does show signs of segregation among races.
Layout 6. Education and crime are both aggregated in this layout. Like poverty, I first isolated all block groups with parks using select by location. A correlation does seem to exist as can be seen by higher percentages of less educated individuals in block groups with higher crime.
4
GEPL 4110: 5110 Geographic Information Systems
Layout 7. These two layouts show population and population density. When I discovered that no real pattern existed between population and crime within the Toledo blocks groups, I calculated population density by adding a field in the block group attributes table and dividing area by total population. Because no set units were ascribed to the AREA field in the attributes table, population density is merely a ratio. I, therefore, labeled the aggregated fields as low, medium-low, medium-high, and high. Alas, there still seemed to be no correlation between population density and crime. While population is a proven factor of crime in urban areas, perhaps that correlation is less effective on a more microscopic level such as block groups. 5
GEPL 4110: 5110 Geographic Information Systems
Layout 8. Poverty and Education (the social factors that seemed to have some correlation) were selected using select by attribute and the resulting selections were exported into a new shapefile. There does not seem to be any patterns present.
Layout 9. The upper map shows aggregated numbers of households without cars and all TARTA bus routes. Using select by attribute, I selected the highest aggregated range of no car households and exported the data to the lower map. I then manually selected certain parks that I know are wellmaintained, have several recreational amenities, and are often visited. With those two shapefiles, I selected by location, first all TARTA routes that intersect no car block groups and from then, that selection, those that intersect selected parks. Considering the results shown in the lower map, it is 6
GEPL 4110: 5110 Geographic Information Systems
possible to manipulate the routes so some routes pass through no car block groups and pass by entrances to parks. For example, the south side entrance to Swan Creek is not accessible to these populations. None of these groups can reach Walbridge Park, a park near the Toledo Zoo, which in turn is inaccessible to these groups. Conclusion In conclusion, I found little evidence to support my hypothesis that presence of parks helped to decrease crime. One reason for this result may be how I analyzed the topic. First, I could have better highlighted differences between block groups without parks versus those with parks. Second, I could have incorporated factors concerning the parks themselves, specifically area and condition of parks. If those two data sets had been more easily attainable, they would have been very useful to my analysis. Another reason for my results may have been other mitigating circumstances. Currently Toledo is in great economic hardship because of industry loss, rising unemployment, rising prices, inflation and more. These factors may have nullified or weakened the other factors used in my case study. In short, I should have incorporated more attribute data. Despite these findings, however, I found GIS to be very useful in analyzing the issue. Because I dealt mainly with proximity and accessibility to parks, a tool for spatial analysis was ideal for this problem. In fact, I had very little problems using GIS to analyze the issue. My main problems arose when preparing the attribute data, specifically finding and modifying the census data that I required for the project. Once that was finished, preparing the data in ArcGIS was relatively simple and methodic. The one success in my research was the analysis of TARTA bus routes. By using GIS to isolate block groups with high numbers of no car households and to find TARTA bus routes that intersect those block groups and various highly frequented parks, one could analyze whether those routes should be shifted in some way to accommodate a larger population of potential customers. In this case, GIS provides a nice visual analysis of the situation. Bibliography American FactFinder. http://www.census.gov. City of Toledo. Division of Parks and Forestry: Park Maintenance. http://www.ci.toledo.oh.us/. Department of Justice. Crime in the United States 2005. http://www.fbi.gov/ucr/05cius/offenses/violent_crime/. Kuo, Frances E. and William C Sullivan. 2001. Environment and Crime in the Inner City: Does Vegetation Reduce Crime? Environment and Behavior 33(3): 343-367. Newman, O. 1980. Defensible space: Crime prevention through urban planning. New York: Macmillan. O'Brien, Elizabeth. 2006. Social Housing and Green Space: A Case Study in Inner London. Forestry 79(5): 535-551. Toledo Police Department. Toledo Crime Data 2005. 7
GEPL 4110: 5110 Geographic Information Systems
Appendix 1: Summary of Techniques Used Inset map (See Layout 1) - Add Lucas County census block group, Ohio State, and Toledo boundary shapefiles - Use clip feature: clip Ohio State shapefile using census block group to make Lucas County boundary - Change colors of Lucas county boundary and Toledo boundary shapefiles. Symbols and colors: Point (See Layout 5) - See High Crime and Mid Crime shapefiles - Open layer properties > Symbology > Quantities > Graduated Symbols - Choose 1 Class - Click "Symbol" title of column - Select "Properties for Symbols" - Select shape, color and size - One right of Layer Properties, click Background - Fill Color and Outline Color: select no color Aggregating attribute fields (See Layout 2, 4, 5, 6, 9) - All census and crime data - Layer properties > Symbology > Quantites/Categories - Select color schemes - To manipulate specific colors, double-click on symbols Creating indices (See Layout 3-9) Appears in high crime, mid crime, poverty ranges, census blocks with parks, sample block groups, no car census blocks, target TARTA routes, and target parks shapefiles After selecting by attribute and/or location, right-click on shapefile title in table of contents Go to Export Data Attribute sub-sets selections: Select by attributes (See Layout 3, 4, 5, 8, 9) Appears in poverty ranges (high, mid, low), crime (high, mid), sample block groups (various selected attributes), no car block groups Go to Select by attribute Layer: block group shapefi...