Citation
The Relationship of Time-of-Day Travel and Built Environment in Southeast Florida

Material Information

Title:
The Relationship of Time-of-Day Travel and Built Environment in Southeast Florida Incorporating Parking Characteristics in Downtown Miami and Fort Lauderdale
Creator:
Rachmat, Shanty Yulianti
Place of Publication:
[Gainesville, Fla.]
Florida
Publisher:
University of Florida
Publication Date:
Language:
english
Physical Description:
1 online resource (130 p.)

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Design, Construction, and Planning
Design, Construction and Planning
Committee Chair:
STEINER,RUTH LORRAINE
Committee Co-Chair:
ZWICK,PAUL D
Committee Members:
BLANCO,ANDRES G
SRINIVASAN,SIVARAMAKRISHNAN
Graduation Date:
5/3/2014

Subjects

Subjects / Keywords:
City centers ( jstor )
Employment ( jstor )
Land use ( jstor )
Parking ( jstor )
Shopping trips ( jstor )
Transportation ( jstor )
Travel ( jstor )
Travel behavior ( jstor )
Travelers ( jstor )
Work trips ( jstor )
Design, Construction and Planning -- Dissertations, Academic -- UF
built -- choice -- downtown -- environment -- florida -- time -- travel
City of Miami ( local )
Genre:
bibliography ( marcgt )
theses ( marcgt )
government publication (state, provincial, terriorial, dependent) ( marcgt )
born-digital ( sobekcm )
Electronic Thesis or Dissertation
Design, Construction, and Planning thesis, Ph.D.

Notes

Abstract:
Urban planners, who are responsible for land use policies in urban areas, seek the capability of built environment factors to influence travelers’ behaviors.  Prior research has given less attention to the time-of-day dimension than to other dimensions.  Time-of-day choice study advances the travel demand modeling in several ways. First, policy makers need detailed information on peak hours. Second, socio-economic changes influence the travel model in specific periods. Third, it helps in the evaluation of travel demand management strategies, i.e. the ways to modify travelers’ behavior by maximizing current resources. Thus, this study seeks to understand the effects of individual built environment factors and the combination of those factors on time-of-day travel choices.  By combining these factors,this study reduces the impact of collinearity of those factors in statistical procedure.  In the Central Business District (CBD) areas, this study examines the potential of time-of-day choice to influence time-sensitive travel demand management strategies, such as time-restriction and time-variable parking pricing.  The National Household Travel Survey (NHTS) 2009 Florida Department of Transportation (FDOT) add-on file is the main database used for this analysis.  Three counties, Palm Beach,Miami-Dade, and Broward, are used as a regional case study, with two CBDs—Fort Lauderdale and Miami—as specific cases. The parking inventory of these CBDs from a previous project at the University of Florida is used for this study. Using the ordered response model, this study seeks the significances of various variables, including socio-economic, trip characteristic, and built environment. This study’s results shows that built environment factors have a greater impact on time-of-day choice when considered in tandem.  Thus, the purpose of the trip—whether travelers are commuting to work or simply shopping—demonstrates the differences between choices during given periods of time. The relationship between time-of-day choice and a trip’s purpose occurs in the CBD level.  The percentage of parking spaces with time-restriction in the CBD has an effect on the time-of-day choice.  These results suggest that integration of built environment factors is important in creating land use policies to influence travelers’ behavior.  Specifically, the model recommends that local governments incorporate time-sensitive parking policies in CBD re-development. ( en )
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.
Thesis:
Thesis (Ph.D.)--University of Florida, 2014.
Local:
Adviser: STEINER,RUTH LORRAINE.
Local:
Co-adviser: ZWICK,PAUL D.
Electronic Access:
RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2016-05-31
Statement of Responsibility:
by Shanty Yulianti Rachmat.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Embargo Date:
5/31/2016
Resource Identifier:
907379469 ( OCLC )
Classification:
LD1780 2014 ( lcc )

Downloads

This item has the following downloads:


Full Text

PAGE 1

THE RELATIONSHIP OF TIME OF DAY TRAVEL AND BUILT ENVIRONMENT IN SOUTHEAST FLORIDA : INCORPORATING PARKING CHARACTERISTICS IN DOWNTOWN MIAMI AND FORT LAUDERDALE By SHANTY YULIANTI RACHMAT A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2014

PAGE 2

201 4 Shanty Yulianti Rachmat

PAGE 3

To my father and mother, who give me endless love, my husband, Deny Dwiantoro, who gives love, spirit, and effort in our challenging student stage life, and to my little Denisha, I borrow your childhood happy times to struggle together here

PAGE 4

4 ACKNOWLEDGMENTS Bismillahirrohmanirrohim, my first and best gratitude is always for Allah SWT because only by His guidance and blessing, I can be in this point of my life. Furthermore, I would like have my gratitude to my chair, Dr. Ruth Steiner. Once she accepted me as her student, she has given me her support and guidance during my PhD studies Furthermore, I thank to Dr. Paul Zwick, Dr. Andres Blanco, and Dr. Siva Srinivasan for their clear advice for the operationalization of variables in my dissertation. I acknowledge Fulbright Foundation and The American Indonesian Cultural & Educational Foundation (AICEF) for funding my study in the US. Then, I thank to Debra Anderson, Carlos Maetzu, Jean McCall and the Fulbright family that helped us in our early adjustment to the University of Florida. Also, I thank Ruoying, Nagendra, and Ben for helping with the modeli ng. Moreover, I thank for my editors: Melody Schiaffino and PhD moms, Nureet Carmel, Wayne Foster, Maya Wospakrik, Hadyan Ramadhan, and Cininta. My special thanks for my supervisors and colleagues at Bandung Institute of Technology (ITB), especially to: Dr. Kusbiantoro, Dr. Ibnu Syabri, Dr. Krishna Pribadi, and Dr. Tommy Firman. Also, I appreciate all the support from Dean Chris Silver. I thank to my Indonesian fellows: Aan and James Morgan,Uli and Hector Dones, Desi Foster, and Ridwan Sutriadi. Also, I appreciate all helps from Desiree Kipuw. Most importantly, I extend my gratitude to my family, my father and mother inlaw for their love and support. Also, I thank to my brothers, sisters, and all my extended family.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................. 4 LIST OF TABLES ............................................................................................................ 8 LIST OF FIGURES .......................................................................................................... 9 LIST OF ABBREVIATIONS ........................................................................................... 11 ABSTRACT ................................................................................................................... 12 CHAPTER 1 INTRODUCTI ON .................................................................................................... 14 Statement of the Problem ....................................................................................... 14 Purpose of the Research ........................................................................................ 16 Significance of the Research .................................................................................. 18 Dissertation Outline ................................................................................................ 20 2 TIME OF DAY CHOICE AND BUILT ENVIRONMENT ........................................... 22 The Relationship of Travel Behavior and Built Environment ................................... 23 Time of Day Choice ................................................................................................ 27 The Import ance of Timeof Day Choice Studies ............................................... 27 Survey approaches .................................................................................... 30 Methods in timeof day studies .................................................................. 31 The Relationship of Timeof Day and Built Environment .................................. 32 Built Environment Dimensions ................................................................................ 35 Problems on Built Environment Studies .................................................................. 38 Built Environment Dimensions .......................................................................... 38 Self selection Bias ............................................................................................ 39 Unit of Analysis and Location ........................................................................... 40 Variable Selection on the Relationship of Timeof Day and Built Environment ....... 41 Built Environment Factors ................................................................................ 41 Socio Economy Variables ................................................................................ 43 Trip Characteristics Variables ........................................................................... 46 Summary of Time Of Day Choice and Built Environment ....................................... 47 3 PARKING IN DOWNTOWN AS BUILT ENVIRONMENT FACTOR ........................ 48 Parking Demand Management in Downtowns ........................................................ 49 Downtown Redevelopment and Parking Demand Management ...................... 49 Peak Hour Parking Strategies .......................................................................... 51 Past Studies on Timeof Day and Parking Focus ............................................. 51

PAGE 6

6 Variables Selection ................................................................................................. 53 Summary of Parking Management in Downtown .................................................... 54 4 METHODOLOGY AND DATA ................................................................................ 55 Research Design .................................................................................................... 55 Regional Level ........................................................................................................ 55 Regional Hypothesis ........................................................................................ 55 Study Area ........................................................................................................ 56 Data Sources .................................................................................................... 57 Unit of Analysis ................................................................................................. 58 The Data Processing ........................................................................................ 59 Data Set ........................................................................................................... 61 Downtown Level ..................................................................................................... 62 Downtown Hypothesis ...................................................................................... 62 CBD Study Area ............................................................................................... 63 Data and Unit of Analysis for CBD Level .......................................................... 64 The Model Framework ............................................................................................ 65 Ordered Choice Model ..................................................................................... 66 LUCIS and Suitability Model ............................................................................. 68 5 ANALYSIS AND RESULTS .................................................................................... 70 Regional Level ........................................................................................................ 70 Descriptive Statistics ........................................................................................ 70 Dependent variable .................................................................................... 70 Descriptive of socioeconomic and demographic variables ....................... 72 Descriptive of trip characteristic variables .................................................. 77 Descriptive of built environment variables .................................................. 79 Regional Model and the Results ....................................................................... 85 Socio economy variables ........................................................................... 87 Trip characteristics variables ...................................................................... 88 Built environment variables ........................................................................ 88 Downtown Level ..................................................................................................... 89 Descriptive Analysis of Parking Variables ........................................................ 89 CBD Model and the Results ............................................................................. 95 Hypotheses Testing ................................................................................................ 97 6 DISCUSSION AND IMPLICATIONS FOR POLICY .............................................. 100 Discussion ............................................................................................................ 100 Policy Implications ................................................................................................ 107 7 CONCLUSION AND AREAS FOR FURTHER RESEARCH ................................. 111 Summary of Findings ............................................................................................ 112 Study Limitations and Future Research ................................................................ 1 13

PAGE 7

7 APPENDIX A DETAILED DATA DESCRIPTION ........................................................................ 116 B MATRIX OF PEARSON CORRELATION ............................................................. 118 C PICTURE OF DOUBLE PARKING VIOLATION ................................................... 119 LIST OF REFERENCES ............................................................................................. 120 BIOGRAPH ICAL SKETCH .......................................................................................... 130

PAGE 8

8 LIST OF TABLES Table page 2 1 Built environment measurements ....................................................................... 34 5 1 Descriptive statistics for arrival time at destinations for work and shopping trips ..................................................................................................................... 70 5 2 Descriptive of socioeconomic variables with scale measures for work trips ...... 76 5 3 Descriptive of socioeconomic variables with scale measures for shopping trips ..................................................................................................................... 76 5 4 The mode of transportation for work trips and shopping trips ............................. 77 5 5 Ta ble for travel time spent by working and shopping travelers ........................... 79 5 6 Descriptive of distance (mile) for work and shopping trips .................................. 79 5 7 Estimation of time of day choice (ordered probit model) .................................... 86 5 8 The average price of parking in Fort Lauderdale CBD ....................................... 93 5 9 Th e average price of parking in Miami CBD ....................................................... 93 5 10 Estimation of time of day choice (ordered probit model) for CBD ....................... 97 A 1 Detailed data description .................................................................................. 116 B 1 The matrix of socioeconomic variables in Pearson correlation ........................ 118

PAGE 9

9 LIST OF FIGURES Figure page 2 1 The Position of timeof day in general travel demand forecasting process (Martin & McGuckin, 1998, p.6) .......................................................................... 32 4 1 Study area (regional) .......................................................................................... 57 4 2 The relationship between four NHTS files (FDOT, 2010, p.9) ............................ 59 4 3 A) The study cases boundaries B) Parking inventories at Fort Lauderdale CBD, and C) Parking inventories at Miami CBD ................................................. 64 4 4 The study fr amework .......................................................................................... 66 4 5 The conflict space diagram of 27 combinations: three categories and three ordinal measures (Carr & Zwick, 2007, p. 147) ................................................... 69 5 1 Time of day variation for work trips and shopping trips ...................................... 71 5 2 Gender of respondent and timeof day of arrival for: A) Work trips. B) Shopping trips. C) Both work and shopping trips ................................................ 72 5 3 Income and timeof day arrival: A) Work trips. B) Shopping trips. C) Both work and shopping trips ...................................................................................... 73 5 4 Worker status and timeof day arrival for: A) Work trips. B) Shopping trips. C) Both work and shopping trips ......................................................................... 75 5 5 Presence of child and timeof day arrival for: A) Work trips. B) Shopping trips. C) Both work and shopping trips ................................................................ 76 5 6 Mode of travel used by respondents for trips and timeof day of arrival: A) Work trips. B) Shopping trips. C) Both work and shopping trips ......................... 78 5 7 Employment density at trip end location: A) Work trip, B) Shopping trip ............ 80 5 8 Built environment variable for population density at home location: A) work trip, B) Shopping trip ........................................................................................... 81 5 9 Land use diversity at trip end locations in the Southeast Florida ........................ 82 5 10 Design dimension of built environment variables: A) Intersection density B) Culdesacs density ............................................................................................ 83 5 11 The combined built environment variables ......................................................... 84 5 12 Trip purposes and time of day arrival in the Fort Lauderdale CBD ..................... 91

PAGE 10

10 5 13 Trip purposes and time of day descriptive at Miami CBD ................................... 91 5 14 The block groups in: A) Fort Lauderdale CBD, B) Miami CBD ........................... 93 5 15 The number of parking spaces based on restriction hours and parking types in Fort Lauderdale CBD ...................................................................................... 95 5 16 The number of parking spaces based on restriction hours and parking types in Miami CBD ...................................................................................................... 95 C 1 Double parking violation in Miami CBD during working hours .......................... 119

PAGE 11

11 LIST OF ABBREVIATIONS CBD Central Business District FDOT Florida Department of Transportation FIPS Federal Information Processing Standards GIS Geographic Information System HBShop H ome Based Shop HBW Home Based Work IIA Independence from Irrelevant Alternatives MNL Multinomial Logit NHB Non Home Based NHTS National Household Travel Survey LEHD Longitudinal Employer Household Dynamics LODES LEHD Origin Destination Employment Statistics LUCIS Land Use Conflict Identification Strategy SPSS Statistical Package for the Social Sciences TAZ Traffic Analysis Zone TOD Transit Oriented Development VKT Vehicle Kilometers Travelled VMT Vehicle Miles Travelled

PAGE 12

12 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy THE RELATIONSHIP OF TIME OF DAY TRAVEL AND BUILT ENVIRONMENT IN SOUTHEAST FLORIDA : INCORPORATING PARKING CHARACTERISTICS IN DOWNTOWN MIAMI AND FORT LAUDERDALE By Shanty Yulianti Rachmat May 2014 Chair: Ruth L. Steiner Co chair: Paul Zwick Major: Design, Construction, and Planning Urban planners, who are responsible for land use policies in urban areas, seek the capability of built environment factors to influence travelers behaviors. Prior research has given less attention to the time of day dimension than to other dimensions. T ime of day choice study advances the travel demand modeling in several ways. F irst, policy makers need detailed information on peak hours S econd, socioeconomic changes influence the travel model in specific periods T hird, it helps in the evaluat ion of t ravel demand management strategies, i.e. the ways to modify travelers behavior by maximizing current resources. Thus, this study seeks to understand the effects of individual built environment factors and the combination of those factors o n time of day tr avel choices. By combining these factors, this study reduces the impact of collinearity of those factors i n statistical procedure. In the Central Business District (CBD) areas, this study examines the potential of timeof day choice to influence time sen sitive travel demand management strategies, such as time restriction and timevariable parking pricing.

PAGE 13

13 The National Household Travel Survey (NHTS) 2009 Florida Department of Transportation (FDOT) addon file is the main database used for this analysis. Three counties, Palm Beach, Miami Dade, and Broward, are used as a regional case study, with two CBDs Fort Lauderdale and Miami as specific cases. The parking inventory of these CBDs f rom a previous project at the University of Florida is used for thi s study Using the ordered response model, this study seeks the significances of various variables, including socioeconomic, trip characteristic, and built environment. This studys results shows that built environment factors have a greater impact on timeof day choice when considered in tandem. Thus, the purpose of the trip whether travelers are commuting to work or simply shopping demonstrates the differences between choices during given periods of time. T he relationship between time of day choice and a trips pur pose occurs in the CBD level. T he percentage of parking spaces with timerestriction in the CBD has an effect on the time of day choice. These results suggest that integration of built environment factors is important in creating land use pol icies to influence travelers behavior. Specifically, the model recommends that local governments incorporate timesensitive parking policies in CBD re development.

PAGE 14

14 CHAPTER 1 INTRODUCTION Statement of the Problem The study of time of day travel has received much attention because it has allowed travel demand analysis to get the detail of activities during specific period, such as peak period. Peak periods the ranges of times when the traffic is dense because people pr efer to travel during these periods have been identified to cause some problems, such as intense traffic congestion, increased energy consumption, and increased emissions of air pollutants. Besides these problems, researchers have considered the time shif ting as one of the strategies for travel demand management. The complete understanding of timeof day travel behavior is required to apply and evaluate travel demand management strategies. T ravel demand management (TDM) strategies are defined as the set of procedures to reshape peoples travel behavior by maximizing current resources (Ferguson, 1990). TDM strategies that include congestion pricing, parking pricing, employer incentives in transit, incentives to change driver behavior and other strategies to either change the mode of travel from driving a vehicle alone to transit, bicycling or other modes of travel, or the flexible working time are sensitive to peak period. Thus, some studies have examined the possibility of timeof day changes regarding the implementation, the feasibility, and the proposal for travel demand strategies (Yamamoto, Fujii, Kitamura, & Yoshida, 2000; Saleh & Farrell, 2005; De Jong et al.,2003; Arnott, de Palma & Lindsey, 1990; Ozbay & Yanmaz Tuzel, 2008; Hensher & King 2001; Lam Li, Huang & Wong, 2006; Atherton et al.,1982; Chin, 1990; He, 2013). Among those TDM strategies, this study focuses on parking management policies.

PAGE 15

15 Parking management policies are parts of TDM framework because they share similar objectives in managing available resources and influencing travel behavior by changing the time, the mode and destination of travel Parking supply and demand managements goals are to maintain and maximize current parking resources through programs and policies (VTPI, 2013 ). Parking integrates transportation and land use issues. Thus, in a broad range of TDM strategies that is showed by VTPI (2013 ) and Kuzmyak (2003), parking management is one of the land use strategies. The fact that parking is also landuse related issue emphasizes the importance to understand the position of parking in broader landuse aspects, which are called the built environment. The built environment is defined as: [a] multidimensional concept th at includes urban design, land use, and the transportation system, and encompasses patterns of human activity within the physical environment ( Handy, Boarnet, Ewing, & Killingsworth, 2002, p.65). TDM, especially parking, is included as one of the built environment dimensions (Ewing & Cervero, 2010). In a wider context, t ransportation and land use coordination may bring benefits in creating city livability (Steiner, 2012). Therefore, this study focuses on finding the connection between built environments as the land use aspects, especially parking, and travel behavior, with the intention of understanding the coordination of transportation and land use. However, because parking is incorporated into the built environment at a neighborhoodscale rather than macro level or regional level, this study takes the built environment factors at regional (counties) level, and parking factors at more specific scope, which is downtown level, to be connected into individual timeof day choice.

