Jump to ContentJump to Main Navigation
The Criminology of PlaceStreet Segments and Our Understanding of the Crime Problem$

David Weisburd, Elizabeth R. Groff, and Sue-Ming Yang

Print publication date: 2012

Print ISBN-13: 9780195369083

Published to Oxford Scholarship Online: January 2013

DOI: 10.1093/acprof:oso/9780195369083.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2017. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see http://www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 25 February 2017

(p.201) Appendix 4 Data Collection

(p.201) Appendix 4 Data Collection

Source:
The Criminology of Place
Publisher:
Oxford University Press

(p.201) Appendix 4

Data Collection

We began the data collection effort with a list of constructs important under opportunity theories and social disorganization theories that we wanted to represent for each street segment in Seattle. Related to opportunity theories, we wanted data to describe the number of motivated offenders, suitable targets, and informal guardians on a street segment as well as place characteristics such as accessibility. For social disorganization theories, we wanted to represent economic deprivation, racial heterogeneity, social ties, collective efficacy, informal social control, and physical deterioration. Using the experience of one of the coauthors (Groff) in identifying and collecting data about micro-level places, we identified likely data sources to represent the constructs just listed.

As we note in the main text, there were some types of information that we were unable to collect. For example, related to opportunity theory, we could not obtain data on automobile parking for all streets in Seattle. This would have been a good indicator of potential targets for auto theft. As discussed in chapter 5, census geographic definitions are not consistent with the street segment unit of our study, and thus census data were not available to us. Limitations of historical land use data made it impossible to accurately distinguish multifamily land use, renter-occupied housing, and real estate sales. For example, without accurate demographic information available at the street level, we had to rely on data collected from public school students to create a proxy to represent population heterogeneity. Such information would have allowed us to better represent informal social control stemming from residential stability and lower housing density. Lack of information collected from residents over time also affected our ability to quantify or to track changes in the strength of social ties and mutual trust among residents. Finally, we were not able to get information on the health of the residents at the individual street segment level. While these omissions were less than ideal, we were prepared to encounter (p.202) difficulties in collecting data at the micro level and were still able to obtain a wide variety of information.

Characteristics of Street Segments: Opportunity Perspectives

Based on the opportunity theory perspective, we collected data on 16 characteristics for each street segment in Seattle (table A4.1). These characteristics were then aggregated to create the final 10 characteristics we focused on for the analysis (table A4.2). The 16 source characteristics are discussed next.

Overall, the geocoding rate for data sets was very good. It varied from a low of 87 percent for the business data to a high of 100 percent for some of the public facilities.1 Vacant land is one non-geocoded data set that contained the highest level of missing data.2 Retrospective data collection was the single most challenging aspect of the research effort.

a) Public School Students

Data about public school students was obtained from Seattle Public Schools. It included all public school students from grades 3–12. The total number of public school students ranged from a low of 35,857 in 1992 to a high of 37,433 in 2004. The number of juveniles was relatively stable over the study period, with 37,029 registered in 2004. Three analysis variables were developed for use in the study: (1) total number of public school students who resided on each street segment, (2) the total number of students who were considered truant (10 or more absences in a school year) and (3) the total number of students who were considered to be low academic achievers.3

The number of students classified as low academic achieving (LAA) was typically three times that of truant students in any given year (mean of 14,366 versus 4,634). The year with the lowest number of LAA students was 2004 (25.11 percent) and the year with the highest number was 1997 when there were 13,563 (38.2 percent) LAA students identified. That was also the year with the highest proportion of students classified as LAA. The lowest number of truant students was in 2004 when there were 3,581 (9.85 percent). The highest number occurred in 1994 when there were 6,489 (18.4 percent) truant students. That was also the year with the highest proportion of students classified as truant.

