<%BANNER%>

Analysis on Correlation between Urban Form Factors and Children's Walkability to School

Permanent Link: http://ufdc.ufl.edu/UFE0043827/00001

Material Information

Title: Analysis on Correlation between Urban Form Factors and Children's Walkability to School
Physical Description: 1 online resource (73 p.)
Language: english
Creator: Yang, Wencui
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2011

Subjects

Subjects / Keywords: children -- connectivity -- density -- school -- street -- urban -- walkability
Urban and Regional Planning -- Dissertations, Academic -- UF
Genre: Urban and Regional Planning thesis, M.A.U.R.P.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Childhood obesity has been constantly increasing in the United Stated. Research studies suggest that one of reasons responsible for childhood obesity is the loss of physical activity. One solution to increase the amount of physical activity children engage in is the provision of environment where children can walk and bicycle to school safely. This study tests the correlation between urban form factors and children's walkability to school. The study is conducted based upon the data from 40 public schools in Orange, Seminole, Pasco and Hillsborough Counties in Central Florida. The study areas are created using School Attendance Zone (SAZ) at the distance of half-mile and one-mile around school. The indicators are developed, representing urban form variables ---school attendance zone geometry, street connectivity and residential density. The dependent variable is represented by children potential walkabiltiy, assuming that all children within study zone could walk or bike to school. The Ordinary Least Squares regression model tool in ArcGIS is used to test the relationship between dependent variable and explanatory variables. The findings from this study suggest that school attendance zone geometry is not significant associated with children's potential walkability; street connectivity and residential density show the different levels of correlation to children's potential walkability. The results from half-mile and one-mile study zones present the different correlations among variables. Due to the limitation of this study, the actual children's walkability to school cannot be predicated based on the results. The future study could take into account other social, economic and family factors to further explore the correlation between the actual children's walkability and urban form factors.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Wencui Yang.
Thesis: Thesis (M.A.U.R.P.)--University of Florida, 2011.
Local: Adviser: Bejleri, Ilir.
Local: Co-adviser: Steiner, Ruth L.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2011
System ID: UFE0043827:00001

Permanent Link: http://ufdc.ufl.edu/UFE0043827/00001

Material Information

Title: Analysis on Correlation between Urban Form Factors and Children's Walkability to School
Physical Description: 1 online resource (73 p.)
Language: english
Creator: Yang, Wencui
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2011

Subjects

Subjects / Keywords: children -- connectivity -- density -- school -- street -- urban -- walkability
Urban and Regional Planning -- Dissertations, Academic -- UF
Genre: Urban and Regional Planning thesis, M.A.U.R.P.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Childhood obesity has been constantly increasing in the United Stated. Research studies suggest that one of reasons responsible for childhood obesity is the loss of physical activity. One solution to increase the amount of physical activity children engage in is the provision of environment where children can walk and bicycle to school safely. This study tests the correlation between urban form factors and children's walkability to school. The study is conducted based upon the data from 40 public schools in Orange, Seminole, Pasco and Hillsborough Counties in Central Florida. The study areas are created using School Attendance Zone (SAZ) at the distance of half-mile and one-mile around school. The indicators are developed, representing urban form variables ---school attendance zone geometry, street connectivity and residential density. The dependent variable is represented by children potential walkabiltiy, assuming that all children within study zone could walk or bike to school. The Ordinary Least Squares regression model tool in ArcGIS is used to test the relationship between dependent variable and explanatory variables. The findings from this study suggest that school attendance zone geometry is not significant associated with children's potential walkability; street connectivity and residential density show the different levels of correlation to children's potential walkability. The results from half-mile and one-mile study zones present the different correlations among variables. Due to the limitation of this study, the actual children's walkability to school cannot be predicated based on the results. The future study could take into account other social, economic and family factors to further explore the correlation between the actual children's walkability and urban form factors.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Wencui Yang.
Thesis: Thesis (M.A.U.R.P.)--University of Florida, 2011.
Local: Adviser: Bejleri, Ilir.
Local: Co-adviser: Steiner, Ruth L.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2011
System ID: UFE0043827:00001


This item has the following downloads:


Full Text

PAGE 1

1 ANALYSIS ON CORRELATION BETWEEN URBAN FORM FACTORS AND CHILDREN S WALKABILITY TO SCHOOL By WENCUI YANG A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER O F ARTS IN URBAN AND REGIONAL PLANNING UNIVERSITY OF FLORIDA 20 11

PAGE 2

2 20 11 Wencui Yang

PAGE 3

3 To my Family and Friends

PAGE 4

4 ACKNOWLEDGMENTS First and foremost, I want to thank my parents for all of their love support, and encouragement I would like to thank my Chair, Ilir Bejleri who provided me valuable instruction, wisdom, and guidance throughout this entire process I also would like to thank C oc hair, Dr. Ruth Steiner for her valuable instruction. Finally, I would like to thank all of the people from the Department of Urban and Regional Planning (faculty, staff, students, and colleagues) that supported me along the way

PAGE 5

5 T ABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURES ................................ ................................ ................................ .......... 8 A BSTRACT ................................ ................................ ................................ ..................... 9 CHAP TER 1 INTRODUCTION ................................ ................................ ................................ ..... 11 Background ................................ ................................ ................................ .............. 11 Problem Statement ................................ ................................ ................................ .. 11 Study Objectives ................................ ................................ ................................ ..... 12 2 LITERATURE REVIEW ................................ ................................ ........................... 14 Childhood Obesity and Physical Inactivity ................................ .............................. 14 Urban Form Factors I ................................ ........... 16 Measures of Walkability ................................ ................................ .................... 17 Issue of Public Schoo l Siting ................................ ................................ ............ 18 Street Connectivity ................................ ................................ ........................... 21 Residential Density ................................ ................................ ........................... 24 Summary ................................ ................................ ................................ ................ 27 3 METHODOLOGY ................................ ................................ ................................ ... 29 Establish Study Zone ................................ ................................ .............................. 29 School Attendance Zone ................................ ................................ .................. 29 Walkable Distance ................................ ................................ ............................ 29 Dependent Variable ................................ ................................ ................................ 31 Independent Variables ................................ ................................ ............................ 32 School Attendance Zone Geometry Indicators ................................ ................. 32 Street Connectivity Indicators ................................ ................................ ........... 33 Resident ial Density Indicators ................................ ................................ .......... 35 Regression Model ................................ ................................ ................................ ... 36

PAGE 6

6 Summary ................................ ................................ ................................ ................ 37 4 RESULTS AND FI NDINGS ................................ ................................ ..................... 44 Results from Regression Model of Half mile Study Zone ................................ ........ 44 Results from Regression Model of One mile Study Zone ................................ ....... 45 Summary ................................ ................................ ................................ ................ 46 5 DISCUSSIONS ................................ ................................ ................................ ....... 55 Discussions and Conclusions ................................ ................................ ................. 55 Limitations and Future Studies ................................ ................................ ............... 57 Limitations ................................ ................................ ................................ ........ 57 Future Studies ................................ ................................ ................................ .. 58 6 CONCLUSIONS ................................ ................................ ................................ ..... 59 APPENDIX INTERPRETATION OF OLS RESULTS ................................ ................................ ........ 60 LIST OF REFERENCES ................................ ................................ ............................... 68 BIOGRAPHICAL SKETCH ................................ ................................ ............................ 73

PAGE 7

7 LIST OF TABLES Table page 4 1 Summary of Results from Half mile Model ................................ ...................... 50 4 2 Summary of Results from One mile Model ................................ ......................... 54