PAGE 16

16 In understanding the conn ection between built environment and travel behavior at regional level, additional method is required. Many of built environment factors have the possibility to be interrelated among themselves and cause statistical problems (Cervero & Radisch, 1996), stated as the collinearity problem (Cervero & Kockelman,1997). Thus, modifying variables into one combination may help to simplify the model and to reduce the collinearity problem without losing the information. Additionally, this combination may show the relative importance of one built environment dimension compared to others dimensions. In downtown areas, parking is a special issue. On one side, the other dimensions of downtowns built environment such as higher densit y development a greater mix of land uses, greater connectivity in design, and shorter distances to transit (the socalled Ds ), are suitable to support sustainable land usetransportation solutions However, in another side, businesses in downtowns and CBDs throughout the US prefer to hav e generous parking supply and low parking pricing because their consideration of competition with businesses in suburban areas (Steiner et al., 201 2 ). Inexpensive parking in downtown that encourages people to keep using automobiles and to park nearby their activities places may hinder the goals of sustainable landuse and transport. As a consequence, parking management should be integrated into transport and land use coordination to develop livable downtown and regional area. Purpose of the Research Thi s study focuses on the relationship of land use policy and travel behavior in general, as well as timeof day choice in its relation with the built environment factors and their connection to parking policies, in particular. The built environment and par king as one of built environment factors has been explored to be a factor in determining the

PAGE 17

17 other dimension of travel behavior; however, few studies have investigated the connection between the built environment variables and the timeof day choice model. By investigating these factors, this study contributes to the existing debates on the relationship between the built environment and travel behavior. Also, this study contributes to the body of literature by using combined measures of the built environment. This method allows the combination of interaction of variables among built environment factors that have previously been identified as collinear in statistical procedures. This study uses the preprocessing in Geographic Information S ystems ( GIS ) of the built environment factors before performing the choice model T his study also tests the other factors such as socioeconomic variables and trip characteristics as they affect the main relationship between built environment factors and timeof day choi ce. This research uses data from the NHTS 2009 Florida Department of Transportation (FDOT) addon from the Southeast Florida. T his study employs a cross sectional approach and utilizes the orderedpropensity based model to assess the probability of travelers choosing specific time periods. Specifically, this study also concentrates on the timeof day choice with its relation to the parking situation in downtown Miami and Fort Lauderdale. Taking two CBDs within the Southeast Florida area, this study explores the average price, the percentage of timerelated restriction parking policy and the availability of parking spaces within block groups in these downtown areas. This study uses data for parking inventory from the project of Impact of parking supply and demand management on the CBD by the University of Florida (Steiner et al., 2012). Also, this study identifies the

PAGE 18

18 possibility of travel demand management applicability, such as variable parking pricing and timerelated parking restriction in those two CBDs. This study evaluates two following hypotheses at the regional level, for the Southeast Florida region: 1. When considering socio economic variables and trip characteristics, there is a relationship between travel behavior (timeof day) and the buil t environment in the Southeast Florida ; and 2. The combinations of built environment factors are more influential than individual factor of built environment on the connection to timeof day choice. Two other hypothes e s are tested for the two Southeast Florida CBDs, Miami and Fort Lauderdale: 1. Trip purpose determines different travel behavior, especially related to timeof day choice 2. T he average parking rate determines timeof day choice at the two CBDs destinations. Significance of the Research In general, this research attempts to enhance the body of knowledge on the relationship between landuse and transportation. By focusing on timeof day study, this research specifically contributes to one travel behavior dimension. Other dimensions are trip frequenc y, trip length, and mode of transportation choice. Understanding this relationship emphasizes the role of land use policies in shaping travel patterns. The results of the study are also expected to provide insights into landuse policies that can be used to shift travelers choices into the intended time ranges. Prior literature focuses on the relationship between built environment and travel behavior; however, four gaps can be identified. First, timeof day choice has not been extensively examined in rel ationship to the built environment factors. Only a few studies

PAGE 19

19 have linked the built environment variables in the timeof day choice model ( Chu, 2009; He, 2013; Chikaraishi, Fujiwara, Zhang & Axhausen, 2009; Lee, Washington, & Frank, 2009; Zhang, 2005) bu t these studies have treated the separate rather than combined effect of the built environment variables This study examines the arrival time and the built environment at destination, while other studies highlight the differences at the trip origin. Thus because the existence of a relationship between time of travel and built environment characteristics at the destination remains imprecise, this study endeavors to enhance previous timeof day choice studies Second, one of the gaps from the built environment research is the lack of methodology that can capture the complexity of built environment interactions. According to Cervero (2003), measuring the built environment variables may be subject to collinearity a statistical problem because variables tend to come in tandem in one area. High density areas frequently have greater street connectivity and great mix of land uses. A combined measure is needed to respond this problem (Krizek, 2003). This study allows interactions of built environment variables by converting each unique value for each combination in GIS Thus, this method overcomes the collinearity problem. Consequently, this study also provides new insights to the body of literature about how to incorporate the complexity of built environment factors and the relative importance of one factor over another in shaping the travel behavior. Third, t his research takes Southeast Florida as empirical study location, where little research has explored this location, especially in the relationship between travel behavior and built environment variables ( Messenger & Ewing 1996 ; Neog, 2009 ; S teiner et al. 2010; Srinivasan, S., Provost, R., & Steiner, R., 2013) Considering the

PAGE 20

20 suggestion from Boarnet and Sarmiento (1998) and Boarnet and Greenwald (2000), who conducted similar study in two different regions, they found different results and highlighted the importance of doing empirical study in different locations. By employing time of day dimension that i s different from those studies, this research may enrich the figure of travel behavior from previous literature in those specific locations. Furthermore, this study extends the variety of empirical travel behavior stud ies with a focus on built environment factors. Fourth, t his research attempts to explore the downtownspecific focus by incorporating average parking price and parking restriction in downtowns and in particular, the CBD of Fort Lauderdale and Miami Local government may experience parking supply dilemma in downtowns. A lack of parking space may create cruising or spillover parking in surrounding areas. Meanwhile, ample or excessive parking may bring negative effect; especially, it hinders the objective of city livability because it increases the incentive to drive and discourages transit or nonmotorized modes of travel. This particular research may provide evidence of the importance of the current parking policies and the possibility to enhance the current practices to improve them as a part of downtown livability initiatives. Dissertation Outline The remainder of this dissertation is organized as follows. Chapter 2 synthesizes literature review about time of day dimension of travel into the broader context of the connection between travel behavior and built environment. Chapter 3 explores existing studies about built environment and parking as one of the built environment variables. Also, this chapter explains the measurement of built environment and the selection of variables Chapter 4 describes the methodology and data, and provides an overview of

PAGE 21

21 the two downtown CBDs that form the case study areas for downtown parking policies Chapter 5 provides the descriptive analysis, the empirical analyses and findings. Chapter 6 discusses the conclusions and recommendations. In addition, this chapter also outlines the limitation of the study and the further research that may enrich this studys findings.

PAGE 22

22 CHAPTER 2 TIME OF DAY CHOICE AND BUILT ENVIRONMENT People often make multiple trips to meet their daily needs and to engage in their ongoing activities. The terms trip and travel both mean movement from one place to another by any means of transportation ( Handy et al., 2002). Currently, automobile users dominate the travel mode for these daily activities According to the 2009 NHTS the percentage of persontrips in the US taken by automobile for all trip purposes is 83.4 percent and 91.4 percent for trip to and from work (Santos, McGuckin, Nakamoto, Gray, & Liss, 2011 p. 9 ). The domination of automobile use matches what Newman and Kenworthy (1996) characterized as the automobile city that has a low density of population and widely spread development. Automobile dependency requires continuous transportation development. However, some issues such as financial cost physical constraints, and environmental awareness (Handy, Cao, & Mokhtarian, 2005) may change the political considerations associated with creating additional transportation infrastructure in coping w ith the demand resulted from automobile dependency. In turn, automobile dependency has generated the wide population dispersal. Thus, people continue to rely heavily on automobiles as their preferred mode of transportation and policy efforts to reduce auto mobile use have been hampered by urban sprawl (Handy et al., 2005). Urban sprawl is characterized as unpleasing development because of the low population or employment density and scattered physical environment and the inability of such development patter ns to serve modes of travel other than the automobile (Ewing, 2008). Seeing twoway relationship between automobile users behavior and urban sprawl, urban planners seek to develop an effective built environment to shape the

PAGE 23

23 travel behavior. Although many interpretations of built environment exist in various disciplines, for the purposes of this study, built environment will refer to various physical and manmade settings that support daily activities of individuals (Brownson, Hoehner, Day, Forsyth, & Sall is, 2009). Many planners and policy makers are convinced that a relationship exists between landuse and travel behavior ; they support designing neighborhoods in such a way that encourages people to walk or use mass transportation. Urban planners call this perspective New Urbanism ( Handy & Clifton, 2001). In general, this perspective highlights the importance of transportation and land use coordination, specifically through built environment factors and their relationship to travel demand behavior. The Relationship of Travel Behavior and Built Environment P rior studies have connected the built environment and travel behavior quite extensively (Ewing & Cervero, 2001 ; Ewing & Cervero, 2010). Most of those studies focus on trip frequency, trip length ( cumulative v ehicle h ours of t ravel or VHT and v ehicle m iles t ravelled or VMT), and mode of tra vel to measure travel demand. Cumulatively, these studies show that a relationship exists between built environment factors and each of these travel dimensions First, some researchers found that built environment factors have a relationship to trip frequency that usually as an intermediate variable to explain travel choice by different modes. For example, a greater number of pedestrian trips is associated with more density, location near schoo ls (Lee & Moudon, 2006), shorter distances to commercial businesses ( Cao, Handy, & Mokhtarian, 2006), better connected street pattern with smaller blocks, and more accessible locations for travelers (Handy et al., 2006). Moreover, the presence of bi cycle facilities, such as trail s, ha s a positive effect

PAGE 24

24 on the number of bicyclists (Moudon et al.,2005). B ikeways and street network connectivity affected the nonmotorized trips ( Guo, Bhat & Copperman, 2007). A ccessibility density, and the proximity to the city center and employment significant ly reduce frequency of automobile trips (Shay & Khattak, 2007). However, some researchers found that socio economic factors of the household or attitude ( Kitamura, Mokhtarian & Laidet 1997) played a greater role than land use variables when they analyzed trip frequencies by various modes. Another argument states that lack of significance on for land use variables exists for nonwork trips ( Boarnet & Sarmiento, 1998). In another case study, various effects sometimes they we re weak and sometimes apparent of built environment variables are associated with the frequency of non work trips ( Boarnet & Greenwald, 2000) The authors conclude that how the built environment variables a re measured and how the geographic unit of analysis i s chosen may contribute to the results. As an example their study uses zip code level data as a unit of analysis that is broader than others whose analysis showed a significant relationship between built environment factors and driving behavior. Second, m ost of the prior studies support the argument that incorporating built environment factors may increase the ability of a model to explain the trip length ( Kockelman, 1997; Frank & Pivo, 1994; Chatman, 2003, Bento et al., 2005; Steiner et al., 2010; Cervero & Murakami, 2010; Cervero & Kockelman, 1997) Meanwhile, some studies stated that the effect of built environment factors is minimal (Brownstone, 2008) Trip length is commonly associated with trip dis tance and travel time. Trip distance is defined as VMT or v ehicle kilometers t ravelled (VKT) as the measurements. Variables that influence trip length are accessibility land use balance (entropy), mix (dissimilarity),

PAGE 25

25 employment density, and population density (Frank & Pivo, 1994 ; Kockelman,1997). Higher employment density at the workplace reduced the VMT (Chatman, 2003) and lower population density increase VMT per year and raises the fuel consumption ( Brownstone & Go lob, 2009). Also, accessibility and gridtype measurement influenced the VMT ( Cervero &Kockelman, 1997). Third, some of studies support the argument that land use variables have a significant effect on mode choice (Cervero, 2002; Cervero, 1996; Chen, Gon g & Paaswell, 2008; Rodriguez & Joo, 2004; Pinjari, Pendyala, Bhat & Waddell 2007; Kockelman, 1997; Cervero & Kockelman, 1997; Cervero & Radis ch, 1996; Ewing & Cervero, 2010; Frank & Pivo, 1994; Silva, Golob, & Goulias, 2006); however, some other of studi es make a contradict ory argument ; either the built environment factors have no significance (Hess, 2001), or little effect on mode choice (Crane & Crepeau, 1998; Neog 2009), or other factors, such as socioeconom ic characteristics attitudes (Kitamura et al., 1997), and lifestyles, influence the selection of mode more than built environment factors. These studies present different results based on detailed built environment factors. For example, higher density and greater land use mix has a greater ef fect on transit mode selection than design dimensions, such as higher sidewalk ratio (Cervero, 2002). Other built environment factors, such as employment density and mixed land use have a negative correlation with single o ccupant v ehicle (SOV) choice and a positive correlation with transit and walking ( Frank & Pivo, 1994) At the destination, higher employment density at the workplace results in less automobile usage; whereas better job accessibility from home plays a role in transit selection for homebased work (HBW) travelers (Chen et al., 2008). Similarly, population and

PAGE 26

26 employment density influenced the nonmotorized mode choice, and land us e mix has an effect on transit choice (Pinjari et al., 2007). Some studies suggest that the built environment has an insignificant effect on mode selection. For example, p edestrian friendliness and the presence of light rail for commuters were insignific ant in mode choice in Portland, Oregon (Hess 2001) Neog (2009) concluded that the workplaces built environments are not significant in the study area of the Southeast Florida. The author found only one factor of the built environment turning significa nt the proportion of railroad in one mile radius. In summary, p rior studies have presented different arguments and specific results from empirical studies. The differences imply that the effect of built environment to travel behavior is still unclear c ontext sensitive, and worth further examination. Three common dimensions of travel behavior that have been investigated to have relationships with built environment factors are trip frequency, trip length, and mode choice Few studies have explored simil ar connection to another travel behavior dimensiontime of day choice. Furthermore, urban planners are concerned about automobile dependency because it causes people to travel longer distances, produces greater carbon emission than other modes of transpor tation, and creates congestion, especially in peak hours. Because so many people travel during peak hours, street networks experience lower levels of service, longer travel times and traffic congestion. Therefore, travelers may change their behaviors by switching to a different mode of transportation, changing the destinations, or shifting the specific time of travel. Because travelers can change the

PAGE 27

27 time of their travel it is important to understand travelers behaviors based the specific and the actual information regarding timeof day periods. Time of Day Choice Peak hour studies have emphasiz ed their importance in travel demand management strategies proposal, such as: congestion pricing ( Saleh & Farrell, 2005) and turnpike pricing policy ( Ozbay & Ya nmaz Tuzel, 2008). In addition, Hensher and King (2001) compared two parking policies parking pricing and supply by time of day showing that shifting departure time is one of the responses to the policies; and Atherton, Scheuernstuhl, and Hawkins (1982) evaluated the compressed workweek policy that results in decreasing peak hour traffic volume. Time of day choice is one element of the travel demand forecasting process. Therefore, time of day studies may have roles in studies of traffic impact studies, trip accumulation, highway volume and capacity, transportation system management, and transport demand management (Martin & McGuckin, 1998, p.82). According to similar study, the position of timeof day studies among the others dimension of travel behavior can be seen on the Figure 21. Furthermore, the following section describes the importance of knowing this dimension, prior studies on time departure choices, and the existing studies about the relationship between time of day and built environment. The Importance of Time o f Day Choice Studies Time of day choice studies help us to understand the following aspects of planning for travel : (1) detailing the polic ies on peak period, (2) responding to socio economy changes, and (3) accommodating the growing i nterest of travel demand management. P olicy makers need the specific and actual information regarding peak periods (Abkowitz, 1981), because traffic in the peak hours has produced more severe

PAGE 28

28 congestion, higher energy consumption, and more emission product ion Moreover, when congestion occurs, the characteristics of street networks result in different travel time and level s of service. Travelers may choose to change their behaviors in an attempt to minimize time and expense by modifying their timeof day travel choice s or other adjustments to their travel P rior research (Ettema & Timmermans 2003; Chin, 1990; Mannering, 1989) suggested that people may choose specific timeof day adjustment to their travel to avoid congestion in peak periods For the sam e reason, Mannering and Hamed (1990) found that half of their respondents temporarily modified their travel time to preventing them from encountering congestion. Then, specific timebased policies is required because dense traffic at peak hour results in h igher energy consumption and more emissions of certain pollutants than at the others times. In his article, Abkowitz (1981) stated that one of the reasons to reduce peak hour travel is the associated fuel economy improvement s. Concurrently, the 1990 Clea n Air Act Amendments urge local and regional policy makers to understand and undertake studies to understand the importance and variability of peak hour travel on current and future travel demand, mode choice and the associated congestion (Bhat, 1998). Ano ther reason to focus on peak hours and timeof day research is that peak hours contribute to the increasing air pollution emissions. For example, Steed and Bhat (2000) explained three reasons for modeling peak hour emissions. First, emission s have been accounted as a variable in the model ing of automobile VMT E ach time period of the day has a distinctive temperature and humidity profile with resulting differences in emissions Second the operations of various modes of travel differ

PAGE 29

29 throughout the day. Cold starts produce more emissions at all times of the day but if they are concentrated during the peak hour they will have a greater impact on the overall emissions. Finally, the ozone formation and dispersion models require an understanding of t he, emission s level during specific times of the day. As the s econd aspect, time of day differences may occur because of the changes in employment and socioeconomic characteristic. Taking as an example, a traveler may modify his or her departure time t o shop because she has a school age child In turns, he or she has to drop and pick them up which may constrain their decision on when to shop during prior time ranges. According to Steed and Bhat (2000), some metropolitan areas have used fixed factor f or the timeof day in their travel demand study; however, this factor may not be able to capture the timeof day shift because of the changes in socioeconomic characteristics. Lastly, applying and evaluating travel demand management strategies requires a complete understanding of timeof day travel behavior. Three possibilities of a policy response are mode changes, route changes, and departure time changes (Mannering, 1989). By referring to these possibilities, some studies have examined specific possibility of time of day changes regarding the implementation, the feasibility or the proposal for travel demand strategies. C ongestion pricing (Yamamoto et al., 2000; Saleh & Farrell, 2005; De Jong et al.,2003) toll pricing (Arnott, de Palma, & Lindsey, 1990; Ozbay & Yanmaz Tuzel, 2008), parking pricing or choice ( Hensher & King 2001; Lam, Li, Huang & Wong, 2006), and flexible work hours (Atherton et al.,1982; Chin, 1990; He, 2013) are examples of timeof day travel demand strategies E merging researc h on activity based and trip chaining behavior require more detailed information

PAGE 30

30 on timeof day travel Since time limits the opportunity in doing activities ( Ettema & Timmermans, 2003) time of day studies give s complete information about trip chaining Survey approaches Researchers have widely incorporated two approaches from travel surveys stated preferences from primary survey and revealed preferences. The first approach stated preferencehas been used by several studies due to some of the research ers desire to accommodate the situation that has yet to be implemented, namely, congestion pricing. Stated preference studies can be differentiated from revealed preference studies, which are based upon actual behavioral responses. The advantage of state d preferences approach is that it can capture the original experience from the respondents (Bellei et al., 2006). However, most of the stated preference studies are from international cases, e.g., studies by Arellana, Daly, Hess, de Dios Ortzar, & Rizzi (2012) in Santiago, Chile; Saleh and Farrell (2005) in Edinburgh, UK; and Brjesson (2008) in Sweden. The latter studies in Sweden have combined state preference and revealed preference data. Another type of survey, revealed preference depends on the readily available household travel survey. Most studies capture the singletrip and some studies took multiple trips and trip chaining to have the tour format, singleday travel diary or cyclic pan els. As an illustration, Lemp et al. (2010) used tour format, Chu (2009) utilized a oneday activity diary, Chikaraishi et al.(2009) conducted a six day travel diary, and Yamamoto et al. (2000) use four wave panel study in understanding the behavior changing because of the proposed congestion pricing.

PAGE 31

31 Methods in time of day studies Most of previous research used the logit model as their approach. Nevertheless, they had applied some different variances and extensions in the model. Besides regular multinomial logit model (Abkowitz, 1981; He, 2013; Mannering & Hamed, 1990; Yamamoto et al., 2000; Kitamura, Chen, & Narayanan, 1998; Okola, 2003; Saleh & Farrell, 2005), other examples of variations and extensions include the mixed logit model (Brjesson, 2008), nested logit model (Bhat, 1998; Chin, 1990; Bellei, Gentile, Meschini, & Papola, 2006; Lemp et al., 2010; Sall & Bhat, 2007; Yang, Zheng & Zhu, 2013;Ozbay & Yanmaz Tuzel, 2008), the error component logit (de Jong et al., 2003), generalized logit (Sasic & Habib, 2013), ordered generalized extreme value (OGEV) (Bhat, 1998; Steed and Bhat, 2000), and dogit ordered generalized extreme value (Chu, 2009). The advantage of methods other than multinomial logit model is the weakness of multinomial logit ( MNL ) mode l that accounts to the i ndependence from i rrelevant a lternatives (IIA) property When experiencing this problem, the model might not be able to capture the similarities and correlations between adjacent time interval choices. According to DeJong et al (2003), the weakness of MNL choice models for departure time choice is the correlation between the choices.

PAGE 32

32 Figure 21 The Position of t ime of d ay in g eneral t ravel d emand f orecasting p rocess ( Martin & McGuckin, 1998, p.6) The R elationship of Time of Day and Built Environment Few studies on timeof day choice have linked the built environment variables in the model. Researchers describe the terms of built environment differently, for example locationrelated variables and spatial variations. The relationship of built environment and timeof day are varied, in the similar way as that relationship with others travel behavior dimensions.