For our final analysis, we created a variable defined as “high-risk juveniles,” which represented the total number of students who were considered either truant (i.e., they had 10 or more unexcused absences) or were categorized as low academic achievers. The number of high-risk juveniles fluctuated from 14,524 in 1992 to a high of 15,908 in 1997 before falling to its lowest level in 2004 (n = 11,230). (p.203)

Table A4.1 Roots of Characteristics Used in the Model

Variable

Geocoding Hit Rate

Data Source

Contributes to:

Total public school students with 10 or more unexcused absences and/or flagged as low academic achievers

97.1%

Seattle Public Schools

High-Risk Juveniles

Total number of employees at businesses located on the block

87.8%1

InfoUSA database of all businesses in Seattle

Employment

Total number of public school students

97.1%

Seattle Public Schools

Residents

Total number of registered voters

99.7%

Labels & Lists Inc.

Residents

Total retail business sales on the block

86.9%

InfoUSA database of all businesses in Seattle

Business Crime Attractors/Crime Generators—Total Sales

Community centers

100%

Fleets and Facilities Department, City of Seattle

Public Crime Attractors/Crime Generators

Hospitals

100%

Yellow pages

Public Crime Attractors/Crime Generators

Libraries

100%

Seattle Public Libraries

Public Crime Attractors/Crime Generators

Parks

97.9%

Fleets and Facilities Department, City of Seattle

Public Crime Attractors/Crime Generators

Middle and high schools

100%

Seattle Public Schools

Public Crime Attractors/Crime Generators

Street type

N/A

Seattle GIS

Type of street (arterial vs. residential), Static across all years

Total number of bus stops

Came as shapefile

Department of Transportation (Metro Transit Division)

Bus Stops

Percentage of vacant land parcels

N/A2

Developed from Historic Assessor's Data (Seattle Planning Department) and parcel boundaries (King County GIS)

Vacant Land

Total number of police stations within 1,320 feet of a street block

100%

Fleets and Facilities Department of the city of Seattle—location source, variable calculated by researchers

Fire and Police Stations

Total number of fire stations within 1,320 feet of a street block

100%

Fleets and Facilities Department, City of Seattle—location source, variable calculated by researchers

Fire and Police Stations

Total amount of watts per street segment

Came as shape file

Seattle Public Utilities

Street Lighting

(1) Since this data set was pulled by zip code there are quite a few records that are outside the city of Seattle but still have Seattle addresses. In addition, some address fields are blank and others have P.O. Boxes and not street addresses.

(2) Getting the historical information joined to the shape file of parcels required several steps and resulted in an average of 81 percent of the parcels having land use information.

Note: The geocoding rate listed represents an average across all years.

(p.204)

b) Businesses

After trying unsuccessfully to obtain business license data, we finally purchased data from InfoUSA. While they had data available from 1998 to 2004, the data were expensive so we purchased every other year (i.e., 1998, 2000, 2002, and 2004). The vendor could only pull data by zip code, which meant we obtained quite a few records from outside of Seattle that still had Seattle addresses.4 We believe the relatively low geocoding rate of 87.8 percent reflects the influence of businesses that were located outside Seattle and thus unmatchable. Unfortunately, it was impossible to distinguish between those records that were unmatched because they were outside of Seattle and those that were genuinely unmatched but inside Seattle. Thus, the inability to identify records outside of Seattle artificially lowered our match rate; the true rate was in all likelihood much higher. The total number of businesses per year ranged from a low of 32,517 in 1998 to 37,916 in 2000 before dropping again to 34,547 in 2004. The average number of businesses per year was 35,573. We aggregated the geocoded records to street segments and calculated the total number of employees.

Because of the important relationship between retail establishments and crime levels, we isolated all businesses that were primarily retail focused and used total retail sales as the measurement of the intensity of retail on a street.5 Roughly 10 percent of all businesses were retail businesses. The number of (p.205) retail businesses declined over the time period. The highest number was 3,333 in 2000 followed by 3,331 in 2002. The earliest and latest years were both lower with 3,251 in 1998 and 3,160 in 2004. We then aggregated the total amount of sales for each retail business to the street segment on which it was located.

c) Registered Voters

Our figures related to voting behavior also were purchased from a vendor, Labels & Lists Inc. (table A4.1). These data were originally collected by the Elections Department for King County, but they do not keep historical records. Thus, we turned to Labels & Lists Inc., which kept historical records of Elections Department data back to 1999. The total number of registered voters increased from 383,216 in 1999 to the highest number of 418,665 in 2002, and then declined to 345,664 in 2004. From these records we developed a variable that contains the total number of registered voters for each year. This information was used later to help estimate adult residential population.

d) Public Facilities

The data on the locations of facilities were primarily obtained from the city departments that run them (table A4.1). Seattle Fleets and Facilities Department was helpful with the police facilities, fire facilities, parks, and community centers. Obtaining the current locations was very straightforward. As mentioned earlier, the challenge came when we tried to establish which facilities had opened, closed (even temporarily), or changed location during our study period.