PAGE 8

8 LIST OF FIGURES F i g u r e page 2 1 Prevalence of Overweight Chi ldren a nd Adolescents ................................ ......... 28 3 1 Illustration of School Attendance Zone ................................ ............................... 39 3 2 Illustration of Study Zone ................................ ................................ .................... 40 3 3 Illustration of Dependent Variable ................................ ................................ ....... 41 3 4 Illustration of School Attendance Zone Geometry ................................ .............. 42 3 5 M athematical Formula of Re gression Model ................................ ...................... 43 3 6 Ordinary Least Squares Tools in ArcGIS ................................ ............................ 43 4 1 Indicators of Half mile Model ................................ ................................ .............. 47 4 2 Pre running Results from Half mile Model ................................ .......................... 48 4 3 Results from Half mile Model ................................ ................................ .............. 49 4 4 Indicators of One mile Model ................................ ................................ .............. 51 4 5 Pre running Results from One mile Model ................................ ......................... 52 4 6 Results from One mile Model ................................ ................................ ............. 53

PAGE 9

9 Abstract of Thesis Presented to the Graduate S chool o f the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Arts in Urban and Regional Planning ANALYSIS ON CORRELATION BETWEEN URBAN FORM FACTORS AND CHILDREN S WALKABILITY TO SCHOOL By Wencui Yang December 2 011 Chair: Name: Ilir Bejleri Cochair: Ruth L. Steine r Major: Urban and Regional Planning Childhood obesity has been constantly increasing in the United Stated. R esearch studies suggest that one of reasons responsible for childhood obesity is the loss of physical activity One solution to increase the amount of physical activity children engage in is the provision of environment where children can walk and bicycle to school safely. T his study tests the correlation between urban form factors and children s walkability to school. T he study is conducted based upon the data from 40 public schools in Orange, Seminole, P asco and Hillsborough Counties in Central Florida. The study areas are created using School Attendance Zone ( SAZ) at the distance of half mile a nd one mile around school. The indicators are developed, representing urban form variables --school attendance zone geometry, street connectivity and residential density. The dependent variable is represented by children potential walkabiltiy, assuming th at all children within study zone could walk or bike to school. The Ordinary Least Squares

PAGE 10

10 regression model tool in ArcGIS is used to test the relationship between dependent variable and explanatory variables. The findings from this study suggest that scho ol attendance zone geometry is not significant associated with children s potential walkability; street connectivity and residential density show the different levels of correlation to children s potential walkability. The results from half mile and one m ile study zones present the different correlations among variables. Due to the limitation of this study, the actual children s walkability to school cannot be predicated based on the results. The future study could take into account other social, economic and family factors to further explore the correlation between the actual children s walkability and urban form factors.

PAGE 11

11 CHAPTER 1 INTRODUCTION Back g round Over las t forty years, planning in United States has put too much attention on automobile us age which created the automobile oriented community where walking has become increasingly undesirable and unsafe. With the more concern on environmental and health issues, people start realizing that their over dependence on automobile has been responsibl e for many health issues The awareness towards the negative consequence of automobile usage has drawn attention to the benefits of walking and the importance of walkable community. W alking is the most basic form of transportation Pedestrian access betwe en home, work, and facilities improves people s quality of life by providing residents with options of daily exercise and healthy lifestyle. T he good connectivity of pedestrian network provide s residents without access to automobile access to employment, r ecreation, and educatio n opportunities in a safe and efficient way. Thus, provision of pedestrian infrastructure that facilitates safe travel is a key issue for urban planners and public policy makers. Problem Statement Elementary and m iddle school st uden ts are unique in their reliance to walking and biking modes of transportation. In the past, walking to /from school used to be one of main options for physical activity available to elementary and middle school students.

PAGE 12

12 However, recent research finds that the number of youth driven to /from school has a constantly increase compared to students in the past. In 1962, through k 12 42% school children walked or bicycled to school while this number declined to 16% in 2001 ( CDC, 2005 ). I t is generally accepted that the loss of physical activity ac quired by walking is one factor responsible for a nationwide rise in childhood obesity rates. One solution to consider in addressing this issue is the encouragement of physical activity by providing safe and accessible oppo rtunities for children to walk and bicycle to school. Urban planners and policy makers have the opportunity to address this issue by designing and providing safe and well connected pedestrian routes to promote the community environment where the children c an improve health by increasing daily physical activity. Study Objectives The time children spend on physical activity could be increased by providing children with safe routes to/from schools One of many aspect s involved in safe routes to/from school is urban form which consists of many factors such as school location, street connectivity, residential density etc. The study on these factors could help to create the urban form where children can walk and bicycle to school safely. The purpose of study is t o understand the correlation between to school and urban form factor s potential walkability, assuming that all students living within study zone could walk to

PAGE 13

13 school Urban form facto rs in this study include street connectivity and residential density within school attendance zones The study is conducted with data from forty public schools in the State of Florida thirty three elementary schools and seven middle schools. Forty schools are respectively from Orange County, Seminole County, Hillsborough Co unty and Pasco County in the State of Florida GIS data is obtained from Geop lan Center at University of Florida including school location, school attendance zone, on and street network y and urban for m are developed. A r egression m odel in ArcGIS was chosen to test if there are correlations between children s walkability and urban form factors. Chapter 2 examines existing research about measures of walkability and urban form factors walkability to school Chapter 3 present s the methodology to establish the study zone and develop indicators to measure urban form factors Chapter 4 sho ws the results from running regression models. Chapter 5 explains the results and discusses the limitation of research.

PAGE 14

14 C HAPTER 2 L ITERATURE REVIEW T his chapter illustrates the connection s be tween urban form factors and children s walkability. T his s ection will start with a review of research on relationship between childhood obesity and physical inactivity Then a summary on measures of walkability are made based on existing research. Then the issue of public school siting associated with children s walkability will be discussed. L ast, some characteristic of urban form will be described by examining the relationship with s walkability to school. Childhood Obesity and Physical Inactivity It is generally accepted that increased physical activ ity promotes good health and increases life expectancy. A major national study found that 42% of men and 28% of women were overweight, and 21% of men and 27% of women were obese and that U.S. adult obesity rates increased from 12.1% to 17.9% between 1991 a nd 1998. In a 1993 study, 14% of all deaths in the United States were attributed to a severe lack of physical activity and poor dietary habits (McGinnis 1993). In a later study, sedentary lifestyles were linked to 23% of deaths resulting from major chronic diseases (Hahn 1998). That is to say, people who exercise reduce their risk of developing or dying from heart disease, diabetes, colon cancer, and high blood pressure. In fact, long term changes in obesity and being overweight are more closely correlated to physical activity than dietary changes (Prentice 1995). Thus, people who exercise tend to have longer lives than less

PAGE 15

15 active individuals (Kushi1997; Lee 1999; Wei 1999). These results suggest policies and programs aimed at increasing physical activity w ill prove to be effective in addressing the current obesity epidemic in the U.S. Although physical activity is a critical component of overall health youth are exercising less today than their counterparts 35 years ago (U.S. Department of Health and Human Services 2002). Over the last 40 years, the rate of childhood overweight has tripled (Figure 2 1 ). Scientific evidence has shown that physical activity plays a critical role in supporting weight loss Unfortunately, many children do not participate in th e recommended amount of physical activity needed to maintain a healthy weight that supports healthy lifestyles (Dellinger & Staunton, 2002). I n many cases, children are not provided with opportunities of engaging in physical activity due to the inadequate of facilities, poor commute system between residential dwellings and service, and the lack of necessary infrastructure such as pedestrian sidewalks and bike lanes. Walking or bicycling to school is one of the easiest ways for school aged children to gain p hysical activity on a daily basis. However, i n the United States, the number of children walking or bicycling to school has declined to 15% in 2001 compared to the percentage of 48% in 1969 ( Steiner, et al 2008 ) D ecrease in the number of children walking and bicycling to school and the reduction in the amount of time children spend in