PAGE 33

33 Some researchers conclude that the built environment variables are insignificant in their relat ionship to timeof day choice. Chu (2009) found that the corridor density between home and workplace is insignificant for the departure time within 3 hours period (6:009:00 a.m.). He (2013) found that employment density at the trip end is not significan t for commuters departure time choice in the morning. In contrast, Chikaraishi et al.(2009) identified spatial variation such as land use, facilities location and the distribution of population, as significant in travelers time choice for mandatory a ctivities such as school, work, and other work related travel. Lee, Washington, and Frank (2009) showed that the built environment factors, such as residential density and the number of commercial parcels within one kilometer radius are significant in rel ation to weekday and weekend timeof day variation. Zhang (2005) found the varied effects of accessibility on nonwork trips and travel, depending on the activity categories. Thus, those inconsistent results for a significant coefficient of built environment to time of day choice show that the existence of a relationship remains imprecise. Built environment may refer to urban forms, landuse variables, urban designs, spatial variations, or locationrelated variables. Despite of considering each term dif ferently, this research allows those terms to be used interchangeably. Several researchers attempt to characterize the built environment variables into groups. Handy et al. (2002) mentioned six dimensions of the built environment: density and intensity, land use mix, street connectivity, street scale, aesthetic qualities, an d regional structure. Detailed measurements of these dimensions are displayed on Table 21. Thereafter, Ewing and Cervero (2010) categorized multidimensionality in the built environment by

PAGE 34

34 grouping them into seven Ds dimensions. The first three D di mensions density, diversity and design (Cervero & Kockelman, 1997) have been broadly used. Various effects of those three dimensions have been found with respect to travel behavior. Trip frequency relates to the socioeconomic factors extensively. Trip length associates with regional accessibility. Mode choices have the relationship with these land use pattern factors. Also, transit choice primarily has connection mainly with local density. Later, Ewing and Cervero (2010) expanded into seven Ds. They added destination accessibility, distance to transit, demand management, which includes parking demand and supply management, and demographics to the prior three Ds. However, most researchers considered demographic as control variable in the relationship of travel behavior and built environment factors. Table 21 Built environment measurements Dimension Definition Examples of Measures Density and Intensity Amount of activity in a given area Person per acre or jobs per square mile Ratio of commercial floor space to land area Land use mix Proximity to different land uses Distance from house to nearest store Share of total land area in different uses Dissimilarity index Street Connectivity Directness and availability of alternative routes through the network Intersection per unit (e.g. sq mile of area ) Ratio of straight line distance of network distance Average block length Street scale Three dimensional space along a street as bounded by buildings Ratio of building heights to street width Average distance from street to buildings Aesthetic Qualities Attractiveness and appeal of a place Percent of ground in shade at noon Number of locations with graffiti per square mile Regional Structure Distribution of activities and transportation facilities across the region Rate of decline in density with distance from downtown Classification based on concentrations of activity and transportation network Source: Handy et al. (2002, p.66)

PAGE 35

35 Built Environment Dimensions This study follows the grouping of built environment dimensions based on Ewing and Cervero (2010). First, density is the most common built environment factor measured for the relationship with travel demand behavior. Several studies believed th at this dimension is the most important one among the others (Chen et al., 2008). Defined as the intensity of land use for housing, employment, and other purpose (Cervero, 2002), density is measured by dividing total population, employment, household, or other variable of interests, with the area. The ratio between floor space per parcel area is also a popul ar measure of density (Handy et al., 2002). Past literature has used density in their model by measuring total population and employment divided by total square miles of the traffic analysis z one or TAZ (Cervero, 2002); average population and employment density at travelers home or work location, the average of maximum population density (person/sq. miles) (Chen et al., 2008); population per develop ed acre, and employment per developed acre (Cervero & Kockelman, 1997) The next dimension is diversity that is defined as the different land use s per unit of area. Two common measurements are: entropy or the uniformity of landuse and diversity of land use (Ewing & Cervero, 2010). Handy et al. (2002) suggested land use mix as a term to give similar insights of diversity that is defined as the relative proximity of different land uses within a given area (p.66). As the measurements, they took a dissimilarity index share of total land area for different uses and distance from each residential unit to shop ping attractions. Measurements of diversity have various degrees of complexity. In his 1996 study, Cervero used the number of retail and commer cial areas within 300 feet of the location of a travelers The mix land use turned

PAGE 36

36 to have positive effect on nonmotorized choice. He concluded that land use environment and mix use have an effect on travel behavior. Several formulas provide the process of calculating the diversity (Frank & Pivo, 1994; Gue et al 2007; Kockelman, 1997; Cervero& Kockelman,1997); Rajamani,et al.,2003), either by calculating an entropy or the index of mixed use in dedicated areas. T he t hird dimension is design that is g eneral term for the various aspects that rel ated to street network features. F or example, the design includes measures of connectivity (e.g. the number of intersections per unit area, and the proportion of the four way street ) ; infrastructures for pedestri an and bicycle, such as sidewalk or bike path availability; and other design features, for instance the proportion of front and sidelot parking, pedestrianfriendly design, and reflecting wide setback (Cervero and Kockelman, 1997). In comparison, Handy et al.(2002) defined these features as connectivity of the street network and scale of street. Their other specification of design dimension aesthetic qualities may also be included in this design category. However, with the lack of data about aesthetic s, such as building design quality window orientation, perception of aesthetics, and decorati ve elements this dimension has been difficult to use as measurement in empirical studies As Handy et al.(2002) mentioned, this dimension tended just to be more explanatory than a measure. Almost similarly, Cervero (2002) took design measurement as the quality of walking environment and the physical configuration of street networks the ratio of sidewalk miles to the centerline miles of roadway. Pinjari et al.(2007) suggest s the design as the network level of service measures for street block density, bicycle facility density, and transit availability.

PAGE 37

37 The fourth and fifth dimensions are distance to transit and destination accessibility. Although these two dimensi ons may be incorporated into the third D or design; s ometimes, researchers differentiate the m from desi gn. For example, Neog (2009) measured trip distance with network distance from home to work in miles. Kockelman (1997, p.117) measured the accessibility using Equation 21 Accessibility= (2 1 ) Where, Aj is the attractiveness of zone j and tij is the travel time from zones i to j. Chen et al.(2008) measures job accessibility of track A as the weighted sum of the number of jobs in every tract, weighted by the distance to tract A. Moreover, Neog (2009) added the measurement of regional accessibility by transit using a gravity model. The gravity type of acces sibility Equation 22 was also presented by Rajamani et al. (2003 p. 161): = (2 2 ) Where f (Cijm) = friction factors between zone I and j by mode m Rj = retail employment in zone j J = total number of zones i = zone for which the accessibility index is being computed m = mode for which the accessibility index is being computed. Accessibility may decrease the work commute time; however, it increases the social activity and shopping tr avel (Zhang, 2005). The sixth and seventh D dimensions are demand management and demographic. D emographic information is commonly used as control variables, especially in socioeconomy variables that have relationship to travel behavior (Ewing & Cerve ro, 2001) The review about demand management i s included in Chapter 3.

PAGE 38

38 Problems on Built Environment Studies Three difficulties in measuring and incorporating built environment factors into travel behavior are the varying dimensions of built environment, the behavior al bias introduced because of self selection, and the unit of analysis chosen for the study. Built Environment D imensions Many of built environment factors have the possibility to be interrelated among themselves and cause statistical proble ms (Cervero and Radisch, 1996). Cervero and Kockelman (1997) stated this as the collinearity problem. As an example, residential density relates to other indicators: mixed uses, shorter block lengths, grid pattern of street, and continuous sidewalk networ k. Additionally, Crane and Crepeau (1998) mentioned that the inability of prior literature is to show individual effect of neighborhoods street design if come together with any other design feature. To overcome this correlation problem, Silva et al. (2006) performed loading factors to choose the variables with the greatest effects. Similarly, Cervero and Kockelman (1997) adopted factor analysis to see the interaction among the built environment factors. Factor analysis proved to be a useful approach to combine collinear variables and to reveal contributions of different attributes of the built environment in explaining travel demand (Cervero & Kockelman, 1997, p.218). Shay and Khattak (2005) offered factor loading and cluster analysis to combine and s elect many of built environment factors. The authors explain the analysis as follows: The factor analysis, which generates indices that combine overlapping simple measures and cluster analysis, which adds a spatial dimension to sort individual neighborhoods into groups of similar composition may be useful tools for better understanding how the environment influences auto ownership and travel behavior (p. 80)

PAGE 39

39 Moreover, Neog (2009) used correlation matrix for their control variables and tested the combined effect from density. The author tested the interaction between density and diversity and compared the model with the one without the combined variable. Additionally, different measurements between built environment factors and socio economic attributes may contribute to the significances in the model. Most of built environment factors is ordinal level, or even sometimes nominal, while socioeconomic attributes have interval ratio that have advantage in the calculation of model (Cervero & Kockelman, 1997). Self selection B ias Recent literature has focused on the influence of self selection in the relationship between travel behavior and built environment (Ewing & Cervero, 2010). Chatman (2009) defined self selection as the behavior that households choose neighborhoods based on their expected travel patterns. This sorting process, if not statistically controlled, confounds the estimation of the effects of the neighborhood built environment upon household travel, because, if variation in the built environment leads to households spatially sorting themselves according to their travel preferences, then those preferences will be highly correlated with built environment characteristics (p. 1072) Self selection of people who like to walk or use transit into highdensity, mixeduse and transit rich neighborhoods, might mean that researchers and policy makers over estimate the impact of built environment variables on travel behavior. The author explained four approaches to overcome the self selection problem: (1) landvalue study, (2) joint choice model of residential and travel behavior, (3) before and after move study, and (4) the reportedattitudes surveys. Several studies have incorporated self selection and attitudes into the model by specifying tr avelers into mover or nonmovers and their preferences into modes (Handy

PAGE 40

40 et al., 2006, Cao, Mokhtarian & Handy 2009; Schwanen & Mokhtarian, 2005). Self selection affects various travel behavior, such as: the frequency of walking (Handy et al., 2006), the frequency of nonwork travel by modes (Cao et al., 2009) and the probability to choose the modes of personal vehicle, rail, bus, and nonmotorized (Schwanen & Mokhtarian, 2005). Those studies concluded that built environment factors still have influence on travel behavior even after controlling for self selection. Unit of Analysis and Location Prior studies employed various units of analysis. Studies using the prototypes of neighborhoods defined a neighborhood as the unit. Studies using travel surveys u se various units of analysis from the individual household to the neighborhood and the region. Using smaller units of analysis can allow researchers to more fully specify their model. However, one problem with smaller units of analysis is the data availability (Cervero & Kockelman, 1997). Researchers find that more detailed the need for data, such as for neighborhood level analysis the more difficult it is to get a large enough sample for any given neighborhood M ost regional travel surveys focus on the neighborhood scale rather than the macro scale Obtaining complete data on the built environment for track level can be difficult ; even if densities, housing and sociodemographic features can be measured, they can be quite variable and based upon a small sample size (Cervero & Kockelman, 1997). Furthermore, researchers should be aware of these dissimilar units of analysis because these may generate varied results of empirical studies ( Zhang, Hong, Nasri, & Shen, 2012; Zegras, 2010; and Boarnet & Crane, 2001). As an example, neighborhood level variables show greater effects on non motorized trips, while regional built environment affec ts auto commuters trip (Zhang et al., 2012).

PAGE 41

41 Several studies agreed that different places may have different natur e of land use and travel behavior relationship (Boarnet & Greenwald, 2000; Zegras, 2010; Neog, 2009). Neog (2009) found contradictive results of density and diversity relationship to mode choice than prior studies due to the nature of study location. Vari able Selection on the Relationship of Time of Day and Built Environment Besides the timeof day as dependent variable, built environment, socioeconomic and trip characteristics variables are included in the model. Specifically, the measurements for each variable are based on the previous studies that are within similar topic. Following sections are the considerations of each variable. Built Environment Factors Density in this research is represented by employment density at the trip end and population density at the home. The unit of measurements is at block group. Most of built environment research that utilized travel survey used census tract as their unit of measurement, such as Frank and Pivo (1994), or Traffic Analysis Zone, for example Hess (20 01), Pinjari et al.(2007), Cervero (2002), and Kockelman (1997). This research takes more detailed unit in measuring density. Although the database is available on really detailed unit block level. However, three level of geocoded were applied: to the nearest road, to the nearest border of blocks, and to the nearest parcel; those caused the difficulty to decide which blocks belong to one geocoded location. Within the scope of timeof day, densitys effects remain unclear. One said that the employment density on trip end location has insignificant relationship to departure choice (He, 2013). The others showed the significance of density on timeof day choice in this case population density (Chikaraishi et al., 2009; Lee et al., 2009). In turns, this

PAGE 42

42 stud y considers the density of population at home location and employment at trip end location. Diversity in the trip end is measured by the count and area measures for retail/commercial parcels within the block groups. Previous study shows the number of comm ercial parcels influenced the departure time choice (Lee et al., 2009). This study refers the diversity measure as Rajamani et al. (200 3 ) and Bhat and Gossen (2004) used. The F ormula 2 3 is as follow. = 1 (2 3 ) Where r = acres in residential use (single and multifamily housing) c = acres in commercial use, i = acres in industrial use, o = acres in other land use, and T = r + c + i + o Bhat and Gossen (2004) found that diversity is significant for type of recreational purpose. However, specifically for timeof day choice, diversity and land use mix have not been much incorporated on the prior literature. Design dimension in this study refers to stre et network features. Available data of this study allow the calculation of the number of intersections per unit area. Not many of prior studies on timeof day choice consider the built environment, especially design dimension, study that considered desig n dimension was Chu (2009). However, the author found that the corridor density between home and workplace failed to give significance to departure choice.

PAGE 43

43 SocioEconomy V ariables Controlling for socioeconomy factors has been a common practice in exam ining travel behavior. Most of the literature both in travel demand and built environment areas agrees that socio economic variables contribute to travel behavior pattern ( Neog 2009; Steed & Bhat ; Sall & Bhat, 2007). Two different data are examined to understand the nature of socioeconomy characteristics from travelers household and person data. The characteristics that follow the household data are the number of household, the number of vehicle, the number of worker in the household, household income, the number of adult, the number of children, and household ethnicity. Personrelated data include age, gender, income, the presence of children, and the status of employment. The total number of household members might influence the time of day choice. For instance, households with more members tend to choose earlier timeof day period than household with fewer members (Lemp et al., 2010). On the contrary, Habib (2012) found that household size (larger number of member) affect s the delay in work start t ime. The existence or the number of vehicle may have an effect on particular time of day choice. He (2013) includes the car availability as one of the factors influencing time of day. The author uses the number of cars per licensed driver. Furthermore, the study concludes that this factor was insignificant because lack of variety, since most of travelers had 1 vehicle per licensed driver (He, 2013) Age may determine different behavior on time of day choice. Many studies on time of day choice have found age as significant variable. For work trip, older people are expected to travel later than younger people have because older workers may have more experiences and have longer adaptation on work force (Abkowitz, 1981).

PAGE 44

44 Meanwhile, Steed and Bhat (2000) found that age influenc es the travel time of older travelers who tend to choose the midday for their recreational and shopping activities. Moreover, Bhat (1998) show s that age had a negative influence on evening shop travel especially for travelers above 65 year old. This result was also found by Okola (2003), who found that old age tends to cause people to make trips in the afternoon peak rather than in the evening. Silva et al.(2006) treated age as an exogenous variable and found that older people preferred the traditional urban areas. D ifferences i n time of day choice exist between male and female. Several studies prove the effects of gender o n time of day choice. For example, Abkowitz (1981) and Lemp et al.(2010) took gender as one influencing variable on timeof day choice and concluded that women travel at a later time period than men. Similarly, Bhat (1998) considered that gender influenced the preference on time for time of travel In the study, women traveled during the a.m./p.m. off peak and p.m. peak rather than in the morning peak or evening period. T his has implications for women s responsibilities for shopping activities which are a part of the study focus. For work trips, Chu (2009) found that women tended to travel during the period surrounding 8 a.m. (7:30 to 7:29 and 7:30 to 7:59). Although not specifically concern about timeof day, Silva et al.(2006) concluded that the gender had negative correlation to land use variables. Income has a relation ship to time of day travel choice. Travelers with high level income have the tendency to choose other time rather than morning time (a.m peak) and midday period for shopping and recreation trips (Steed and Bhat, 2000). Also, high income traveler tended to choose the later period of time for work trip such as 8:00 to 8:29 and to avoid the earlier times (Chu, 2009). In the same study, middle income

PAGE 45

45 trave l l er s were less likely to t ravel at this time and more likely to travel earlier (7:30 to 7:59). Lemp et al.(2010) found that the high income workers are likely to choose a later time for travel Hypothetically, the y have this choice is because they have more flexibility i n their w ork schedule. The presence of children may a ffect the choice of when people travel ; although, the age of children creates different effects. Steed and Bhat (2000) showed that age of children influence the preference to choose certain timeof day periods; for example, in households with children age 5 or under, travelers tend to choose to travel during the a.m. off peak, and p.m. off peak, while adults with children age 6 to 15, tend to choose a.m. peak to do shopping. For work trip, the presence of children under 16 years resulted in an earlier departure time choice (Chu, 2009). In the study, the effect is most pronounced in the period between 7:00 to 7:29. Similar argument was presented by He (2013). The study presented the presence of children as lifecycle component that turned significant in timeof day choice. Individuals with children tended to leave home during earlier periods than the peak. Most li kely, the childrens school schedule may have had an effect on the individual s with children. However, Okola (2003) found this variable insignificant, most likely because the focus of his study was on elderly people living with other adults or grown up children. Also, when the effect was tested on join dimensions between timeof day and mode choice, the presence of children significantly affect s the mode choice (Bhat, 1998). The status of employment affects the tendency to choose certain timeof day per iods. Steed and Bhat (2000) found that travelers that are employ ed by others and

PAGE 46

46 have long working hours are choosing evening period for shopping trips. In contrast, self employed travelers are more likely to do shopping and recreational activity during mid day period. Similarly, Bhat (1998) found that employed respondents preferred to shop in the evening period rather than a.m. off peak and p.m. off peak. Those effects might be caused by the work schedules Likewise, Abkowitz (1981) proved that work schedule flexibility was one of factors on departure time choice. Moreover, this study also concluded that occupation with categories such as professional/ technical or management/administration influenced the timeof day choice. In the same way, Chu (2009) considers work duration, full time employee, and the occupation categories as the factors on timeof day choice for workers. Those employment characteristics turned significant. Sasic and Habib (2013) found that office and professional workers tended to choose early morning travel times because of regular office time. Trip Characteristics V ariables Trip characteristics variables affected departure time choice for work trips (Chu, 2009). This study also found that travel time had negative relationship to the choice, which implie s that the workers chose later periods of time s when the travel time is less. Abkowitz (1981) used home and work locations of travelers as variables. The location was characterized as whether it had good access to transit as transit users. Steed and Bhat (2002) found trip travel time impacting recreational purpose. Therefore, the authors predict that shopping travelers try to choose timeof day that shortened the travel time. Distance is a significant variable in the connection to the variation in days of a week (Kumar & Levinson, 2008). Ozbay and Yanmal Tuzel (2008) considered distance

PAGE 47

47 as one trip related variable that influences departure time choice. Longer distance influence travelers to choose earlier times of travel especially for work trips. Distance is significant variable for departure time choice for trips with congestion charges (Saleh & Farrell, 2005). The choice of mode affects the timeof day travel choice (Abkowitz,198 1). Travelers who drive alone tend to choose earlier time period than those who use transit (Chu, 2009). The author suggested that possible reason was because transi t users try to avoid the off peak transit schedule and the longer waiting and transfer time. Summary of Time Of Day Choice and Built Environment Many studies have been found focusing on timeof day travel. Those studies utilized various methods, survey approaches, cases, and the stratification of trip purposes. However, the relationship of this travel behavior dimension with built environment variables has received less attention. Only a few studies have attempted to investigate the connection between built environment variables and the time of travel. From those studies, work trip and nonwork trip have different results of significance. Accordingly, this topic is worthy of further exploration. Built environment is a multidimensional issue. Accordingly, it includes many dimensions and measurements. For the same reason, dissimilar results of prior studies may be caused by the complexity of built environments. Researchers need to d etermine the proper unit of analysis, account for self selection t o different environments and understand the interrelation among the built environment factors as a part of conducting research. Furthermore, as is described in the following chapter, they need to focus on parking as an element of the built environment, especially in downtowns.