As it turned out, the number of each type of public facilities was relatively small and very stable. There were 26 community centers from 1989 to 2003. Two were added in 2004 for a total of 28 community centers at the end of the study period. There were 13 hospitals in Seattle over the entire study period. There were 17 libraries in 1989 and 21 in 2004. Middle and high schools had some minor fluctuations over the years (e.g., closing for remodeling) but in general they showed an increase from 28 facilities in 1989 to 30 in 2004. Except for hospitals, public facilities were distributed evenly across Seattle.

To represent the crime-generating effect of public and quasi-public facilities on nearby street segments, we calculated a spatial variable using a geographic information system (GIS). This variable represents the number of public and quasi-public facilities (i.e., community centers, hospitals, libraries, parks, and middle and high schools) within a 1,320-foot distance (i.e., a quarter mile) of each street segment. Distance was measured along the street network using the ArcGIS™ Network Analyst extension. Using street network distance was especially important in a city like Seattle, which is trisected by waterways. Since (p.206) these waterways must be crossed using bridges, they represent significant physical barriers to travel.

Police and fire stations were considered separately because they represent the potential guardianship effect of police and fire personnel on the street segment and on nearby street segments (table A4.1). From 1989 to 2001, there were four police stations. Another station was added in 2002 to increase the number to five stations. There were 33 fire stations throughout the time period. To capture the effect of police and fire stations on nearby street segments, we used the same methodology as for public facilities: we calculated a spatial variable using a GIS. This variable captured the number of police and fire stations within a 1,320-foot distance (i.e., a quarter mile) of each street segment. Once again, distance was measured along the street network using the ArcGIS Network Analyst extension.

e) Bus Stops

The data describing bus stops were obtained from the Metro Transit Division of the Department of Transportation (table A4.1). The number of bus stops in Seattle decreased over the study period (mean = 4,160). The highest number existed in 1998 (n = 4,287), and by 2004 there were 4,053 bus stops. While this was a relatively minor drop, it is mirrored in the number of street segments with a bus stop, which had also fallen from 3,106 in 2008 to 2,989 in 2004. About one-third of those streets experienced a change over the time period (some gained or lost service completely).

f) Vacant Land

Assembling vacant land information required two separate data sources (table A4.1). Historical data related to land use codes and value from 1989–1999 came from the Planning Department. More recent data was obtained from the King County Tax Assessor's web site (2000–2004). From these data, we calculated the percentage of the total number of parcels on each street that were vacant land.

Relatively few streets in Seattle contained vacant land. Between 1989 and 2004, only about 5 to 8 percent of street segments per year had any vacant parcels. The percentage of streets with at least one vacant parcel declined over the time period, reaching a low of 4.7 percent in 2004.

g) Street Lighting

Information on street lighting was supplied by Seattle Public Utilities and was used to create a total number of watts per street value for 1997–2004 (table A4.1).6 The number of street poles and their associated lights increased steadily (p.207) over the time period (from 68,725 in 1997 to 83,709 in 2004). This overall increase was in contrast to variability at the individual street segment level, where there was change in both directions. The street lighting wattage decreased on 540 streets (413 were residential and 127 were arterial). Almost all these streets were concentrated in one suspiciously rectangular area in the northeast part of the city. The utility department had no explanation for this “dark” area. Wattage increased on 5,420 streets (4,059 were residential and 1,361 were arterial). Street segments with decreasing lighting were spread throughout the city with the exception of west Seattle in which there were no increasing streets.