PAGE 16

16 It could be challenging to assess correctly one certain factors without taking in to account other built, social, and economic environment factors because they may occur and influence children's physical activity jointly. However, r ecent research suggests that children s weight is influenced by a number of built environment factors. For instance, a higher level of physical activity in children is associated with better sidewalks ( Jago et al., 2006 ), higher quality recreational facilities (Romero, 2005), easier access to recreational facilities ( Gomez et al ,2004), greater housing density ( Roemmich et al., 2006), and higher neighborhood walkability(Kerr et al., 2006). There are many methods that encourage children to engage in physical activity one of which is provision of safe access and route to school. Davison et al(2006),through t hirty three quantitative studies concluded that children's participation in physical activity is positively associated with publicly provided recreational infrastructure (access to recreational facilities and schools) and transport infrastructure (presence of s idewalks and controlled intersections, access to destinations and public transportation). Therefore, children s physical activity could be increased by improving the walkability to school. Urban Form Factors I nfluencing There have be en numerous studies on walkability in the past. In this section of the report, an effort is made to emphasize some of these studies that form the base for

PAGE 17

17 identifying the measures of walkability to be used in formulating the model developed in this study Measures of Walkability A n effective way to provide younger populations the opportunity to engage in physical activity is to provide safe and accessible means for children to travel between home and school by walking or bicycling Provision of safe and ac cessible ways could be fulfilled by improving the safety and connectivity of phy sical environment in the street scape To better understand the link between walkability and physical environment, a review on measures of walkability would be helpful. I n his r eport Coffee (2005) concluded that studies on walkability show a consistent emphasis on connectivity, proximity, land use mix, density and safety. Due to the characteristic of this research, the author will focus on urban form factors. Although there are differences between adult and children in travel patterns, some research found that there is a similarity in the factors influencing bo th walkability. Frank et al(2007 ) concluded that the same indicators of walkability that are related to active transport ation and physical activity in adults (i.e., street connectivity, residential density, and mixed land use) are related in similar ways to walking for transportation in children and especially adolescents. Apart from these indicators, school location and si ting are also proved to be related to school children s travel. Ewing et al(2004) examine d the relationship between

PAGE 18

18 mode of travel to school and the full range of factors related to school location. In this study, students with shorter walk and bike times to school proved significantly more likely to walk and bike. I n another study, Steiner and Bejleri(2008) investigated the relation of the school siting with children s potential walkabiltiy throughout Growth Management periods from the 40 public school in Central Florida. This research emphasized the role of public school siting in determining the children s walkability to school. Given that the unique nature of children s travel between school and home and the limitation on data availability this study wi ll focus on the issue of public school siting, street connectivity and residential density. The following sections discuss these factors in detail. Issue of Public School Siting There are many issues associated wit h school siting. This study briefly discus s ed enrollment capacity, school site location and site size School enrollment c apacity Since the 1980 s, average school size (measured by enrollment capacity) has g r own with the increasingly larger school facilities size and longer distance from the neigh borhoods they serve Florida public school enrollment in 1980 for P re kindergarden t hrough G rade 12 school aged students was approximately 1,510,000 (American Fact Finder, 2008). This number increased by 76.62% based on the approximated 2,667,000 student s enrolled in 2007 (American Fact Finder, 2008).

PAGE 19

19 During about 30 years 1,157,000 students across the entire State. Furthermore, the National Center for Education Statistics reported that from 1930 to 2001, public school enrollment in the U.S. nearly doubled, from 26 to 48 million students (across all grade levels), yet the number of public school buildings decreased 60 percent in the same period, from 247,000 to 93,000 (ICMA, 2008). This statistic indicates a shift from a n average of 105 students to nearly 516 students per school building. As the average size of a school has grown, so to have the distances between schools and the neighborhoods they serve. This trend not only relates to growth in average enrollment size, bu t is also causative of the policies and practices that encourage large site locations and discourage expansion and renovation of existing school sites. School enrollment capacity is partially decided establishment of School Attendance Zones (SAZ). School attendance zones are the geographical boundaries that institute which communities (area) a school will serve. SAZs are established based on a series of factors, of which enrollment capacity plays a critical role. SAZs are als o developed with the purpose of integrating a diversity of students into school facilities. However, balancing the structure of a student body is quite difficult, due to residential development patterns which often produce neighborhoods that are

PAGE 20

20 demographi cally and socio economically unbalanced. The result is a set of attendance boundaries that vary considerably in shape and size (Steiner, et al., 2008). The location of a school site relative to the neighborhood it is intended to serve may be quite distant, which in turn gives rise to n transportation issue concerning the mode choice of children travel from home to school. While recognizing this problem and understanding its implications this study is focused on the location of school and students in relati on to the center of SAZ. Sch ool site location and site s ize The increase in school site size also plays a critical role in determining school walkability. Over the past thirty years the acreage required for school sites has been increased so much that it is becoming more difficult to consider walkability as a critical factor in school location siting In many case s, over grown school site is resultant of concerning on costs and availability of land within existing urban areas. With scarce budgetary resour ces under a weakened economy, government dollars are being spent to cover the most crucial needs, particularly wages walkability as a priori ty is typically an afterthough t ( Schmucker, 2009). Due to the increased distances between many schools and the neighborhoods they serve, walking and bicycling is simply not feasible and the school bus is typically acknowledged as the solution to the problem. In the State of Florida, d istrict bussing is not provided within a

PAGE 21

21 two mile radius around school sites, unless hazardous walking conditions exist preventing safe and accessible access for children to walk or bicycle to school (e.g., road construction). Under these circumstances, co urtesy bussing is provided for those students until such problems hindering access from home to school are remedied. Besides, due to other factors such as safety parents still drive their children to school even close enough to school ( Campbell & Wang, 20 08) Another study, examining the impact of neighborhood walkability (based on street connectivity and traffic exposure) within 2 km of public primary schools on children regularly walking to school reveals that c onnected street networks provide direct r outes to school ( Georgina et al,2010) However, when connected street networks are designed for heavy traffic, the potential for children to walk to school is reduced. This highlights the importance of carefully considering school siting and, particularly, street design in school neighborhoods Street Connectivity A central point of contention among urban planners and transportation engineers is the issue of street network design and pedestrian travel options. In designing road networks with the primary goal of increasing automobile efficiency, critics argue transportation planners have built mode choice out of the built environment equation. The development of cul de sacs, for example, represent an approach to design

PAGE 22

22 efficiency for automobile transportation, but they have the opposite effect on pedestrian access and efficiency; pedestrians often have to take out of the way, circuitous routes because direct routes are truncated by cul de sacs, and transit vehicles cannot efficiently serve curvilinear neighborho ods or branch roads. Therefore, many modern suburbs limit pedestrian and transit access in exchange for increased auto mo bility (Cervero 1997 ). Reform minded urban designers argue that walking will increase in neighborhoods designed with more pedestrian fr iendly features, such as connected sidewalk layouts, increased mixed use development, and high density commercial and residential development (Duany 2001). Street design is one example of measures commonly used to assess neighborhood walkability research ers also frequently employ provision of sidewalks, streetscape design, miles of street, and access to activities. There have been many studies of the relationship between active transportation and urban form in adults, but the associated factors for adults may differ from those for children. F ew studies have investigated the relationship between street connectivity and children s walkability and physical activity and results are mixed with some indicating a relationship and others finding no associations Norman et al(2006) investigating c ommunity de sign and access to recreational facilities variables derived using geographic information systems (GIS) for799 adolescents found limited evidence that street connectivity( intersection density as

PAGE 23

23 indicator) w as associated with adolescence activity. However, this study shows the results that intersection density inversely related to girls p hysical activity which implies that the impact of street connectivity on children s physical activity may differ by gende r. Center for Design, Methods, and Analysis (2004) evaluates the relationship between neighborhood design and rates of students walking and biking to elementary school in 34 California communities The results from this study shows that children s biking an d walking rates were not associated with intersection density. I t is worthy noted that i n another study street con nectivity was found to be inversely related to physical activity Timperio A, Ball K, Salmon J, et al (2006) conducted a Cross sectional study of 235 children aged 5 to 6 years and 677 children aged 10 to 12 years from 19 elementary schools in Melbourne, Australia and found that Good connectivity en route to school was negatively associated with walking or cycling to school among older children. On the other hand, The City of Raleigh, North Guidelines for Pede strian (2008) outlining thirteen main characteristics of the area within one quarter mile radius of a school that indicate high potentia l for walking and bicycling which includes street connectivity as indicator. Schlossberg et al (2006) conducted a study by evaluating the effects of urban form and distance on travel mode to school among mi ddle school students in Oregon which found that urban form travel options to school. Schlossberg and colleagues (2006)