PAGE 48

48 CHAPTER 3 PARKING IN DOWNTOWN AS BUILT ENVIRONMENT FACTOR Demand Management is the sixth D of built environment dimensions according Ewing and Cervero (2010). TDM performs as mobility management and provides numerous strategies that are grouped into four categories : improved transportation options, incentives to shift mode, land use management and policies and programs (Litman, 2006), in which shifting the time of travel is also a part of each of those strategies (Stei ner, 1992). Furthermore, parking management is part of land use strategies Thus, parking demand management becomes one factor of the built environment. Parking supply and demand management refers to variety of parking management solutions that deal wi th both supply and demand, as what researchers in parking management, such as Shoup (2005) and Litman (2006) propose. The goals of parking supply and demand management are to maintain and maximize current parking resources through programs and policies ( Litman, 2006) Various strategies are identified, such as: pricing, benefit districts, fines, parking reduction and exemptions, incentives for alternative modes, the regulations for parking providers, the improvement of parking design and technology, public education, and institution coordination (Litman, 2006). This study focuses on three strategies: pricing based on time, parking restriction and parking availability. Due to intensive problems of parking in downtown, parking demand management for downtowns should be considered, especially for the ones that are autocentric and supply oriented.

PAGE 49

49 Parking Demand Management in Downtowns The auto oriented city era as what has been mentioned earlier has changed the activities in downtown from centralized into decentralized locations to suburban areas. Jakle and Sculle (2004) in their book Lots of Parking: Land Use in A Car Culture explained the changes in downtowns development and in downtowns parking orientation. T he decentralized activities happened because shoppers and workers continue to depend on automobile. Downtowns need to provide parking spots to resolve the demand. However, many parking spots in downtown have caused downtowns had a decrease share of regional retail floor space. Characteristics of retail declin e in downtowns are the high number of empty retail lots and the neglected downtown area in the night (Edwards, 1996). In turns, the decline of retail may stimulate specific downtown parking characteristics, such as long term parking and decreased peak hour parking around noon time (Edwards, 1996). The author compared thirty two small downtowns historical data of floor space, employments, and residential unit to understand the changes in traffic and parking characteristics. The study suggest s traffic calming strategies by lowered the speed limits, manag ing the signal timing, relocat ing truck movement, and improv ing onstreet parking provision. These strategies are intended to support the downtown development goals. Downtown Redevelopment and Parking Demand Management Downtown decline has been becom e the setting of many studies for downtown redevelopment (Jakle & Sculle, 2004; Balsas, 2004; and Robertson, 1995, 1997). The i deal downtown is characterized by Having retail businesses that are open to pedestrian and vehicular traffic; easy access; low amount of congestion during shopping hours; a leisurely pace that encourages window shopping; pedestrian amenities such as

PAGE 50

50 wide sidewalks, attractive paving, and street furniture; convenient parking; and a clean, well lit, and safe environment (Edward, 1996, p.109) This ideal form of downtown has underlined the proposal of downtown redevelopment. For example, Robertson (1995; 1997) proposed that the strategies for downtown redevelopment include: promoting pedestrian oriented development, indoor shopping centers, historic preservation, waterfront development, office development, and special activity generators. Along with the light of downtown redevelopment, the efforts to integrate land use and transportation have been introduced by several concepts, such as smart growth and transit oriented development. Parking demand and supply manag ement has role in determining the success of those downtown redevelopment strategies. The decline of downtown that characterized by many self sufficient high rise buildings with attached parking garages within the building and surrounded with openlot park ing in the transition zones around downtown, has promoted downtown fragmentation (Jakle & Sculle, 2004) and has reduced downtown street livability significantly such as losing its pedestrian traffic. As a consequence, having downtown redevelopments along w ith the strategies requires parking related policy or management in the design. Willson (2005) performed a study focusing on t ransit o riented d evelopment (TOD) and parking policy and suggested that in order to make a TOD neighborhood functional, it should consider parking supply and parking policy other than solely the proximity to transit. Since many strategies are under the umbrella of parking demand and supply management, the following section highlights those that have connection to peak hour management.

PAGE 51

51 Peak Hour Parking Strategies The parking providers and pertinent governments may find challenges in examining how to provide parking spaces. One of the challenges is emphasizing the peak hours. When people travel in similar range of hours, it will bring two sides of parking provisions. In the one side, when demand of parking space is predicted from nonpeak hours, the shortage of parking space occurs in the peak hours. As consequences, this shortage causes inefficient searching time, cruising for parking illegal parking and congestion. On the other side, when demand was predicted based on the peak hours, ample parking spaces happen. This causes underutilized parking spaces at nonpeak hours, and dispersed development, or sprawl (Shoup 2005; Litman 2006). Thus, variable parking pricing and parking space restriction based on time are parking strategies focusing in the peak hours. First, one determinant o f parking demand is cost, which facilitates parking pricing a s an effective way to manage parking demand. Variable parking based on time charges the parkers based on the popularity of hours from existing demand. Second, besides charging fix prices for all day, off street parking providers and onstreet parking meters adjust higher price or apply the restricti on of those parking spaces in the peak hour times, such as morning from 7 am to 9 a.m. and afternoon from 3 pm to 6 pm. To provide the information of peak hour popularity time of day parking study is required. The f ollowing section reviews prior timeof d ay study focusing in parking. Past Studies on Time o f Day and Parking Focus Studies that highlighted the timeof day dimension on parking focus are Lam et al. (2006), Hess (2001), Shiftan and BurdEden (2001) and Hensher and King (2001).

PAGE 52

52 Lam et al. (2006) emphasized the important of timedependent analysis on parking and travel choice model. The authors suggested that departure time choices have the relationship with parking activities, e.g. the choice of parking location, searching time delay for parking, and parking charges. The study proposed a network equilibrium model that considered the departure time, route, parking location, and parking duration. Second, Hess (2001) did not examine directly on timeof day dimension; instead, he used multinomial logi t to seek the significant variables for commuter mode choice at Portland, Oregon. The author distinguishes the commuters based on the morning peak hours. This study suggested insignificant land use variables and concludes that parking cost and the travel t ime are two determinants o f mode choice. Third, Shiftan and BurdEden (2001) test two proposed parking policies: the increase of parking cost and the decrease of parking availability. This study suggest s that in Haifa, Israel, workers tend to switch their mode of transportation and to modify their time of day in responding parking policies. Also, they found that nonworkers may consider not only those two responses but also all other possible responses such as to continue as car users, to shift to taxi, to walk, to cancel the trip, or to change destination. However, this study found that nonworkers tend to change their destination. Last ly Hensher and King (2001) compared two parking policies parking pricing and supply by time of day and showed that shifting departure time is one of the responses to the policies. Actually, the study uses a stated preferences study to understand the travelers parking location choices in Sydney CBD during 1998. As the result, the authors conclude that only 3 percent of respondents chose parking location

PAGE 53

53 regarding the supply by time of day while the other 97 percent of travelers chose the location based on parking price. Variables Selection Three parking variables on downtown level are the number of available parking lots, the average parking prices, and the percentage of timerestriction parking. First, the number of available parking lots has been considered as one variable that connects built environment and travel behavior. Cervero and Kockelman (1997) incorporated parking variables in mode choice model. In this article, they calculate the proportion of commercial retail and service parcel with off street parking; off street parking between the store and curb; onstreet; drive ins or drivethroug h. The variable of parking availability for retail may give the probability of 56% shoppers choosing to drive alone. Another study also used this variable in their timeof day model in the previous section ( Shiftan & Bur d Eden, 2001). When they included the decrease of parking availability in the model, workers tend to choose their behaviors by changing the mode or their timeof day choice. Second, the average parking prices can be considered as a variable in the model as other studies suggested on prev ious research of travel behavior and built environment. Paid parking may urge people to walk if they shop or do nonwork activities (Cervero & Kockelman,1997). Some others studies about timeof day have incorporated parking pricing as one variable in the model ( Hensher & King, 2001 ; Shiftan & Burd Eden 2001; Lam et al. 2006) All of those studies agree that parking pricing has the connection to peoples travel behavior either choosing the parking location, changing their mode of transportation, or changi ng their departure time.

PAGE 54

54 Third, time restriction parking means the supply varies by time of day. Some of parking lot can be used only for short duration, around 2 to 3 hours; while, some others can be used for long durations. This variable is considered as one of parking variables in the model because people choose their location based on supply by timeof day (Hensher & King, 2001 ). Actually, other parking variables may be included in the model, such as duration of parking. However, because the secondar y data does not cover the duration information, this study cannot incorporate this variable Summary of Parking Management in Downtown Automobile dependence has influenced the decline of downtown that is characterized by the rate of retail vacant lots and abandoned downtown in the night. Parking conditions are also typified downtown declinationintensive parking garage within self sufficient high rise buildings and openspace parking lots surrounding downtowns. Furthermore, this downtown decline has encourag ed the downtown revitalization efforts, e.g. by promoting pedestrianoriented development and other related strategies to promote downtown livability. Thus, parking should be also integrated in those strategies. Parking demand and supply management has been introduced to have efficient means of parking in downtown; variable parking pricing based on peak hour popularity and parking restriction on some time periods are two such strategies As a consequence, timeof day choice study has an important role in u nderstanding the peak hour popularity. However, only few studies have focused on timeof day and parking. Therefore, it is plausible to do further investigation in this area of interest

PAGE 55

55 CHAPTER 4 METHODOLOGY AND DATA Research Design Taking the type of data into account, this study utilizes the cross sectional research design. The cross sectional study design takes place when the data have one reference period of the study and one population (Kumar, 2011). This study design is also simple and cheap in analysis procedure; however, it is not able to measure the differential changes (Kumar, 2011). Despite the weakness, the cross sectional study design is best suited for study aiming to understand overall picture of the study area (Kumar, 2011), in this study, overall pictures of timeof day choice in the Southeast Florida region and the relationship with parking in two downtowns. Regional Level Regional Hypothesi s There is a relationship between travel behavior (timeof day) and the built environment in the Southeast Florida This study seeks a position within different existing arguments about the relationship of travel behavior and built environment. More specifically, from the literature review, less attention has been given to the effect of built enviro nment on timeof day choice. Therefore, identifying the correlation of built environment variables and various travel behavior dimensions becomes important in order to capture the overall understanding of the relationship between those two variables. Furt hermore, the finding of this hypothesis may reveal what the built environment factors that influence the travel behavior are. By employing the logistic probit regression, the analysis shows the correlation and the significance of each built

PAGE 56

56 environment factor. Specific method used to analyze the relationship is the ordered response model. The combinations of built environment factors have greater effect than individual factor of built environment on the connection to time of day choice This study attem pts to understand the overall effect of combined built environment factors and compare to that of individual factors. As previous study mentioned (Krizek, 2003; and Cervero, 2003) built environment factors come with interaction among them, the following analysis compares the model with combined factors using ordinal measures on GIS calculation and individual based factors that have been tested on hypothesis 1. The ordered probit model is still used to test the overall relationship. Prior to put on overal l regression, combined built environment factors are calculated using LandUse Conflict Identification Strategy (LUCIS) ( Carr & Zwick, 2007) This method enables to capture more than two built environment factors. Study Area This study chooses those cou nties for several reasons First, the state of Florida has specific data about individual parcels. To specify the area, this study focuses on three of counties within the state of Florida. Second, these study areas have given less attention in regard to the relationship of built environment and travel behavior; only three studies at the South Florida are found with different focus on travel behavior (Noeg, 2009, Steiner et al., 2010, and Ewing & Messenger, 1996). None of them focuses on time of day behavior in the relationship with built environment variables. Meanwhile, there have been many studies on the relationship in California ( Abkowitz, 1981; Bhat, 1998; He, 2013; Lemp et al. 2010; Sall & Bhat, 2007; Kitamura et al., 1998. Third, two downtowns t hat have parking lots information are within these counties. This study

PAGE 57

57 considers Palm Beach County because it is still located in the influenced area of travelers who visit those downtowns. Figure 41 shows the study area and the home locations of travelers in the NHTS Florida add on data. Figure 41 Study a rea ( r egional) According to US Census Bureau (2013), total population of the South Florida for 2010 is 5,564,635 people. The population of each county is as follows : 2,496,435 people in Miami Dade County, 1,748,066 people in Broward County, and 1,320,134 people in Palm Beach County. Data Sources The m ain data for time of day information is ga ther ed from the 2009 NHTS. Specifically, Florida Department of Transportation (FDOT) paid for add on data that collected a larger sample of Florida households to allow metropolitan planning organizations ( MPOs ) and other agenc ies to develop transportation models and other

PAGE 58

58 planning activities Each traveler socioeconomy characteri stics and trip related information is attached to a unique ID and its location. Furthermore, each location of travelers is linked to the built environment properties that are derived from other sources. Various sources are identified to have proper data on the built environment factors. First, this study uses parcel data from the Florida Geographic Data Library (FGDL) website. This parcel data completes the information about the variety of land use for each parcel to be calculated in the next step. Sec ond, block groups level data, which is also obtained from the same website, are used to get the population density, employment density, and diversity of land use. Not all block groups are used, but only the ones that intersect the geocoded location of res pondents. Third, this study also utilizes the Longitudinal Employer Household Dynamics (LEHD) 2009 data. Specifically, the employment data per block group in three counties are used to capture the employment density. Specific file that is used in this s tudy is all jobs in the year 2009. The l ast source is 2010 Florida Traffic Information and Highway Data from Florida Department of Transportation and NAVTEQ network map. These sources give information on the details of roadway network for the calculation of the number of intersections and cul desacs using short distance toolbox in GIS. Unit of Analysis Mainly, the persontrip is used as the unit of analysis. This data cover s the information of gender, age, and worker status. NHTS 2009 data consist of p ersontrip, household, trip, vehicle and location data. Figure 4 2 illustrates the relationship among those data. This study does not use any of vehicle data. Moreover, this study takes specific variables within those groups of data as follows. First, within the household

PAGE 59

59 data, this study uses income, the number of vehicle, and the number of children Second, the trip data show s the information of a unique ID for each trip that contains the information about trip purpose, location, and mode used. Last ly, the location data, which gives information about the geocodes of longitude and latitude of home and trip end f or each persontrip. Figure 42 The r elationship between f our NHTS files (FDOT, 2010, p.9) For the built environment variables, this study chooses block groups as unit of analysis. The f irst reason for this choice is that it is the smallest geographical area available for calculationblock may be too small and confusing since the geocoding proc ess was referred to different places such as nearby intersections, or streets, or parcels between blocks boundaries. The s econd reason is that bigger geographical area such as T AZ or Census Track may not contain some detailed information The Data Processi ng This study sorts the database into three counties based on trip purposes. Then, the sorted database is merged with built environment variables. First, for travel data, this study uses Microsoft Excel to focus on three counties before the data are imported to SPSS. From location data, three county of trip destination ( Federal Information

PAGE 60

60 Processing Standard ( FIPS ) code are selected: 011 for Broward County, 086 for Miami Dade County, and 099 for Palm Beach County. This process extracts 3,980 households 8,040 persons, and 29,274 trips for Southeast Florida from total of 15,884 households; 32,065 persons; and 114,910 trips for the state of Florida. Furthermore, the variables are sorted into the requirements based on literature using SPSS software. Thi s study selects two trip purposes work trip and shopping trip for which the data contains 2,574 and 7,029 trips, respectively. Furthermore, the data cleaning processes include missing value, refused answer, appropriate skip ( 1 value), and not ascertained ( 9) value. Because this study considers that workdays have more problems related to the peak hours, weekend trips are excluded. Then, t h e se exclusions reduce the total number of shopping trip s to 3,468 and the number of work trips to 2,069. Moreover, for the purpose of incorporating the built environment factors, this study only uses data with valid locations. First, this study utilizes GIS to place the x and y coordinates for home, trip end, and working place locations. Features that are used to link those locations with FGDL data are spatial join, join table, and relate tables. Then, the locations for trip end are linked into employment data from LEHD data. Meanwhile, home locations are linked into population density from FGDL block group 2010. Third, the diversity variable is calculated using formula 38. To produce total area per different land use categories, this study rearranges the land use categories of parcels into four types: residential, retail and commercial parcels, offices, and others. Using spatial join and summary statistic tools in GIS, total areas of each category are calculated and ready for further calculation of diversity.

PAGE 61

61 Network data are used for calculating the distance for each traveler and for finding the number of intersection and cul desacs. Data processing for these variables includes the network analyst tool in GIS. The process gives the distance for each pair of home location and end trip location. Furthermore, total intersections and cul de sacs are united by block groups using summary statistics. The l ast process for built environment variable is to combine the variables using basic calculation of LUCIS procedure. Instead of using the raster, this study uses simple field calculation with this formula: = ( 100 ) + ( 10 ) + ( 1 ) (4 1 ) Data Set The time of day choice model is specified as the function of built environment variables, socioeconomic variables, and level of service variables. Final specification of variables includes one dependent variable categorical choice of end time. Independent variables are (complete speci fication can be seen on the table A 1): Built environment variables include population density at home location, employment density at destination, diversity of land use the number of intersections, and the number of cul de sacs. Socio economic variables consist of gender, age, income, the presence of child, working status, household vehicles, household size, employment status, and number of workers in the household. Trip characteristics variables contain mode of transportation, trip purpose, calculated tr ip distance in miles, and reported travel times.

PAGE 62

62 Downtown Level Downtown Hypothesis Trip purpose determines different travel behavior, especially related to time of day choice. Previous studies have found that generally work trip and nonwork trip have di fferent choices on travel behavior, and specifically on timeof day choice. Thus, this study performs the analysis that based on those purposes; however, despite the classif ication of trips into nonwork trip, this study chooses shopping trips. S hopping trips are chosen because: (1) the occurrences of shopping trips are more frequent than other nonwork trips, and (2) like work activities, shopping activities are a basic need, so the comparison between those two may be reasonable. The method to underst and these trip purposes is by incorporating this variable within the model. Also, trip purpose is included within the model for the two CBDs of Fort Lauderdale and Miami The average parking rate determines timeof day choice at two CBDs destinations Miami and Fort Lauderdale. This study tests the relationship between the average parking rate in downtown and timeof day specifically at downtown destinations. The relationship includes the different trip purposes as one variable. Additionally, two others parking variables parking availability and the percentage of parking space with timerestriction are examined in the model. Thus, the result of this test may reveal the parking demand management strategies that are appropriate for each tri p purposes in downtown area, especially the ones with timesensitive needs, such as variable parking pricing and time restriction policies. However, because of the limited data for downtown areas (less than 200 respondents), this study also explores the r elationship with descriptive statistics.

PAGE 63

63 CBD Study Area This study selects two downtowns Fort Lauderdale and Miami. The main consideration in choosing these downtowns is because they have information of parking rate and availability. The CBD of these downtowns were included in a previous study on parking supply and demand management in which the author participated. Fort Lauderdale CBD is about 0.57 square miles, with the boundaries: NE 6th Street on the north, New River on the southwest, SE 7th Street on the southeast, NE/SE 5th Terrace on the east, NW 2nd Avenue on the west, and the boundary of SW 7th Avenue and SW 2nd Avenue from north to south (Steiner et al., 2012). The Miami CBD area encompasses 1.7 square miles. In this study, the boundaries are : NE 9th Street on the North, Miami River on the South, Biscayne Boulevard and Bay front Park on the east, and SE 1st Avenue on the West (Steiner et al., 2012). The Figure 4 3 displays the area for study cases. In addition, this study also checks the parc els of parking spaces using Google maps. Figure 4 3 shows parking i nventories for both downtowns.

PAGE 64

64 Figure 43 A) The study cases b oundaries B) P arking i nventories at Fort Lauderdale CBD, and C) Parking i nventories at Miami CBD Data and Unit of Analysis for CBD Level The main data for the CBDs is similar to the regional level. F or downtown Miami and Fort Lauderdale the transportation supply data was gathered through a field survey. Since the parking inventory is based on parcel level data, the unit of analysis for downtown refers to the built environment factors, i.e. block groups. The unit of analysis for downtown data is the same as for the regional level, the person data. However, parking information is gained from parcel level calculation. In the analysis, parking data is grouped into block groups.