h) Street Type

We obtained street type information as part of the street centerline file (table A4.1).7 Seattle GIS provided their 2006 street centerline file, which we used to develop the units of analysis and to obtain information on street type. We considered two types of streets in the study: arterial and residential (which includes walkways/stairs). Arterial streets are higher traffic streets that have higher speed limits.8 They collect traffic flowing from residential streets and provide for movement within areas of the city while still enabling access to abutting land uses. Residential streets also provide access to land uses, but they have lower speed limits and are designed to carry less traffic.

i) Final Characteristics for Analysis

These characteristics were then used to create the final characteristics for analyses reported in chapters 5 and 7 (table A4.2). Each of the final 10 characteristics represents one of four constructs: motivated offenders, suitable targets, guardianship, or accessibility/urban form. All the final data sets have a geographic extent that includes the entire city of Seattle. However, their temporal extent varies. We were able to get only the crime data and the public facilities data over the entire study period. The temporal period of the other characteristics range from four years of coverage to 13 years of coverage.

Characteristics of Street Segments: Social Disorganization

Based on social disorganization theories, we collected nine characteristics for each street segment in Seattle at the address level of analysis (table A4.3). These characteristics were then aggregated to create the final eight characteristics we focused on for the analysis (table A4.4). The geocoding rate varied from a low of 93.3 percent for the physical disorder incidents to a high of 100 (p.208)

Table A4.2 Sources and Extents of Opportunity Theory Variables and Means and Standard Deviations for Starting Values

Variable

Definition

Source

Temporal Extent

Years

Starting Mean (S.D.)

High Risk Juveniles

Total number of public school students with 10 or more unexcused absences and/or flagged as low academic achievers

Seattle Public Schools

13 years

1992–2004

0.615 (2.102)

Employment

Total number of employees at businesses located on the street segment

InfoUSA database of all businesses in Seattle

4 years over a 7-year period

1998, 2000, 2002, 2004

14.914 (127.245)

Residents

Composite variable combining the total number of public school students and the total number of registered voters

Seattle Public Schools (public school students), Labels & Lists Inc. (voter registration)

6 years

1999–2004

17.787 (27.213)

Business Crime Attractors/Crime Generators

Total retail business sales on the street segment

InfoUSA database of all businesses in Seattle

4 years over a 7-year period

1998, 2000, 2002, 2004

1.750 (27.670)

Public Crime Attractors/Crime Generators

Calculated variable capturing the total number of Public Facilities within 1,320 feet of a street street segment

Fleets and Facilities Department, City of Seattle (Community centers), Seattle Public Libraries, Seattle School District

16 years

1989–2004

0.534 (0.845)

Street type

Type of street (arterial vs. residential), Static across all years

Seattle GIS

Static

2006

0.270 (0.442)

Bus Stops

Total number of bus stops

Department of Transportation (Metro Transit Division)

8 years

1997–2004

0.176 (0.510)

Vacant Land

Percentage of vacant land parcels

Developed from Historic Assessor's Data (Seattle Planning Department) and parcel boundaries (King County GIS).

6 years over a 14 years

1991, 1993, 1995, 1997, 1998, 2004

0.019 (0.103)

Fire and Police Stations

Calculated variable capturing the total number of police or fire stations within 1,320 feet of a street street segment

Fleets and Facilities Department

16 years

1989–2004

0.068 (0.260)

Street Lighting

Total amount of watts

Seattle Public Utilities

8 years

1997–2004

3.952 (6.937)

(p.209) (p.210) percent for the public housing units. One notable deficiency in the data representing characteristics of parcels was related to the relatively lower join rate (about 80 percent) for the data regarding residential property values and type of land use.

a) Residential Property Value

Assembling residential property values for Seattle required two separate data sources (table A4.3). Historical data related to land and building values for 1989–1999 came from the Planning Department, and more recent data was obtained from the King County Tax Assessor's web site (2000–2004). From these two data sets, we calculated a weighted index variable to represent the

Table A4.3 Roots of Characteristics Used in the Model

Variable

Geocoding Hit Rate

Data Source

Contributes to:

Residential property value

N/A1

Developed from Historic Assessor's Data (Seattle Planning Department) and parcel boundaries (King County GIS).

Socioeconomic Status (as represented by residential property value)

Type of land use

N/A1

Developed from Historic Assessor's Data (Seattle Planning Department) and parcel boundaries (King County GIS).

Mixed Land Use

Total number of illegal dumping and litter incidents

93.3%

Seattle Public Utilities

Physical Disorder

Total number of public housing units

100%

Seattle Housing Authority

Housing Assistance

Total number of Section 8 housing vouchers

99.7%

Seattle Housing Authority

Housing Assistance

Total number of public school students with 10 or more unexcused absences

97.1%

Seattle Public Schools (public school students)

Truant Juveniles

Racial Heterogeneity of public school students

97.1%

Seattle Public Schools (public school students)

Racial Heterogeneity

Percent of active voters represented on each street

99.7%

Labels & Lists Inc. (voter registration)

Percent of Active Voters

Distance from geographic center of Seattle

N/A2

Seattle GIS

Urbanization

(1) Getting the historical information joined to the shape file of parcels required several steps and resulted in an average of 81 percent of the parcels having land use information.

(2) No street centerline file was available for 1989, so we used the same one throughout the study period.

(p.211) ranked property value on each street segment. To create this variable, we first ranked all the property values in the city from 1 to 10, with 10 being the highest value. In order to separate the single-family housing from multifamily dwellings, we ranked these two groups separately. Then we combined the ranks of single-family housing (SF) and multifamily housing (MF) into a final number that represents the property value of a street. We also weighted the ranks by percentage of housing type in the given street so the composite score also reflects the proportion of the type of property. The index ranges from 0 to 10, with zero representing a street with no residential property. This variable contributed to our measure of the socioeconomic status of each street (table A4.4).

b) Land Use Type

Similarly, assembling land use information required two separate data sources (table A4.3). Historical data related to land use codes from 1989–1999 came from the Planning Department. No historical data related to land use was available from the King County Tax Assessor's web site; 2004 data were available, however, and those were used. From these two data sources we calculated the percentage of the total of each type of land use on each street segment. Finally, we created a dichotomous variable representing those streets with percent residential land use between 25 percent and 75 percent that also have nonresidential land uses present (e.g., commercial, institutional, industrial, but not solely water or vacant land). Streets meeting this criterion were coded as mixed land use. This variable contributed to Mixed Land Use (table A4.4).

c) Physical Disorder

Data documenting physical disorder incidents were collected from Seattle Public Utilities (tables A4.3 and A4.4). The incidents in this database were generated from problems noticed by inspectors (self-initiated), reports from other agencies, and from citizens calling the hot line or emailing the agency to report physical disorder problems. The physical disorder measure includes: illegal dumping, litter, graffiti, weeds, vacant buildings, inoperable cars on the street, junk storage, exterior abatement, substandard housing, and minor property damage. The types of dumping and litter items recorded in the database consist of things like tires, appliances, yard waste, mattresses, and freezers, to list just a few. This database covers the time period from 1992–2004, but the information was not consistently gathered for 1992. Therefore, this study only uses information from 1993 to 2004. There were 42,331 incidents in the original database and 93.3 percent of them were successfully geocoded.

(p.212) d) Public Housing

The locations of public housing communities and the total number of units in each community were collected from the Seattle Housing Authority (table A4.3). Several of the large communities reported one total number of units for the entire complex (High Point, New Holly, and Rainier Vista). For these, we divided the total number of units by the number of street segments that participated in the development and allocated the resulting number of units to each street in the development. There were 5,857 public housing units from 1989–1997. In 1997, the number dropped to 5,299 units and stayed there until 2002, when it dropped again to 4,218. Two subsequent changes occurred in the last two years of the study period: first a reduction to 3,838 units in 2003 and then a slight increase to 3,896 units in 2004. This variable contributed to the composite variable of Housing Assistance (table A4.4).

e) Section 8 Vouchers

We obtained information on the allocation of Section 8 vouchers in Seattle from the Seattle Housing Authority (table A4.3). Section 8 housing vouchers can be used to rent any apartment for which the management will accept the vouchers. Vouchers allow individuals to rent market-rate apartments for a reduced cost, with the voucher bridging the gap between what the individual can pay and market-rate rent. The number of vouchers increased 34.4 percent from 1,674 to 2,250 between 1998 and 2004. The minimum number of vouchers on a given street was zero across all years, while the maximum ranged from a low of 125 in 2003 to a high of 152 in 1999. This variable contributed to the composite variable of Housing Assistance (table A4.4). The presence of public housing units and Section 8 voucher holders show where the disadvantaged populations are located.

f) Housing Assistance

The variable called Housing Assistance consists of the total number of public housing units plus the total number of Section 8 vouchers in use on a street segment. We created this variable to capture the total number of housing units on each street segment receiving some type of housing assistance (table A4.4).

g) Truant Juveniles

Data about truant juveniles were obtained from Seattle Public Schools as part of the public school's student database (tables A4.3 and A4.4). Truant juveniles (p.213) were defined as the total number of students with 10 or more unexcused absences in a school year (see earlier section on public school data for more information). The lowest number of truant students in any given year was 3,581 (9.85 percent) in 2004. The highest number occurred in 1994 when there were 6,489 (18.4 percent) truant students. That was also the year with the highest proportion of students classified as truant.

h) Racial Heterogeneity

Racial heterogeneity was estimated using information that was part of the Seattle Public Schools’ student database (tables A4.3 and A4.4). The data contain racial identification of all students enrolled in Seattle's public schools from 1992 to 2004. Four racial groups were identified in this study: white, black, Asian, and Hispanic. The probabilities of each racial group encountering another out-group member were then computed and averaged to form an overall racial heterogeneity index. The detailed computation process of the variable is described in chapter 6.

i) Percent of Active Voters

The variable Percent Active Voters was drawn from the voting database (tables A4.3 and A4.4). The data include not only the registered voters’ voting behaviors (whether they voted or not in the given year) for the current year but also their past voting frequency dating back to 1990. To differentiate active voters from the rest, we compared each registered voter's short-term average voting behavior to the population's short-term voting average in the most recent two years. In any given year, if a person had an average short-term voting behavior greater than the mean of Seattle's registered voters, then we assigned this person an active voter status. On each street, the number of active voters is divided by the number of total registered voters to get the percentage of active voters. This value is used as a proxy of residents’ willingness to participate in public affairs.

j) Distance from City Center

The distance from the city center was calculated from each street segment to the geographic center of Seattle (tables A4.3 and A4.4).9 This measure represents the degree of urbanization of each street. The geographic center of Seattle was located at 331 Minor Avenue N.10 Distance was measured along the street network using the ArcGIS Network Analyst extension. Using street network distance was especially important in a city like Seattle, which is trisected by waterways. Since these waterways must be crossed using bridges, they represent significant physical barriers to travel.

(p.214) k) Final Characteristics for Analysis

The preceding characteristics were then used to create the final characteristics for analyses reported in later chapters (table A4.4). Each of the final eight characteristics represents one of two theoretical dimensions. The structural dimension includes socioeconomic status, mixed land use, urbanization, housing assistance, physical disorder, and racial heterogeneity. The intermediary dimensions of the effects of structural factors on crime include unsupervised teens and willingness to intervene in public affairs.

Table A4.4 Sources and Extents of Social Disorganization Variables and Means and Standard Deviations for Starting Values

Variable

Definition

Temporal Extent

Years

Starting Mean (S.D.)

Socioeconomic Status (as represented by residential property value)

To create this variable, we first ranked all the property in the city from 1 to 10, with 10 being the highest value. In order to separate the single-family (SF) housing from multifamily (MF) dwellings, we ranked these two groups separately. Then we combined the ranks of SF and MF into a final value that represents the property value of a street. We also weight the ranks by percentage of housing type on the given street so the composite score of SES also reflects the proportion of the type of property.

6 years over a 14-year period.

1991, 1993, 1995, 1997, 1998, 2004

4.491

(3.608)

Mixed Land Use

Dichotomous variable representing those streets with nonresidential land use (e.g., commercial, institutional, industrial, but not solely water or vacant land) and between 25% and 75% residential land use

6 years over a 14-year period.

1991, 1993, 1995, 1997, 1998, 2004

0.04 (.207)

Physical Disorder (count)

Total number of reported physical disorder incidents

12 years

1993–2004

0.112

(0.379)

Housing Assistance

Composite variable of the total number of public housing units and the total number of Section 8 vouchers in use.

7 years

1998–2004

0.375

(4.615)

Truant Juveniles

Total number of public school students with 10 or more unexcused absences

13 years

1992–2004

0.244

(0.834)

Racial Heterogeneity (students)

The probabilities of each racial group to encounter another out-group member

13 years

1992–2004

0.017

(0.038)

Percent of Active Voters

Percent of active voters represented on each street

6 years

1999–2004

0.375

(0.317)

Urbanization

Distance from geographic center of Seattle

Static

2006

255.712

(117.415)

(p.215) The spatial and temporal extent of each characteristic as well as its definition is included in table A4.4. All the final data sets have a geographic extent that includes the entire city of Seattle. However, their temporal extent varies. We were not able to obtain any of these characteristics for the entire study period. The temporal coverage of other characteristics ranged from a low of six years of coverage for percent of active voters to a high of 13 years for physical disorder.

Notes:

(1.) These were above the commonly accepted minimum acceptable geocoding rate of 85 percent (Ratcliffe, 2004).

(2.) The historical data related to real property (i.e., value and land use type) was the most problematic to assemble. The final data set had about 20 percent missing data.

(3.) Data describing the number of public schools students were provided at the hundred-block level and geocoded by the researchers to the street segment. The data were for academic years. We refer to each academic year by its earlier calendar year (e.g., data for 1989–1990 are used to represent 1989).

(4.) The following zip codes were used to define Seattle: 98101, 98102, 98103, 98104, 98105, 98106, 98107, 98108, 98109, 98112, 98115, 98116, 98117, 98118, 98119, 98121, 98122, 98125, 98126, 98133, 98134, 98136, 98144, 98146, 98168, 98177, 98178, 98195, and 98199.

(5.) The following North American Industrial Classification (NAIC) codes were used to identify retail businesses: 441–Motor Vehicle and Parts Dealers; 442–Furniture and Home Furnishings Stores; 443–Electronics and Appliance Stores; 444–Building Material and Garden Equipment and Supplies Dealers; 44612–Cosmetics, Beauty Supplies, and Perfume (p.235) Stores; 44613–Optical Goods Stores; 44619–Other Health and Personal Care Stores; 448–Clothing and Clothing Accessories Stores (Retail); 451–Sporting Goods; Hobby, Book, and Music Stores (Retail); 452–General Merchandise Stores (Retail); 453–Miscellaneous Store Retailers (Retail).

(6.) When representing lighting related to a street, we only included light poles that were within 90 feet of the street centerline for residential roads and within 300 feet for arterial roads. The two different thresholds were used because of the difference in the average width of an arterial street and a residential street. After establishing the street poles with light that might reach the edge of the street, we then found the total wattage for the street lights associated with each pole and aggregated the total watts by street segment.

(7.) We used the 2006 centerline file since the Seattle Planning Department and Seattle GIS department verified there had been no significant changes in the street configuration, nor had there been any annexations during the study period.

(8.) Information on street classifications was retrieved from the King County Department of Transportation web site at http://www.kingcounty.gov/transportation/kcdot/Roads/TransportationPlanning/ArterialClassificationSystem.aspx. Briefly, arterial streets are those that carry larger volumes of traffic. Residential streets run through neighborhoods and are designed to carry lower volumes of local travel at slower speeds. Walkways are non-vehicular paths or stairways that typically connect two residential streets.

(9.) We also considered using the cultural center of Seattle, but we could find no documentation regarding a cultural center. The librarians at the Seattle Public Library identified the Westlake Center (4th Avenue and Pine Street), which opened in 1988, as the cultural center (personal conversation, 2008). Since the two addresses are only 3,650 feet (a little less than three-quarters of a mile) apart as the crow flies, we went with the geographic center, which was verifiable.

(10.) The geographic center of Seattle is located at N 47° 37.271 W 122° 19.986, which translates to 331 Minor Avenue N (see http://www.waymarking.com/waymarks/WM29A8).