PAGE 24

24 utilize several indicators of street connectivity, including street density, intersection density and pedestrian route directness. T he findings from this s tudy also reveal s that students are willing to walk at distances greater than the accepted standard of one quarter mile which support s further research investigating the impact of urban form at greate r distances around school sites I n addition, this stud y found that while holding distance and other urban form measures constant, intersection density was a strong indicator which influenced overall walkability (Schlossberg, et al., 2006). In support of the findings from this study, Frank et al( 2005) suggest that areas with densities equal to or greater than 30 intersections per square kilometer have been associated with greater overall connectivity and increased levels of physical activity. Schmucker(2009) investigated how urban form impacts children s poten tial walkability to school by conducting a case study of Orange and Seminole Counties in Central Florida. This study, using half mile and one mile as walkable distance, reveals that good street connectivity reflects the well connected urban form which prov ides a great potential for children walking or bicycling to school. Residential Density The U.S. Census compiles data on the characteristics and locations of citizens across the count r y. Therefore, density measures such as population, housing units, and

PAGE 25

25 e mployees per unit area are the most readily accessible and oft used urban form variable in neighborhood accessibility research. Neighborhood accessibility research frequently relies on household survey results and personal daily trip diaries (Cervero 1997; Audirac 1999). These data collection methods are designed to spatially locate In her 1999 article, Ivonne Audirac (1999) explored the likelihood that housing consumers would trade off living on smaller lots for pe destrian proximity to community amenities. Her analysis of the University of Florida, Bureau of Economic and Business Research (BEBR) consumer attitude survey found residents of single family homes were willing to trade smaller lot sizes for improved pedes trian access to 2 of 5 types of neighborhood amenities. Residents of apartments and condos, for whom the spatial costs of reduced lot size are minimal, were willing to accept smaller lots for improved access to any community facility. These results suggest higher residential densities may instill a greater appreciation of walkable neighborhoods. Many studies have been published relating walkability and increased levels of physical activity to various characteristics of urban form including, residential dens ity, mix of land uses, street connectivity, and aesthetics & safety (Saelens, Sallis, Black, & Chen, 2003). Although little research has been conducted specifically examining the impact of

PAGE 26

26 to walk and bicycle to sch ool two studies give a look at the relationship between them. Kerr et al(2007)conducted a study looking at pedestrian travel in Atlanta by US youths aged 5 18 years Relati onships between five urban form variables and walking in specific demographic subgr oups are assessed using stratified logistic models and controlling for participant demographics R esidential density and recreation space were strongly related to walking in the highe st income group and residential density was a stronger factor in the lar ger households. I n another study, Center for Design, Methods, and Analysis (2004) evaluates the relationship between neighborhood design and rates of students walking and biking to elementary school in 34 California communities The results from this study supported that the walking and biking rates are higher in denser neighborhoods and to smaller schools In this research study, the author will examine the impact of residential density on school walkability. A key issue to consider when using density as a measurable characteristic of walkability is the differences in calculating density types. Depending on the type of research one might be conducting, it is important to differentiate the classification of density calculations being used, as results can so metimes be confusing or misinterpreted (Forsyth, 2003). For instance when calculating either Gross or Net

PAGE 27

27 This is due to a difference in mathematical equating. Net density refers to densities where the base land area calculation focuses only on the parcel or, if covering larger areas, excludes certain uses (e.g.: commercial and retail uses). Gross densities do not have such exclusions (calculations include all la nd uses) (Forsyth, 2003). Research suggests that schools be located in neighborhoods with a minimum net residential density of 5 dwelling units per acre (City of Raleigh, N.C., 2008). However, other research has shown that more walkable areas have a net re sidential density equal to or greater than 6 dwelling units per residential acre while less walkable areas experienced less than 4 dwelling units per residential acre (Frank, et al., 2005). I n addition, Student s Residential ratio is a good indicator repres enting density within all dwelling units. Summary Childhood obesity has been a national trend in the United States. A number of research has proved that the decline in physical activity could contribute to childhood obesity. W alk and bicycle to s chool could be an effective way to encourage children to engage in physical activity. Throughout the literature, t her e are a number of factors affecting walkabilit y. Studies on measures of walkability show a consistent focus on connectivity, proximity, lan d use mix, density and safety Due to the characteristic of children s travel between school and home, several studies about urban forum factors

PAGE 28

28 put an emphasis on school location, size and siting, street connectivity and residential density. M ost research agreed with the findings that both school location and residential density are associated with children s walkability. However, it is contentious among literature that if street connectivity has a relationship with children s biking and walking to school. This research will investigate if there are correlations between these urban form factors and children s walkability to scho ol. Figure 2 1 Prevalence of Overweight Ch ildren and Adolescents (CDC,2011)

PAGE 29

29 C HAPTER 3 M ETHODOLO GY This chapter describes the methods used to examine the correlation between urban form factors and children s walkab i lity. potential walkability to school, assuming that all students living within study zo ne could walk to school. The measures of urban form factors included in this research are street connectivity, residential density and school attendance zone location. Establish Study Zone and urban form factors the research establishe s the study zone by taking into account school attendance zone and walkable distance. School Attendance Zone Given that the students who are zoned within school attendance zone are eligible for attending certain school the research mapped the school attendance zone for each school and identified the students who are eligible for certain schools accordingly ( Figure 3 1 ). Walkable Distance The school attendance zone defines a boundary by which the study measures could be picked and developed. However, on one hand the fact that the different school

PAGE 30

30 attendance zones vary in size gives rise to the difference in the size of study measures. For example, the number of residential units largely depends on the size of school atte ndance zone. The larger school attendance zone would be, the more residential units could be produced. On the other hand, in certain case some students who are zoned within school attendance zone may live beyond the reasonable walk distance as we discussed before i n Chapter 2. For these students who have least potential to walk to/from schools, the research cannot take them as study subjects. In order to eliminate the effect of size of school attendance zone on study, the conception of reasonable walk dista nce is used to adjust the school attendance zone. Some research on walkability before has cited one quarter mile (or 400 meters) as an acceptable distance an individual might be expected to walk to any given destination (Atash, 1994). However, in the Stat e of Florida (where this research study takes place), a two mile radius shed has been established around school sites, defining the boundary by which parents are responsible for getting their children to school. On one hand, a s mentioned above the accepta ble distance a person might be expected to walk is approximately one quarter mile. This suggests that the two mile radius established in the State of Florida is not a reasonable distance to expect a child to walk or bicycle to school. On the other hand, ta king into account the real condition (two mile radius shed) in the State of Florida, one quarter mile boundary would exclude too many students out of the

PAGE 31

31 category that has potential to walk to school. To investigate the potential walkability of children wh o walk or bicycle to school in the State of Florida one half mile would be a reasonable distance for most students to walk or bicycle to school and even one mile would be still acceptable for biking or for res earch will focus on urban form factors within half mile and one mile sheds. Due to the geographic scale at which it is most potential for students to walk to/from schools, the measures of this research are developed at two scales : 0.5mile radius distance a nd 1 mile radius distance. Using school location point as center, a circle is drawn at a radius of 0.5 mile and 1 mile. T he overlapping part of school attendance zone and cir cle is the study zone. ( Figure 3 2 ) Thus the study zone defines a boundary by which students eligible for attending school live within a n acceptable walk /bicycle distance. Dependent Variable urban form factors, this research used Regression Model in ArcG IS to test the potential correlations between walkability and urban forum variables. S tudent s walkability is establish ed as dependent variable and urban form f actors as independent variables A s noted in last section, it is accepted that s tudents living w ithin study zone have most potential to walk or bike to school Therefore the number of students within study

PAGE 32

32 zone could represent the potential walkability. Taking into account the variation in the shape of different school attendance zones, this study us es the density of students who reside within study zone as the measure of dependent variable ( Figure 3 3 ) Measure of dependent variable= the number of students within study zone/ the acreage of study zone. Independent Variables The study utilized three ch aracteristics of urban form as independent variables: school attendance zone geometry street connectivity and residential density Based on the literature review, three of these characteristics are measured by quantitative indicators illustrated in the fo llowing sections. School Attendance Zone Geometry I ndicators This research used three geographic characteristics to calculate the indicators of school attendance zone location : school location point, geographic mean center of geographic mean center of study zone. Based on these three points, the following indicators are developed : ( Figure 3 4 ) Indicator 1 : The Euclidean distance between school location point and geographic (DSS P) DSSP s hows the geographic relationship between school site and students residing within study zone.

PAGE 33

33 Indicator 2 : The Euclidean distance between school location point and geographic mean center of study zone (DSSZ) DSSZ shows the centralization of school site i n the relation to study zone. Indicator 3 : The Euclidean distance between location points and geographic mean center of study zone (DSPSZ) DSPSZ shows the shows the geographic relationship between study zone and studen ts residing within study zone. Street Connectivity Indicators To measure street connectivity three walkability indicators are used: (1) Street Density (the total number of linear miles of street per square mile); (2) Intersection Density (the total numbe r of intersections per square mile); and (3) Pedestrian Route Directness (the ratio of the network distance to the Euclidean (straight line) distance between two points). Indicator 1 : Street Density is used as an indicator to provide a quantitative measur e of the number of available pathways (available miles of streets) a child might be able to travel between home and school. This study will utilize the street network as the surrogate pathway for measuring these connections. These pathways are identified b y a calculation measuring the distance along the centerline of each street. A greater street density within the proximity of a school site presumably means that there are more roads

PAGE 34

34 available thus increasing the potential for a child to walk or bicycle to school. Street density is a calculation derived from dividing the total number of street miles by the total area (square miles) within a specified range Indicator 2 : Intersection D ensity is correlated to street patterns in that intersections rely on the p resence of street networks. Intersection density is also used to provide information of connectivity by illustrating nodes of intersection (junction between streets and roadways). A higher degree of intersection density presumably indicates higher levels o f connectivity, thus providing environments that support a greater potential for walkability. Intersection density is calculated by dividing the total number of intersections by the linear miles of street within a specified range. Again this calculation wi ll be conducted using the half and one mile analysis zones as boundaries. Indicator 3 : Pedestrian Route Directness (PRD) is a value representing the directness of travel between two points. More specifically, it is a ratio between the straight line distan ce of two points divided by the network distance between those same two points (an origin and a destination). In this study, PRD is measured using schools as destination points and using individual residential dwelling units as origin points. PRD is then t allied for each analysis zone using the average network and straight line distances between residential dwelling units to the corresponding school they are zoned. Although

PAGE 35

35 a formative indicator of connectivity, PRD is not always representative of walkable environments (Dill, 2004). Residential Density Indicators Residential density is also measured, using three indicators: (1) Gross Residential Density (total number of dwelling units per gross acre); (2) Net Residential Density (total number of dwelling un its per residential parcel acre). (3) Students Residential Ratio. Residential density information used for this study was provided from data prepared in previous research (Steiner, et al., 2008; Bejleri, et al., 2008). Three of them are calculated by assig ning dwelling unit counts to residential parcels within each study zone. This information was created using land use codes acquired by the Department of Revenue, dwelling unit counts for multifamily parcels using data from the 2006 American Community Surve y, and additional information provided by the Bureau of Economic and Business Research, and county apartment complex records (Steiner, et al., 2008; Bejleri, et al., 2008). Indicator 1 : Gross Residential Density provides a value representing the total numb er of dwelling units per gross acre. Gross acreage includes all land use designations (e.g.: residential, commercial, industrial, etc. ). Indicator 2 : Net Residential Density provides a value representing the total number of dwelling units per residential p arcel acre. The residential parcel acre is the

PAGE 36

36 acreage containing only residential land uses, excluding such land uses as mentioned in the calculation of gross residential densities (e.g.: commercial). Net residential density will be compared to acceptable standards of net residential density to determine levels of walkability within the established half and one mile analysis zones. Indicator 3 : S tudent R esidential R atio is a value representing the number of students per residential unit. It is calculated as students number within analysis zo ne divided by total residen tial units within analysis zone. Compared to the gross residential density and net residential density, student residential ratio reveals the generation of students population in study zone. A higher value indicates a greater children s potential walkability. Regression Model Regression analysis allows you to model, examine, and explore spatial relationships, and can help explain the factors behind observed spatial patterns. Regression analys is is also used for prediction ( ArcGIS Desk Help 9. 1 ). O rdinary Least Squares(OLS) is the best known of all regression techniques. It is also the proper starting point for all spatial regression analyses. It provides a global model of the variable or pro cess you are trying to understand or predict (early death/rainfall) .(ArcGIS Desk Help9. 1 ) it creates a single regression equation to represent

PAGE 37

37 that process. In this study, a single regression equation is created to represent how urban form factors correlat e with the school children s walkability. A ccording to ArcGIS Desk Help 9. 1 a mathematic al formula of regression model could be expressed as Figure 3 5 Y regression coefficients In this research, Y is students density within study zone, the X s are all indicators representing urban form factors described above, and the are regression coefficient produced by OLS Thus, the equation applying to this study could be expressed as: Students Density within Study Z one = 0+ 1(DSSP) + 2(DSSZ) + 3(DSPSZ) + 4(Street Density) + 5(Intersection Density) + 6(PRD) + 7(Gross Resid ential Density) + 8(Net Residential Density)+ 9(Student Residential Ratio)+ Applying this equation to the Ordinary Least Squares Tools in ArcGIS ( Figure 3 6 ) will produce the results demonstrated in the Chapter 4. The interpretation of OLS re sults is re presented on APPENDIX Summary T he methodology described in this chapter provides a quantitative way to measure home. Dependent variable is value d by students density within study zone, assuming

PAGE 38

38 that all children living within study zone could walk and bicycle to school. The total of nine indicators created in this chapter represents three urban form factors: school attendance zone geometry, street connectivity and resi dential density OLS r egression m school and urban form factors, which helps to understand the effect of urban form on childr

PAGE 39

39 Figure 3 1 Illustr ation of School Attendance Zone

PAGE 40

40 Figure 3 2 Illustration of Study Zone

PAGE 41

41 Figure 3 3 Illustration of Dependent Variable

PAGE 42

42 Figure 3 4 Illustration of School Attendance Zone Geometry

PAGE 43

43 Figure 3 5 M athematical Formula of Regression Model (ArcGIS Desktop Help 9.1) Figure 3 6 O rdinary Least Squares Tools in ArcGIS

PAGE 44

44 C HAPTER 4 RESULTS AND FINDINGS T his chapter demonstrates the results and find ing s from the regression model based on the me thodology described in C hapter 3 The whole chapter consists of 3 sections T he first section presents the results for the half mil e study zone. The second s ection presents the results f or the one mile study zone. I n last section, the results are summarized based on the interpretation of variables and indicators described in the methodology, and compared between two different scales o f study zone. Results from Regression Model of Half mile Study Zone Figure 4 1 shows those indicators of half mile study zone containing one dependent variable and nine explanatory variable indicators from all 40 school study zones Pre running of OLS reg ression model ( Figure 4 2 ) reveals that explanatory variable redundancy exists between street density and intersection density.( value of VIF > 7.5, see the interpretation of OLS results at APPENDIX A ) Therefore intersection density is removed to refine the m odel. After refining the model, th e results from OLS regression model shows in Figure 4 3 The Adjusted R Squared value of 0 .71309 indicates that regression model in this research explains approximately 71.3% of the variation in the dependent vari able, or