PAGE 65

65 To get downtown information, this study delineates the respondents of NHTS that have tripend location in two downtowns. Because of the limited numbers of respondents, thi s study does not conduct similar data processing as the regional level. In the regional level this study uses only respondents with shopping trips and work trips. In downtown analysis, this study uses the all of the trips of respondents. After the cle aning processes which include missing value, refused answer, appropriate skip ( 1 value), and not ascertained ( 9) value, 120 trips are included in the Miami CBD and 83 trips in the Fort Lauderdale CBD For CBD level, t he timeof day choice model is speci fied as a function of parking as a built environment variable and socio economic variables. This study does not consider other built environment factors since the condition of these factors within the CBD are relatively similar. The f inal specification o f the variables includes one dependent variablecategorical choice of end time. The i ndependent variables are: Parking as built environment variables include the average number of parking lots the average of price, and the percentage of parking restriction. Socio economic variables and t rip characteristics variables similar to regional level The Model Framework In general, the relationship between hypotheses and model used in this study can be seen in the Figure 44 This study examines the relationship of built environment and timeof day using Ordered Choice Model. Subsequently, this study also tests the combined built environment rather than individual effects as previous step using the LUCI S model. The results of thes e models are then compared. T rip purposes are also included in the timeof day choice model and are segmented based on the type of trip work trip and shopping trip. In addition to those analyses, this study explores the cent er

PAGE 66

66 of regional destinations for each trip purposes and descriptive statistics for the relationship of timeof day and average parking price. The following sections describe the Ordered Choice Model, and LUCIS procedures. Figure 44 The study f ramework Ordered Choice Model This study utilizes an ordered response model or propensity based model to understand which attributes of socioeconomic, trip characteristics, and built environment variables have an effect on travel time departure choice. Several considerations are underlined in the chosen model. First, the focus on this study is ordinal value in nature; the departure time choice is shown by the consecutives values, especially when the times are divided into discrete choices. The ordered response model or propensity based is properly suitable for ordinal choice. Second, the consideration is also based on the weakness of MNL model that accounts to the IIA property. This might be a problem to choices that have ordinal value in nature. As a

PAGE 67

67 consequence, the Ordered Response Model is considered more appropriate than the MNL. Lastly, based on popularity, the propensity model is favorable. The theory of Ordered Response Model follows McKelvey and Zavoina (1975) work. Given that the departure time choices for person q are: = { 0 1 2 3 , 1 } 6 (4 2 ) Furthermore, the total of propensity function is = + (4 3 ) Where the Vq is the systematic or observed propensity and q is random or unobserved, which is assumed as the normal distribution. Since when the alternative is chosen: = (4 4 ) ( ) = (4 5 ) Then, if the assumption is q is standard logistic. The probability of person q to choose the alternatives is on the following options: ( ) = (4 6 ) ( )= 1 1 + (4 7 ) For the operationalization of this ordered response methodology, this study uses SPSS software that will provide the threshold and the t significance for each variable. Because this st udy utilizes 6 choices, the SPSS give the number of thresholds: (1) the number of = k 1, with k is 6 means 5 thresholds are given (first threshold is 0, then variables.

PAGE 68

68 LUCIS and S uitability Model This study employs a part of suitability analysis, specifically in combine calculation using these steps. First, combine calculation is useful to response the one of weaknesses in incorporating the built environment factors into the model, i.e. the built environment factors come together or be highly correlated, in the field. Furthermore, the preprocessing in GIS for the built environment factors is done before the choice mod el analysis. Built environment factors are ordinal or nominal in the statistical measurements, such as 0 and 1 for entropy or low medium high for density. Second, this study is calculating the unique values that represent all nominal/ordinal measurements. The method is defined as conflict mapping that is an adaptation from LandUse Conflict Identification Strategy (LUCIS). According to Carr and Zwick (2007), LUCIS has three conflict categories, i.e. urban, conservation, and agriculture, where each of the categories has highmedium low preferences. When those preferences and categories are combined, there are 27 combinations (33). If these combinations are displayed on the raster cells, the conflict space diagram will look as the Figure 45 In this stu dy, the three conflicts, or convergence, categories are defined as density, diversity and design, or street connectivity. This study attempts to measure the built environment factors as a combination. The conflict mapping helps simplifying the model and r educes the parameters of built environment. By introducing this conflict approach, following advantages are defined. First, it may represent the need of combining the built environment variables. According to Cervero (2003, p.120), dimensions of built environments tend to operate in tandem and synergistically . As consequences, collinearity problem appears in the statistical analysis. Modifying variables into one combination may help to reduce the collinierity

PAGE 69

69 problem without losing the information. Second, it may simplify the model. Having all built environment factors in a model is complex. Since the model has to incorporate other factors besides built environment, such as socioeconomic factors, and trip characteristic factors, testing all built environment variables in one model seems exhaustive and requires long process. Besides the weakness explained in previous point, those factors should come in unity. Long specification of built environment factors in a model is not preferable. Lastly, it gives a proposed input for built environment research to specify the combination in the relationship of built environment and travel behavior model. By testing the different approaches in calculating combination of built environment, this study may enrich the dimension of built environment and travel research. Figure 45 The conflict space diagram of 27 combinations: three categories and three ordinal measures (Carr & Zwick, 2007, p.147)

PAGE 70

70 CHAPTER 5 ANALYSIS AND RESULTS Regional Level This section describes the dependent variable and independent variables of socio economy, trip characteristics, and built environment. The cross tabulation of the relationship between each group of variables is also provided. Descriptive S tatistics Dependent variable The dependent variable in this study is discrete choice of timeof day period that indicates the time of arrival at destination for each traveler. Available NHTS data on endtime military variable is converted into a discrete variable that represents the travel time periods of choice. This study uses six periods: a.m. peak (6:00 to 8:59), a.m. midday (9:00 11:59), p.m. mid day (12:00 14:59), p.m. peak (15:00 17:59), p.m. early evening (18:00 20:59), and other (21:00 5:59). Table 5 1 provides the number of travelers at a given time period and its corresponding percentages. The travelers are classified based on two trip purposes working and shopping. Table 51 Descriptive statistics for arrival time at destinations for work and shopping trips Time Period Work trips Shopping Trips Frequency Percentage Frequency Percentage Other (21:00 5:59) 58 2.8 62 1.8 a.m. peak (6:00 to 8:59) 955 46.2 193 5.6 a.m. mid day (9:00 11:59) 443 21.4 1055 30.4 p.m. mid day (12:00 14:59) 405 19.6 1013 29.2 p.m. peak (15:00 17:59) 173 8.4 804 23.2 p.m. early evening (18:00 20:59) 35 1.7 341 9.8 Total 2069 100 3468 100

PAGE 71

71 The t wo trip purposes i n this study clearly have different tripend arrival time as can be seen on the Figure 5 1 Figure 51 Time of day variation for work trips and shopping trips As shown in Figure 5 1 most work travelers arrive at work during the a.m. peak. The second highest period for travelers to arrive at work is a.m. mid day with approximately 21.4% of arrival travel, and the third highest, is the p.m. midday, with approximately 19.6% of travelers arriving during that period. In contrast, travelers with shopping as a trip purpose arrive at their destination beginning during the mid day period, with almost equal percentages in the a.m. and p.m. midday with 30.4% and 29.2%, respectively The next highest period of arrival for shopping is in the p.m. peak, with 23.2%. The arrival time at work is at its lowest during the p.m. early evening, while t he least chosen period for arriving for both trip purposes is other (21:00 to 5:59). The independent variables in this study are classified into three subgroups. First, the socioeconomy variables include gender, working status, number of drivers in the household, household income, number of household members, number of vehicles in the household, number of adults, presence of child, and age of res pondent Second,

PAGE 72

72 the trip characteristics variables cover trip purpose, mode of transportation, the distance of travel and calculated travel time. Third, the built environment variables are density, diversity, and design. Descriptive of socio economic and demographic variables This study differentiates socioeconomy variables based on the level of measurements. First, nominal measures are used for gender (male, female) income, working status (yes, no) and presence of child (yes, no) Second, scale m easures are used for the number of drivers, number of household members, number of vehicles, number of adults, and age of respondent. This study predicts that gender may have an effect on time of day choice. Figure 5 2 displays the difference between male and female in time of day arrival at their destination for work and shopping trips. In work trips, both male and female respondents dominate in arrival during the a.m. peak time. Females are more likely than males to arrive at work during the a.m. peak time. For shopping trips, male and respondents arrive in almost equal proportions during the a.m. mid day, p.m. mid day, and p.m. peak periods. Figure 52 Gender of respondent and timeof day of arrival for : A) Work trips. B) Shopping trips C) Both work and shopping trips

PAGE 73

73 This study categorizes income into three levels: low, medium, and high. The l ow category is used when the income is below $30,000. The m edium category is used when the income is on the range between $30,000 and $55,000. High category is used when the income is above $55,000. There is slight difference between r espondents with low, medium, and high income in choosing to arrive d uring the a.m. peak for work trips with middle income worker most likely to arrive during this time. This is not suprising because 46% of arrivals occur during the morning peak. For shopping trips, a high percentage of low income respondents choose a.m. midday. The late morning (a .m. mid day ) and early afternoon ( p.m. mid day ) are chosen by most of the medium income respondents. The percentages of high income respondents choosing a.m. mid day, p.m. mid day, and p.m. peak are almost equal. Figure 5 3 displays the percentage of time of day choice in three income levels. Figure 53 I ncome and time of day arrival : A) Work trips. B) Shopping trips C) Both work and shopping trips

PAGE 74

74 Hypothetically, the worker status affects the time of day choice. The a .m. peak is a dominant choice when respondents arrive for work trips regardless of their worker status. Most of respondents who are not in worker status choose a.m. midday for shopping trips. A high percentage of respondents in worker status choose the p.m. peak for shopping trips, while the percentages of a.m. midday and p.m. mid day are almost similar. A bout 37 respondents whose status is a nonw orker but hav e work trips rather than shop trips. These quantities might be because they have a temporary status of workers but they have work trips. Figure 5 4 provides distribution of the arrival time based on worker status. This study obtains the information regarding the presence of child from a life cycle classification. Out of ten categories in the life cycle data, four categories indicate no children in the household. After recategorizing the data, this study considers four categories of the presence of child: (1) no child, (2) child aged 05, (3) child aged 615, and (4) child aged 1521. Most travelers choose a.m. peak for arriving a work regardless of they have children or not. Travelers with no children mostly ch oose to arrive at their destination during the a.m. mid day for shopping trips. The p.m. peak period has the highest percentage of travelers arriving at their destination with 0 to 5 year old children and travelers with 16 to 21 year old children. Meanwhile, the percentages of travelers with children of 6 to 15 years old choosing to arrive at their destination for shopping during the a.m. mid day, p.m. mid day, and p.m. peak are almost similar. In both work trip and shopping trip, total percentages of tr avelers with 0 to 5 and 6 to 15 year old children choosing to arrive at their destination during the a.m.

PAGE 75

75 peak are higher than other periods of time. Figure 5 5 displays the presence of child information. Table 5 2 and 5 3 display the socioeconomy variabl es with scale measures for work trips and shopping trips respectively. Although all of these variables have s imilar ranges except the age of respondent, the mean of variables are different The range of age for work trips is between16 and 91, while the ra nge of age for shopping trips is between 16 and 97. The mean differences include the number of driver, people, vehicle, adults, and worker in the household. F or example, mean value for the number of driver in a household is 2.19 for work trip; while, the m ean value of that for shopping trip is 1.91. Another example is observed on the number of workers in the household. The mean value, at 1.70, of the number of working family members for work trips is higher than that for shopping trips which is only 0.83. Figure 54 W orker status and timeof day arrival for : A) Work trips. B) Shopping trips C) Both work and shopping trips

PAGE 76

76 Figure 55 P resence of child and timeof day arrival for : A) Work trips. B) Shopping trips C) Both work and shopping trips Table 52 Descriptive of socio economic variables with scale measures for work trips Variables N Range Minimum Maximum Mean Std. Deviation HHdriver 2069 7 0 7 2.19 .805 HHSize 2069 9 1 10 2.83 1.259 HHVeh 2069 12 0 12 2.30 1.043 HHAdult 2069 9 1 10 2.23 .839 Age 2069 75 16 91 49.24 12.838 HHworkers 2069 4 0 4 1.70 .702 Table 53 Descriptive of socio econom ic variables with scale measures for shopping trips Variables N Range Minimum Maximum Mean Std. Deviation HH driver 3468 7 0 7 1.91 .835 HH Size 3468 9 1 10 2.37 1.250 HH Veh 3468 12 0 12 1.89 .977 HH Adult 3468 9 1 10 2.04 .887 Age 3468 81 16 97 60.16 16.348 HH workers 3468 4 0 4 .83 .892

PAGE 77

77 Descriptive of trip characteristic variables Mode of transportation may determine the timeof day choice for travelers in relation with the schedules. This study categorizes modes of transportation into five groups: automobiles, nonmotorized vehicles, transit systems, motorcycles, and others. Automobiles are the majority of mode used by travelers for both work and shop purposes as we can see from Table 5 4 Other than automobile, travelers use nonmotorized and transit as the second and third biggest number of respondents; however, the number of respondents on both those modes account for less than 5 percent of total respondents. Table 54 The mode of transportation for work trips and shopping trips Work trip Shopping trip Total Auto 1925 3262 5187 Non motorized 79 156 235 Transit 39 35 74 Motorcycle 15 11 26 Others 4 3 7 2062 3467 5529 Overall automobile users have equal preference on a.m. peak, a.m. midday, and p.m. mid day. However, most automobile users that have work trip choose a.m. peak; while most of automobile users that have shopping trip choose a.m. midday or p.m. mid day. Mo torcycle users choose p.m. midday over other periods for work trips and a.m. mid day for shopping trips. Higher percentage of transit users choose a.m. peak as the period for work trips and either a.m. midday or p.m. midday as the period for shopping t rips. Figure 56 shows the descriptive for mode of transportation.

PAGE 78

78 Figure 56 M ode of travel used by respondents for trips and time of day of arrival : A) Work trips. B) Shopping trips C) Both work and shopping trips This study uses the reported travel time from the travelers. Table 5 5 displays the range distribution of travel time for work and shopping trips Although the maximum time spent on travel for a shopping trip is 330 minutes, which is longer than th e long est work trip by 100 minutes the mean of time spent for travel for shopping trips is less than the time spent on traveling to work. The mean time spent for a shopping trip is 14.03 minutes, while the mean time spent for a work trip is 23.56 minutes. Thi s research calculates the distance for each traveler using the shortest distance tool in GIS. The distance is calculated from the pair of coordinates between home and the end trip. For work trips, the minimum distance is 0.05 miles and the maximum distance is 80.022 miles. The mean distance for work trips is 10.12 miles. For shopping trips, the minimum distance travelled is 0.005 mile and the maximum distance travelled is 82.64 miles. The mean value for shopping trips is 4.33 miles. Table 5 6 shows th e complete descriptive of distance variable.

PAGE 79

79 Table 55 Table for travel time spent by working and shopping travelers N Range Minimum Maximum Mean Std. Deviation Travel time in minutes for work trips 2069 219 1 220 23.56 17.7 9 Travel time in minutes for shopping trips 3468 329 1 330 14.03 12.5 5 Table 56 Descriptive of distance (mile) for work and shopping trips N Range Minimum Maximum Mean Std. Deviation Distance 2052 79.97 0.050 80.02 10.118 8.53 Valid N (listwise) Work Trip 2052 Distance 3459 82.6 4 0.005 82.64 4.336 5.22 Valid N (listwise) Shopping trip 3459 Descriptive of built environment variables The f irst built environment variable is density. This study measures the employment density at the destination (trip end) and population density at home location as the measure of density. The employment density is between 0 and 145.69 workers per acre at shopping trip destinations is and between 0.001 and 467.47 workers per acre at work trip destinations. There are 913 block groups as trip destinations for working purpose and 1217 block groups for shopping purpose. The means of density are 7.945 and 4.998 workers per acre for shopping and work trips respectivel y. Figure 5 7 shows the employment density with three classifications based on geometrical interval ; the lowest density block groups are recategorized into the 1 category while the highest are recategorized into the category Category 2 has the interval between categories 1 and 3.

PAGE 80

80 Figure 57 E mployment density at trip end location: A) Work trip, B) Shopping trip This study incorporates the population densities at the origin (i.e., home ) locations for both work and shopping trips. For both trips, the density at the home location ranges between 0 and 180. 60 people per acre. A total of 840 and 1103 block groups for home locations are represented in the sample for working and shopping trips respectively The means of population densities for home origins are 10.45 (work trips) and 10.62 people per acre for (shopping trips). In the analysis, this study classifies population density into three classes based on the geometric interval. Figure 58 displays the population density throughout the study area; the green categories (value = (A) (B) (A) (B)

PAGE 81

81 1 ) show the lowest population density, while the red category shows the highest density of population (value = 3 ) Figure 58 Built environment variable for population density at home location: A) work trip, B) Shopping trip This study incorporates the diversity dimension by calculating proportion of four different land use areas. This study gather s the information of land use by summing up the parcel database into block groups. As a consequence, the range of diversity is between 0 and 0.907, with the highest numbers having the greatest diversity Figure 5 9 shows the land use diversity with three classifications based on geometric interval. (A) (B)

PAGE 82

82 Figure 59 Land use diversity at trip end locations in the Southeast Florida This study uses intersection density and cul de sac density as design dimension. For work trips, the range of intersection density is between 0 and 1.0038 intersections per acre and the range of cul de sac density is between 0 and 0.33 cul desacs per acre at the work destinations Shopping trips have the range between 0 and 0.928 intersections per acre and the range between 0 and 0.434 cul de sacs per acre at shopping destinations The trip end associated block groups have the number of intersections ranging from 0 to 526 and the number of cul desacs ranging from 0 to

PAGE 83

83 132. The means for intersecti on density are 0.239 (work trips) and 0.245 (shopping trips). Moreover, the means for cul desacs density are 0.037 (work trips) and 0.043 (shopping trips). Figure 5 10 displays design variables of built environment ; the locations with the lowest intersection density have been reclassified as 1 while the locations with the highest density are reclassified as 3. Figure 510. Design dimension of built environment variables: A) Intersection density B) Culdesacs d ensity This study uses the field calculator to produce combined built environment variables. The formula to obtain the combined values is : = [( ) 100 + ( ) 10 + ( ) 1 ] (5 1 ) As a result, this variable has 27 combinations. Figure 410 displays the descriptive of combined built environment variables. Few block groups hav e all built environment A B

PAGE 84

84 variables in the high categories (code: 333), i.e. 22 out of 1217 for shopping locations, and 9 out of 913 block groups for working locations. For the least favorable built environment category (code: 111), 68 out of 913 are assoc iated with work trip locations. For shopping trips, 17 out of 1217 block groups are of category of 111. T he most common combination for respondent destination is medium employment density, high diversity and medium intersection density (code =232), which is 13.8 percent of all respondents. The l east combination of built environment chosen by respondents is high employment density, high diversity and low intersection density (code =331) or only 0.2 percent. Figure 511 shows locations of combined built environment variables. Figure 511. The combined built environment variables

PAGE 85

85 T his study compares the models involving individual built environment variables and combined built environment variables as the analysis at the regional level using ordered probit models. The models predict the probability of timeof day choice for arrival time. At the CBD level, this study examines the trips that have trip end locations in the CBDs study cases. The results of the regional mod els are i n the Table 48 and the CBD models are on Table 49. T h e next section explores the factors determining significant and insignificant variables in these models Regional Model and the Results Before performing Ordered Choice Model, this study us es correlation analysis to see the connectivity among variables. Variables that correlate with other and have coefficient more 0.5 are number of driver, number of household members, number of adult, number of vehicles, number of workers in households, age, the presence of child, low income, and medium income. This matrix of correlation analysis can be seen on Table B 1. Then, some representatives among these variables are selected to avoid the collinearity. After running the model, significant socio eco nomic variables include income, age, and the presence of child. T rip purpose and travel mode variables are significant among the group of trip characteristics (TR) variables. T his analysis treats the built environment variables differently. In the first estimation shown in the first column of Table 5 7 built environment (BE) variables are treated individually. Then, this analysis finds only one variable is significant for timeof day choicehigh cul desac density. When this analysis uses combined buil t environment variables, three combined variables are as significant as high cul de sac density.

PAGE 86

86 In order to compare the goodness of fit from those models, the T able 5 7 mentions the log likelihood at convergence and equal share. For the first model, the log likelihood value at convergence is 4077.3145, while at equal share is 4629.483. This gives the chi square value of 1104.338. Since the degrees of freedom ( df ) or the restriction variable between constant only model and final model is 7 and the st andard chi square value for 95% confidence interval is 14.07, the first model has better performance than the equal share model or constant only model. The second model has the log likelihood value at convergence of 4353.17 and at equal share of 4914.55. The chi square value for this model is 1122.753. The standard chi square value for df of 10 is 18.31. This proves the second model has better performance than the constant only model. Table 57 Estimation of time of day choice (ordered probit model) Variables SE + TR + BE (Individual) SE + TR + BE Combined Estimate Value t value Estimate Value t value Constant 1.933 19.45 1.976 19.735 Income high 0.138 4.669 0.139 4.738 Shopping trip 1.026 29.858 1.031 31.181 Age 0.003 3.036 0.003 3.095 Presence of c hild 0.078 2.149 0.077 2.111 Mode a uto 0.238 3.556 0.262 4.034 Mode t ransit 0.376 2.35 0.387 2.805 Cul de sac density high 0.1 3.471 0.098 3.446 Combination of BE all high 0.206 2.192 Combination 123 0.176 2.256 Combination 322 0.426 3.014 or mu (1) 1.423 5.466 1.425 5.861 or mu (2) 0.857 3.717 0.858 3.272 or mu (3) 0.756 11.743 0.757 11.265 or mu (4) 0.862 20.472 0.864 19.976 N 5537 5537 Log likelihood at convergence 4077.3145 4353.17 Log likelihood at equal shares 4629.483 4914.55 Chi square v alue 1104.338 1122.753 df 7 10

PAGE 87

87 In order to know which model is better, this study compares the chi square values of both models. Model 2 has a greater value of 2 (chi square) than that of model 1. The 2 value of model 2 is 1122.753, whereas the 2 value of model 1 is 1104.338. This comparison shows that model 2 explains the condition of population better than model 1. Furthermore, the following sections discuss the variable results by variable group. Socioeconomy variables Significant variables include high income, age of responde nt, and the presence of child. First, h igh income has a coefficient of 0.138, which is the highest among the socio economy variables for the final model that treats the built environment variables individually. Also, high income variable is significant for the time of day choice in all attempted models. The p ositive sign o n high income shows that people in high income category tend to have the propensity of later arrival time at the destination. The second significant socio economic variable is age of r espondent The coefficient of for age is 0.003. The negative value shows that as age increases, the propensity of choosing later time decreases. In the other words, older people tend to choose to travel earlier in the day As expected, the presence of child affects the travel time choice. This variable has coefficient of 0.078 in the first model and 0.077 in the second model. The estimated coefficients may imply that travelers having children tend to choose to travel earlier in the day.

PAGE 88

88 Trip characteristics variables The two significant trip characteristics variables are trip purpose and mode preference of automobiles and transit systems. First, the dummy variable of shopping trip purpose turns significant for timeof day choice with coefficients of 1.026 and 1.031. These positive coefficients imply that travelers with shopping related purpose tend to choose to travel later in the day. The second significant variable on trip characteristic group is mode preference f or travel by automobiles and transit. The coefficients of automobile choice are 0.238 and 0.262 in the first and second model respectively, while the coefficients of transit choice are 0.376 and 0.387. Negative coefficients indicate that travelers who choose those modes tend to choose earlier travel time s during the day. The condition that coefficient of transit is greater than coefficient of automobile shows a logical relationship. Built environment variables Culdesac density is significant across the models with coefficient of 0.1 for the first model and 0.098 for the second model. Positive values of those coefficients indicate that travelers choose later timeof day travel if their destinations have a higher dens ity of cul desacs. Since denser cul desacs in one area refl ect more autooriented characteristics, travelers may choose a later travel time because they do not want to wait in congestion during peak hours. E mployment density is significant in some models only if cul desac density is not included in the model. T his condition may result from the collinearity of the se two variables. In the second model, this condition does not occur. The cul de sac density is still significant even after all combined variables are included and significant.

PAGE 89

89 As expected, the combin ed built environment variable is significant for the second model. A coefficient of 0.206 indicates that the effect of built environment exists and is relatively highly explanatory of the travel time choice. Negative value means that travelers with thei r trips end s in locations with high employment density, high diversity, and high intersection density areas tend to choose earlier travel time periods for shopping and working activities. The reason underlining this preference may be that the travelers an ticipate congestion and competition for parking spaces. Besides the combined built environment variables with all high value of built environment factors, two other combinations of built environment factors are significant ly related to the travel time choice. Those are combination of 123 and combination of 322. The f irst combination reflects low employment density, medium diversity, and high intersection density ; these locations may include older residential neighborhoods with a mix of land uses The s econd combination indicates high employment density, medium diversity, and medium intersection density which would include older employment centers with adjacent residential neighborhoods The coefficients of those combinations are 0.176 and 0.4 26 respectively. Similar to previous combined variables, negative value indicates that travelers tend to choose to travel during earlier time periods. The coefficient of 322 combinations, which is bigger than that of other combined variables, points out greater employment density effects on time of day choice than other variables in combination. Downtown Level Descriptive Analysis of Parking Variables Overall, 120 trips and 83 trips were taken to the Miami CBD and Fort Lauderdale CBDs, respectively A fter linking these trips in the trip database, 9 trips to the Miami

PAGE 90

90 CBD and 2 trips to the Fort Lauderdale CBD were eliminated from the analysis because the home locations of those trips are not in Southeast Florida. Most of the trips to these CBDs are work trips. Although Miami CBD is the destination for only 18 percent of Home Based Work (HBW), most of the n on h ome b ased (NHB) trips are for work purposes28 out of 54 trips. The second largest proportion of the trips to the Miami CBD is for shop ping From a total of 97 trips, the shopping purpose accounts for 17.5 percent. A s imilar situation happens in the Fort Lauderdale CBD. The percentage of NHB trips is 38. 3 percent, while the percentage of HBW trips is 29.6 percent. From those NHB trips, 11 out of 31 trips are for work related purposes. The cross tabulation analysis shows that most of the HBW trips in Fort Lauderdale CBD have a.m. peak as the arrival time and the rest of them occur during the a.m. mid day period. In the Miami CBD, most HBW trips occur during the a.m. peak. The second period most popular travel time for work trips is the p.m. mid day. Most shopping travelers choose the p.m. mid day and the p.m. peak as their time of ar rival in the Miami CBD. F igure s 5 1 2 and 5 13 show the distribution of arrival times of various trip purpos e s for the Fort Lauderdale and Miami CBD s, respectively.

PAGE 91

91 Figure 512. Trip purposes and timeof day arrival in the Fort Lauderdale CBD Figure 513. Trip purposes and timeof day descriptive at Miami CBD This study incorporates three variables related to parking in the CBD into the model : the number of available parking spaces the average parking prices, and the percentage of timerestriction parking in each block group. These variables are joined

PAGE 92

92 with travelers locations on two CBDs. First, this study utilizes the split analysis to calculate the number of onstreet parking spaces per block group. The Fort Lauderdale CBD has six block groups and the Miami CBD has seven block groups. The location and the number of block groups within the two CBDs are shown i n Figure 5 14. In Fort Lauderdale CBD, the number of onstreet par king lots ranges between 34 and 323 spaces. The ranges of public parking and commercial off street parking are 0 to 4,673 and 0 to 8,753 respectively for the Fort Lauderdale and Miami CBDs Parking prices vary within the CBD with the averages of onstree t parking and public off street parking less than that of commercial off street parking. The cheapest average price is on the west northern block group, which is $ 0.368 per hour In the Miami CBD, the number of onstreet parking lots is within the range of 51 to 154 spaces per block group, while the number of public off street parking spaces is ranged from 0 to 2,150 spaces per block group. Commercial off street parking accounts for the most available parking lots in Miami CBD. The range of this type o f parking lot is from 437 to 5,087 spaces per block group. The average price of onstreet parking is cheaper than off street parking for each block group. Some block groups have cheaper average prices for public off street parking lots than the rest of block groups. In the southeast block groups, commercial off street parking spaces have more expensive average prices per block group than the rest of block groups The number of parking lots and average prices for each type by block groups are shown in Ta ble s 5 7 and 58 for the Fort Lauderdale and Miami CBDs, respectively.

PAGE 93

93 Figure 514. The block groups in: A) Fort Lauderdale CBD B) Miami CBD Table 58 The average price of parking in Fort Lauderdale CBD GEOID On street Public Off street Commercial Off street Space Price ($/hour) Space Price ($/hour) Space Price ($/hour) 120110425001 62 1.504 66 0.924 1550 5.234 120110416001 117 1.000 130 0.368 0 0.000 120110419003 34 2.210 0 0.000 0 0.000 120110425002 323 1.654 4673 1.231 8753 2.960 120110425003 88 1.741 169 0.750 1119 4.274 120110426005 85 1.747 20 1.250 0 0.000 Table 59 The average price of parking in Miami CBD GEOID On street Public Off street Commercial Off street Space Price ($/hour) Space Price ($/hour) Space Price ($/hour) 120860037051 51 1.452 828 5.891 437 4.119 120860037024 88 1.500 0 0.000 721 1.712 120860037041 154 1.500 1543 3.000 5087 6.094 120860037061 148 1.500 1972 4.000 1587 4.658 120860036011 150 1.469 2150 3.887 2728 2.912 120860037031 97 1.500 631 5.648 3694 8.000 120860037071 122 1.500 1627 3.884 3516 3.824 (A) (B)

PAGE 94

94 The maximum time allowed for parking in the Fort Lauderdale CBD varies in number. Various locations have time restrictions for onstreet parking that include : one, two, three, four, and six hours. About 36 percent of parking lots have four to six hour parking restriction s. For public parking, two, three, four, six, and twelvehour restrictions are enforced About 66.5 percent of parking lots have more than six hour restriction, which is categorized as no restriction. Commercial off street parking also has most lots having six hour or longer restriction. Also, 35 percent of total commercial off street parking does not have any time restriction. F igure 414 shows the proportion of each time restriction category by parking type. In the Miami CBD, off street parking lots account for most of the parking hour restriction s, although the percentage for tim e period is low: 19.1 percent for one to three hour s and 16.4 percent for four to six hour period. This study does not find any short period, i.e. one to three hour, on commercial off street parking. Also, the hour restriction for onstreet parking is low. The percentages on both periods that are identified are less than five percent: 3.1 percent for one to three hours and 0.6 percent for four to six hours. The classification of th e hour restrictions can be difficult because public operator also offers parking meter rent, flat rate, and monthly rates that support the low percentage of hour restriction in Miami CBD. Figure 5 15 and Figure 516 displays the number of parking lots based on the period of restriction hour for the Fort Lauderdale and Miami C BDs, respectively Overall, the blue figure, f or no hour restriction category dominates the statistics for every parking type: onstreet parking, off street public parking, and off street commercial parking.

PAGE 95

95 Figure 515. The number of parking spaces based on restriction hours and parking types in Fort Lauderdale CBD Figure 516. The number of parking spaces based on restriction hours and parking types in Miami CBD CBD Model and the Results Without separating the trip purposes as what as was done for the regional level, this study finds 192 travelers completed their trip s in the one of the CBDs. A similar

PAGE 96

96 ordered probit model is used to understand the relationships between socioeconomic, tr ip characteristics and built environment factors and time of travel to the CBD. Among various variables tested on the model, only three variables are significant : working status, trip purpose dedicated to work, and the percentage of parking without restri ction. Table 5 10 displays the model for CBD. Surprisingly, none of the socioeconomy variables is significant except for working status. The coefficient of this variable is 0.592. The n egative value of this variable shows that if travelers have wor king status, they tend to choose to travel earlier in the day. The second significant variable is trip purpose dedicated to work. When the trip purpose is HBW or other types of trips that are related to working, the variable is coded as 1. The second si gnificant variable is trip purpose dedicated to work. When the trip purpose is HBW or other types of trips that are related to working, the variable is coded as 1. The result of this model is significant with coefficient of 0.689. Similar to the working status, the negative value means that travelers with work trips tend to choose to travel earlier in the day. Meanwhile, travelers with other purposes may choose later timeof day travel Two possible reasons that cause shoppers to choose later time of day are because: (1) they prevent the conflict with workers or avoiding peak hours, and (2) they find that later timeof day support the flexibility, such as in terms of parking availability. The third significant variable is the percentage of parking lots without restriction. A p ositive value applies for this variable. It means that when the percentage is higher, travelers tend to choose to travel later in the day. The coefficient for this variable is 1.160. Two other variables, i.e. the average park ing prices and the availability of

PAGE 97

97 spaces per block group, are not significant. The result may also imply that parking spaces are relatively available and the price per hour does not affect the choice of travel time. Additionally, it also may indicate that many of the parking providers charge a flat rate or a rate that is not sensitive to difference in the demand for parking at various times of day Table 510. Estimation of time of day choice (ordered probit model) for CBD Variables Estimate Value t value C onstant 1.581 4.168 Worker s tatus 0.592 2.863 Work trip p urpose 0.689 3.710 Percentage of parking s paces without e nforcement 1.160 2.678 or mu (1) 1.298 0.781 or mu (2) 0.660 1.037 or mu (3) 0.862 3.360 or mu (4) 0.986 5.573 N 192 Log likelihood at convergence 244.489 Log likelihood at equal shares 286.221 Chi square Value 41.732 df 3 Hypotheses Testing From the above results, this study responds to these several hypotheses The f irst hypothesis is that built environment factors have an effect on the time of day travel choice. Significant built environment variables in the first and the second models show that they have the relationship with timeof day travel choice. In the firs t model, only one built environment factor, the cul desac density is significant. This variable is consistent across the tested model s. The second model also supports the first hypothesis F our built environment variables cul desac density, a ll high v alue of built

PAGE 98

98 environment factors (code = 333), low employment density medium diversity high intersection density (code = 123), high employment density m edium diversity medium intersection density (code = 322), are found to be significant. These significa nt variables contribute to the second hypothesis. The second hypothesis proposes that the combined factors of built environment have a more significant effect on time of day travel choice Based on the comparison between the first and second model on regional level, the second model that includes the combined measures of built environment perform s better in terms of the goodness of fit in the model. None of the built environment variables is significant in the first model except the cul de sac density. Since this cul desac density is also significant in the second model, the combined variables tend to have a greater ability to explain the time of travel in this case study. The third hypothesis that questions whether the trip purpose is significant variable in explaining timeof day travel choice was tested by segmenting trip purposes into working and shopping trips. Both models of regional and CBD levels show the significance of these purposerelated variables. At the regional level, shoprelated trips are taken later in the day, while at the CBD level, work related trip affects the earlier time of day. The last hypothesis proposes that the average parking price rate and availability o f parking a ffect s the time of day travel choice for CBD destinations; t he average price is not a significant variable for the model tested in the CBD s. This study tests the hypothesis by calculating the average parking prices at the block group level. A long with the price, this study tests the total spaces available and the percentage of parking

PAGE 99

99 spaces with and without time parking restriction. The time parking restriction variable is significant.

PAGE 100

100 CHAPTER 6 DISCUSSION AND IMPLICATIONS FOR POLICY This research aimed to comprehend the relationship between landuse and transportation, especially the relationship of built environment factors and time of day choices. By understanding this relationship, this study attempts to take a position i n the cu rrent debate about the relationship of those variables and to suggest policy implications at the regional level. Discussion This study supports prior studies argument that a relationship between the built environment and travel behavior exists. By per forming time of day dimension as a part of travel behavior procedures, this study finds several significant built environment factors using a time of day choice model. Built environment factors include cul desac density, high levels of built environment combinations, the combination of high employment density medium diversity medium intersection density, and the combination of low employment density medium diversity high intersection density. This result suggests the possibility that land use policy may influence time of day preferences and travel behavior. The combined effect of built environment variables, which are density, diversity, and design, meets prior expected findings on their influence on time of day travel choice. Many researchers have argued that built environment affect travel behavior; however, most built environment variables are correlated to each other (Cervero & Kockelman, 1997; Crane & Crepeau 1998; Silva et al., 2006; Shay & Khattak, 2005). However, those studies did not examine the built environment to time of day choice relationship. Negative values of built environment variables found in this research

PAGE 101

101 indicate travelers with shopping and work purposes tend to choose earlier time periods for arrival if their end trips are loc ated in high employment density, high diversity, and high intersection density areas This relationship contradicts the finding by Chu (2009) and He (2013) that shows the insignificanc e of corridor density and employment density for work trips In simila r study area with this research, Neog (2009) found one contradict ory result that she did not find any effect or combined effect from the built environment factors on mode choice travel behavior in the Southeast Florida. Meanwhile, the combined factors are significant in this current study. One possible explanation for the lack of significan ce in Neog s study may be in part due the lack of variability for the model considering the domination of automobile in the study case, which is more than 90 percent. A nother reason, current research measures three combined built environment factors that are different to Neogs study that bond two built environment factors. Changing the time period of travel may be one option for traveler in response to any transportati on policies, such as increasing pricing of travel, land use policies, or any restriction. As an example, Hensher and King (2001) found that people will also consider changing locations besides using public transport when higher parking pricing or time res tricted parking is imposed. Lam et al. (2006) performed an analysis that relates the departure time choice with parking location, time delay because of parking searching, and parking charges. This study implied that people may change their preference of time because of parking policy.

PAGE 102

102 Moreover, the characteristics of trip end destination appear to affect the time of day travel choice more than the origin, or home, start ing point of travel. Although only one variable, population density, is attributed to the origin location, it is not significant across the models. This may also be attached to the characteristics of study case in the Southeast Florida that has a strong mode preference for automobile usage. For example, this study finds that the mean of distance travelled for work trip s is 10 miles and for shopping trips is 4 mile. T he variability of home location may be low because most start or home locations are in low population density areas or in single family land use s. Moreover, people may choose their home location based on individual preferences, as prior literature calls self selection (Chatman, 2009; Handy et al., 2006; Cao et al., 2009). On the other side, the built environment attributes at the trip end locations may have more variation than home location. T rip end location, especially for work location, is not chosen based on individual preferences. Furthermore, this study supports the finds of prior research in terms of several socio e conomic variables that are significant to time of day travel choice. Th e se include high income, age of respondent and travelers with children. With understanding these significant socio economic, policy makers can consider the proper strategies regarding socio econom ic changes and related policies or further model for time of day travel For example, Steed and Bhat (2000) highlight the importance of predicting trends between socioeconomic changes and transportation and air quality analysis. High income variable is significant to the model of timeof day choice. This result is consistent with prior study focusing similar topic, such as Steed and Bhat (2000). Their result suggests that travelers with high income travel less during the a.m. peak

PAGE 103

103 and midd ay periods. While their study shows that this situation only applies to shopping and recreation trips, this study suggests that later time preference also applies for work trips. The reason for shopping travelers to choose later time may be because the s trict schedules they have during earlier time (Steed and Bhat, 2000) and other constraints related to the earlier time, such as congestion and later hours for shopping; while, working travelers with high income may have higher positions that provide them with greater flexibility to go to their office at the time they desire, which generally do not require them to come at the earlier periods. Then, the significance for age of respondent is consistent to the past literature (Steed & Bhat, 2000; Okola, 200 3) that agrees that age is a predictor of timeof day travel choice, although the results are contradictory. The reason of the difference is due to both past literatures having specific focus, e.g. shopping and recreation, while this study combines working and shopping. Work trips may contribute to the effect of earlier time because the trips are relatively strict to the early time schedules. However, the coefficient for the effect of age on time of day choice is relatively small The presence of child i s significant to the model. This result confirms the result of a previous study by Steed and Bhat (2000) that shows that travelers with children tend to choose a.m. off peak and p.m. off peak for shopping and recreational trip. Also, it confirms the study of Chu (2009) and He (2012) that found travelers with children travel earlier for work trips. This study finds many reasons explaining the tendency of travelers with children choosing earlier travel times. First, childrelated activities, such as droppi ng children off at school or day care, may cause travelers with children to choose earlier periods of travel. Second, travelers who shop with children tends to

PAGE 104

104 choose periods earlier in the day due to the early bedtimes of children (Steed and Bhat 2000). The trip purpose shows a significant effect on timeof day travel choice. This result is not surprising, since much of the literature discussed travel behavior differences between work trips and nonwork trips. One of study (Kumar and Levinson, 2008) tha t verifies this result shows that working and nonworking travelers having different behaviors. As well as different types of behavior, working travelers tend to have inflexibility in terms of their schedules as the basic reason of why most working travel ers tend to choose earlier periods. The main reason for this different may be time flexibility; most workers have a limited flexibility in their work schedule. Prior research focusing on work trip and timeof day choice shows that the flexibility of work schedule affects the travel time choice (Abkowitz ,1981; Arnott et al.,1990); Arellana et al. ,2012; Bellei et al. ,2006; Borjesson ,2008; Chin, 1990; De Jong et al.,2003; Habib, 2012); He, 2013; Lemp et al., 2010; Mahmassani & Chang, 1986; Mannering, 198 9, Mannering & Hamed, 1990; Noland & Small, 1995; Sall & Bhat, 2007; Sasic &Habib, 2013; Yamamoto et al., 2000 ; and Yang et al. 2013, Ettema & Timmermans, 2003; Chu, 2009; Saleh & Farrell ,2005). The result of this study confirms this schedule limitation. At the regional level, travelers with work related trips tended to choose earlier periods in the day for travel. Conversely, travelers with shoprelated trips favored a traveling at later in the day. Almost similar result applies to CBD travelers: when the trip is work related, travelers tend to choose to travel the earlier in the day. In fact, this result also provides us with insight into to the relationship between timeof day travel choice and parking restrictions and costs in CBDs.