PAGE 45

45 said another way: the model of half mile study zone tells approximately 71.3 % of the school children s walkability "story". Probability shows that both Gross Residential D ensity and Student Residential Ratio are statistically significant. Coefficient s hows that both Gross Residential D ensity and Student Residential Ratio are positively correlated to dependent variable ( school children s potential walkability). (Table 4 1.) Results from Regression Model of One mile Study Zone Figure 4 4 shows d ata of One mile study zone containing one dependent variable and nine explanatory variables from all 40 school study zones Pre running of OLS regression model gives rise to the same issue of explanatory variable redundancy as half mile model does ( Figure 4 5 ) Ther efore i ntersection Density is removed for the purpose of refining the model. After refining the model, the results from OLS regression model shows in Figure 4 6 The Adjusted R Squared value of 0 .757944 indicates that the model of one mile study zone expla ins approximately 75.79% of school children s walkability in the one mile study zone. Probability values present that four independent variables are statistically significant: PRD, Gross Residential Density, Street Density and Student Residential Ratio Ba sed on Coefficient values PRD is negatively correlated to dependent variable while other three independent variables are positively correlated to dependent variable ( Table 4 2)

PAGE 46

46 Summary I n both models, the majority of findings from OLS reveal that there ar e correlations between dependent variable and some independent variables. In half mile model, it is apparent that Gross Residential Density and Student Residential Ratio are statistically correlated to school children s potential walkability. Unlike half m ile model, one mile model presents that PRD and Street Density are also statistically correlated to dependent variable except both Gross Residential Density and Student Residential Ratio It is worthy noted that neither of model produces the statistic sign ificance on school attendance zone location indicators. Besides, the result that explanatory variable redundancy exists between street density and intersection density implies that these two indicators explain the same characteristics of street connectivit y. Although two model s returned different results, the value of Adjusted R Squares, 0.713019 and 0.757944 respectively, has proved that the selections of explanatory variables and indicators are effective. F indings did not reveal any contradictions making it reasonable to explain the results. The next chapter provide s discussion of the findings presented here especially interpreting the difference in the results from both models. Chapter 5 also i ncludes the limitations of this study and provides recommen dation s for future research

PAGE 47

47 Figure 4 1 Indicators of Half mile Model NE T_DU_DNS: Net Residential D ensity GRS_DU_DN: Gross Residential Density STRLN2NTWD: PRD. INT_DNSTY: Intersection D ensity STU_DNSTY: Students density within study zone. RD_DNS: Street Density. STU_GEN: S tudent Residential Ratio. DIST _1: Euclidean di stance between school location point and the geographic center point of study zone DIST_2: Euclidean distance between the mean center of zone and the geographic center of study zone DIST_3: Euclidean distance between the mean center point of zone and school location points

PAGE 48

48 Figure 4 2 Pre running Results from Half mile Model

PAGE 49

49 Figure 4 3 Results fr om Half mile Model

PAGE 50

50 Table 4 1 Summary of R esults from Half mile Model Variables Coefficient Probability (95%) Net Residential D ensity 7.206927 0.520606 PRD 228.735 0.068123 Gross Residential Density 60.4862 0.039739 Street Density 5.972002 0.112063 S tudent Residential Ratio 901.5548 3.92E 09 DSSP 0.040154 0.483574 DSPSZ 0.08607 0.239525 DSSZ 0.00473 0.94136

PAGE 51

51 Figure 4 4 Indicators of One mile Model NE T_DU_DNS: Net Residential D ensity GRS_DU_DN: Gross Residential Density STRLN2NTWD: PRD. INT_DNSTY: Intersection D ensity STU_DNSTY: Students density within study zone. RD_DNS: Street Density. STU_GEN: S tudent Residential Ratio. DIST _1: Euclidean distance between school location point and the geographic center point of study zone DIST_2: Euclidean distance between the mean c enter of and the geographic center of study zone DIST_3: Euclidean distance between the mean center point of zone and school location points

PAGE 52

52 Figure 4 5 Pre running Results from One mile Model

PAGE 53

53 Figure 4 6 Results from One mile Model

PAGE 54

54 Variables Coefficient Probability (95%) Net Residential Density. 3.884205 0.606127 PRD 306.409 0.012255 Gross Residential Densit y 59.40819 0.005127 Street Density 8.619387 0.019781 Student Residential Ratio 860.75 3.31E 08 DSSP 0.027695 0.16508 DSPSZ 0.00516 0.838033 DSSZ 0.01224 0.547367 Table 4 2. Summary of R esults from One mile Model

PAGE 55

55 CHAPTER 5 D ISCUSSIONS I n th is cha pter of this research, the author will discuss how the findings from regression models support the research question of how the urban form factors is correlated to children s walkability. Discussions and Conclusions The study chose to use school attendance zone location, street connectivity and residential density as urban form factors to investigate the children s walkability around public s chool sites. Based on the prior literature, it is confident to say that these three variables are best representative of children s potential walkability. However, the findings from both models of half mile and one mile study zone revealed that school attendance zone geometry indicators are not associated with school children s walkability, which might impl y geometry of school attendance zone does not affect children s walkability within half mile and one mile sheds. This finding does not support the studies ( S teiner,et al, 2008)claiming that school location and siting play a key role in determining the children s walkabi lity I n the results from the model of half mile st udy zone, the fact that both statistic al significant indicators respond to dependent variable positively indicates that higher residential density and percentage of residential units that have school child ren could produce higher children s walkability. T his findings support the discussion of former

PAGE 56

56 literature about residential density is positively related to children s walkability. However, it is unexpected that street connectivity did not show any correl ation to the dependent variable in this case. In the results from model of one mile study zone, the indicators of residential density responded to the dependent variable the same as half mile model did. Apart from this, PRD presents the negative correlatio n to the dependent variable while street density presents the positive one, indicating that the better street connectivity shows the higher children s walkability. T he results support the studies that found both residential density and street connectivity ha ve positive correlations with children s walkability. It is worthy noted that Students Residential Ratio and PRD have stronger effect on dependent variable based on the values of coefficient The fact that variables present the different relations in two models suggests that children s walkability to school i s affected by the distance to school. For children who live close enough to school ( within half mile in this study), street connectivity is not a key concern for walking and biking to school. Street c onnectivity is taken into account by children who live within one mile from school However, this implication casts a contradiction with the findings that school attendance zone geometry indicators are not associated with school children s walkability. One of reason for this contrary could be that

PAGE 57

57 OLS regression model is not the best tool to test geographic relationship. Further study needs to explore the geometry of school attendance zone. Nonetheless, with the confidence, the conclusion could be made that there is a correlation between urban form factors and children s walkability to school under certain circumstances PRD, Gross Residential Density, Students Residential Ratio and Street Density present the statistical correlation to children s potential w alkability. T he correlations present a difference in half mile and one mile models. T he findings that Students Residential Ratio and PRD have stronger effect on children s walkability demonstrate that the improvements on residential density and street conn ectivity could facilitate the children s walk or bicycle to school, which in turn may decrease the childhood obesity by increasing the children s daily physical activity. Limitations and Future Studies L imitations T his research study is conducted based up on data from 40 elementary and middle schools. S ample of 40 schools are too small to tell the truth about children s walkability. In addition, the selected schools are not geographically weighted, which in part explains no significance on school attendance zone geometry indicators S econdly t here is a difference of travel model between middle school students and elementary school students, which makes it different in the acceptable walk

PAGE 58

58 distance.( middle school students who ride bicycle can reach a distance larger than 1 mile) Besides, some research also shows urban form factors have different impact on children at different age group and different gender. The finding s cannot tell which students act ually do walk or bike to school, because this research evalu ated the children s potential walkability, assuming that all children within study zone could walk to school. W hether a chi ld would walk or bike to school could be determined by other social, economic or family factors. Although a survey was conducted on t hese 40 schools to collect the number of children walking or biking to school during three surveyed days, it cannot help to tell the exact child who actually walks or bikes either. Future Studies D ue to the limitation of this research, future study could c hoose the larger sample that locates close to each other geographically, using Geographically Weighted Regression (GWR) after OLS to further explore the geographical relation between children s walkability and urban form factors. Given that the actual trav el model choice of children could be determined by a number of social, economic, family and physical environment factors, future research could add dummy variables that representing other factors other than urban form to refine the study model, such as gen der, family income and ethnic

PAGE 59

59 C HAPTER 6 C ONCLUSIONS This research testified that there is a correlation between urban form factors and children s walkability under certain circumstances by using OLS regression model tool in ArcGIS. T he findings support the prior theory about the impact of street connectivity and residential density on children s walkability to school. Although the study is conducted upon a relatively small sample of forty public elementary and middle schools, the findings could help to i ncrease children s physical activity by improving physical e nvironment related to urban for m factors, especially street connectivity and residential density, considering the findings that PRD and Students Residential Ratio have stronger relations to childr en s potential walkability. Due to the limitation on availability of data and research model, the findings cannot tell the actual children s walkability

PAGE 60

60 A PPENDIX INTERPRETATION OF OLS RESULTS You will need to provide an input feature class with a unique I D field, the dependent variable you want to model/explain, and all of the explanatory variables. You will also need to provide a pathname for the output feature class, and optionally, pathnames for the coefficient and diagnostic output tables. As the OLS t ool runs, statistical results are printed to the screen. (B) Examine the statistical report using the numbered steps described below under "Dissecting the Statistical Report": Dissecting the Statistical Report

PAGE 61

61 Assess model performance. Both the Multip le R Squared and Adjusted R Squared values are measures of model performance. Possible values range from 0.0 to 1.0. The Adjusted R Squared value is always a bit lower than the Multiple R Squared value because it reflects model complexity (the number of va riables) as it relates to the data, and consequently is a more accurate measure of model performance. Adding an additional explanatory variable to the model will likely increase the Multiple R Squared value, but decrease the Adjusted R Squared value. Suppo se you are creating a regression model of residential burglary (the number of residential burglaries associated with each census block is your dependent variable, y ). An Adjusted R Squared value of 0.84 would indicate that your model (your explanatory vari ables modeled using linear regression) explains approximately 84% of the variation in the dependent variable, or said another way: your model tells approximately 84% of the residential burglary "story". Assess each explanatory variable in the model: Co efficient, Probability or Robust Probability, and Variance Inflation Factor (VIF). The coefficient for each explanatory variable reflects both the strength and type of relationship the explanatory variable has to

PAGE 62

62 the dependent variable. When the sign assoc iated with the coefficient is negative, the relationship is negative (e.g., the larger the distance from the urban core, the smaller the number of residential burglaries). When the sign is positive, the relationship is positive (e.g., the larger the popula tion, the larger the number of residential burglaries). Coefficients are given in the same units as their associated explanatory variables (a coefficient of 0.005 associated with a variable representing population counts may be interpreted as 0.005 people) The coefficient reflects the expected change in the dependent variable for every 1 unit change in the associated explanatory variable, holding all other variables constant (e.g., a 0.005 increase in residential burglary is expected for each additional pe rson in the census block, holding all other explanatory variables constant). The T test is used to assess whether or not an explanatory variable is statistically significant. The null hypothesis is that the coefficient is, for all intents and purposes, equ al to zero (and consequently is NOT helping the model). When the probability or robust probability is very small, the chance of the coefficient being essentially zero is also small. If the Koenker test (see below) is statistically significant, use the robu st probabilities to assess explanatory variable statistical significance. Statistically significant probabilities have an asterisk "*" next to them. An explanatory variable associated with a statistically significant coefficient is important to the regress ion model if theory/common sense supports a valid relationship with the dependent variable,

PAGE 63

63 if the relationship being modeled is primarily linear, and if the variable is not redundant to any other explanatory variables in the model. The variance inflation factor (VIF) measures redundancy among explanatory variables. As a rule of thumb, explanatory variables associated with VIF values larger than about 7.5 should be removed (one by one) from the regression model. If, for example, you have a population variab le (the number of people) and an employment variable (the number of employed persons) in your regression model, you will likely find them to be associated with large VIF values indicating that both of these variables are telling the same "story"; one of th em should be removed from your model. Assess model significance. Both the Joint F Statistic and Joint Wald Statistic are measures of overall model statistical significance. The Joint F Statistic is trustworthy only when the Koenker (BP) statistic (see below) is not statistically significant. If the Koenker

PAGE 64

64 (BP) statistic is significant you should consult the Joint Wald Statistic to determine overall model significance. The null hypothesis for both of these tests is that the explanatory variables in the model are not effective. For a 95% confidence level, a p value (probability) smaller than 0.05 indicates a statistically significant model. Assess Stationarity. The Koenker (BP) Statistic (Koenker's studentized Bruesch Pagan statistic) is a test to det ermine if the explanatory variables in the model have a consistent relationship to the dependent variable (what you are trying to predict/understand) both in geographic space and in data space. When the model is consistent in geographic space, the spatial processes represented by the explanatory variables behave the same everywhere in the study area (the processes are stationary). When the model is consistent in data space, the variation in the relationship between predicted values and each explanatory vari able does not change with changes in

PAGE 65

65 explanatory variable magnitudes (there is no heteroscedasticity in the model). Suppose you want to predict crime and one of your explanatory variables in income. The model would have problematic heteroscedasticity if th e predictions were more accurate for locations with small median incomes, than they were for locations with large median incomes. The null hypothesis for this test is that the model is stationary. For a 95% confidence level, a p value (probability) smaller than 0.05 indicates statistically significant heteroscedasticity and/or non stationarity. When results from this test are statistically significant, consult the robust coefficient standard errors and probabilities to assess the effectiveness of each expla natory variable. Regression models with statistically significant non stationarity are especially good candidates for GWR analysis.

PAGE 66

66 Assess model bias. The Jarque Bera statistic indicates whether or not the residuals (the observed/known dependent variab le values minus the predicted/estimated values) are normally distributed. The null hypothesis for this test is that the residuals are normally distributed and so if you were to construct a histogram of those residuals, they would resemble the classic bell curve, or Gaussian distribution. When the p value (probability) for this test is small (is smaller than 0.05 for a 95% confidence level, for example), the residuals are not normally distributed, indicating model misspecification (a key variable is missing from the model). Results from a misspecified OLS model are not trustworthy. Assess residual spatial autocorrelation. Always run the Spatial Autocorrelation ( Moran's I) tool on the regression residuals to ensure they are spatially random. Statistically significant clustering of high and/or low residuals (model under and over predictions) indicates a key variable is missing from the model (misspecification). OL S results cannot be trusted when the model is misspecified.

PAGE 67

67 Finally, review the section titled "How Regression Models Go Bad" in the Regression Analysis Basics document as a check that your OLS regression model is properly specified. Notice, too, that there is a section titled "Notes on Interpretation" at the end of the OLS statistical report to help you remember the purpose of each statistical test.

PAGE 68

68 LIST OF REFERENCE S Atash, F (1994). Redesigning suburbia for walking and transit: Emerging concepts. Journal of Urban Planning and Development, 120(1), 48 57. Audirac Ivonne (1999). Stated Preference for Pedestrian Proximity: An Assessment of New Urbanist Sense of Community. Journ al of Planning Education and Research. September 1999 vol. 19 no. 1 53 66 ArcGIS Desktop Help 9.1 Bejleri, I, Steiner, R.L., Provost, R.E., Fischman, A., & Arafat, A.A. (2008). Understanding lity to Walk and Bicycle to School: A Case Study of Two Tampa Bay Counties. Unpublished document, University of Florida. Campbell, E.K. & Wang, Q. (2008). Pupil Transportation: Factors Affecting Mode Choice and the Amount of Parent driven Trips to School, University of Vermont, Department of Community Development and Applied Economics, No: 09 3753. Center for Design, Methods, and Analysis, US Government Accountability Office, Washington, DC 20548, USA. Centers for Disease Control and Prevention (2005). B arriers to Children Walking to or from School --United States, 2004. Retrieved August 7,2011, from http://www.cdc.gov/mmwr/preview/mmwrhtml/mm5438a2.htm Centers for Disease Control and Prevention (2011). Prevalence of Obesity Among Children and Adolescents: United States, Trends 1963 1965 Through 2007 2008 Retrieved August 7, 2011, from http:/ /www.cdc.gov/nchs/data/hestat/obesity_child_07_08/obesity_child_07_08.ht m Cervero, R. & Kockelman, K. (1997). Travel Demand and the 3Ds: Density, Diversity, and Design. Transportation Research Design Part D, 2(3), 199 219. City of Raleigh, North Carolin a (2008). Design Guidelines for Pedestrian Friendly Neighborhood Schools. Retrieved June 16, 2008, from

PAGE 69

69 http://www.raleighnc.gov/publicat ions/Planning/Guides,_Handbooks_and_Manuals /School_Design_Guidelines.pdf Diss., University of Adelaide, 2005 Dellinger, A. & Staunton, C. (2002). Barriers to Chil dren Walking and Biking to School in the United States. 1999. Morbidity and Mortality Weekly Report, 51(32), 701 704. Dill, J. (2004). Measuring Network Connectivity for Bicycling and Walking. Presented at the Annual Transportation Research Board meeting, Washington D.C. Duany, Andrs and Plater Zyberk, Elizabeth. 2001. The Lexicon of the New Urbanism. Duany Plater Zyberk & Company. Travel Analysis of Factors Affecting Mo Forsyth, A. (2003). Measuring Density: Working Definitions for Residential Density and Building Intensity. Design Center for American Urban Landscape, College of Architecture and Landscape Architecture, University of Minnesota, Design Brief, No 8. Frank, L., Kerr, J., Chapman, J., & Sallis, J. (2007). Urban Form Relationships with Walk Trip Frequency and Distance Among Youth. American Journal of Health Promotion, 21(4Supp), 305 311. Frank, L., Sallis, J., Saelens, B., Bachman, W., & Washbrook K., (2005). Travel Behavior, Environmental, & Health Impacts of Community Design & Transportation Investment, LUTAQH: A Study of Land Use, Transportation, Air Quality, and Health, King County, Washington Georgina S. A. Trapp, Billie Giles Corti, Hayley E. Christian, Max Bulsara, Anna F. Timperio, Gavin R. McCormack and 11 October 2011 Karen P. Villanueva,2010. Behav published online 11 October 2011

PAGE 70

70 Gomez, J.E., Johnson, B.A., Selva, M., Sallis, J.F., 2004. Violent crime and outdoor physical activity among inner city youth. Prev. Med. 39, 876 881. Hahn, R. A., S.M. Teuesch, R.B. Rothenberg, et. al. (1998). "Ex cess deaths from nine chronic diseases in the United States, 1986." Journal of the American Medical Association 264(20): 2554 59. International City/County Management Association (2008). ICMA IQ Report Special Edition, Item No. E 43527 (40). Jago,R.,Baran owski, ., Baranowski, J.C.,(2006). Observed, GIS, and self reported environmental features and adolescent physical activity. Am. J. Health. Promot. 20, 422 428. Kerr Jacqueline Frank Lawrence, Sallis James F. Chapman Jim(2007). Urban form correlates o f pedestrian travel in youth: Differences by gender, race ethnicity and household attributes. Transportation Research Part D 12 (2007) 177 182 Kerr, J., Rosenberg, D., Sallis, J.F., Saelens, B.E., Frank, L.D., Conway, T.L., 2006. Active commuting to scho ol: associations with environment and parental concerns. Med. Sci. Sports Exerc. 38, 787 794. environment influence children's physical activity? A review of the liter International Journal of Behavioral Nutrition and Physical Activity 2006, 5868 3 19 Kushi, L. H., R.M. Fee, A.R. Folsom, et al. (1997). "Physical activity and mortality in postmenopausal women." Journal of the American Medical Association 277: 12 87 92. Lee, C. D., S.N. Blair, A.S. Jackson (1999). "Cardiorespiratory fitness, body composition, and all cause and cardiovascular disease mortality in men." American Journal of Clinical Nutrition 69(3): 373 80. McGinnis, J. M., W.H. Foege (1993). "Actua l causes of death in the United States." Journal of the American Medical Association 270(18): 207 12. Norman GJ, Nutter SK, Ryan S. Community design and access to recreational facilities correlates of adolescent physical activity and body mass index. J Ph ys Act Health. 2006;3(suppl 1):S118 S128.

PAGE 71

71 Prentice, A., Susan Jebb (1995). "Obesity in Britain: Gluttony or sloth?" British Medical Journal 311: 437 39. Roemmich, J.N., Epstein, L.H., Raja, S., Yin, L., Robinson, J., Winiewicz, D., 2006. Association of a ccess to parks and recreational facilities with the physical activity of young children. Prev. Med. 43, 437 441. Romero, A.J., 2005. Low income neighborhood barriers and resources for adolescents' physical activity. J. Adolesc. Health 36, 253 259. Saelen s, B.E., Sallis J.F., Black, J.B., & Chen, D. (2003). Neighborhood Based Differences in Physical Activity: An Environment Scale Evaluation. American Journal of Public Health, 93(9), 1552 1558. Schlossberg, M., Greene, J., Phillips, P.P., Johnson, B., & Pa rker, B. (2006). School Trips: Effects of Urban Form and Distance on Travel Mode. Journal of the American Planning Association, 72(3), 337 346. Schmucker Jeffrey M. How Does Urban Form Impact the Potential for Children to Walk and Bicycle to School: A Cas e Study of Orange and Seminole Counties in Central Florida. Master of Arts Thesis, University of Florida,2009 Steiner, R.L., Bejleri, I., Wheelock, J.L., Boles, G., Cahill, M., & Perez, B.O. (2008). Understanding and Mapping Institutional Impediments to Walking and Bicycling to School: A Case Study of Hillsborough County. Transportation Research Record, 2074, 3 11. Timperio A, Ball K, Salmon J, et al. Personal, family, social, and environmental correlates of active commuting to school. Am J Prev Med. 200 6;30:45 5 1. U.S. Department of Health and Human Services (2002). Physical Activity Fundamental to Preventing Disease, Office of the Assistant Secretary for Planning and Evaluation: 1 19. US Census Bureau; 2008 American Community Survey, 1 Year Estimates; generated using American Factfinder; Retrieved October 7, 2011, from http://www.census.gov/newsroom/releases/archives/facts_for_features_speci al_ed itions/cb09 ff14.html

PAGE 72

72 Wei, M., J.B. Kampert, C.E. Barlow, et al., (1999). "Relationship between low cardiorespiratory fitness and mortality in normal weight, overweight, and obese men." Journal of the American Medical Association 282(16): 1547 53.

PAGE 73

73 BIOGRAPHICAL SKETCH Yang Wencui was born October 1984 in Chuxiong Yunnan, China. Before attending graduate school at University of Florida, Wencui received his Bachelor of Science degree in u rban and r egional p lanning at Nanjing University, China. After r eceiving his degree of Master of Art s in u rban and r egional p lanning, Wencui will be pursuing his degree of D octor of P hilosophy in u rban and r egional p lanning at the University of Florida.