PAGE 105

105 As di scussed in the descriptive analyses, the single purpose of work dominates the trips to CBDs. Likewise, two significant variables impacting time of day travel choice with the CBD as a destination are work related; the work trip purpose and the worker status. This result may imply that the behavior of work related travelers can bring in CBDs time of day peak period. In fact, this situation is not surprising and can be seen in the example of parking behavior. Characteristics regarding parking behavior of work trips are long term parking, self sufficient office buildings with parking garages, and many off street parking lots in the downtown periphery. This study highlights two CBDs in the study case that demonstrate these parking characteristics. From the inventory, Fort Lauderdale CBD has 17,211 parking spaces that consist of on street about 4.2 percent, off street public about 29.4 percent, and off street commercial about 66.36 percent. This number does not account for the self sufficient office buildings in the downtown area because this study gets only parking lots that are open for public use. However, most of these off street parking lots (both commercial and public) are located in the center of CBD. One block group dominates the location f or these off street parking lots. It means that the characteristic of off street around the downtown periphery does not apply for Fort Lauderdale CBD. However, long term parking may happen in this CBD since the 74.7 percent of commercial off street parki ng lots do not have time restrictions, in terms of maximal hours allowed. Meanwhile, Miami CBD has 27,331 parking spaces with the following parking types: onstreet represents 2.9 percent; off street public is 32 percent; and off street commercial at 65 percent. This data indicate that off street commercial parking dominates parking provision types in both CBD. Also, the Fort Lauderdales location is

PAGE 106

106 centralized, different with off street commercial parking lots in Miami CBD that are more dispersed, especially in the periphery area of CBD (northern part and western part of CBD). Also, all types of parking are available across all block groups within the Miami CBD. Commercial off street parking lots without the time restriction account for 83.7 percent of commercial lots. Considering these numbers also does not include privateowned parking lots, Miami CBD may have the potential to experience the undesired negative characteristics of long term parking more profoundly. The estimation model of time of day choice supports to the result of descriptive analysis. Trip purpose, worker status, and the percentage of parking spaces with time restriction were significant variables. Travelers tend to choose earlier time of day travel when they have work related tr ips purpose and have worker status. This result confirms the finding of Sasic and Habib (2013) that shows that office and professional workers tend to choose early morning travel because of regular office work hours. Because this variable is significant only at CBD trip end, this result may relate to the trip purpose variable. This result is consistent to the condition on regional level. This result also supports prior literature that argued different trip purposes determine different timeof day choices (Kitamura et al., 1998; Steed and Bhat, 2000; Chikairaishi et al., 2009). The directions of the relationship are confirmed, such as travelers with shopping trips tend to choose later timeof day. Travelers tend choose later timeof day travel when the percentage of time restricted parking was low. This may imply that earlier timeof day travel may be dense for people who are working and searching for parking, even when the time restriction is high. This situation may conflict with people who are shopping in downtown. They

PAGE 107

107 must then compete to find parking with people who are working in the morning and dealing with a similar desire to get parking. However, the total percentage of publicly owned and privately owned parking lots that have no timerestriction is similar for both the Fort Lauderdale and Miami CBD 78 percent. The Fort Lauderdale CBD has fewer parking lots with time restrictions of one to three hours than Miami CBD does, 4 percent and 7 percent respectively. Parking lots with restriction of four to six hours are 18 percent in Fort Lauderdales CBD versus 15 percent in Miamis CBD. This implies that both CBDs have the potential to increase their time restriction policy to influence travel behavior, especially in the morning peak choice tim es. However, since the majority of travelers are workers who have limited flexibility in terms of time schedules, this time restriction policy should be accompanied with other supporting policies to accommodate long term needs. The following sections exp lore policy implication in specific terms. Policy Implications The abovementioned conclusions suggest the following policy implications at both the regional and CBD levels. First, in order to coordinate land use policies with travel behavior, the integrat ion of built environment factors should be considered in addition to individual factors. The integration here does not need to be limited to the three variables included in this study density, diversity, and design. It may also encompass other built envi ronment factors, such as distance to transit infrastructures, accessibility, pedestrian and bicycle amenities, and demand management strategies. Although the method to combine those factors is more complicated if the number of variables is more than what is used in this study, it may be applicable to understand which combinations are more important than others for specific travel behavior studies.

PAGE 108

108 Second, trip purpose segregation to understand specific travel behavior is valid because this study also finds that different trip purposes have different behaviors and characteristics. Moreover, understanding different trip purpose behavior may enable proper policy development in relation with time of day behavior, especially for demand management during peak hours. As an example, policy makers can focus on work travelers behavior when they want to apply different fees for vehicles at the time congestion occurs, usually called congestion charging, or different parking fees based on time periods, or to promote high occupancy vehicle policies in early time of day Also, understanding how far the workers may negotiate their time flexibility may also facilitate the research of time of day elasticity for workers. It gives information about whether the travelers m ay change their locations of working or parking places, whether they may be willing to pay more as a policy is imposed, or may change their time of day Third, this study concludes that understanding regional time of day travel behavior may give insight in to parking policy at specific locations such as downtown Miami and Fort Lauderdale, in this study Parking spaces that are relatively available, cheap, and have no time restriction do not help the ultimate goals of multi modal transportation or landuse a nd transportation coordination. Those parking conditions match what has been identified as the decline of downtowns (Edward, 1996). In downtown revitalization efforts, local governments should be able to understand travel behavior for specific trip purpose as this study presented. In fact, the current tendency of CBDs to have a single work related purpose explains why early mornings dominate the travelers preferences. Local government should be able to coordinate with private

PAGE 109

109 companies in downtown about this through parking policies for workers. Without any parking policy or any other multimodal supporti ve policies dedicated for workers, all efforts to improve downtown or to redevelop the CBD will not reach optimal results. Fourth, besides outreach to employers, local government should be able to manage collaborations with retailers associations or any other shopping related and downtownoriented stakeholders. As regional characteristics show, shop ping related travelers have shorter trip distances than work related travelers. Meanwhile, the distance travelled for shopping to the CBD, which is 10 miles, is greater than the distance traveled for shopping throughout the region. This may indicate that shopping downtown may not be attractive due to the relatively longer distance. There is also the possibility that the desire to shop in downtown is worsened by the congestion travelers could face, or by parking policies that are more oriented to worker s. A n example of worker oriented parking policies ar e the flat rates and the lots without time restrictions This may result in double parking violations for other downtown users who are wanting to park for shorter time periods Suggestion is on more parking restriction or pricing based on peak time in downtown. Appendix B shows a picture that illustrates an example of this violation. By having a discussion with the stakeholders, local governments can identify parking policies that promote shopping and downtown revitalization. To name a few, these polici es are: shared parking, timerestriction, and variable parking pricing based on time of travel Additionally, this policy does not include standalone parking lots for shopping areas. Currently, variable parking pricing has not been applied in both CBDs In fact, the average prices per hour for public parking are relatively cheaper than those for

PAGE 110

110 commercial parking. As a comparison, the prices for public parking are: $1.50 for onstreet and $1.20 for off street in Fort Lauderdale CBD; and $1.50 for onstreet and $4 for off street in Miami CBD. Meanwhile, the commercial parking providers charge more; the average price in Fort Lauderdale is $3.30 per hour, and for Miami it is $5.20. As a consequence, not only is the average price of parking still low, especially for public providers, but also the possibility of variable parking pricing exists. Additionally, if this parking policy is applied, policies should be imposed to commercial parking as well. Regulations for these commercial parking providers should be compatible with the overall goals of parking and broader transportation policies. Last, there are supporti ve policies that help the ultimate goals of landuse and transportation coordination, particularly to support CBD redevelopment, parking poli cies, smart growth policies, and transit oriented development. Examples of such policies are: 1) transit passes for workers to encourage the use of public transportation, 2) park and ride, where travelers park their car outside CBD areas and travel using transit system, 3) nonmotorized transportation infrastructures, for example bicycle parking spots or bicycle paths, and 4) integration of all modes into a single integrated network that reduces the transfer time and increases the convenience for travelers in downtown.

PAGE 111

111 CHAPTER 7 CONCLUSION AND AREAS FOR FURTHER RESEAR CH Coordinating transportation and l and use has been considered as a way in building the livable city. One of the efforts to coordinate transportation and land use is by understanding the relationship between those two topics and the implications of this connection for travel. However, past studies have not been in agreement about that relationship due to the broad topics of transportation and land use, the complexity of built environment variables, and the involvement of many variables in the relationship. Accordingly, the focus of this study is on time of day travel choice and its relation to built environment factors and parking policies, in particular This study uses data from the National Household Travel Survey (NHTS) 2009 FDOT for the Southeast Florida region and parking inventory from t he project of i mpact of parking supply and demand management on CBD transportation system outcome by the University of Florida (Steiner et al., 2012) to include in the model of timeof day travel choice. This study examines the relationship of built envir onment and timeof day using o rdered choice m odel s. Subsequently, this study also tests the combined built environment rather than individual effects as previous step using the LUCIS model. Furthermore, this study responds to these following hypotheses at the regional level: (1) w hen considering socioeconomic variables and trip characteristics, there is a relationship between travel behavior (timeof day) and the built environment in the Southeast Florida, and (2) t he combinations of built environment factors are more influential than individual factor of built environment on the connection to timeof day choice At CBD level, two downtowns are chosenFort Lauderdale and Miami. Hypotheses regarding the CBD level are: (1) t rip purpose determines different travel

PAGE 112

112 behavior, especially related to timeof day choice, and (2) t he average parking rate determines timeof day travel choice at two CBD destinations. Working within these hypotheses may provide insights into landuse policies that can be used to shift travelers choices into the intended time ranges In the second hypothesis of regional level, this study converts each unique value of built environment factors combination. Then, the relative importance of one factor to another may be known in the relationship to time of day choice. Specifically for CBD level, these hypotheses may give information about the position of parking as built environment factor in the relationship with timeof day travel choice for travelers to downtown destination. By employi ng this hypothesis, the study may give significan ce to illustrate the current situation of parking policies and the possibility to enhance the current practices into making improvement that also consider the livability of downtown. Summary of Findings This research has several key findings for the regional level at the Southeast Florida ( three counties: Broward, Palm Beach, and Miami Dade ): Time of day travel choice has the relationship with socioeconomic and demographic variables, including income, age, and the presence of children; Time of day choice also has the relationship with trip purpose and travel mode of transportation. When the built environment variables are treated individually, only cul desacs density is significant. Three combinations of built environment variables are significant in the relationship with timeof day travel choice. These are the combination of high employment density, high diversity, and high intersection density ; the combination of low employment density, medium diversity, and high intersection density ; and the combination of high employment density, medium diversity, and medium intersection density

PAGE 113

113 Negative value of the combined variables in the relationship with timeof day travel choice indicates that travelers tend to choose earlier travel time periods when they travel to these characteristics of built environment. The coefficient of 322 combination ( high employment density, medium diversity, and medium intersection density ) is greater than the other s combined variables It may point out greater employment density effects on timeof day travel choice than other variables in combination. Furthermore, the key findings for the downtown level at Fort Lauderdale and Miami are: Three variables are significant for the relation ship with time of day choice in CBDs. They are working status, trip purpose dedicated to work, and the percentage of parking without restriction. Work related trip purposes have a coefficient of 0.689 in the relationship to timeof day travel in these CB Ds T he negative value means that travelers with work trips tend to travel earlier in the day T ravelers with the purpose of shopping tend to travel later in the day T he percentage of parking lots without restriction is a significant variable for the rel ationship with timeof day travel to the CBDs. With the coefficient of 1.160, it means that when the percentage is higher, travelers tend to choose later timeof day travel T he average parking prices and the availability of spaces per block group are in significant to time of day travel choice in the CBDs These results imply that price per hour does not affect the timeof day travel choice and parking spaces are relatively available. Study Limitations and Future Research This study notes limitations as follows. First, because this study relies on data from the NHTS travel survey Florida addon it uses a limited sample of traveling in any given neighborhood including travelers to the downtown CBD. However, this limitation may be addressed in future res earch, especially research with a focus on downtown areas or by performing primary collection among downtown workers and shoppers Other limitations are due to the nature o f the national level travel survey related to the

PAGE 114

114 geographical units. Testing using several different units may g enerate different results. Buffers or smaller units may give more neighborhood specific characters in built environment variables. This study uses readily available data that the geocoding locations do not allow for blocks level measurement. This limitation gives the possibility for future research to expand the scope of this research into more detailed unit of measurements or buffer area, either by circular areas network distance, or accessibility associated with indi vidual parcels Also, the location choices of regional travel for specific purposes may be a different topic in an extended version of this study. Additionally, different choice in classifying built environment variables may give different result in the model. This study uses geometric interval; meanwhile, other classification methods, such as natural breaks, equal interval, and quantile, could be used in other similar studies. Second, this study selects a limited number of built environment variables a mong many potential built environment variables. Accessibility is one of the built environment variables that were not incorporated in this study. In fact, the insignificance of individual built environment variables may open the possibility for other built environment variables to be significant. Future research can incorporate many of these built environment variables either individually or together in combination to test the relationship with time of day choice or other travel behavior dimensions. Third, since this study considers lack of variability from travelers in terms of transit users, this study has not incorporated transit specific variables. One variable that was included is transit user ; it is significantly associated with earlier time of day travel choice. This result opens the possibility that transit system may be included in

PAGE 115

115 the model, especially when such timesensitive polic ies are imposed. Most workers hypothetically have fix ed work time schedules; they are more likely to choose other locations or other modes of transportation if such transportation demand management is imposed. As a consequence, the transit system may be essential to be included for future study if we want to explore the elasticity of transit users and conduct more detailed surveys. Fourth, with regard to parking, this study has limitations that can be addressed by future research. First, the information about how long the travelers spend in the CBDs may complete the information about whether the travelers are long term or short term travelers and parkers. Also, by having more data on time of day choice for downtown travelers, future study may draw the relationship of time of day choice and parking policies more accurately than this current research. In additi on, the elasticity of response to price or time restricted policies into the time of day travel choice can be examined. This proposed future research may help local government in determining proper parking policies that meet the objective of downtown development based on the charact eristics of travelers. Third, i n the relation to accommodate interest i n retail development, future research can focus on shopping preference based upon the timeof day Asking different shopping areas and types may provide in formation about specific time s t ravelers prefer. Also, the parking and built environment at home or start location variables may be incorporated into the survey.

PAGE 116

116 APPENDIX A DETAILED DATA DESCRIPTION Following table is the detailed data sources, variable used in the analysis, description and detailed coding. Table A 1 Detailed d ata d escription Data Sources Variable Description Detail Coding NHTS 2009 Person File Time Choice Discrete variable that represents the time periods of choice 1 A.M. peak (6:00 to 8:59), 2 A.M. mid day (9:00 11:59), 3 PM mid day (12:00 14:59),4 PM peak (15:00 17:59), 5 PM early evening (18:00 20:59), and 6 other (21:00 5:59) Gender Traveler's gender 1 if male and 0 if female Worker Whether traveler has a worker status 1 if yes and 0 if no Age Traveler's age Numerical value of age Ethnicity Traveler's ethnicity 1 white, 2 African americans, 3 Hispanics, 4 Others NHTS 2009 Household File Presence of Child Whether household has a child or children; data were processed from life cycle variable 1 if yes and 0 if no Income The category of income from travelers; three categories income low equals to 1 if < $30,000; income medium equals to 1 if $30,000 $49,999; income high equals to 1 if income is $55,000 or more HH Driver The number of driver in the household Numerical number of drivers HH Size The number of household members Numerical value of members HH Vehicle The number of vehicles in the household Numerical value of vehicles HH Adult The number of adults in the household Numerical value of adults HH Workers The number of workers in the household Numerical value of workers NHTS 2009 Trip File Mode Mode of transportation used in the travel; data were processed by recategory (TRPTRANS variable in NHTS 2009) 1 automobiles, 2 nonmotorized vehicles, 3 transit systems, 4 motorcycles, and 5 others Travel Time in minutes Total time for travel in minutes; data are from TRVLCMIN variable Numerical value of times in minutes Location of travelers Start and End locations longitude and latitude; geocode to GIS Software X and Y values Trip Purpose Working trip s and Shopping trips 1 for shopping trip s; 0 for working trips; CBD coding: 1 for working trips, 0 for others (not only shop trips)

PAGE 117

117 Table A 1 Continued Data Sources Variable Description Detail Coding FGDL Administration Boundaries State Boundaries; Block Boundaries; Block Group Boundaries File: cntbnd_jul11.shp; cenblkgrp2010_aug11.shp; cenblk2010_aug11.shp Area Calculated Areas for Block Groups Numerical value of areas in meter square Parcel Land Use Land Use Categories and Areas of each parcel 1 residential, 2 retail and commercial parcels, 3 offices, and 4 others Diversity Calculated diversity of land use; summarized from parcel land use and calculated using formula Numerical value of diversity (0 to 1); categorical values of low medium high are based on geometric in terval CBD Whether traveler's location is within CBDs 1 yes; 0 no Population Population in the block group Numerical value of population within block groups Population Density Calculated of the number of population / Areas of block group Numerical value of population density; categorical value (low medium high) were calculated using geometric interval Combined Built Environment Variables Calculated using formula (first digit employment density, second digit diversity, third digit intersection density) 27 combinations LODES Employment The number of employment per block group; summarized from geocoded blocks level Numerical value of workers in the block group Employment Density Calculated the number of employment / areas of block group Numerical value of employment density; low, medium, high categories are based on geometric interval 2010 Florida Traffic Information and Highway Data from Florida Department of Transportation and NAVTEQ network map The Number of Intersections The number of intersections per block group; calculated using network analysis in GIS, summarized using summary statistic for block groups Numerical value of intersections The Number of Cul desacs The number of cul de sacs per block group; calculated using network analysis in GIS, summarized using summary statistic for block groups Numerical value of cul de sacs Intersection density Calculated the number of intersections / areas of block group Numerical value of intersection density; categories are based on geometric interval Cul de sacs density Calculated the number of cul desacs/ areas of block groups Numerical value of cul de sacs density; categories are based on geometric interval Distance Calculated distance travelled by travelers; using pair of start geocoded location and end geocoded location, the distance was calculated by network distance (short distance in GIS) Numerical value of distance

PAGE 118

118 APPENDIX B MATRIX OF PEARSON CORRELATION Table B 2 The matrix of socioeconomic variables in Pearson correlation Socio economic variables gender worker hhdriver hhsize hhveh hhadult age hhworkers low income medium income high income Gender dedicated to male The presence of child gender 1 .115 .076 .051 .095 .062 .017 .081 .076 .010 .071 1.000 .011 worker .115 1 .190 .200 .237 .107 .451 .699 .226 .085 .258 .115 .241 hhdriver .076 .190 1 .718 .624 .827 .357 .560 .228 .049 .228 .076 .391 hhsize .051 .200 .718 1 .452 .755 .492 .487 .126 .075 .168 .051 .706 hhveh .095 .237 .624 .452 1 .499 .307 .459 .300 .083 .317 .095 .254 hhadult .062 .107 .827 .755 .499 1 .282 .482 .088 .028 .096 .062 .265 age .017 .451 .357 .492 .307 .282 1 .516 .158 .107 .222 .017 .540 hhworkers .081 .699 .560 .487 .459 .482 .516 1 .260 .104 .303 .081 .362 low income .076 .226 .228 .126 .300 .088 .158 .260 1 .296 .560 .076 .109 medium income .010 .085 .049 .075 .083 .028 .107 .104 .296 1 .625 .010 .077 high income .071 .258 .228 .168 .317 .096 .222 .303 .560 .625 1 .071 .156 Gender dedicated to male 1.000 .115 .076 .051 .095 .062 .017 .081 .076 .010 .071 1 .011 The presence of child .011 .241 .391 .706 .254 .265 .540 .362 .109 .077 .156 .011 1

PAGE 119

119 APPENDIX C PICTURE OF DOUBLE PARKING VIOLATION Following picture is example of double parking violation in Miami CBD during working hours Figure C 1. Double parking violation in Miami CBD during working hours

PAGE 120

120 LIST OF REFERENCES Abkowitz, M. D. (1981). An analysis of the commuter departure time decision. Transportation, 10(3), 283297. Akiva, M. E. B., & Lerman, S. R. (1985 ). Discrete choice analysis: theory and application to predict travel demand (Vol. 9). The MIT press. Arnott, R., de Pal ma, A., & Lindsey, R. (1990). Departure time and route choice for the morning commute. Transportation Research Part B: Methodologica l ,24(3), 209228. Arellana, J., Daly, A., Hess, S., de Dios Ortzar, J., & Rizzi, L. I. (2012). Development of surveys for study of departure time choice. Transportation Research Record: Journal of the Transportation Research Board, 2303(1), 918. Athe rton, T. J., Scheuernstuhl, G. J., & Hawkins, D. (1982). TransportationRelated Impacts of Compressed Workweek: The Denver Experiment .Transportation Research Record, (845). Bagley, M. N., & Mokhtarian, P. L. (2002). The impact of residential neighborhood type on travel behavior: a structural equations modeling approach. The Annals of Regional Science 36(2), 279297. Balsas, C. J. (2004). Measuring the livability of an urban centre: an exploratory study of key performance indicators. Planning, Practice & Research, 19(1), 101110. Bellei, G., Gentile, G., Meschini, L., & Papola, N. (2006). A demand model with departure time choice for withinday dynamic traffic assignment. European journal of operational research, 175 (3), 1557 1576. Bento, A. M. Cropper, M. L., Mobarak, A. M., & Vinha, K. (2005). The effects of urban spatial structure on travel demand in the United States. Review of Economics and Statistics 87(3), 466478. Bhat, C. R. (1998). Analysis of travel mode and departure time c hoice for urban shopping trips. Transportation Research Part B: Methodological 32(6), 361371. Bhat, C. R., & Gossen, R. (2004). A mixed multinomial logit model analysis of weekend recreational episode type choice. Transportation Research Part B: Met hodological 38(9), 767787. Bhat, C. R., & Steed, J. L. (2002). A continuous time model of departure time choice for urban shopping trips. Transportation Research Part B: Methodological 36(3), 207224.

PAGE 121

121 Boarnet, M., & Crane, R. (2001). The influence of land use on travel behavior: specification and estimation strategies. Transportation Research Part A: Policy and Practice, 35(9), 823 845. Boarnet, M. G., & Greenwald, M. J. (2000). Land use, urban design, and nonwork travel: reproducing other urban areas' empirical test results in Portland, Oregon. Transportation Research Record: Journal of the Transportation Research Board, 1722(1), 2737. Boarnet, M. G., & Sarmiento, S. (1998). Can landuse policy really affect travel behaviour? A study of the link between nonwork travel and landuse characteristics. Urban Studies 35(7), 11551169. Brjesson, M. (2008). Joint RP SP data in a mixed logit analysis of trip timing decisions. Transportation Research Part E: Logi stics and Transportation Review 44(6), 10251038. Brownson, R. C., Hoehner, C. M., Day, K., Forsyth, A., & Sallis, J. F. (2009). Measuring the built environment for physical activity: state of the science. American Journal of Preventive Medicine, 36 (4 ), S99 S123. Brownstone, D. (2008). Key relationships between the built environment and VMT. Transportation Research Board, 7. B rownstone, D., & Golob, T. F. (2009). The impact of residential density on vehicle usage and energy consumption. Journal o f Urban Economics 65(1), 9198. Carr, M. H., & Zwick, P. D. (2007). Smart Land Use Analysis: The LUCIS Model Landuse Conflict Identification Strategy ESRI, Inc. Cao, X., Handy, S. L., & Mokhtarian, P. L. (2006). The influences of the built envir onment and residential self selection on pedestrian behavior: evidence from Austin, TX. Transportation, 33(1), 120. Cao, X. J., Mokhtarian, P. L., & Handy, S. L. (2009). The relationship between the built environment and nonwork travel: A case study of Northern California. Transportation Research Part A: Policy and Practice, 43(5), 548559. Cervero, R. (1996). Mixed landuses and commuting: evidence from the American Housing Survey. Transportation Research Part A: Policy and Practice, 30(5), 36137 7. Cervero, R. (2002). Built environments and mode choice: Toward a normative framework. Transportation Research Part D: Transport and Environment 7(4), 265284. Cervero, R. (2003). The built environment and travel: evidence from the United States. European Journal of Transport and Infrastructure Research, 3(2), 119137.

PAGE 122

122 Cervero, R., & Kockelman, K. (1997). Travel demand and the 3Ds: density, diversity, and design. Transportation Research Part D: Transport and Environment 2(3), 199219. Cervero R., & Murakami, J. (2010). Effects of built environments on vehicle miles traveled: evidence from 370 US urbanized areas. Environment and Planning A 42(2), 400418. Cervero, R., & Radisch, C. (1996). Travel choices in pedestrian versus automobile o riented neighborhoods. Transport Policy 3(3), 127141. Chatman, D. G. (2003). How density and mixed uses at the workplace affect personal commercial travel and commute mode choice. Transportation Research Record: Journal of the Transportation Researc h Board, 1831(1), 193201. Chatman, D. G. (2009). Residential choice, the built environment, and nonwork travel: Evidence using new data and methods. Environment and Planning. A 41(5), 1072. Chen, C., Gong, H., & Paaswell, R. (2008). Role of the built environment on mode choice decisions: additional evidence on the impact of density. Transportation, 35(3), 285299. Chikaraishi, M., Fujiwara, A., Zhang, J., & Axhausen, K. W. (2009). Exploring Variation Properties of Departure Time Choice Behavior by Using Multilevel Analysis Approach. Transportation Research Record: Journal of the Transportation Research Board 2134 (1), 10 20. Chin, A. T. (1990). Influences on commuter trip departure time decisions in Singapore. Transportation Research Part A: General 24(5), 321333. Chu, Y. L. (2009). Work departure time analysis using dogit ordered generalized extreme value model. Transportation Research Record: Journal of the Transportation Research Board, 2132(1), 4249. Crane, R., & Crepeau, R. (199 8). Does neighborhood design influence travel?: A behavioral analysis of travel diary and GIS data. Transportation Research Part D: Transport and Environment 3(4), 225238. De Jong, G., Daly, A., Pieters, M., Vellay, C., Bradley, M., & Hofman, F. (2003 ). A model for time of day and mode choice using error components logit. Transportation Research Part E: Logistics and Transportation Review 39(3), 245268. Edwards, J. D. (1996). Downtown traffic and parking needs related to downtown economic trends. Transportation Research Record: Journal of the Transportation Research Board, 1552(1), 107 111.

PAGE 123

123 Ettema, D., & Timmermans, H. (2003). Modeling departure time choice in the context of activity scheduling behavior. Transportation Research Record: Journal of the Transportation Research Board, 1831(1), 3946. Ewing, R., & Cervero, R. (2001). Travel and the built environment: a synthesis. Transportation Research Record: Journal of the Transportation Research Board 1780(1), 87114. Ewing, R. H. (2008). Ch aracteristics, causes, and effects of sprawl: A literature review. In Urban Ecology (pp. 519 535). Springer US. Ewing, R., & Cervero, R. (2010). Travel and the built environment: A metaanalysis. Journal of the American Planning Association 76(3), 265 294. Ferguson, E. (1990). Transportation demand management planning, development, and implementation. Journal of the American Planning Association, 56(4), 442 456. Florida Geographic Data Library (FGDL). (2012). various data. Accessed from http ://www.fgdl.org/metadataexplorer/explorer.jsp Florida Department of Transportation.(2010). Florida Traffic Information and Highway Data CDROM. Florida Department of Transportation.(2009). NHTS (National Household Travel Survey) 2009 FDOT File Addon dat a CDROM. Florida Department of Transportation.(2010 ). 2009 National Household Travel Survey User Guide FDOT File Add on data. CDROM. Frank, L. D., & Pivo, G. (1994). Impacts of mixed use and density on utilization of three modes of travel: Single occupant vehicle, transit, and walking. Transportation Research Record, 44 52. Guo, J. Y., Bhat, C. R., & Copperman, R. B. (2007). Effect of the built environment on motorized and nonmotorized trip making: Substitutive, complementary, or synergistic?. Tr ansportation Research Record: Journal of the Transportation Research Board 2010(1), 111. Habib, K. M. N. (2012). Modeling commuting mode choice jointly with work start time and work duration. Transportation Research Part A: Policy and Practice, 46(1 ), 3347. Handy, S. L., Boarnet, M. G., Ewing, R., & Killingsworth, R. E. (2002). How the built environment affects physical activity: Views from urban planning. American Journal of Preventive Medicine, 23(2), 6473.

PAGE 124

124 Handy, S. L., & Clifton, K. J. (2001). Evaluating neighborhood accessibility: Possibilities and practicalities. Journal of Transportation and Statistics, 4(2/3), 6778. Handy, S., Cao, X., & Mokhtarian, P. (2005). Correlation or causality between the built environment and travel behavior? Evidence from Northern California. Transportation Research Part D: Transport and Environment 10(6), 427444. Handy, S., Cao, X., & Mokhtarian, P. L. (2006). Self selection in the relationship between the built environment and walking: Empirical evidence from Northern California. Journal of the American Planning Association, 72(1), 55 74. He, S. Y. (2013). Does flexitime affect choice of departure time for morning homebased commuting trips? Evidence from two regions in California. Transport Po licy 25, 210 221. Hensher, D. A., & King, J. (2001). Parking demand and responsiveness to supply, pricing and location in the Sydney Central Business District. Transportation Research Part A: Policy and Practice 35(3), 177196. Hess, D. B. ( 2001). Effect of free parking on commuter mode choice: Evidence from travel diary data. Transportation Research Record: Journal of the Transportation Research Board 1753(1), 3542. Jakle, J. A., & Sculle, K. A. (2004). Lots of parking: Land use in a car culture. Charlottesville: University of Virginia Press. Okola, A. (2003). Departure time choice for recreational activities by elderly nonworkers. Transportation Research Record: Journal of the Transportation Research Board 1848(1), 8693. Kitamu ra, R., Mokhtarian, P. L., & Laidet, L. (1997). A microanalysis of land use and travel in five neighborhoods in the San Francisco Bay Area. Transportation 24(2), 125158. Kitamura, R., Chen, C., & Narayanan, R. (1998). Traveler destination choice behav ior: Effects of time of day activity duration, and home location. Transportation Research Record: Journal of the Transportation Research Board 1645(1), 7681. Kockelman, K. M. (1997). Travel behavior as function of accessibility, land use mixing, and l and use balance: evidence from San Francisco Bay Area. Transportation Research Record: Journal of the Transportation Research Board 1607(1), 116125.

PAGE 125

125 Krizek, K. J. (2003). Residential relocation and changes in urban travel: does neighborhoodscale urban form matter?. Journal of the American Planning Association 69(3), 265 281. Kumar, A., & Levinson, D. (2008). Temporal variations on the allocation of time. Available at SSRN 1091810.Retrieved from http://rational.ce.umn.edu/Papers/Temporal.pdf Kumar, R. (2011). Research methodology: A stepby step guide for beginners Los Angeles: SAGE. Lam, W. H., Li, Z. C., Huang, H. J., & Wong, S. C. (2006). Modeling time dependent travel choi ce problems in road networks with multiple user classes and multiple parking facilities. Transportation Research Part B: Methodological 40(5), 368395. Lee, C., & Moudon, A. V. (2006). The 3Ds+ R: Quantifying land use and urban form correlates of walk ing. Transportation Research Part D: Transport and Environment 11(3), 204 215. Lee, Y., Washington, S., & Frank, L. D. (2009). Examination of relationships between urban form, household activities, and time allocation in the Atlanta Metropolitan Regio n. Transportation Research Part A: Policy and Practice,43(4), 360373. Litman, T. (2006). Parking management: Strategies, evaluation and planning Victoria, BC: Victoria Transport Policy Inst itut Lemp, J. D., Kockelman, K. M., & Damien, P. (2010). The continuous cross nested logit model: Formulation and application for departure time choice. Transportation Research Part B: Methodological 44(5), 646661. Ozbay, K., & Yanmaz Tuzel, O. (2008). Valuation of travel time and departure time choice in the presence of time of day pricing. Transportation Research Part A: Policy and Practice, 42(4), 577590. Pinjari, A. R., Pendyala, R. M., Bhat, C. R., & Waddell, P. A. (2007). Modeling residential sorting effects to understand the impact of the built environment on commute mode choice. Transportation, 34(5), 557573. Mahmassani, H. S., & Chang, G. L. (1986). Experiments with departure time choice dynamics of urban commuters. Transportation Research Part B: Methodological, 20(4), 297320. Mannering, F. L. (1989). Poisson analysis of commuter flexibility in changing routes and departure times. Transportation Research Part B: Methodological ,23(1), 5360.

PAGE 126

126 Mannering, F. L., & Hamed, M. M. (1990). Occurrence, frequency, and duration of commuters' w ork to home departure delay. Transportation Research Part B: Methodological 24(2), 99109. Martin, W. A., & McGuckin, N. A. (1998). Travel estimation techniques for urban planning (Vol. 365). Washington, DC: National Academy Press. McKelvey, R. D., & Zavoina, W. (1975). A statistical model for the analysis of ordinal level dependent variables. Journal of mathematical sociology 4 (1), 103 120. Messenger, T., & Ewing, R. (1996). Transit Oriented Development in The Sun Belt. Transportation Research Record: Journal of the Transportation Research Board, 1552(1), 145153. Moudon, A. V., Lee, C., Cheadle, A. D., Collier, C. W., Johnson, D., Schmid, T. L., & Weather, R. D. (2005). Cycl ing and the built environment, a US perspective. Transportation Research Part D: Transport and Environment ,10(3), 245261. Neog, D. (2009). Evaluating the effect of the work place built environment on travel behavior of employees in Southeast Florida.(O rder No. 3399224, The Florida State University). ProQuest Dissertations and Theses, 123. Retrieved from http://search.proquest.com/docview/304884675?accountid=10920. (UMI No. 304884675). Newman, P. W., & Kenworthy, J. R. (1996). The land usetra nsport connection: An overview. Land Use Policy 13(1), 122. Noland, R. B., & Small, K. A. (1995). Travel time uncertainty, departure time choice, and the cost of the morning commute. Institute of Transportation Studies, University of California, Ir vine. Rajamani, J., Bhat, C. R., Handy, S., Knaap, G., & Song, Y. (2003). Assessing impact of urban form measures on nonwork trip mode choice after controlling for demographic and level of service effects. Transportation Research Record: Journal of the Tra nsportation Research Board, 1831(1), 158 165. Robertson, K.A. (1995): Downtown redevelopment strategies in the United States: An endof the century assessment, Journal of the American Planning Association 61:4, 429437 Robertson, K. A. ( 1997). Downtown retail revitalization: A review of American development strategies. Planning Perspectives 12(4), 383401. Rodriguez, D. A., & Joo, J. (2004). The relationship between nonmotorized mode choice and the local physical environment. Transportation Research Part D: Transport and Environment 9(2), 151173.

PAGE 127

127 Saleh, W., & Farrell, S. (2005). Implications of congestion charging for departure time choice: Work and nonwork schedule flexibility. Transportation Research Part A: Policy and Practice, 39(7), 773791. Sall, E. A., & Bhat, C. R. (2007). An analysis of weekend work activity patterns in the San Francisco Bay Area. Transportation, 34(2), 161175. Santos, A., McGuckin, N., Nakamoto, H. Y., Gray, D., & Liss, S. (2011).Summary of travel tr ends: 2009 National Household Travel Survey (No. FHWA PL ll022). Sasic, A., & Habib, K. N. (2013). Modelling departure time choices by a Heteroskedastic Generalized Logit (Het GenL) model: An investigation on homebased commuting trips in the Greater Toronto and Hamilton Area (GTHA). Transportation Research Part A: Policy and Practice 50, 1532. Schwanen, T., & Mokhtarian, P. L. (2005). What affects commute mode choice: Neighborhood physical structure or preferences toward neighborhoods?. Journal of Transport Geography 13(1), 8399. Shay, E., & Khattak, A. J. (2005). Automobile ownership and use in neotraditional and conventional neighborhoods. Transportation Research Record: Journal of the Transportation Research Board, 1902(1), 18 25 Shay, E., & Kh attak, A. J. (2007). Automobiles, trips, and neighborhood type: Comparing environmental measures. Transportation Research Record: Journal of the Transportation Research Board, 2010(1), 73 82. Shoup, D. C. (2005). The High Cost of Free Parking (Vol. 7). Chicago: Planners Press, American Planning Association. Shiftan, Y., & Burd Eden, R. (2001). Modeling response to parking policy.Transportation Research Record: Journal of the Transportation Research Board 1765(1), 2734. Silva, J. D. A., Golob, T F., & Goulias, K. G. (2006). Effects of land use characteristics on residence and employment location and travel behavior of urban adult workers. Transportation Research Record: Journal of the Transportation Research Board, 1977(1), 121 131. Sriniva san, S., Provost, R., & Steiner, R. (2013). Modeling the landuse correlates of vehicle trip lengths for assessing the transportation impacts of land developments. Journal of Transport and Land Use, 6 (2), 59 75. Steed, J. L., & Bhat, C. R. (2000). On modeling departuretime choice for homebased social/recreational and shopping trips. Transportation Research Record: Journal of the Transportation Research Board, 1706(1), 152159.

PAGE 128

128 Steiner, R. L. (1992). Lessons for Transportation Demand Management from U tility Industry DemandSide Management (Abridgment). Transportation Research Record, 1346, 1417. Steiner, R. (2012). Does Land Use and Transportation Coordination Really Make a Difference Creating Livable Communities? In F. Wagner & R. Caves (Eds.), C ommunity Livability: Issues and Approaches to Sustaining the Well Being of People and Communities. New York: Routledge Press, 223243 Steiner, R. L., Srinivasan, S., Provost, R. E., Mackey, J., Arafat, A. A., Anderson, N., & DeLarco, L. (2010). VMT B ased Traffic Impact Assessment: Development of a Trip Length Model (No. CMS Project No. 2008007). Steiner, R. L., Jourdan, D., Blanco, A. G., Mackey, J. L., Hanley, G. R., Sucar, V. A., Rachmat, S.Y, Perch, S.R. & Whiteford, W. C. (2012). Impact of parking supply and demand management on Central Business District (CBD) traffic congestion, transit performance and sustainable land use: A study of the impact of parking supply and demand management U.S. Census Bureau. (2013, September). Longitudinal Employer Household Dynamics (LEHD) Origin Destination Employment Statistics (LODES). Retrieved September 29, 2013, from http://lehd.ces.census.gov/data/. U.S. Census Bureau. (2013, September). State & Count y Quickfacts. Retrieved September 29, 2013, from http://quickfacts.census.gov/qfd/states/12000.html Victoria Transportation Policy Institute (VTPI). (2013) Online TDM Encyclopedia. Retrieved from http://vtpi.org/tdm/. Willson, R. (2005). Parking policy for transit oriented development: lessons for cities, transit agencies, and developers. Parking Policy for Transit Oriented Development 79 94. Retrieved from http://www.nctr.usf.edu/jpt/pdf/JPT%208 5%20Willson.pdf Yamamoto, T., Fujii, S., Kitamura, R., & Yoshida, H. (2000). Analysis of time allocation, departure time, and route choice behavior under congestion pricing. Transportation Research Record: Journal of the Transportation Research Board, 1725(1), 95101. Yang, L., Zheng, G., & Zhu, X. (2013). Cross nested logit model for the joint choice of residential location, t ravel mode, and departure time. Habitat International 38, 157166. Zegras, C. (2010). The built environment and motor vehicle ownership and use: evidence from Santiago de Chile. Urban Studies 47(8), 17931817. Zhang, M. (2005). Exploring the rela tionship between urban form and nonwork travel through time use analysis. Landscape and Urban Planning, 73(2), 244261.

PAGE 129

129 Zhang, L., Hong, J. H., Nasri, A., & Shen, Q. (2012). How built environment affects travel behavior: A comparative analysis of the c onnections between land use and vehicle miles traveled in US cities. Journal of Transport and Land Use,5(3), 40 52

PAGE 130

130 BIOGRAPHICAL SKETCH The author was awarded a Bachelor Degree in City and Regional Planning by Institute of Technology Bandung (ITB), Indonesia in 2002. She then obtained a double Master Degree in 2005 in Infrastructure Management and Environmental Planning, from the University of Groningen, Netherlands. She received scholarships from ITB and the StuNed (Studeren in Netherland) for her study and was also awarded Cum Laude. After finished her study, she has been an academic assistant at ITB. Her research interests are transportation planning and policy, specifically in transportation demand management. Her thesis topic was road pricing st rategy. She started PhD program in Fall 2009. She received her Ph.D from the University of Florida in the spring 2014


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E593M8QLI_ZWT240 INGEST_TIME 2014-10-03T21:38:29Z PACKAGE UFE0046162_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES