USF Libraries
USF Digital Collections

An analysis of characteristics of long and short commuters in the United States

MISSING IMAGE

Material Information

Title:
An analysis of characteristics of long and short commuters in the United States
Physical Description:
Book
Language:
English
Creator:
Vaddepalli, Srikanth
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla.
Publication Date:

Subjects

Subjects / Keywords:
commuter behavior
socio-demographic characteristics
job access
residence and workplace location
transportation equity
social isolation
Dissertations, Academic -- Civil Engineering -- Masters -- USF   ( lcsh )
Genre:
government publication (state, provincial, terriorial, dependent)   ( marcgt )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Summary:
ABSTRACT: An in-depth-analysis was carried out on short, medium and long commuters using the National Household Travel Survey (NHTS) of 2001 and American Community Survey (ACS) of 2000 to determine the role of individual,household, trip and area related characteristics on commute length. The individuals with commute time less than or equal to 15 min were considered as short commuters and individuals with commute time greater than 15 min but less than 60 min were considered as medium commuters and the individuals with commute time 60 min or more were considered as long commuters. The commute time is considered as a link joining the residence and workplace locations. The availability of the desired mode used is considered as flexibility in moving the location of these points in the area. As the jobs get dispersed the lower income people face more and more transportation problems in linking the residence and workplace. There is a potential threat in their social, physical and economic isolation in the society. The individual, household, and area related characteristics are assumed to influence both the commute time and location of these points. The descriptive analysis using NHTS 2001 and ACS 2000 revealed that the characteristics of short and long commuters are different in nature. A commuter type choice model and commute length measurement models were used to estimate the influence of socio-demographic characteristics on the residential and workplace separation. Multinomial Logit Model (MNL) methodology was adopted to develop the commuter type choice model and Structural Equations Model methodology (SEM) was adopted with commute time and commute distance as endogenous variables to estimate the commute length on a continuous scale. The models confirmed the importance of demographic variables in explaining commuter length.
Thesis:
Thesis (M.S.C.E.)--University of South Florida, 2004.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: World Wide Web browser and PDF reader.
System Details:
Mode of access: World Wide Web.
Statement of Responsibility:
by Srikanth Vaddepalli.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 103 pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001469423
oclc - 55731559
notis - AJR1177
usfldc doi - E14-SFE0000324
usfldc handle - e14.324
System ID:
SFS0025019:00001


This item is only available as the following downloads:


Full Text

PAGE 1

An Analysis of Characteristics of Long a nd Short Commuters in the United States by Srikanth Vaddepalli A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Civil Engineering Department of Civil and Environmental Engineering College of Engineering University of South Florida Major Professor: Ram M. Pendyala, Ph.D. Steven E. Polzin, Ph.D., P.E. Jian J. Lu, Ph.D., P.E. Date of Approval: March 30, 2004 Key Words: Commuter behavi or, socio-demographic char acteristics, job access, residence and workplace location, tran sportation equity, social isolation. Copyright 2004, Srikanth Vaddepalli

PAGE 2

ACKNOWLEDGEMENTS I wish to express my sincere gratitude to Dr. Ram Pendyala, my major professor, for his continued support and able guidance in my re search efforts. I would like to thank Dr. Steven E. Polzin, Dr. Jian John Lu, and Ju an Pernia for serving on my committee and providing their valuable suggestions. I al so thank the Department of Civil and Environmental Engineering fo r providing with such excelle nt facilities and research environment. I would also like to thank the faculty of Center for Urban Transportation Research for the excellent coursework th ey provided during the Master’s program. Finally, I would like to thank my friends and colleagues Siva, Abdul, Ravi, Sashi, Uttam, Sharath, Buggana, Vipan, Ashish, Ram Nehra, Amlan, Xin Ye, Consta ntinos, Isidro and Challa for their support and enc ouragement throughout my research.

PAGE 3

i TABLE OF CONTENTS LIST OF TABLES iii LIST OF FIGURES vi ABSTRACT vii CHAPTER 1: INTRODUCTION 1 1.1 Background 1 1.2 Problem Definition 2 1.3 Commute Time and Commute Mode 2 1.4 Commuter Behavior 3 1.5 Market Segments 3 1.6 Objectives of the Study 4 1.7 Approach of the Study 4 1.8 Outline of the Thesis 5 CHAPTER 2: LITERATURE REVIEW 6 2.1 Commute Time and Mode Choice 6 2.2 Residential and Workplace Location 7 2.3 Transportation Equity 8 2.4 Access to Job Opportunities 9 2.5 Travel Time Expenditure 9 CHAPTER 3: DATA DESCRIPTION 11 3.1 National Household Travel Survey Data of 2001 11 3.2 American Community Survey Data of 2000 13 CHAPTER 4: DESCRIPTIVE ANALYSIS OF COMMUTERS 14 4.1 Background 14 4.2 Person Characteristics 15 4.3 Household Characteristics 22 4.4 Trip Characteristics 27 4.5 Area Related Characteristics 46 4.6 Summary of Person, Household and Area Characteristics 60 4.7 Afro-American, Poor and Bus Users 61 4.8 Range of Short and Long Commuters 63

PAGE 4

ii CHAPTER 5: METHODOLOGY 65 5.1 Theory of Multinomial Logit Models 65 5.2 Test Statistics for Multinomial Logit Models 68 5.3 Theory of Structural Equations Models 70 5.2 Test Statistics for Stru ctural Equations Models 74 CHAPTER 6: MODEL ES TIMATION RESULTS 77 6.1 Commuter Type Choice Model 77 6.2 Model of Commute Length 82 CHAPTER 7: CONCLUSIONS AND FURTHER RESEARCH 85 7.1 Conclusions 85 7.2 Further Research 86 REFERENCES 87

PAGE 5

iii LIST OF TABLES Table 4.1 Sample Size and Weight ed Population of Commuters 14 Table 4.2 Person Characteristic s of Commuters (NHTS) 17 Table 4.3 Person Characteris tics of Commuters (ACS) 19 Table 4.4 Distribution of Commuters by Standard Occupational Category (ACS) 20 Table 4.5 Percentage of Commuter Type within each Standard Occupational Category (ACS) 21 Table 4.6 Household Characteris tics of Commuters (NHTS) 23 Table 4.7 Household Characteris tics of Commuters (ACS) 25 Table 4.8 Distribution of Long Commuters by Household Property Values (ACS) 26 Table 4.9 Distribution of Long Co mmuters by Duration of Status by Household Ownership Type (ACS) 26 Table 4.10 Commute Time Distribu tion by Commuter Type (NHTS) 29 Table 4.11 Commute Time Distri bution by Commuter Type (ACS) 30 Table 4.12 Commute Dist ance Distribution by Commuter Type (NHTS) 30 Table 4.13 Average Trip Rate by Pu rpose by Commuter Type (NHTS) 31 Table 4.14 Average Trip Length Travel ed by Purpose by Commuter (NHTS) 32 Table 4.15 Total Trip Length Trav eled by Purpose by Commuter Type (NHTS) 33 Table 4.16 Average Trip Duration by Purpose by Commuter Type (NHTS) 34

PAGE 6

iv Table 4.17 Total Travel Time Expend iture by Purpose by Commuter Type (NHTS) 35 Table 4.18 Average VMT by Purpos e by Commuter Type (NHTS) 36 Table 4.19 Total VMT by Purpose by Commuter Type (NHTS) 37 Table 4.20 Mean Departure Time by Pu rpose by Commuter Type (NHTS) 38 Table 4.21 Drive Alone vs Carpooling (NHTS) 39 Table 4.22 Drive Alone vs. Carpooling (ACS) 40 Table 4.23 Mode Share by Trip Pu rpose by Commuter Type (NHTS) 41 Table 4.24 Trip Length of Long Co mmuters by Job Specialization 42 Table 4.25 Distributi on of Commuter Type by MSA Size (NHTS) 47 Table 4.26 Percentage of Commute r Type by MSA Size (NHTS) 47 Table 4.27 Distribution of Commuter Type by Urban Area Type (NHTS) 48 Table 4.28 Percentage of Commuter Type by Urban Area Type (NHTS) 48 Table 4.29 Average Commute Time by Co mmuter Type by MSA Size (NHTS) 48 Table 4.30 Average Commute Distance by Commuter Type by MSA Size (NHTS) 49 Table 4.31 Average Commute Time by Commu ter Type by Urban Area (NHTS) 49 Table 4.32 Average Commute Distance by Commuter Type by Urban Area (NHTS)49 Table 4.33 Distribution of Co mmuter Type by State (ACS) 52 Table 4.34 Percentage of Commuter Type within each State (ACS) 54 Table 4.35 Average Commute Time by State by Commuter Type (ACS) 56 Table 4.36 Average Commute Time by CMSA (NHTS) 58 Table 4.37 Average Commute Distance by CMSA (NHTS) 59 Table 4.38 Summary of the Person and Household Characteristics 60

PAGE 7

v Table 5.1 Identification Rules for Stru ctural Equations with Observed Variables Assuming No Measurement Error (y = By + x + ) 72 Table 6.1 Commuter Type Choice Model (NHTS) 79 Table 6.2 Commuter Type Choice Model (ACS) 81 Table 6.3 Structural Equations Model for Commute Length 83

PAGE 8

vi LIST OF FIGURES Figure 4.1 Work Trip Departure Ti me Distribution by Commuter Type 43 Figure 4.2 Work Related Trip Depart ure Time Distribution by Commuter Type 44 Figure 4.3 School Trip Departure Ti me Distribution by Commuter Type 45 Figure 4.4 Commute Length by MSA Size 50 Figure 4.5 Commute Length by Area Type 51 Figure 4.6 Proportions of Co mbination of Afro-American, Poor and Bus user groups 61 Figure 4.7 Percentage of Combination of Afro-American, Poor and Bus user groups in Long Commuters 62 Figure 4.8 Distribution of Commuters by Time 63 Figure 4.9 Share of Short Commuters by Upper 64 Figure 4.10 Share of Long Commuters by Lower Limit 64 Figure 5.1 Direct and Indirect Effects 76 Figure 6.1 Structural Equations Model of Commute Length 84

PAGE 9

vii AN ANALYSIS OF CHARACTERISTICS OF LONG AND SHORT COMMUTERS IN THE UNITED STATES SRIKANTH VADDEPALLI ABSTRACT An in-depth-analysis was carried out on s hort, medium and long commuters using the National Household Travel Survey (NHTS) of 2001 and American Community Survey (ACS) of 2000 to determine the role of in dividual, household, trip and area related characteristics on commute length. The indivi duals with commute time less than or equal to 15 min were considered as short comm uters and individuals with commute time greater than 15 min but less than 60 min were considered as medium commuters and the individuals with commute time 60 min or more were considered as long commuters. The commute time is considered as a link joini ng the residence and wo rkplace locations. The availability of the desired mode used is cons idered as flexibility in moving the location of these points in the area. As the jobs get di spersed the lower income people face more and more transportation problems in linking the re sidence and workplace. There is a potential threat in their social, physical and economic isolation in the society. The individual, household, and area related characteristics ar e assumed to influence both the commute time and location of these points. The de scriptive analysis using NHTS 2001 and ACS 2000 revealed that the characteri stics of short and long commut ers are different in nature. A commuter type choice model and commute length measurement models were used to estimate the influence of socio-demographi c characteristics on the residential and workplace separation. Multinomial Logit Mo del (MNL) methodology was adopted to develop the commuter type choice model a nd Structural Equations Model methodology (SEM) was adopted with commute time and commute distance as endogenous variables to estimate the commute length on a cont inuous scale. The models confirmed the importance of demographic variable s in explaining commuter length.

PAGE 10

1 CHAPTER 1 INTRODUCTION 1.1 Background Over 60% of the U.S population lives in th e metropolitan areas-ten of these areas have five million or more people and many other ar eas have experienced a rapid growth of about 25% or more in a decade (Census 2000). The growth of these metropolitan areas in both population and area resulted in continue d dispersion of jobs and population to the outer fringes. The lower income people who are mostly city residents have greater difficulty in accessing job opportunities in subu rban areas. The significant challenge of lower income people is overcoming transportati on problems to reach their jobs. There is a potential threat to their so cial, physical and economic is olation in the society. The policies that are meant to suppress the us e of automobile could have unintentional influence on lower income people who are ju st on the verge of obtaining mobility. Some of the factors mentioned above explain the reason why 56 percent of the lower income people are short commuters. Women who play a major role in taking care of young children and household maintenance are also rest ricted to the job oppor tunities and prefer to work very close to home. On the other ha nd, higher income people li ke to live in lower density areas, generally suburban areas and ar e capable of traveling great distances to reach the jobs they satisfy. Commute time reflects an individual’s preference for residential and workplace location. The prefer ences are influenced by many other factors that explain an individual’s choice for be ing a long or short distance commuter. An in-depth analysis is nece ssary to evaluate the above hyp othesis about short and long commuters. Analyzing the restri ctions and preferences that lead an individual to be a particular type of commuter would help for making better policies to provide mobility options and job accessibility for those in real need. This research work is an attempt to provide thorough analysis to the policy makers for clear understanding of the commuter choice (short, medium or long) from the av ailable databases. This study uses commute time, a measurement of accessibility and mobil ity to define commuter choice. The main purpose of the study is to anal yze different charac teristics and find their effect on the commuter length choice. Hence, the market segmentation would be based on commuter choice rather than on income, gender, race or other individual characteristics. This study is critical in identifying th e role of transportation in offering different opportunities through commuting choices to people for reducing the socio-economic disparities.

PAGE 11

2 1.2 Problem Definition Commute time reflects the work and home separation. The individual, household, and area related characteristics are assumed to influence both the commute time and location of work and home. The mode choice is assumed to be the next immediate or the simultaneous step in this decision making pro cess (commuting to work). The disutility in the length of commute would result in a change of either the reside nce or the workplace location. This freedom to change the location according to the utility differs for different sections of the people. This freedom is less for lower income people because, of the budget constraints. This will affect their choi ce of short, medium or long commute, which is considered as a measure of accessibility and mobility. As the jobs get dispersed the lower income people face more and more transportation problems in linking the residence and workplace. In this attempt they move their residence location closer to transit-serviced areas and get engaged in the jobs closer to these transit serviced areas. This is resu lting in their social, physical and economic isolation in the society. The following section gives an overview of the changing commute time and commute mode over the decade. This will give an idea about the aggravating problem. The remaining sections of this chapter desc ribe the subjects “commuter behavior” and “market segments” in the direction of st udy and finally the objective, approach and outline of the thesis are presented. 1.3 Commute Time and Commute Mode According to the United States Census Bu reau (U.S Census 2000 Journey to Work) and United States Department of Transportati on (U.S DOT, FHWA Journey to Work) the average commute time has increased from 21.7 minutes in 1980 to 25.5 minutes in 2000. The increase in the commute time from 1990 to 2000 is more than four times the increase from 1980 to 1990. There are nearly ten million workers in the United States who spend 60 or more minutes to reach their jobs. Thes e facts have important implications for the congestion, urban sprawl, growth, deterior ating transportation infrastructure, aging population, auto ownership, vehicle utilization and transit rider ship. There is also a change in the commute mode over the years as seen from the US Census Journey to Work report. It shows the percentage of co mmuters who carpool has decreased from 19.7 in 1980 to 11.3 in 2000. The mode share of public transportati on and non-motorized travel also follow a downward trend leaving th e drive alone mode as the dominant choice of the commuters. Commute time and commute mo de indicates an individuals access to j ob opportunities defined by urban structure, individual’s lif e style and transportation system. Commute time is an essential part and an alternative measurement of total travel time expenditure (Mokhtarian and Chen, 2002) inhere nt to an individual in orde r to engage in activities

PAGE 12

3 and satisfy their needs. More importantly, it is compulsory and regular time expenditure because of the mandatory type of activity at the destination. All the individual decisions and inherent qualities about commute time and commute mode result in aggreg ate behavior called as “Commuter Behavior”. Hence, to understand the commuter behavior, a disaggregate study of commuters with different commute times would help in implementing policies and adopting strategi es specifically focused on certain groups to control urban spra wl, congestion, pollution, excessive energy consumption and other adve rse impacts on the society. 1.4 Commuter Behavior Commuter behavior is about the awareness, attitudes, perceptions and options of the people who travel regularly on a daily basis to perform a desired or necessary activity. The study of commuter behavior is essentia l in understanding how people prefer to commute under certain circumstances in orde r to maximize the perc eived utility of a commute alternative to perform a par ticular or sequence of activities. Commuter behavior interests researchers in both public and private agencies in transportation and related fields in implem enting policies and adopting strategies. In public agencies, policy makers study the commu ter behavior to get feedback about the existing policies and make future decisions to minimize transportation costs on the society. Policy makers and planners are inte rested in knowing the behavior and analyzing the reason for such behavior. 1.5 Market Segments Commute time, a regular travel time-expend iture to work gives inference of an individual’s value for time and is a measurem ent of total travel time-expenditure. Hence, in this study the market segmentation is based on the commute time to study their individual, household and commut e characteristics. The market is segmented into three categories-short commuters, medium commuters and long commuters. Short commuters are defined as those indivi duals who commute 15 minutes or less to work. Medium commuters are defined as thos e individuals who commu te more than 15 minutes but less than 60 to work. Finally long commuters are defined as those who commute 60 minutes or more to work. These three market segments are considered for this study but the main focus will be on th e short and long commute rs. The terms-‘market segments’ and ‘groups’ are used interchangeably hereafter.

PAGE 13

4 1.6 Objectives of the Study The objectives of the study are To analyze the socio-economic characte ristics of different commute lengths To verify if there is any significant influence of socio-economic characteristics on the commute time To compare the work trip characterist ics like departure time, distance and commuting mode (drive alone or carpool) of different groups To analyze and compare the variation of mode share by purpose within each group and between the groups To study the other trip characterist ics of different commute lengths To construct models that estimate th e effect of socio-economic and area characteristics on the commute time To construct a model that predicts the proba bility of an individual to be a short, medium or long commuter 1.7 Approach of the Study The American Community Surv ey (ACS) of 2000 and National Household Travel Survey (NHTS) of 2001 were used for this study. Both the NHTS and ACS have detailed information about the person and household ch aracteristics. However, information about trip characteristics is only available in NHTS 2001 data and not in ACS 2000 data. Hence, they were studied using only the NHTS 2001 data. However, as anyone could expect, the analysis of the data was done se parately without merging the two datasets. The analysis of person, household and area related ch aracteristics were carried out at the person level by adding the household characte ristics to the pers on file. The trip characteristics were analyzed by aggregating th e trip file to the person level. Only those individuals who reported non-zero travel time to work were considered for the study. The sample used for the study did not include all the workers but only those workers who commute to work. Hence the workers who wo rk at home daily are not included in the study. The commuters are classified into three mark et segments based on th eir reported travel time to work as short, medium and long co mmuters. The commuter type definitions are defined in the previous section 1.5. The ma in focus of the study is on short and long commuters. In this study a commuter type choice mode l was developed to estimate the influence of socio-demographic characteristics in making a commuter type choice. Multinomial Logit Model (MNL) methodology was adopted to develop the commuter type choice model. A Structural Equations Methodology (SEM) was also developed to estimate the influence of socio-demographic characteristics on the commute length on a continuous scale. Commute time and commute distance we re used as endogenous variables.

PAGE 14

5 1.8 Outline of the Thesis The thesis is composed of seven chapters. This chapter has provided an introduction about the background, problem definition, commute time, commute mode, commuter behavior, market segments, objective and a pproach of the study. Chapter 2 presents a review of literature related to this study. Chapter 3 provides a description of the data from National Household Survey of 2001 and Am erican Community Survey of 2000. Chapter 4 provides descriptive analysis of the three market segments. Chapter 5 explains the methodology for the Multinomial Logit Modeling (MNL) and Structural Equations Modeling (SEM). Chapter 6 explains the model estimation and results. Chapter 7 concludes the analysis of s hort and long commuters and prov ides scope for further study.

PAGE 15

6 CHAPTER 2 LITERATURE REVIEW Commuter behavior is of central interest to many researchers in transportation and other related fields to measure the performa nce of transportation system and its interrelationships with area characteristics th at define urban spatial structure. Lot of research has been done on the commuter beha vior and is difficult to cover the whole body of literature. This chapter provides liter ature review on commut er behavior related to this study. 2.1 Commute Time and Mode Choice Commute time and mode choice are central in explaining the commuter behavior and other travel behavioral characteristics. Th e study of mode choice modeling to work has long been central to the evaluation of the effo rts to mitigate traffic congestion (Palma and Rochat, 2000). Recent studies have examined the role of non-work travel mode on commute mode as well as other residential loca tion and land use (Bhat, 1997; Anas et. al, 1996; Ben-Akiva and Bowman, 1998; Boarnet and Sarmiento, 1998). Cevero et. al, 1998 has found that decentralization has reduced tra ffic congestion and trav el distances and has contributed to a weakening of transit systems. Mannering et al, (1985) found that the number of autos in the house hold influences the commute m ode and suggested that it should be considered as an e ndogenous variable in models. Some researchers claim that basic commute mode choice encountered by an individual is between automobile and public transit and explain that modal split shoul d be based in this binary choice (Train, 1980; Hensher and Johnson, 1982). However a wide variety of structured mode choice models have been developed over the y ears (Train, 1980; Mannering and Winston, 1985; Thobani, 1984; Berkovec and Rust, 1985; Henshe r et al., 1991; Ben-Ak iva et al., 1994). The changing commute pattern has strong in fluence on commute time and other travel time both in space and time. Research on cros s sectional data acro ss the world suggested that the commute time mostly range betw een 25 to 35 minutes (Kenworthy and Laube, 1999). On the other hand, some researchers ha ve found that it varies with space and time (Gordon et. al, 1989; Cevero et.al, 1988). Le vinson (1998) found that the variation in commute time in space and time is less when compared to other characteristics like distance traveled and mode choice. Variat ions are caused by individual and household characteristics, the spatial context of the commute, access to transportation and factors related to the activity and travel patterns of worker s (Turner and Niemeier, 1997).

PAGE 16

7 Many studies have analyzed the influence of work duration on the commute time and found different commute times for same work duration (Golob and McNally, 1997; Golob et.al, 1995; Golob, 2000; Lu and Pas 1999; Dijst and Vidakovic, 2000). Schwanen and Martin (2001) developed a theoretical framework to addr ess the relationship between commuting time and duration of the workplace visit. Number of researchers have suggested or proven that commute time a nd work activity duration are positively correlated (Hamed and Mannering, 1993; Ki tamura, 1990 and Kitamura, 1998; Levinson, 1999). The following section discusses the lite rature on the effect of commute time on residential and workplace location. 2.2 Residential and Workplace Location The connection between the residential locati on and work place location is a central part of the theory for defining the urban spatia l structure. Many economic models have emphasized the trade-off between commuting costs and housing costs and placed this trade-off at the core of models of reside ntial location (Wingo, 1961; Kain, 1962; Alonso, 1964; Muth, 1969). The dispersal of job op portunities has created a much more complicated behavioral response to the linkage between work and residence (William et. al, 2002). Researchers have found that there is an “indifference zone” that exists for workers within which the changes to em ployment opportunities do not have much influence on residential loca tion. Beyond this zone the commuting distance has influence on an individual’s relocation of household (Getis, 1969; Brown, 1975). A study by Cevero and Wu (1997) in San Fran cisco Bay area, a polycentric city found evidence that suburban employment tends to generate shorter commutes than central city employment. Many studies have examined the impact of commuting times on the relocation to suburbs (Doom et. al, 1990; Bell, 1991; Cevero et. al, 1992; Wachs et. al, 1993) and found that commuting patterns are ad justing to metropolitan dispersal to avoid congestion and long commut es. Levinson (1998) attempted to study the dynamic behavior by including resident ial duration and job duration and found the newly relocated individuals have shorter than average commutes. He feels that that long residential durations and long employment durations will have short commutes, as they are spatial stable for a long time. A number of studies have examined the in terrelations between the job changes and residential changes. Van et. al (1997) and Rouwendal (1999) have found that increasing commute time increases the probability to acce pt a change and the job change is sensitive to residential location than the reverse, due to high costs of residential change. Some researchers have examined a sequential re sidential and workplace choice and found strong correlation between them (Waddell, 1993; Gordon et. al, 1982; Linneman et. al, 1983). Crane (1996) explained that connection between home and workplace is not static and is dependent on the future opportunities and aspirations. Some researchers have studied the residen ce and job location changes for dual worker households. Abraham et, al, (1997) found that th e probability of moving is more strongly

PAGE 17

8 related to commuting distance for women than men. Sermons et al (1999) found that the work place location of women is not an exogeno us variable in explaining the location of household. Some researchers have studied the gender differences in journey to work and found that women’s commute times are shorte r because of low wages and dual role or mother and worker they play. This section has reviewed the literature that explains the relationships between the commute times and residential and job locations. The following section will present the literature on the rela tionship of commute time with total travel time expenditure. 2.3 Transportation Equity The deficiencies present in urban growth and transportation systems lead to the patterns of social and urban segregation. The main obs tacles to the people, notably the poor in daily travel weigh heavily on schedules, compli cate access to services ever further, limit the use of urban space, and place considerable pressure on household budgets. Consequently, the low income people tend to retreat into thei r neighborhood where the low-quality urban facilities are unable to assist in the deve lopment of human and social capital and economic opportunities, the alleviation of poverty or the prevention of social exclusion (Olvera et al, 2003). The concep t of social exclusion highlights the deterioration of the employment market and mo re generally the crisis that affects social links in the various spheres (economic, politi cal, social, and spatial) of community life (Baker, 2001). High-density neighborhoods of low-income h ouseholds lack local facilities, and poor and/or expensive transport considerably incr eases social exclusion. This issue was given very little attention in deve loping countries (Vasconcellos, 2001) and most research in this area deals with poverty and transport (Godard and Di az Olvera, 2000, Grieco et al., 1996 and Turner and Kwakye, 1996). However, in the largest cities of these countries, several factors like explosive population grow th, increasing poverty, rapid and disorderly expansion of the urbanized zone, reinforcemen t of the spatial split between residential, employment and service areas, poor supply of urban services and infrastructures, and the deregulation process combine to make social exclusion problems even more severe. The spatial dispersion of residential areas is a sour ce of particular problems as jobs and the main urban facilities are highly concentrated in the CBD and here transportation becomes a key issue (Olvera et al, 2003). Transpor tation difficulties reduce the number of accessible jobs even further, and also, for tr ips to and from work, longer commutes in transit or the difficulty in walking long co mmutes can reduce the productivity of workers, notably the poor. This population is physically more vulnerable and more affected by greater fatigue caused by difficult daily travel than people with better living standards. “Low productivity, low income and low capital formation ar e some of the economic factors in the vicious circle of poverty” (A djibolosoo, 2000).To ensure equity as a whole, “Two interdependent aspects of public policy must then be a ddressed to alleviate poverty and prevent social exclusion: the improvement of accessibility throughout the city and the availability of basic serv ices locally” (Werlin, 1999).

PAGE 18

9 2.4 Access to Job Opportunities Research has provided evidence that the pr ocess of suburbanization creates new job opportunities that are not e qually exploited by all work ers (Pinto, 2001). The most important explanation is the spatial mismat ch hypothesis, first developed by John Kain (1968). According to the hypothesis, the importa nt job growth in the suburbs combined with serious constraints on Af rican-Americans’ residential c hoices have created a surplus of workers relative to the number of ava ilable jobs in inner-city neighborhoods, where African-Americans are concentrated. The main assumption is that discrimination in the suburban housing market generates this hostil ity, preventing a natural relocation of work. So as a result African-Americans will have relatively poor access to job opportunities, and they may also have a longer commute to wo rk when compared to whites in a city. As a consequence, poor labor-market outcomes should be expected for African-American households. Many researchers have done studies concerni ng the spatial mismat ch hypothesis (Kain, 1992; Ihlanfeldt et. al, 1998). They investigated the relati onship between employment, wages, or labor force participation and meas ures of job accessibility. The results suggest that poor job access worsens labor-market out comes, confirming the argument. Some researchers have tested the spatial mism atch hypothesis by examining commuting times of African-Americans and whites (Gabriel et al, 1996). Their study showed that AfricanAmericans have significantly longer commuting times than whites. There are also some investigations that and that job decentraliz ation leaves low-wage jobs in the central business district (CBD) (Straszheim, 1980; Vrooman et. al, 1980; Re id, 1985; Ihlanfeldt, 1988; Ihlanfeldt et. al, 1991). 2.5 Travel Time Expenditure Commute time is an essential part and an alte rnative measurement of total travel time. “In particular, because of the regularity, fre quency and importance of the commute trip, responses to a question about the ideal commute time can be considered reasonably informative” (Mokhtarian et al, 2002). Lot of research has been done on travel time over the forty years and has been a variable of central importance to many researchers to understand its demand for travel (Pas, 1998). The behavioral theory assumes that peopl e have a certain amount of time that they are willing to spend on travel; this concept is called “travel time budget”. Several studies on TTB reveal that on average, taken over the regional or na tional scale is constant over space and time: a universal cons tant of 1.1-1.3 hours (per travel er) per day (Bieber, et al., 1994; Zahavi and Ryan, 1980; Zahavi and Talvitie, 1980; Hupkes, 1982; Schafer and Victor, 2000; Vilhelmson, 1999). A common observation in this theory is that as the transportation system is improved due to the advances in technology or additions of capacity to the systemtravel distance tends to increase so as to keep travel times constant (Zahav i and Ryan, 1980; Hupkes, 1982; Marchetti, 1994). Recent research showed that the African villages, while almost entirely

PAGE 19

10 pedestrian based, did not genera te different daily travel times than cities in developing Asia, which were not on average different fr om Japan or Europe or the U.S. (Barnes, 2001). The TTB is also linked to induced tr avel debate, people taking advantage of improvement to travel (Mokhtarian et al., 2002) Many researchers have also incorporated TTB in to the travel behavioral models (Zahavi, 1979; Golob, et al., 1981; Goodwin, 1981; Gu nn, 1981). Recently researchers used the concept of TTB to study the mobility as inco mes rise and slower modes are replaced by faster modes (Schafer, 1998, 2000; Schafer and Victor, 2000). The following section discusses the research on commute time consider ed as a central part of total travel time expenditure. Researchers who argued for the stability of tr avel time expenditures at the aggregate level found that there was considerable variation at the disaggregat e level (Zahavi and Talvitie, 1980). Analysts have attempted to relate these variations to a number of potential explanatory characteristic s as explained below. Research on travel time expenditure reveals that an individual’s travel time expenditure is strongly related to person and household charact eristics, attributes and activities at the destination, and characterist ics of the residential areas (Mokhtarian and Chen, 2002). The above literature suggests that the study of commuter be havior at the disaggregate level is vital in understanding the strong in terrelationship that exists between the commute time and the set of characteristi cs explained above. Most of the studies discussed above are done with market segm entation based on individual, household or area related characteristics like income, age, gender, race, household size, autoownership, area-population etc., For example, low income vs. high income groups, male vs. female, single worker household vs. multi-worker households and soon. The present study is based on the market segmentation based on commute time an indicator of transportation system performance and ur ban growth. The present study focuses on identifying whom the short and long commut ers and exploring the reasons behind their behavior rather than on how di fferent sections of people beha ve in transportation system. This type of study would help in identifying the sections of the pe ople who are affected by or affect the transportation system more directly and gives a broader look at the transportation system in forming policies.

PAGE 20

11 CHAPTER 3 DATA DESCRIPTION The National Household Travel Survey (NHT S) data of 2001 and American Community Survey (ACS) data of 2000 were considered for this study. This chapter provides detailed discussion of the survey and description of th e datasets. The datasets are discussed in two main separate sections as follows. 3.1 National Household Travel Survey Data of 2001 The National Household Travel Survey (N HTS) provides detailed information on demographic characteristics of households, people, vehicles, and detailed information on daily and longer-distance travel for all pur poses. The NHTS is the combination of Nationwide Personal Transportation Survey (NPTS) and American Travel Survey. It assists transportation planners and others who need comprehensive data on travel and transportation patterns in the United States The NHTS survey data are collected from a sample of U.S. households and expanded to provide national estimates of trips and miles by travel mode, trip purpose, and a host of household attributes. Previously the daily travel surveys were conducted in 1 969, 1977, 1983, 1990 and 1995. The series of daily travel surveys conducted over the years can pr ovide detailed information about the person travel patterns in the United States. The in formation about both daily travel and longer distance travel is collected in single survey. The NHTS collected travel data from a national sample of the civilian, noninstitutionalized population of the United St ates. People living in college dormitories, nursing homes, other medical institutions, prisons, and military bases were excluded from the sample. The survey was conducted using Computer-Assisted Te lephone Interviewing (CATI) technology. Each household in the sa mple was assigned a specific 24-hour “Travel Day” and kept diaries to record all travel by all household members for the assigned day. A 28-day “Travel Period” was a ssigned to collect l onger-distance travel (over 50 miles from home) for each househol d member, and includes information on long commutes, airport access, and overnight stays. The assigned travel day was the last day of the assigned travel period. The NHTS 2001 interviews were conducted from April 2001 through May 2002 The 2001 NHTS data can be used to inves tigate topics in transportation safety, congestion, mobility of various population grou ps, the relationship of personal travel to economic productivity, the impact of travel on the human and natural environment, and

PAGE 21

12 other important subjects. These data provide planners and decision makers with up-todate information to assist them with e ffectively improving the mobility, safety, and security of our Nation’s transportation systems. The 2001 NHTS data set includes household data on the relationship of household members, education level, income, housing characteristics, and other demographic information. It provides Information on each household vehicle, including year, make, model, and estimates of annual miles travele d. The driver information is also collected for every trip. Detailed information about th e trips made during the 24 hr assigned day is collected. For example, the time the trip be gan and ended, length of the trip, composition of the travel party, mode of transportation, purpose of the trip, and the specific vehicle used. Information on long distance travel dur ing an assigned four-week period was also collected in the same survey. if no long-dist ance trips were made during the four-week travel period, data on the most recent long-di stance trip by any mode and the most recent long-distance train trip. Re spondents were asked to men tion round-trips taken during a four-week period (the househol d’s travel period) wh ere the farthest point of the trip was at least 50 miles from home, including the farthest destination, access and egress stops and overnight stays on the way to and from the farthest destination, mode, purpose, and travel party information. The NHTS also provides geographic area specific ch aracteristics of the household and workplace. Information about telecommuting, public perceptions of the transportation system, data on Internet us age and the typical numbe r of transit, walk and bike trips made over a pe riod longer than the 24-hour tr avel day was also collected. The survey data is made available in the form of four separate files, which are household file, person file, vehicle file and daily trip file. Household file contains information about household characteri stics of people in the sample. A household identification number identifies each household. Each record in household file is a househol d. There are 26,038 household records in the household file. Person file contains information about the person characteristics of people in the sample. Household identification and pe rson identification number together identifies a person. Each record in th e person file is a person. There are 60,282 person records in the person file. Vehicle file contains information about each vehicle owned by households in the sample. Household identification and ve hicle identification number together identify a vehicle. Each record in the vehicle file is a vehicle. There are 53,278 vehicle records in the vehicle file. Trip file contains information about th e trip characteristics of people in the sample. Each record in the trip file is a trip. There are nearly 248,517 records in the trip file. The household, person, and vehicle identif ication numbers are used to connect information in file to another. Weights given in the data files are used to get the national estimates. In this study only the household, person and trip files were used for analyzing the different commuter types. All the hous ehold characteristics were merged to the

PAGE 22

13 person file using household and person identifi cation number. The trip characteristics in the trip file were aggregat ed, restructured to the pers on level. There are 60,282 records present in the person file each record repr esents an individual identified by both the household and person identification number. Now, all the individual, household, area specific and trip characteristics are available at the person level. From this master data file the commuters are selected using commute time. 3.2 American Community Survey Data of 2000 The American Community Survey is a nationw ide survey and a cri tical element in the Census Bureau's reengineered 2010 census. The American Community Survey is a way to provide the data communities need every year instead of once in ten years. It is an ongoing survey that the Census Bureau plans will replace the long form in the 2010 Census. Information from the long form is used for the administration of federal programs and the distribution of billions of federal dollars. Th e American Community Survey is conducted under the authority of Title 13, United Stat es Code, Sections 141 and 193. The Census Bureau may use this information only for stat istical purposes. Full implementation of the American Community Survey is planned in every county of the United States, pending Congressional funding. The survey would in clude three million households. Data are collected by mail and Census Bureau staff fo llows up with those who do not respond. The ACS provides information about demogr aphic, housing, social, and economic characteristics every year for all states. AC S has information about journey to work and can be used to study the commuter behavior. American community survey in the present form gives area specific characte ristics up to the state level. For smaller areas, it will take three to five years to accumulate sufficient sample to produce data for areas as small as census tracts. The ACS 2000 data set cons ists two files-household and person file Household file contains information about the household charac teristics of people in the sample. A household identificati on number identifies each household. Each record in household file is a househ old. There are 514,779 household records in the household file. Person file contains information about the person characteristics of people in the sample. Household identification and pe rson identification number together identifies person. Each record in the person file is a person. There are 1,192,206 person records in the person file. The data is present for all states together and also separately. The information about the area specific characteristics is limited to the st ate level. The availability of information at the county and city level is expected in the future. In this study all the household characteristics were merged to the person f ile using household and person identification number. There are 1,192,206 records present in th e person file each record represents an individual identified by both the household and person iden tification number. Now, all the individual, household charac teristics are available at the person level. From this master data file the commuters are selected using commute time.

PAGE 23

14 CHAPTER 4 DESCRIPTIVE ANALYSIS OF COMMUTERS 4.1 Background This chapter provides detailed data descrip tion about the person, household, trip and area characteristics of short, medium and l ong commuters using NHTS data of 2001 and ACS data of 2000 discussed in the prev ious chapter. All th e data analysis was carried out at the person level by merging the household characteri stics to the person file and aggregating and restructuring the trip file to person le vel. As our study is about the commuters, only those individuals who reported non-zero travel time to work were considered. Hence, the workers who work at home were not incl uded in the study. The person and household characteristics of the commuters are discus sed in detail for both the data sets. The commute characteristics are di scussed mostly using NHTS 2 001 dataset, as information about commute characteristics is not elaborat ely present in ACS 2000 dataset. The area characteristics of the commuters are discu ssed more using the NHTS 2001 data than ACS 2000 data, as geographic information and ar ea-specific informati on is limited in ACS 2000 data. The characteristics ar e studied for three types of commuters-short, medium and long commuters. Short commuters are defi ned as those indivi duals who commute 15 minutes or less to work. Medium commuter s are defined as those individuals who commute more than 15 but less than 60 minut es to work. Finally long commuters are defined as those who commute 60 minutes or more to work. The study is mainly focused on the short and long commuters. The followi ng table below shows the sample size and weighted population of the commu ters for the two data sets. Table 4.1 Sample Size and Weighted Population of Commuters Data Set Year Short Commuters Medium Commuters Long Commuters All Commuters NHTS 2001 Sample Size 11,876 (47.6%) 11,641 (46.7%) 1,431 (5.7%) 24,948 (100%) Weighted Population 54,462,391 (46.3%) 55,736,358 (47.5%) 7,207,110 (6.1%) 117,405,859 (100%) PUMS ACS (USA) 2000 Sample Size 249,880 (47.4%) 242,264 (45.9%) 35,602 (6.7%) 527,746 (100%) Weighted Population 56,067,997 (45.2%) 58,929,736 (47.5%) 8,941,813 (7.2%) 123,939,546 (100%)

PAGE 24

15 The NHTS 2001 data shows that there are 46.3% of short commuters and 6.1% of long commuters. The ACS 2000 data shows that th ere are 45.2% of short commuters and 7.2% of the long commuters. The difference may be attributed to the missing data. Both the data sets were used to study all possible ch aracteristics and were compared wherever it was possible. 4.2 Person Characteristics Person characteristics like gender, age, race, ed ucation, income and driver status of the individual are analyzed for short, medium and long commuters for the two data sets. The Table 4.2 and Table 4.3 show the above mentio ned person characteristics for the NHTS and ACS data sets respectively. The NHTS 2001 data shows that 50.5% of the short commuters are males and 49.5% of the short commuters are females, which implies almost equal share of males and females. Whereas in long commuters, 63.6% are males and 36.4% of them are females. ACS 2000 data also shows exactly the same shares for males and females in short commuters and almost same for long commuters-64.5% of the long commuters are males and 35.5%. The share of males in long commuters is 27-29 % more than females. This means that men are traveling longer distances than female s to reach their jobs. This may be because, men relatively have more ability or free dom to make long commutes to access job opportunities than females. This could also be at tributed to the lead role females’ play in household maintenance and childcare, which re strict them to work close to work. In married couple households, when the male s are household heads and main source of income, the females try to access jobs in the near vicinity (Hjorthol, 2000). The NHTS 2001 data shows that 11. 2 % of the short commuters are 16 to 20 years of age, while only 4.0% of the long commuters belong to this age group. The share of individuals aged between 21 to 24 y ears is almost same in both s hort and long commuters (7.8% in short and 7.7% in long commuters). The shar e of individuals with age between 25 to 44 years is more in long commuters than in short commuters-47.5 % of this age group in short commuters and 51.9% in long commuters. This difference in share between short and long commuters is also observed in age gr oup of 45 to 65 years. Some difference is observed in ACS 2000 data but the same tre nd is present. The trend shows that an individuals’ share in long commuters is in creasing with age up to 65 years and finally dropping thereafter. This indicates that when individuals are young (16-20years) and just enter the work force they work very close to home and as they gain experience and achieve financial freedom, their preferences for life style tend to increase and this influences their ability to commute long distan ces to satisfy their ne eds. Individuals after reaching 65 years, (retirement age) because of low wages or because of older age, restrict their commute to shorter distances. Two major groups show interesting behavior in commuting to work as seen from the NHTS and ACS data sets. NHTS 2001 data show s that individuals who are white make

PAGE 25

16 81.1 % of the short commuters and 73.6% of the long commuters. Black or African American individuals are 9.2 in short comm uters and 13.4% in long commuters. This might be because of the Black or African Am erican people use transit for traveling to work, which is usually slower than automobiles. The same trend is observed in ACS but there is a difference in the magnitude of the shares between the datasets, which is attributed to the different classification adopted. The educational attainment is the next importa nt indicator of job c hoice of an individual. Highly educated people have more preferen ces in job choice and residential location. Highly educated people try to achieve the jobs that meet their qualifications and that satisfy their requirements. Long commuting ca n be considered as a tradeoff for their preferences. Individual income s are generally proportional to educational attainment and so highly educated people have more free dom for household locati on in low-density and suburban lifestyle, which are generally far aw ay from the main city. This is clearly observed in both NHTS 2001 and ACS 2000 data sets. The share of moderately and highly educated people is more in long co mmuters than in shor t commuters. The NHTS data shows that the share of individuals with bachelors’ degr ee is 22.8% in long commuters and 16.3% in short commutes. The ACS data shows that 20.0% of long commut ers hold bachelors’ degree where as only 16.0% of them hold in short commuters. The sh are of masters’ degr ee holders in long commuters is 6.9% and 5.6% in short commuters The share of professional and doctorate degree holders show the same trend but the difference is not so significant (In both NHTS and ACS). Both the data sets show that most of the lower income pe ople are making short commutes. The share of moderate to higherincome people is more in long commuters than in short commuters. The NHTS data show s that people with annual income less than $15,000 make 41% of the short commuters and only 12% of the long commuters. ACS data of 2000 shows that lower income people make 28% of the short commuters and only 15.6% of long commuters. The huge difference in the two data sets is because of the large amount of missing data (95%) about person’s income in NHTS. The ACS shows that as income increase there share in long commuters is increasing. The information about driver status is pres ent only in the NHTS dataset and not in ACS dataset. The data shows that share of th e drivers in long commuters are less (91.1%) when compared to short commuters (95.2%). This may be because regular transit users who are long commuters fall into this “not a driver” category. Table 4.4 and Table 4.5 present the distri bution and percentage of commuters by Standard Occupational Category (SOC). The pe rcentage of long commuters is more for individuals working in Comput er and Mathematical occupa tions, Legal occupations, and Construction and Extraction occupations. This variable needs further investigation before making any conclusions.

PAGE 26

17 Table 4.2 Person Characteristics of Commuters (NHTS) Sample Size 11,876 (47.6%) 11,641 (46.7%) 1,431 (5.7%) 24,948 (100%) Weighted Population 54,462,391 (46.3%) 55,736,358 (47.5%) 7,207,110 (6.1%) 117,405,859 (100%) Characteristic Short Commuters Medium Commuters Long Commuters All Commuters Gender Male 50.5 55.1 63.6 53.5 Female 49.5 44.9 36.4 46.5 Age Under 16 years 0.0 0.0 0.0 0 16 to 20 years 11.2 5.0 4.0 7.9 21 to 24 years 7.8 7.4 7.7 7.6 25 to 44 years 47.5 53.1 51.9 50.5 45 to 64 years 30.0 31.9 34.0 31.1 65 years and over 3.4 2.5 2.5 2.9 Race White 73.0 70.5 64.0 71.3 African American, Black 10.5 12.3 13.6 11.6 Asian Only 2.3 2.7 3.5 2.6 American Indian, Alaskan 0.4 0.4 0.6 0.4 Native Hawaiian, Pacific Isld 0.3 0.3 0.5 0.3 Hispanic/Mexican Only 5.9 5.4 6.7 5.7 White & African American 0.1 0.1 0.0 0.1 White & Asian 0.1 0.2 0.1 0.1 White & American Indian 0.9 1.0 1.0 0.9 White & Hispanic 4.3 4.9 6.1 4.7 African American & Hispanic 0.2 0.2 0.7 0.2 American Indian & Hispanic 0.2 0.1 0.2 0.1 Other Combination 2 Races 0.6 0.6 1.8 0.7 Other Combination 3 Races 0.2 0.2 0.2 0.2 Other multiracial 0.1 0.1 0.3 0.1 Other specify 0.8 1.0 0.8 0.9 Education Less then high school gra duate 11.3 7.3 9.2 9.3 High school graduate, inc GED 30.7 27.9 27.5 29.2 Vocational/technical training 3.5 4.0 3.8 3.8 Some college, but no degree 19.3 18.0 14.5 18.4 Associate’s degree 7.1 8.9 6.7 7.9 Bachelor’s degree 16.3 19.8 22.8 18.4 Graduate/professional school 1.9 2.0 2.1 2.0 Graduate/professional de gree 9.7 12.2 13.3 11.1

PAGE 27

18 Table 4.2 (Continued) Sample Size 11,876 (47.6%) 11,641 (46.7%) 1,431 (5.7%) 24,948 (100%) Weighted Population 54,462,391 (46.3%) 55,736,358 (47.5%) 7,207,110 (6.1%) 117,405,859 (100%) Characteristic Short Commuters Medium Commuters Long Commuters All Commuters Income less than $ 15,000 41 28 12 33.2 $15,000 $19,999 11.0 10.0 15.9 10.8 $20,000 $24,999 12.3 10.7 14.7 11.6 $25,000 $49,999 24.3 30.7 33.6 27.8 $50,000 $74,999 5.5 12.7 7.5 9.0 $75,000 $99,999 3.7 4.6 4.1 4.2 $100,000 and above 2.4 3.7 12.3 3.5 Driver status driver 95.2 95.2 91.1 95.0 not a driver’ 4.8 4.8 8.9 5.0

PAGE 28

19 Table 4.3 Person Characteristics of Commuters (ACS) Sample Size 249,880 (47.4%) 242,264 (45.9%) 35,602 (6.7%) 527,746 (100%) Weighted Population 56,067,997 (45.2%) 58,929,736 (47.5%) 8,941,813 (7.2%) 123,939,546 (100%) Characteristic Short Commuters Medium Commuters Long Commuters All Commuters Gender Male 50.5 56.1 64.5 54.2 Female 49.5 43.9 35.5 45.8 Age Under 16 years 0.0 0.0 0.0 0.0 16 to 20 years 8.9 4.5 3.4 6.4 21 to 24 years 8.8 7.5 6.6 8.0 25 to 44 years 46.4 52.0 53.6 49.6 45 to 64 years 32.4 33.6 34.2 33.1 65 years and over 3.5 2.5 2.2 2.9 Race White alone 81.1 77.1 73.6 78.7 Black or African American alone 9.2 11.7 13.4 10.7 American Indian alone 0.5 0.4 0.5 0.5 Alaska Native al one 0.0 0.0 0.0 0.0 American Indian & Alas ka Native 0.1 0.1 0.2 0.1 Asian alone 3.3 4.3 5.1 3.9 Native Hawaiian/Pacific Islander 0.1 0.1 0.1 0.1 Some other race alone 3.9 4.5 5.1 4.3 Two or more major race groups 1.7 1.7 2.0 1.7 Education Less than High school graduate 15.50 13.1 14.3 14.1 High school gradua te 29.6 27.3 26.2 28.3 Some college but no degree 7.3 7.2 6.9 7.3 Vo/Tech/Bus school degree 15.7 15.4 15 15.5 Associate degree in college 7.2 7.8 7.3 7.5 Bachelor's degree 16 19.1 20 17.7 Master's degree 5.6 6.9 6.9 6.4 Professional school degree 2 2.2 2.2 2.1 Doctorate degree 1.1 1.1 1.2 1.1 Income less than $ 15,000 28 18.1 15.6 22.4 $15,000 $19,999 10.5 8.9 7.2 9.5 $20,000 $24,999 10.6 10.1 8.3 10.2 $25,000 $49,999 33.1 37.8 36.5 35.6 $50,000 $74,999 10.6 14.8 17.8 13.1 $75,000 $99,999 3.1 4.9 6.6 4.2 $100,000 and above 4.2 5.4 7.9 5

PAGE 29

20 Table 4.4 Distribution of Commuters by Standard Occupational Category (ACS) Sample Size 249,880 (47.4%) 242,264 (45.9%) 35,602 (6.7%) 527,746 (100%) Weighted Population 56,067,997 (45.2%) 58,929,736 (47.5%) 8,941,813 (7.2%) 123,939,546 (100%) Occupational Category Short Commuters Medium Commuters Long Commuters All Commuters Management Occupati ons 8.8 9.5 11.0 9.3 Business and Financial Operations Occupations 3.3 4.7 5.3 4.1 Computer and Mathematical Occupations 1.7 3.1 4.2 2.6 Architecture and Engineering Occupations 1.5 2.6 2.6 2.1 Life, Physical, and Social Scie nce Occupations 0.8 1.1 1.1 1.0 Community and Social Services Occupations 1.7 1.4 1.2 1.5 Legal Occupations 0.9 1.2 1.5 1.1 Education, Training, and Librar y Occupations 6.3 4.6 2.8 5.2 Arts, Design, Entertainment, Sports, and Media Occupations 1.5 1.7 1.9 1.6 Healthcare Practitioners and Techni cal Occupations 4.6 5.0 3.6 4.7 Healthcare Support Occupations 2.2 2.0 1.7 2.1 Protective Service Occupations 2.0 2.1 2.4 2.1 Food Preparation and Serving Rela ted Occupations 6.9 3.6 2.5 5.0 Building and Grounds Cleaning and Maintenance Occupations 3.7 3.3 3.4 3.5 Personal Care and Service Oc cupations 3.0 2.0 1.8 2.4 Sales and Related Occupa tions 12.8 9.9 9.3 11.1 Office and Administrative Support Occupations 15.6 15.8 13.7 15.5 Farming, Fishing, and Forestry Occupations 0.8 0.5 0.6 0.6 Construction and Extraction O ccupations 4.1 6.4 11.5 5.7 Installation, Maintenance, and Re pair Occupations 3.7 4.2 4.6 4.0 Production Occupations 7.6 8.8 6.4 8.1 Transportation and Material Movi ng Occupations 6.4 6.3 6.8 6.4 Military Specific Occupations 0.2 0.2 0.1 0.2

PAGE 30

21 Table 4.5 Percentage of Commuter Type wi thin each Standard Occupational Category (ACS) Sample Size 249,880 (47.4%) 242,264 (45.9%) 35,602 (6.7%) 527,746 (100%) Weighted Population 56,067,997 (45.2%) 58,929,736 (47.5%) 8,941,813 (7.2%) 123,939,546 (100%) Occupational Category Short Commuters Medium Commuters Long Commuters All Commuters Management Occupati ons 42.7 48.7 8.6 100 Business and Financial Operatio ns Occupations 36.1 54.6 9.3 100 Computer and Math ematical Occupati ons 31.0 57.3 11.8 100 Architecture and Engineering Occupations 32.7 58.4 8.9 100 Life, Physical, and Social Scie nce Occupations 37.6 54.4 8.0 100 Community and Social Services Occupations 49.9 44.5 5.7 100 Legal Occupations 38.0 52.1 10.0 100 Education, Training, and Libr ary Occupations 54.3 41.8 3.8 100 Arts, Design, Entertainment, Sports, and Media Occupations 42.5 49.1 8.4 100 Healthcare Practitioners and Tec hnical Occupations 43.9 50.5 5.5 100 Healthcare Support Occupations 47.6 46.4 6.0 100 Protective Service Occupations 43.2 48.6 8.2 100 Food Preparation and Serving Rela ted Occupations 62.1 34.3 3.6 100 Building and Grounds Cleaning and Maintenance Occupations 48.1 45.0 6.9 100 Personal Care and Service Occupations 55.1 39.5 5.4 100 Sales and Related Occ upations 51.9 42.1 6.0 100 Office and Administrative Support Occupations 45.4 48.3 6.3 100 Farming, Fishing, and Forest ry Occupations 54.9 37.9 7.2 100 Construction and Extraction Occupations 32.5 53.0 14.6 100 Installation, Maintenance, and Re pair Occupations 41.5 50.2 8.3 100 Production Occupations 42.7 51.6 5.7 100 Transportation and Material Mo ving Occupations 45.4 46.9 7.7 100 Military Specific Occ upations 50.5 45.0 4.5 100

PAGE 31

22 4.3 Household Characteristics Household characteristics like household size, family structure, income, number of workers, number of vehicles and number of children are analyzed for short, medium and long commuters for the two datasets and are presented in the Table 4.6 and Table 4.7. The share of 3 or more person households is more in long commuters when compared to short commuters. The NHTS data shows that th e share of individuals in long commuters increases with the household size. But, when we observe the number of children and the age of the children in the household the shar e of individuals is increasing in long commuters with the number of children and decreasing with the increase in children’s age. The ACS data also shows the same trend but the family structure variable is classified in different way. The ACS data s hows that the share of individuals in long commuters increases with number of children. The existence of a family in the household increases an individuals’ share in long commuters. Both the NHTS and ACS data show that individuals belonging to higher income households have larger share in long commuters than in short commuters. The NHTS data shows that higher income households ($100,000 and above) have a share of 22.7% in long commuters and only 14.5% in shor t commuters. The similar difference was observed in ACS data with hi gher income household individu als share of 24.5% in long commuters and 16.9% in short commuters. The NHTS data shows that the share of indivi duals with zero vehicl e households is high in long commuters. The ACS data shows the sa me trend and individua ls with 5 or more vehicles also have bigger share in long commuters. This could be because the long commuters are generally lower income households who do not afford a vehicle and travel in transit (slow modes) and very higher in come household who live in suburbs and travel great distances to access jobs in the other areas.

PAGE 32

23 Table 4.6 Household Characteristics of Commuters (NHTS) Sample Size 11,876 (47.6%) 11,641 (46.7%) 1,431 (5.7%) 24,948 (100%) Weighted Population 54,462,391 (46.3%) 55,736,358 (47.5%) 7,207,110 (6.1%) 117,405,859 (100%) Characteristic Short Commuters Medium Commuters Long Commuters All Commuters Number of people 1 person 9.0 8.2 6.7 8.5 2 person 28.5 29.7 26.5 28.9 3 person 22.8 23.5 23.9 23.2 4 person 23.0 23.3 25.7 23.3 5 or more 16.8 15.4 17.3 16.1 Family structure one adult, no children 8.7 7.9 6.2 8.1 2+ adults, no children 29.0 31.4 30.2 30.2 one adult, youngest child 0-5 1.2 0.9 1.5 1.1 2+ adults, youngest child 0-5 18.7 20.7 22.5 19.9 one adult, youngest child 6-15 1.9 1.9 2.0 1.9 2+ adults, youngest child 6-15 21.6 21.3 21.9 21.5 one adult, youngest child 16-21 1.5 1.2 1.2 1.3 2+ adults, youngest child 16-21 11.6 8.9 9.7 10.2 one adult, retired, no children 0.2 0.1 0.0 0.2 2+ adults, retired, no children 5.6 5.6 4.9 5.6 Total Income less than $ 15,000 6.2 4.7 6.5 5.5 $15,000 $19,999 4.5 3.9 4.2 4.2 $20,000 $24,999 4.8 3.6 2.6 4.1 $25,000 $49,999 33.4 29.3 25.9 31.0 $50,000 $74,999 22.2 23.8 21.6 22.9 $75,000 $99,999 14.4 16.8 16.5 15.7 $100,000 and above 14.5 17.9 22.7 16.6

PAGE 33

24 Table 4.6 (Continued) Sample Size 11,876 (47.6%) 11,641 (46.7%) 1,431 (5.7%) 24,948 (100%) Weighted Population 54,462,391 (46.3%) 55,736,358 (47.5%) 7,207,110 (6.1%) 117,405,859 (100%) Characteristic Short Commuters Medium Commuters Long Commuters All Commuters Number of workers no worker 0.0 0.0 0.0 0.0 1 worker 23.5 24.8 27.5 24.4 2 workers 51.7 54.7 51.3 53.1 3 workers 17.2 14.1 15.5 15.6 4 workers 5.8 5.1 4.7 5.4 5 workers 1.7 1.3 1.0 1.5 Vehicle availability 0 2.5 3.5 7.7 3.3 1 17.9 16.7 17.8 17.3 2 41.5 43.0 40.2 42.1 3 22.5 22.0 22.2 22.2 4 10.0 9.9 7.7 9.8 5 or more 5.6 5.1 4.5 5.3 Number of children 0 50.8 51.2 46.6 50.7 1 21.3 20.9 22.3 21.2 2 18.7 18.6 19.3 18.7 3 6.5 6.5 8.8 6.6 4 or more 2.8 2.8 3.0 2.8

PAGE 34

25 Table 4.7 Household Characteri stics of Commuters (ACS) Sample Size 249,880 (47.4%) 242,264 (45.9%) 35,602 (6.7%) 527,746 (100%) Weighted Population 56,067,997 (45.2%) 58,929,736 (47.5%) 8,941,813 (7.2%) 123,939,546 (100%) Characteristic Short Commuters Medium Commuters Long Commuters All Commuters Number of People 1 person 12.3 11.3 10.4 11.7 2 person 30.4 30.5 29.0 30.3 3 person 21.4 21.9 21.8 21.6 4 person 20.3 20.6 21.3 20.5 5 or more 15.6 15.7 17.5 15.8 Family structure married couple (f) 62.3 63.6 64.4 63.1 male hholder, no wife (f) 5.1 5.4 6.0 5.3 female hholder, no husband (f) 12.0 11.9 12.4 12.0 male hholder, living alone (nf) 6.4 5.9 6.0 6.2 male hholder, not living alone (nf) 4.7 4.6 4.0 4.6 female hholder, living alone (nf) 5.9 5.4 4.4 5.6 female hholder not living alone (nf) 3.6 3.2 2.9 3.4 Income less than $ 15,000 5.5 3.8 4.3 4.6 $15,000 $19,999 4.0 3.0 2.8 3.5 $20,000 $24,999 4.9 4.0 3.8 4.4 $25,000 $49,999 29.2 26.5 24.1 27.6 $50,000 $74,999 25.2 25.7 24.7 25.4 $75,000 $99,999 14.2 16.2 15.8 15.3 $100,000 and above 16.9 20.7 24.5 19.3 Number of workers No workers 0.2 0.2 0.3 0.2 1 worker 19.1 20.7 24.1 20.2 2 workers 55.2 56.5 55.0 55.8 3 or more workers 25.4 22.6 20.6 23.7

PAGE 35

26 Table 4.8 Distribution of Lo ng Commuters by Household Property Values (ACS) Table 4.9 Distribution of Long Commuters by Duration of Status by Household Ownership Type (ACS) Duration of Status OwnedRental 12 months or less 8.8 31.8 13 to 23 months 3.2 6.6 2 to 4 years 25.0 33.4 5 to 9 years 23.8 15.6 10 to 19 years 23.2 8.6 20 to 29 years 10.6 3.0 30 years or more 5.3 1.0 Total 100 100 Property Value Percentage Less than $ 10000 1.1 $ 10000 $ 14999 0.7 $ 15000 $ 19999 0.7 $ 20000 $ 24999 1.0 $ 25000 $ 29999 0.9 $ 30000 $ 34999 1.0 $ 35000 $ 39999 1.2 $ 40000 $ 49999 2.6 $ 50000 $ 59999 2.7 $ 60000 $ 69999 3.6 $ 70000 $ 79999 3.9 $ 80000 $ 89999 4.7 $ 90000 $ 99999 4.6 $100000 $124999 10.1 $125000 $149999 10.2 $150000 $174999 9.4 $175000 $199999 7.0 $200000 $249999 11.5 $250000 $299999 5.9 $300000 $399999 8.3 $400000 $499999 3.9 $500000 $749999 3.2 $750000 $999999 0.9 $1000000+ 0.8

PAGE 36

27 4.4 Trip Characteristics Commute time distribution of commuters ar e presented in Table 4.10 and Table 4.11 for NHTS 2001 and ACS 2000 data respectively. The tables also give information about the average travel times by purpose. The NHTS 2001 data shows that average commute time for all commuters in United States is 23.5 minutes. The ACS 2000 data shows the average commute time as 24.3 minutes. Th e Table 4.12 shows commute distance distribution of short, medium and long co mmuters by trip purpose for NHTS 2001 data. This table shows that there are some commut ers who travel 20 to 39.99 miles to work in less than or equal to 15 mi nutes and on the other side there are some commuters who spend 60 or more minutes traveling to work to cover a distance of 1 mile. This behavior is strongly attributed to area specific attributes and modal level of service variables. The average commute distance trav eled for all commuters in United States is 13.2 miles. Table 4.13 shows the average trip rates for di fferent commuter types. The average trip rate for short commuters is more than long commuters because, long commuters spend most of the travel time in commuting to wo rk and so this decreases their average trip frequency for other activities. Th e short commuters on the othe r hand have time for other activities on the way to work and return home like dropping off kids, shopping trips etc., The interesting trend is that ev en if the average trip rates vary with commute length, the proportions of work/work related/return to wo rk/return home and ot her travel in total trips do not vary significantly with commut e length. This indicates that commute length affects the participations rate for work/work related/return to work /return home and other travel but not the percentage of participati on (in terms of number of trips). The short commuters do more social recreational, se rve passenger and return home trips when compared to long commuters. Table 4.14 and Table 4.15 provide average trip length and total trip lengths by purpose for different type of commuters. The trends in these tables are similar to that of commute times and trip rates. This trend is highly reflected in the average and total travel time expenditu res provided in Table 4.16 and Table 4.17. The table 4.15 shows the travel time expenditure by purpose by co mmuter type. It can be seen that travel time expenditure of long commuters is more for both work/work related/return to work/return home and other travel. Long commuters spend more time on daily total travel than short and medium commuters. Within each commuter type the proportions of work/work related/return to work/return home tr avel time and other travel time in “total travel time” also vary significantly with co mmute length. This could have significant influence on out of home and in home ac tivity durations. The travel time expenditures and activity durations by commute length n eeds further research. The VMT is another important variable that explains the comm uter behavior. The average and total VMT traveled are shown in Table 4.18 and Table 4.19 respectively. The commute time influences other trip characteris tics of commute trip and othe r non-work travel and this is shown in Table 4.20, which shows the mean departure time by purpose for different commuters. The departure time for work trip for long commuters is 1 hr 17 minutes earlier than short commuters. Other trips lik e work related, religi ous, shopping, eat meat and serve passenger are also starting early for long commuters than short commuters. The trips like school, medical, family and personal social recreational and return home trips

PAGE 37

28 are starting later for long commuters than short commuters. Figures 1 to 3 show the departure time distributions for different trips purposes. Figure 4.1 and Figure 4.2 shows that long commuters start early to work and work related trips when compared to short and medium commuters. Figure 4.3 shows th at the peak of the departure time distributions of short and long commuters for sc hool trip are separate d significantly in the time of day. Carpooling is another important aspect related to commuter behavior. Table 4.21 and Table 4.22 show the drive alone vs. carpooling behavior among different type of commuters using NHTS 2001 data and ACS 2000 data respectively. NHTS 2000 data shows that the share of carpool ing and transit is more in long commuters than in short commuters. This could be attributed to th e commute mode itself. Travel time is “alternative attribute” of mode. So, the peopl e who are carpooling or traveling by transit might be spending more time in reaching their work than it usually takes by drive alone or personal vehicle and could be the main reason for their commute being long. The directional relationship between the commuter choice and mode choice needs further research. The NHTS 2001 data shows that 84.7 % of short commuter s drive alone, which only 57.5% of the long commuters do. The sh are of transit user s is 26.3% in long commuters whereas it is only 0.94% in shor t commuters. The ACS 2000 data shows that 82.25% of short commuters drive alone, wh ich only 58.58% of the long commuters do. The share of transit users is 23.82% in long co mmuters where as it is only 1.07% in short commuters. The Table 4.23 shows mode-share by trip purpose for different commuters. It shows that 14.3% of long commuters travel to work by rail transit. Nearly 47% of the long commuters travel to work related trips using pickup and ot her truck. It is interesting to note that 29.3 % of the long commuters ride on rail to school whereas only 0.7% of the short commuters do. The same is the case w ith religious trips, 28.4 % of the long commuters walk to religious trips whereas only 0.6% of the short commuters do. Nearly 33% of the long commuters walk to eat/meal trips while only 9.8% of short commuters do.

PAGE 38

29 Table 4.10 Commute Time Distribut ion by Commuter Type (NHTS) Sample Size 11,876 (47.6%) 11,641 (46.7%) 1,431 (5.7%) 24,948 (100%) Weighted Population 54,462,391 (46.3%) 55,736,358 (47.5%) 7,207,110 (6.1%) 117,405,859 (100%) Trip length (min) Short Commuters Medium Commuters Long Commuters All Commuters 1min-4 min 10.7 0.0 0.0 5.0 5min-9min 26.8 0.0 0.0 12.4 10min-15min 62.5 0.0 0.0 14.4 16min-29min 0.0 48.3 0.0 37.5 30min-44min 0.0 36.6 0.0 17.4 45min-59min 0.0 15.1 0.0 7.2 60min-74min 0.0 0.0 58.3 3.6 75min-89min 0.0 0.0 14.4 0.9 90min-104min 0.0 0.0 16.4 1.0 105min-119min 0.0 0.0 1.3 0.1 120min or 149min 0.0 0.0 5.0 0.3 150min or more 0.0 0.0 4.5 0.3 Average Commute Time 9.8 29.7 78.3 23.5

PAGE 39

30 Table 4.11 Commute Time Distribut ion by Commuter Type (ACS) Weighted Population 56,067,997 (45.2%) 58,929,736 (47.5%) 8,941,813 (7.2%) 123,939,546 (100%) Trip length (min) Short Commuters Medium Commuters Long Commuters All Commuters 1min-4 min 8.4 0.0 0.0 3.8 5min-9min 25.6 0.0 0.0 11.6 10min-15min 66.0 0.0 0.0 29.8 16min-29min 0.0 44.3 0.0 21.0 30min-44min 0.0 40.4 0.0 19.2 45min-59min 0.0 15.4 0.0 7.3 60min-74min 0.0 0.0 61.0 4.4 75min-89min 0.0 0.0 9.6 0.7 90min-104min 0.0 0.0 15.2 1.1 105min-119min 0.0 0.0 1.2 0.1 120min or 149min 0.0 0.0 6.4 0.5 150min or more 0.0 0.0 6.5 0.5 Average Travel Time 9.9 29.7 79.3 24.3 Table 4.12 Commute Distance Distribut ion by Commuter Type (NHTS) Weighted Population 54,462,391 (46.3%) 55,736,358 (47.5%) 7,207,110 (6.1%) 117,405,859 (100%) Trip Length (miles) Short Commuters Medium Commuters Long Commuters All Commuters Less than 1 mile 11.0 0.4 0.6 5.3 1 to 1.99 miles 9.8 0.7 0.4 5.0 2 to 2.99 miles 13.7 1.1 0.4 7.0 3 to 4.99 miles 21.2 3.4 2.1 11.7 5 to 9.99 miles 33.1 15.2 2.6 22.8 10 to 14.99 miles 9.3 23.6 5.7 15.8 15 to 19.99 miles 1.4 19.5 4.5 10.1 20 to 39.99 miles 0.4 32.2 31.5 17.3 40 to 99.99 miles 0.0 3.9 47.3 4.6 100 to 199.99 miles 0.0 0.0 3.9 0.2 200 or more miles 0.0 0.0 1.1 0.1 Average Commute Distance 4.8 17.3 46.4 13.2

PAGE 40

31 Table 4.13 Average Trip Rate by Purpose by Commuter Type (NHTS) Sample Size 6,386 (45.7%) 6,826 (48.8%) 764 (5.5%) 13,976 (100%) Weighted Population 28,192,662 (44.5%) 31,412,074 (49.6%) 3,738,126 (5.9%) 63,342,863 (100%) Purpose Short Commuters Medium Commuters Long Commuters All Commuters To work 1.04 1.04 1.07 1.04 Work-related 0.17 0.17 0.16 0.17 School 0.06 0.04 0.02 0.05 Religious 0.02 0.02 0.01 0.02 Medical/dental 0.04 0.04 0.02 0.04 Shopping 0.38 0.36 0.34 0.37 Other family & personal 0.34 0.29 0.19 0.30 Social Recreation 0.30 0.26 0.18 0.27 Eat meal 0.28 0.31 0.27 0.30 Serve passenger 0.35 0.32 0.30 0.33 Return to work 0.33 0.26 0.21 0.29 Return home 1.69 1.45 1.29 1.55 Other trip purpose 0.01 0.02 0.04 0.02 Total 5.03 (100%) 4.59 (100%) 4.10 (100%) 4.8 (100%) Work/work related/return to work/return home 3.23 (64.3%) 2.94 (64.0%) 2.73 (66.5%) 3.06 (64.3%) Other travel 1.80 (35.7%) 1.65 (36.0%) 1.37 (33.5%) 1.70 (35.7%)

PAGE 41

32 Table 4.14 Average Trip Length Travele d by Purpose by Commuter (NHTS) Sample Size 6,386 (45.7%) 6,826 (48.8%) 764 (5.5%) 13,976 (100%) Weighted Population 28,192,662 (44.5%) 31,412,074 (49.6%) 3,738,126 (5.9%) 63,342,863 (100%) Purpose Short Commuters Medium Commuters Long Commuters All Commuters To work 4.99 16.17 34.69 12.29 Work-related 1.48 1.94 1.85 1.73 School 0.42 0.31 0.33 0.36 Religious 0.13 0.15 0.06 0.13 Medical/dental 0.36 0.39 0.30 0.37 Shopping 1.26 1.64 2.90 1.55 Other family & personal 1.25 1.46 1.69 1.38 Social Recreation 2.12 2.38 1.95 2.24 Eat meal 1.03 1.19 0.87 1.10 Serve passenger 1.20 1.53 3.00 1.47 Return to work 1.69 1.52 0.96 1.56 Return home 5.44 12.64 25.72 10.21 Other trip purpose 0.09 0.12 0.23 0.11

PAGE 42

33 Table 4.15 Total Trip Length Traveled by Purpose by Commuter Type (NHTS) Sample Size 6,386 (45.7%) 6,826 (48.8%) 764 (5.5%) 13,976 (100%) Weighted Population 28,192,662 (44.5%) 31,412,074 49.6%) 3,738,126 (5.9%) 63,342,863 (100%) Purpose Short Commuters Medium Commuters Long Commuters All Commuters To work 5.27 17.32 36.37 13.08 Work-related 2.71 3.05 3.20 2.91 School 0.45 0.33 0.33 0.38 Religious 0.13 0.16 0.08 0.14 Medical/dental 0.37 0.43 0.34 0.40 Shopping 1.65 2.12 3.83 2.01 Other family & pers onal 1.70 2.00 2.01 1.87 Social Recreation 2.68 2.75 2.08 2.68 Eat meal 1.24 1.44 0.99 1.33 Serve passenger 1.95 2.48 4.86 2.39 Return to work 1.94 1.79 0.99 1.81 Return home 9.06 17.20 31.81 14.44 Other trip purpose 0.10 0.13 0.24 0.12 Total 29.25 (100%) 51.21 (100%) 87.11 (100%) 43.6 (100%) Work/work related/return to work/return home 18.98 (64.9%) 39.37 (76.9%) 72.37 (83.1%) 32.24 (74.0%) Other travel 10.28 (35.1%) 11.84 (23.1%) 14.74 (16.9%) 11.32 (26.0%)

PAGE 43

34 Table 4.16 Average Trip Duration by Purpose by Commuter Type (NHTS) Sample Size 6,386 (45.7%) 6,826 (48.8%) 764 (5.5%) 13,976 (100%) Weighted Population 28,192,662 (44.5%) 31,412,074 (49.6%) 3,738,126 (5.9%) 63,342,863 (100%) Purpose Short Commuters Medium Commuters Long Commuters All Commuters To work 11.90 29.72 65.84 23.92 Work-related 2.67 3.13 2.75 2.90 School 0.96 0.75 1.05 0.86 Religious 0.29 0.30 0.22 0.29 Medical/dental 0.66 0.80 0.62 0.73 Shopping 3.18 3.77 6.87 3.69 Other family & personal 3.11 3.47 3.53 3.32 Social Recreation 4.97 4.99 4.68 4.96 Eat meal 2.56 3.14 2.82 2.87 Serve passenger 2.60 3.20 5.05 3.04 Return to work 3.69 3.32 2.53 3.44 Return home 13.06 25.74 51.58 21.62 Other trip purpose 0.17 0.28 0.73 0.26

PAGE 44

35 Table 4.17 Total Travel Time Expenditure by Purpose by Commuter Type (NHTS) Sample Size 6,386 (45.7%) 6,826 (48.8%) 764 (5.5%) 13,976 (100%) Weighted Population 28,192,662 (44.5%) 31,412,074 (49.6%) 3,738,126 (5.9%) 63,342,863 (100%) Purpose Short Commuters Medium Commuters Long Commuters All Commuters To work 12.47 30.93 68.06 24.90 Work-related 4.53 5.06 4.85 4.81 School 1.05 0.78 1.05 0.92 Religious 0.31 0.32 0.25 0.31 Medical/dental 0.70 0.88 0.72 0.79 Shopping 4.16 4.86 8.50 4.77 Other family & personal 4.29 4.68 4.21 4.48 Social Recreation 6.18 5.87 5.13 5.96 Eat meal 3.07 3.71 3.34 3.41 Serve passenger 4.31 5.17 8.56 4.99 Return to work 4.22 3.87 2.63 3.95 Return home 21.81 35.45 61.64 30.92 Other trip purpose 0.19 0.34 0.91 0.30 Total 67.27 (100%) 101.94 (100%) 169.83 (100%) 90.5 (100%) Work/work related/return to work/return home 43.02 (64.0%) 75.31 (73.9%) 137.18 (80.8%) 64.59 (71.4%) Other travel 24.25 (36.0%) 26.63 (26.1%) 32.66 (19.2%) 25.93 (28.6%)

PAGE 45

36 Table 4.18 Average VMT by Purpose by Commuter Type (NHTS) Sample Size 6,386 (45.7%) 6,826 (48.8%) 764 (5.5%) 13,976 (100%) Weighted Population 28,192,662 (44.5%) 31,412,074 (49.6%) 3,738,126 (5.9%) 63,342,863 (100%) Purpose Short Commuters Medium Commuters Long Commuters All Commuters To work 4.60 14.57 27.52 10.90 Work-related 1.22 1.62 1.57 1.44 School 0.34 0.28 0.15 0.30 Religious 0.09 0.12 0.06 0.10 Medical/dental 0.32 0.34 0.25 0.33 Shopping 1.08 1.43 2.55 1.34 Other family & personal 0.98 1.25 1.13 1.12 Social Recreation 1.41 1.62 1.21 1.50 Eat meal 0.75 1.00 0.71 0.87 Serve passenger 1.05 1.31 2.42 1.26 Return to work 1.48 1.25 0.88 1.33 Return home 4.77 11.31 19.78 8.90 Other trip purpose 0.06 0.09 0.14 0.08

PAGE 46

37 Table 4.19Total VMT by Purpose by Commuter Type (NHTS) Sample Size 6,386 (45.7%) 6,826 (48.8%) 764 (5.5%) 13,976 (100%) Weighted Population 28,192,662 (44.5%) 31,412,074 (49.6%) 3,738,126 (5.9%) 63,342,863 (100%) Purpose Short Commuters Medium Commuters Long Commuters All Commuters To work 4.85 15.11 28.22 11.32 Work-related 2.40 2.57 2.49 2.49 School 0.36 0.29 0.15 0.32 Religious 0.09 0.12 0.07 0.11 Medical/dental 0.34 0.37 0.28 0.35 Shopping 1.43 1.86 3.40 1.76 Other family & personal 1.35 1.73 1.40 1.54 Social Recreation 1.75 1.90 1.36 1.81 Eat meal 0.88 1.21 0.80 1.04 Serve passenger 1.71 2.14 3.97 2.06 Return to work 1.71 1.48 0.91 1.55 Return home 7.89 15.28 23.73 12.49 Other trip purpose 0.06 0.10 0.16 0.09 Total 24.83 (100%) 44.15 (100%) 66.95 (100%) 36.9 (100%) Work/work related/return to work/return home 16.84 (67.8%) 34.43 (78.0%) 55.34 (82.7%) 27.84 (75.5%) Other travel 7.98 (32.2%) 9.72 (22.0%) 11.61 (17.3%) 9.06 (24.5%)

PAGE 47

38 Table 4.20 Mean Departure Time by Purpose by Commuter Type (NHTS) Sample Size 6,386 (45.7%) 6,826 (48.8%) 764 (5.5%) 13,976 (100%) Weighted Population 28,192,662 (44.5%) 31,412,074 (49.6%) 3,738,126 (5.9%) 63,342,863 (100%) Purpose Short Commuters Medium Commuters Long Commuters All Commuters To work 8.45 AM 8.12 AM 7.35 AM 8.30 AM Work-related 11.55 AM 11. 52 AM 11.20 AM 11.46 AM School 10.33 AM 12.11 PM 2.21 PM 11.08 AM Religious 5.14 PM 5.36 PM 4.29 PM 5.18 PM Medical/dental 12.56 PM 12.31 PM 1.34 PM 12.44 PM Shopping 3.17 PM 3.16 PM 2.31 PM 3.13 PM Other family & personal 2.14 PM 2.27 PM 2.28 PM 2.23 PM Social Recreation 4.03 PM 4.1 PM 4.44 PM 4.20 PM Eat meal 2.16 PM 2.01 PM 1.45 PM 2.09 PM Serve passenger 1.01 PM 12.35 PM 12.57 PM 12.45 PM Return to work 1.22 PM 1.11 PM 1.22 PM 1.17 PM Return home 4.39 PM 5.02 PM 5.20 PM 4.50 PM Other trip purpose 4.11 PM 2.29 PM 3.07 PM 2.55 PM

PAGE 48

39 Table 4.21 Drive Alone vs. Carpooling (NHTS) Sample Size 11,876 (47.6%) 11,641 (46.7%) 1,431 (5.7%) 24,948 (100%) Weighted Population 54,462,391 (46.3%) 55,736,358 (47.5%) 7,207,110 (6.1%) 117,405,859 (100%) Commute Pattern Short Commuters Medium Commuters Long Commuters All Commuters Drive alone 84.69 82.18 57.27 81.81 Car 57.16 55.69 37.31 55.25 Van 5.83 4.96 3.22 5.26 SUV 7.21 7.40 5.19 7.18 Pickup truck 13.76 13.54 10.38 13.45 Other truck 0.42 0.24 1.05 0.37 RV 0.01 0.01 0.00 0.01 Motorcycle 0.30 0.34 0.13 0.31 Carpool 8.29 9.75 14.52 9.37 Car 5.92 6.77 7.86 6.44 Van 0.73 0.92 2.20 0.91 SUV 0.50 0.62 0.65 0.57 Pickup truck 1.11 1.43 3.61 1.42 Other truck 0.03 0.02 0.19 0.03 Transit 0.94 6.65 26.56 5.22 Local public transit bus 0.73 3.22 9.12 2.43 Commuter bus 0.03 0.14 1.24 0.16 Charter/tour bus 0.00 0.04 0.41 0.05 City to city bus 0.00 0.05 0.50 0.05 Amtrack/inter city train 0.00 0.18 1.20 0.16 Commuter train 0.00 0.70 6.31 0.72 Subway/elevated rail 0.14 2.25 7.24 1.57 Street car/trolley 0.03 0.05 0.28 0.06 Passenger line/ferry 0.00 0.01 0.27 0.02 Sailboat/motorboat/ya cht 0.00 0.00 0.00 0.00 Other 6.08 1.42 1.65 3.59 Commercial/charter plane 0.00 0.00 0.49 0.03 Private/corporate airplane 0.00 0.00 0.00 0.00 School bus 0.11 0.07 0.47 0.12 Taxicab 0.10 0.05 0.00 0.07 Hotel/airport shuttle 0.00 0.00 0.07 0.00 Bicycle 0.71 0.36 0.20 0.51 Walk 5.03 0.90 0.37 2.78 Other 0.12 0.03 0.05 0.07 Total 100 100 100 100

PAGE 49

40 Table 4.22 Drive Alone vs. Carpooling (ACS) Sample Size 249,880 (47.4%) 242,264 (45.9%) 35,602 (6.7%) 527,746 (100%) Weighted Population 56,067,997 (45.2%) 58,929,736 (47.5%) 8,941,813 (7.2%) 123,939,546 (100%) Commute Pattern Short Commuters Medium Commuters Long Commuters All Commuters Drive alone Car, Van, or Truck 82.25 80.07 58.80 79.52 Carpool Car, Van, or Truck 9.83 11.84 14.57 11.13 Transit 1.07 6.12 23.82 5.11 Bus or trolley bus 0.90 3.55 11.00 2.89 Streetcar or trolley car 0.01 0.10 0.19 0.07 Subway or elevated 0.13 2.08 7.02 1.56 Railroad 0.02 0.36 5.21 0.56 Ferryboat 0.00 0.03 0.40 0.04 Other 6.85 1.97 2.82 4.24 Taxicab 0.20 0.09 0.03 0.13 Motorcycle 0.14 0.12 0.08 0.13 Bicycle 0.61 0.31 0.19 0.44 Walked 4.92 0.84 0.30 2.65 Worked at home 0.00 0.00 0.00 0.00 Other method 0.99 0.60 2.23 0.89

PAGE 50

41 Table 4.23 Mode Share by Trip Purpose by Commuter Type (NHTS) Trip Purpose Type Auto Truck Walk/ Bike Airplane Bus Rail Ship Taxi Other To work Short 77.5 17.6 4. 1 0.0 0.5 0.1 0.0 0.1 0.13 Medium 74.7 17.6 2.2 0. 1 2.7 2.5 0.0 0.1 0.15 Long 56.0 16.4 3.6 0.0 9.2 14.3 0.3 0.0 0.19 Workrelated Short 61.1 33.5 3.7 0. 1 0.7 0.2 0.0 0.3 0.94 Medium 62.1 30.7 5.0 0. 2 0.9 0.7 0.0 0.2 0.31 Long 44.6 47.6 4.1 0.0 0.0 3.6 0.0 0.0 0.00 School Short 77.2 11.1 5.4 0.0 5.5 0.7 0.0 0.0 4.51 Medium 71.2 9.0 11.9 0. 0 6.5 1.3 0.0 0.0 5.88 Long 51.9 8.7 10.1 0.0 0.0 29.3 0.0 0.0 0.00 Religious Short 87.8 10.6 0. 6 0.0 0.6 0.4 0.0 0.0 0.00 Medium 93.1 6.2 0.0 0. 0 0.7 0.0 0.0 0.0 0.00 Long 50.5 0.0 28.4 0.0 7.0 14.1 0.0 0.0 0.00 Medical/ dental Short 88.6 8.4 3.0 0. 0 0.0 0.0 0.0 0.0 0.00 Medium 81.7 11.9 1.4 0. 0 0.3 1.2 0.0 1.2 2.30 Long 70.1 7.0 6.4 0.0 0.0 9.7 0.0 6.8 0.00 Shopping Short 81.7 13.8 4. 3 0.0 0.1 0.1 0.0 0.0 0.00 Medium 79.9 14.3 5.0 0. 0 0.3 0.5 0.0 0.0 0.00 Long 62.7 18.8 11.9 0.0 5.1 1.6 0.0 0.0 0.00 Family & personal Short 77.2 14.0 8.3 0. 0 0.1 0.1 0.0 0.0 0.05 Medium 75.5 11.9 11.2 0. 1 0.6 0.6 0.0 0.1 0.05 Long 62.9 14.6 16.6 0.0 3.9 1.3 0.7 0.0 0.00 Social Short 64.6 12.8 21. 3 0.1 0.3 0.0 0.0 0.0 1.11 Medium 60.9 11.4 23.1 0. 4 0.5 1.0 0.0 0.5 2.19 Long 49.9 11.7 28.2 0.4 2.5 2.9 0.0 0.0 5.32 Eat meal Short 74.4 15.2 9. 8 0.0 0.4 0.0 0.0 0.2 0.18 Medium 71.8 13.3 13.7 0. 0 0.3 0.7 0.0 0.3 0.00 Long 55.3 9.4 33.3 0.0 0.0 1.5 0.0 0.0 0.51 Serve passenger Short 87.2 10.7 1.6 0. 1 0.2 0.0 0.0 0.0 0.33 Medium 87.3 9.9 1.9 0. 0 0.5 0.5 0.0 0.0 0.18 Long 81.6 11.1 3.3 0.0 3.6 0.4 0.0 0.0 0.00 Return to work Short 68.1 20.4 10.6 0. 0 0.4 0.1 0.0 0.1 0.55 Medium 63.9 18.4 16.1 0. 0 0.5 0.5 0.0 0.4 0.37 Long 49.0 8.9 39.3 0.0 0.0 2.0 0.0 0.8 0.00

PAGE 51

42 Table 4.23 (Continued) Trip Purpose Type Auto Truck Walk/ Bike Airplane Bus Rail Ship Taxi Other Return home Short 77.8 15.7 5.9 0. 0 0.3 0.1 0.0 0.0 0.15 Medium 75.5 15.4 5.6 0. 0 1.8 1.4 0.0 0.2 0.16 Long 61.9 14.6 6.3 0.2 7.7 8.5 0.1 0.6 0.27 Other trip purpose Short 78.8 12.8 4.6 0. 0 0.8 1.6 0.0 0.0 2.53 Medium 80.7 7.3 9.4 0. 0 0.4 0.9 0.0 0.0 1.87 Long 52.8 8.6 29.5 2.0 0.8 6.4 0.0 0.0 0.00 All Trips Short 76.6 16.1 6. 6 0.0 0.4 0.1 0.0 0.0 0.31 Medium 74.4 15.4 7.0 0. 1 1.4 1.3 0.0 0.2 0.32 Long 59.5 15.5 11.0 0. 1 5.8 7.4 0.1 0.3 0.37 All Trips All 74.6 15.7 7. 0 0.0 1.2 1.0 0.0 0.1 0.32 Table 4.24 Trip Length of Long Commuters by Job Specialization Occupation Type Commute Time (min) Commute Distance (mile) Sales or Service 78.94 43.91 Clerical or administrative support 70.43 31.90 Manufacturing, construction, main tenance, or farming 80.45 53.11 Professional, managerial or technical 78.85 47.21

PAGE 52

43 Figure 4.1 Work Trip Departure Time Distribution by Commuter Type 0.00 5.00 10.00 15.00 20.00 25.00 30.000:00 AM to 0:59 AM 1:00 AM to 1:59 AM 2:00 AM to 2:59 AM 3:00 AM to 3:59 AM 4:00 AM to 4:59 AM 5:00 AM to 5:59 AM 6:00 AM to 6:59 AM 7:00 AM to 7:59 AM 8:00 AM to 8:59 AM 9:00 AM to 9:59 AM 10:00 AM to 10:59 AM 11:00 AM to 11:59 AM 12:00 PM to 12:59 PM 1:00 PM to 1:59 PM 2:00 PM to 2:59 PM 3:00 PM to 3:59 PM 4:00 PM to 4:59 PM 5:00 PM to 5:59 PM 6:00 PM to 6:59 PM 7:00 PM to 7:59 PM 8:00 PM to 8:59 PM 9:00 PM to 9:59 PM 10:00 PM to 10:59 PM 11:00 PM to 11:59 PMTime of day (hr)Percent of commuters short medium long

PAGE 53

44 Figure 4.2 Work Related Trip Departure Time Distribution by Commuter Type 0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.000:00 AM to 0:59 AM 1:00 AM to 1:59 AM 2:00 AM to 2:59 AM 3:00 AM to 3:59 AM 4:00 AM to 4:59 AM 5:00 AM to 5:59 AM 6:00 AM to 6:59 AM 7:00 AM to 7:59 AM 8:00 AM to 8:59 AM 9:00 AM to 9:59 AM 10:00 AM to 10:59 AM 11:00 AM to 11:59 AM 12:00 PM to 12:59 PM 1:00 PM to 1:59 PM 2:00 PM to 2:59 PM 3:00 PM to 3:59 PM 4:00 PM to 4:59 PM 5:00 PM to 5:59 PM 6:00 PM to 6:59 PM 7:00 PM to 7:59 PM 8:00 PM to 8:59 PM 9:00 PM to 9:59 PM 10:00 PM to 10:59 PM 11:00 PM to 11:59 PMTime of day (hr)Percent of commuters short medium long

PAGE 54

45 Figure 4.3 School Trip Departure Ti me Distribution by Commuter Type 0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.000:00 AM to 0:59 AM 1:00 AM to 1:59 AM 2:00 AM to 2:59 AM 3:00 AM to 3:59 AM 4:00 AM to 4:59 AM 5:00 AM to 5:59 AM 6:00 AM to 6:59 AM 7:00 AM to 7:59 AM 8:00 AM to 8:59 AM 9:00 AM to 9:59 AM 10:00 AM to 10:59 AM 11:00 AM to 11:59 AM 12:00 PM to 12:59 PM 1:00 PM to 1:59 PM 2:00 PM to 2:59 PM 3:00 PM to 3:59 PM 4:00 PM to 4:59 PM 5:00 PM to 5:59 PM 6:00 PM to 6:59 PM 7:00 PM to 7:59 PM 8:00 PM to 8:59 PM 9:00 PM to 9:59 PM 10:00 PM to 10:59 PM 11:00 PM to 11:59 PMTime of day (hr)Percent of commuters short medium long

PAGE 55

46 4.5 Area Related Characteristics The percentage of long commuters in Metropol itan Statistical Area (M SA) increases with the population. This is because, the populati on influences the congestion and urban sprawl and in turn they together incr ease the commute time. The Table 4.25and Table 4.26 shows the distribution and percentage of commuters by population of the MSA. The Table 4.27 and Table 4.28 show the same aspect of commuters with respect to type of urban area commuter belongs. The tables show that nearly 70 % of the commuters belong to an urban area. The share of the individuals in long commu ters increases as we move from urban cluster to lower density areas. The proportion of long commuters is more in an area surrounded by urban areas. The average commute time of all commuters is increasing with the MSA population as shown in Table 4.29. But with in long commuters the commu te time is very high, 107.26 min for MSAs of population less than 250,000. This is because of their long distance travel shown in Table 4.30. The average travel distance is 73 miles for long commuters in MSA with population less than 250,000 whereas only 37.4 miles in MSA with population equal to or more than 3 million. The average commute time of all commuters is more in areas surrounded by urban areas as shown in Table 4.31. The average commute di stance also shows the same trend, it is 16.53% in areas surrounded by urban areas and it is 17.11% in non-urban areas from Table 4.32. For non-urban areas the commute di stances are more and commute times are less when compared to areas surrounded by urban areas because of the low congestion and due to wide sprawl present in the non-urban areas. Table 4.33 and Table 4.34 show the distribution and percent of commuters by each state. The tables show that Alaska, Iowa, Montan a, Nebraska, North Dakota, South Dakota and Wyoming have the high percentage of short commuters. The North Dakota has the highest percentage, 75.5% of short commut ers followed by Wyoming at 73.5%. Illinois, Maryland, New Jersey, New York and West Virginia show high percentage of long commuters. New York has the highest percen tage, 15.2% of long co mmuters followed by New Jersey at 13.3%. Table 4.35 shows the average commute time of commuters for each state. The important identification is that, the states that have low percentage of l ong commuters have the longest commutes for them and at the same time this behavior is not observed for the whole set of commuters. The average commute time is high for Illinois, Maryland, New Jersey and New York. New York State has the highest commute time of 30.6 min and is followed by Maryland State at 29.1 min. The Table 4.36 and Table 4.37 show the average commute time and average commute distance of different commute rs by CMSA. The average commute time for the short commuters is low for Chicago, Cleveland, Ne w York and Seattle. The average commute time for long commuters is high for Denver, Milwaukee, New York, and San Francisco.

PAGE 56

47 The average commute time is highest for Ne w York at 31.7 min followed by Miami at 30 min. The average commute distance is highe st for Dallas at 16.32 miles and followed by Philadelphia at 16.31 miles. The longest commu te distance for long commuters is in Milwaukee, a distance of 70 miles. The shorte st commute distance for short commuters is for New York at 3.47 miles. Table 4.25 Distribution of Commuter Type by MSA Size (NHTS) Sample Size 11,876 (47.6%) 11,641 (46.7%) 1,431 (5.7%) 24,948 (100%) Weighted Population 54,462,391 (46.3%) 55,736,358 (47.5%) 7,207,110 (6.1%) 117,405,859 (100%) MSA Size by Population Short Commuters Medium Commuters Long Commuters All Commuters In an MSA (CMSA) of Less than 250,000 9.3 4.7 2.7 6.7 In an MSA (CMSA) of 250,000 – 499,999 9.1 7.2 4.0 7.9 In an MSA (CMSA) of 500,000 – 999,999 8.7 8.2 4.0 8.2 In an MSA (CMSA) of 1,000,000 2,999,999 21.5 24.2 11.9 22.2 In an MSA (CMSA) of 3 million or more 29.9 40.1 61.0 36.7 Not in MSA or CMSA 21.4 15.5 16.3 18.3 Table 4.26 Percentage of Commuter Type by MSA Size (NHTS) Sample Size 11,876 (47.6%) 11,641 (46.7%) 1,431 (5.7%) 24,948 (100%) Weighted Population 54,462,391 (46.3%) 55,736,358 (47.5%) 7,207,110 (6.1%) 117,405,859 (100%) MSA Size by Population Short Commuters Medium Commuters Long Commuters All Commuters In an MSA (CMSA) of Less than 250,000 64.1 33.4 2.5 100 In an MSA (CMSA) of 250,000 499,999 53.8 43.1 3.1 100 In an MSA (CMSA) of 500,000 999,999 49.4 47.6 3.0 100 In an MSA (CMSA) of 1,000,000 2,999,999 44.9 51.8 3.3 100 In an MSA (CMSA) of 3 million or more 37.9 51.9 10.2 100 Not in MSA or CMSA 54.2 40.3 5.5 100

PAGE 57

48 Table 4.27 Distribution of Commuter Ty pe by Urban Area Type (NHTS) Sample Size 11,876 (47.6%) 11,641 (46.7%) 1,431 (5.7%) 24,948 (100%) Weighted Population 54,462,391 (46.3%) 55,736,358 (47.5%) 7,207,110 (6.1%) 117,405,859 (100%) Urban Area Type Short Commuters Medium Commuters Long Commuters All Commuters In an Urban Cluster 13.3 7.0 8.7 10.0 In an Urban Area 68.3 70.3 71.0 69.4 In an Area Surrounded by Urban Areas 0.0 0.1 0.1 0.1 Not in Urban Area 18.4 22.6 20.2 20.5 Table 4.28 Percentage of Commuter Type by Urban Area Type (NHTS) Sample Size 11,876 (47.6%) 11,641 (46.7%) 1,431 (5.7%) 24,948 (100%) Weighted Population 54,462,391 (46.3%) 55,736,358 (47.5%) 7,207,110 (6.1%) 117,405,859 (100%) Urban Area Type Short Commuters Medium Commuters Long Commuters All Commuters In an Urban Cluster 61.4 33.3 5.3 100 In an Urban Area 45.6 48.1 6.3 100 In an Area Surrounded by Urban Areas 24.8 62.2 13.0 100 Not in Urban Area 41.7 52.3 6.0 100 Table 4.29 Average Commute Time by Commuter Type by MSA Size (NHTS) Sample Size 11,876 (47.6%) 11,641 (46.7%) 1,431 (5.7%) 24,948 (100%) Weighted Population 54,462,391 (46.3%) 55,736,358 (47.5%) 7,207,110 (6.1%) 117,405,859 (100%) MSA Size by Population Short Commuters Medium Commuters Long Commuters All Commuters In an MSA (CMSA) of Less than 250,000 9.57 26.45 107.26 17.6 In an MSA (CMSA) of 250,000 – 499,999 9.55 27.09 86.00 19.4 In an MSA (CMSA) of 500,000 – 999,999 10.19 27.46 89.38 20.6 In an MSA (CMSA) of 1,000,000 – 2,999,999 10.15 28.18 78.04 21.5 In an MSA (CMSA) of 3 million or more 10.11 31.43 74.39 27.6 Not in MSA or CMSA 8.40 29.54 82.38 21.2 Average Commute Time for all Areas 9.8 29.7 78.3 23.5

PAGE 58

49 Table 4.30 Average Commute Distance by Commuter Type by MSA Size (NHTS) Sample Size 11,876 (47.6%) 11,641 (46.7%) 1,431 (5.7%) 24,948 (100%) Weighted Population 54,462,391 (46.3%) 55,736,358 (47.5%) 7,207,110 (6.1%) 117,405,859 (100%) MSA Size by Population Short Commuters Medium Commuters Long Commuters All Commuters In an MSA (CMSA) of Less than 250,000 4.58 16.90 73.40 10.29 In an MSA (CMSA) of 250,000 499,999 4.88 16.95 61.98 11.88 In an MSA (CMSA) of 500,000 999,999 4.87 16.82 61.06 12.26 In an MSA (CMSA) of 1,000,000 2,999,999 4.79 16.36 48.05 12.12 In an MSA (CMSA) of 3 million or more 5.12 16.56 37.40 14.20 Not in MSA or CMSA 4.54 21.33 65.07 14.55 Average Commute Distance fo r all Areas 4.8 17.3 46.4 13.2 Table 4.31 Average Commute Time by Commuter Type by Urban Area (NHTS) Sample Size 11,876 (47.6%) 11,641 (46.7%) 1,431 (5.7%) 24,948 (100%) Weighted Population 54,462,391 (46.3%) 55,736,358 (47.5%) 7,207,110 (6.1%) 117,405,859 (100%) Urban Area Type Short Commuters Medium Commuters Long Commuters All Commuters In an Urban Cluster 8.13 30.43 85.35 19.50 In an Urban Area 10.14 29.34 76.48 23.43 In an Area Surrounded by Urban Areas 7.39 31.34 93.53 33.43 Not in Urban Area 9.35 29.52 80.25 24.28 Average Commute Time for all Areas 9.8 29.7 78.3 23.5 Table 4.32 Average Commute Distance by Co mmuter Type by Urban Area (NHTS) Sample Size 11,876 (47.6%) 11,641 (46.7%) 1,431 (5.7%) 24,948 (100%) Weighted Population 54,462,391 (46.3%) 55,736,358 (47.5%) 7,207,110 (6.1%) 117,405,859 (100%) Urban Area Type Short Commuters Medium Commuters Long Commuters All Commuters In an Urban Cluster 3.77 21.89 60.39 12.67 In an Urban Area 4.80 15.67 39.61 12.08 In an Area Surrounded by Urban Areas 4.54 14.78 47.84 16.53 Not in Urban Area 5.69 20.97 62.94 17.11 Average Commute Distance fo r all Areas 4.8 17.3 46.4 13.2

PAGE 59

50 Figure 4.4 Commute Length by MSA Size 107.26 89.38 78.04 74.39 82.38 73.40 61.06 48.05 37.40 65.07 86.00 61.98 0.00 20.00 40.00 60.00 80.00 100.00 120.00 In an MSA (CMSA) of Less than 250,000 In an MSA (CMSA) of 250,000 499,999 In an MSA (CMSA) of 500,000 999,999 In an MSA (CMSA) of 1,000,000 2,999,999 In an MSA (CMSA) of 3 million or more Not in MSA or CMSA 123456 MSA SizeTime (min) or Distance (mile ) Commute Time Commute Distance

PAGE 60

51 Figure 4.5 Commute Length by Area Type 85.35 76.48 93.53 80.25 60.39 39.61 47.84 62.94 0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 90.00 100.00 In an Urban ClusterIn an Urban AreaIn an Area Surrounded by Urban Areas Not in Urban Area Area TypeTime (min) or Distance (min) Commute Time Commute Distance

PAGE 61

52 Table 4.33 Distribution of Commuter Type by State (ACS) Sample Size 249,880 (47.4%) 242,264 (45.9%) 35,602 (6.7%) 527,746 (100%) Weighted Population 56,067,997 (45.2%) 58,929,736 (47.5%) 8,941,813 (7.2%) 123,939,546 (100%) State Name N (Sample) N (Population) Short Commuters Medium Commuters Long Commuters All Commuters Alabama 6,657 1,811,511 1.6 1.4 1.0 1.5 Alaska 4,335 273,359 0.3 0.1 0.2 0.2 Arizona 7,576 2,130, 871 1.7 1.8 1.3 1.7 Arkansas 4,058 1,116, 363 1.2 0.7 0.5 0.9 California 47,338 14, 293,902 10.4 12.1 14.9 11.5 Colorado 7,871 2,059, 058 1.7 1.7 1.2 1.7 Connecticut 6,202 1, 569,885 1.3 1.3 1.2 1.3 Delaware 4,612 361, 564 0.3 0.3 0.3 0.3 District of Columbia 3,458 258,527 0.1 0.3 0.3 0.2 Florida 23,975 6,778,674 5.1 6.0 4.6 5.5 Georgia 13,033 3,661, 077 2.7 3.2 3.4 3.0 Hawaii 5,049 526,881 0.4 0.5 0.4 0.4 Idaho 4,028 554,924 0.6 0.3 0.3 0.4 Illinois 20,807 5,597, 098 4.0 4.7 6.5 4.5 Indiana 11,007 2,794,209 2.5 2.2 1.3 2.3 Iowa 9,591 1,348,296 1.5 0.8 0.5 1.1 Kansas 7,708 1,272, 298 1.4 0.7 0.5 1.0 Kentucky 10,094 1,670, 616 1.5 1.3 1.0 1.3 Louisiana 8,856 1,778, 972 1.6 1.3 1.3 1.4 Maine 3,912 593,782 0.6 0.4 0.4 0.5 Maryland 10,871 2,472, 095 1.4 2.4 3.2 2.0 Massachusetts 11,858 3, 023,476 2.2 2.6 3.0 2.4 Michigan 17,670 4,377,200 3.6 3.6 2.5 3.5 Minnesota 10,514 2,466, 115 2.2 2.0 1.1 2.0 Mississippi 8,781 1,144, 638 1.1 0.8 0.6 0.9 Missouri 9,995 2,481, 971 2.1 2.0 1.5 2.0 Montana 4,151 383,352 0.5 0.2 0.1 0.3 Nebraska 7,414 810,569 1.0 0.4 0.2 0.7

PAGE 62

53 Table 4.33 (Continued) Sample Size 249,880 (47.4%) 242,264 (45.9%) 35,602 (6.7%) 527,746 (100%) Weighted Population 56,067,997 (45.2%) 58,929,736 (47.5%) 8,941,813 (7.2%) 123,939,546 (100%) State Name N (Sample) N (Population) Short Commuters Medium Commuters Long Commuters All Commuters Nevada 4,265 945,361 0.7 0.9 0.3 0.8 New Hampshire 5,138 627,842 0.5 0.5 0.5 0.5 New Jersey 14,341 3, 920,068 2.7 3.2 5.8 3.2 New Mexico 3,587 750, 450 0.8 0.5 0.3 0.6 New York 27,879 8,003, 123 5.1 6.6 13.6 6.5 North Carolina 13,210 3,534,703 3.0 2.9 2.0 2.9 North Dakota 4,720 302,715 0.4 0.1 0.1 0.2 Ohio 20,478 5,079,864 4.3 4.1 2.3 4.1 Oklahoma 5,281 1,460, 340 1.5 1.0 0.6 1.2 Oregon 5,831 1,496,406 1.4 1.1 0.8 1.2 Pennsylvania 21,539 5, 342,327 4.5 4.2 3.9 4.3 Rhode Island 4,772 470, 137 0.4 0.4 0.3 0.4 South Carolina 6,446 1, 767,891 1.5 1.4 1.0 1.4 South Dakota 7,114 349,254 0.4 0.2 0.1 0.3 Tennessee 9,570 2,508, 523 2.1 2.1 1.4 2.0 Texas 27,958 9,070,833 7.2 7.6 6.2 7.3 Utah 4,990 999,329 1.0 0.7 0.5 0.8 Vermont 4,239 292,138 0.3 0.2 0.1 0.2 Virginia 13,159 3,304, 550 2.4 2.9 2.7 2.7 Washington 10,379 2,631, 623 2.0 2.2 2.3 2.1 West Virginia 6,300 685,156 0.6 0.5 0.8 0.6 Wisconsin 10,948 2,557, 671 2.4 1.9 1.0 2.1 Wyoming 4,181 227,959 0.3 0.1 0.2 0.2

PAGE 63

54 Table 4.34 Percentage of Commuter Type within each State (ACS) Sample Size 249,880 (47.4%) 242,264 (45.9%) 35,602 (6.7%) 527,746 (100%) Weighted Population 56,067,997 (45.2%) 58,929,736 (47.5%) 8,941,813 (7.2%) 123,939,546 (100%) State Name N (Sample) N (Population) Short Commuters Medium Commuters Long Commuters All Commuters Alabama 6,657 1,811,511 48.6 46.3 5.1 1.5 Alaska 4,335 273,359 63.9 31.1 4.9 0.2 Arizona 7,576 2,130, 871 45.3 49.4 5.2 1.7 Arkansas 4,058 1,116, 363 57.8 37.9 4.2 0.9 California 47,338 14, 293,902 40.8 49.9 9.3 11.5 Colorado 7,871 2,059, 058 45.9 49.1 5.0 1.7 Connecticut 6,202 1, 569,885 45.2 48.2 6.6 1.3 Delaware 4,612 361, 564 47.6 46.2 6.2 0.3 District of Columbia 3,458 258,527 28.5 62.7 8.9 0.2 Florida 23,975 6,778,674 42.2 51.7 6.0 5.5 Georgia 13,033 3,661, 077 40.7 51.0 8.2 3.0 Hawaii 5,049 526,881 41.7 50.6 7.6 0.4 Idaho 4,028 554,924 59.6 36.0 4.3 0.4 Illinois 20,807 5,597, 098 39.8 49.9 10.3 4.5 Indiana 11,007 2,794,209 50.0 45.8 4.1 2.3 Iowa 9,591 1,348,296 63.9 32.8 3.3 1.1 Kansas 7,708 1,272, 298 62.2 34.6 3.2 1.0 Kentucky 10,094 1,670, 616 50.2 44.5 5.3 1.3 Louisiana 8,856 1,778, 972 50.7 42.8 6.5 1.4 Maine 3,912 593,782 52.4 41.9 5.7 0.5 Maryland 10,871 2,472, 095 31.9 56.4 11.7 2.0 Massachusetts 11,858 3, 023,476 41.2 50.1 8.7 2.4 Michigan 17,670 4,377,200 46.6 48.3 5.1 3.5 Minnesota 10,514 2,466, 115 49.0 46.9 4.1 2.0 Mississippi 8,781 1,144, 638 51.9 43.0 5.0 0.9 Missouri 9,995 2,481, 971 46.4 48.1 5.6 2.0 Montana 4,151 383,352 71.8 25.7 2.5 0.3 Nebraska 7,414 810,569 66.1 31.3 2.6 0.7

PAGE 64

55 Table 4.34 (Continued) Sample Size 249,880 (47.4%) 242,264 (45.9%) 35,602 (6.7%) 527,746 (100%) Weighted Population 56,067,997 (45.2%) 58,929,736 (47.5%) 8,941,813 (7.2%) 123,939,546 (100%) State Name N (Sample) N (Population) Short Commuters Medium Commuters Long Commuters All Commuters Nevada 4,265 945,361 43.3 53.6 3.2 0.8 New Hampshire 5,138 627,842 46.9 45.3 7.8 0.5 New Jersey 14,341 3, 920,068 38.6 48.1 13.3 3.2 New Mexico 3,587 750, 450 57.0 38.9 4.0 0.6 New York 27,879 8,003, 123 35.9 48.8 15.2 6.5 North Carolina 13,210 3, 534,703 47.0 48.1 5.0 2.9 North Dakota 4,720 302,715 75.5 21.7 2.8 0.2 Ohio 20,478 5,079,864 47.9 48.1 4.0 4.1 Oklahoma 5,281 1,460, 340 57.9 38.4 3.7 1.2 Oregon 5,831 1,496,406 52.5 43.0 4.6 1.2 Pennsylvania 21,539 5, 342,327 47.3 46.2 6.5 4.3 Rhode Island 4,772 470, 137 50.2 44.5 5.3 0.4 South Carolina 6,446 1, 767,891 48.8 46.3 4.9 1.4 South Dakota 7,114 349, 254 71.9 25.9 2.2 0.3 Tennessee 9,570 2,508, 523 46.2 48.6 5.2 2.0 Texas 27,958 9,070,833 44.3 49.6 6.1 7.3 Utah 4,990 999,329 55.3 40.3 4.3 0.8 Vermont 4,239 292,138 53.6 42.1 4.3 0.2 Virginia 13,159 3,304, 550 41.1 51.5 7.4 2.7 Washington 10,379 2,631, 623 43.6 48.6 7.8 2.1 West Virginia 6,300 685,156 46.6 43.4 10.0 0.6 Wisconsin 10,948 2,557, 671 53.7 42.9 3.4 2.1 Wyoming 4,181 227,959 73.5 19.8 6.7 0.2

PAGE 65

56 Table 4.35 Average Commute Time by State by Commuter Type (ACS) Sample Size 249,880 (47.4%) 242,264 (45.9%) 35,602 (6.7%) 527,746 (100%) Weighted Population 56,067,997 (45.2%) 58,929,736 (47.5%) 8,941,813 (7.2%) 123,939,546 (100%) State Name N (Sample) N (Population) Short Commuters Medium Commuters Long Commuters All Commuters Alabama 6,657 1,811,511 10.0 29.1 87.1 22.8 Alaska 4,335 273,359 9.0 26.6 94.2 18.7 Arizona 7,576 2,130, 871 10.1 29.1 80.7 23.2 Arkansas 4,058 1,116, 363 9.7 28.0 85.5 19.8 California 47,338 14, 293,902 10.4 30.0 77.5 26.4 Colorado 7,871 2,059, 058 10.1 29.3 80.9 23.1 Connecticut 6,202 1, 569,885 10.3 28.7 83.3 24.0 Delaware 4,612 361, 564 10.5 28.2 83.2 23.1 District of Columbia 3,458 258,527 11.4 30.1 72.1 28.5 Florida 23,975 6,778,674 10.3 29.5 79.5 24.5 Georgia 13,033 3,661, 077 10.1 30.9 77.9 26.3 Hawaii 5,049 526,881 10.2 30.6 72.1 25.2 Idaho 4,028 554,924 9.4 28.0 87.0 19.5 Illinois 20,807 5,597, 098 9.7 31.4 74.8 27.2 Indiana 11,007 2,794,209 9.8 28.4 83.2 21.3 Iowa 9,591 1,348,296 9.0 28.0 83.9 17.7 Kansas 7,708 1,272, 298 8.8 28.1 80.6 17.8 Kentucky 10,094 1,670, 616 10.0 28.9 86.4 22.5 Louisiana 8,856 1,778, 972 9.9 28.6 85.9 22.9 Maine 3,912 593,782 9.2 29.4 91.7 22.4 Maryland 10,871 2,472, 095 10.6 31.1 75.2 29.7 Massachusetts 11,858 3, 023,476 9.9 30.8 74.6 26.0 Michigan 17,670 4,377,200 10.0 29.6 81.7 23.1 Minnesota 10,514 2,466, 115 9.4 29.3 79.3 21.6 Mississippi 8,781 1,144, 638 9.7 28.7 90.3 22.0 Missouri 9,995 2,481, 971 9.7 29.7 79.8 23.2 Montana 4,151 383,352 8.7 28.1 88.2 15.7 Nebraska 7,414 810, 569 8.9 26.3 93.4 16.6

PAGE 66

57 Table 4.35 (Continued) Sample Size 249,880 (47.4%) 242,264 (45.9%) 35,602 (6.7%) 527,746 (100%) Weighted Population 56,067,997 (45.2%) 58,929,736 (47.5%) 8,941,813 (7.2%) 123,939,546 (100%) State Name N (Sample) N (Population) Short Commuters Medium Commuters Long Commuters All Commuters Nevada 4,265 945,361 10.8 26.8 82.1 21.6 New Hampshire 5,138 627,842 9.6 29.8 76.9 24.0 New Jersey 14,341 3, 920,068 9.9 30.8 79.0 29.1 New Mexico 3,587 750, 450 9.4 28.1 81.8 19.6 New York 27,879 8,003, 123 10.0 31.7 75.7 30.6 North Carolina 13,210 3, 534,703 10.2 29.2 82.8 22.9 North Dakota 4,720 302,715 8.6 27.9 93.7 15.2 Ohio 20,478 5,079,864 9.9 28.6 82.3 21.8 Oklahoma 5,281 1,460, 340 9.7 27.8 86.4 19.5 Oregon 5,831 1,496,406 9.7 28.7 82.6 21.2 Pennsylvania 21,539 5, 342,327 9.7 30.1 82.2 23.8 Rhode Island 4,772 470, 137 10.0 28.6 78.7 21.9 South Carolina 6,446 1, 767,891 10.3 28.7 86.0 22.5 South Dakota 7,114 349, 254 8.6 26.4 94.9 15.1 Tennessee 9,570 2,508, 523 10.3 29.2 82.0 23.2 Texas 27,958 9,070,833 10.1 29.6 77.9 23.9 Utah 4,990 999,329 9.8 28.0 81.9 20.3 Vermont 4,239 292,138 9.1 29.1 85.4 20.8 Virginia 13,159 3,304, 550 10.4 29.8 80.5 25.6 Washington 10,379 2,631, 623 10.0 29.5 81.8 25.1 West Virginia 6,300 685,156 9.4 30.0 88.5 26.2 Wisconsin 10,948 2,557, 671 9.5 28.2 93.4 20.4 Wyoming 4,181 227,959 8.4 28.6 92.1 18.0 Average Time 9.9 29.7 79.3 24.3

PAGE 67

Table 4.36 Average Commute Time by CMSA (NHTS) Sample Size 3,401 (39.0 %) 4,484 (51.6) 813 (9.4%) 8,698 (100%) Weighted Population 17,706,776 (39.0%) 23,751,852 (38.6%) 4,374,181 (9.4%) 45,832,809 (100%) CMSA Name N Sample N Population Short Commuters Medium Commuters Long Commuters All Commuters Boston--Worcester--Lawrence, MA--NH--ME--CT CMSA 607 3,091,566 10.55 32.30 69.68 26.28 Chicago--Gary--Kenosha, IL--IN--WI CMSA 726 3,713,079 9.15 32.50 72.69 28.76 Cincinnati--Hamilton, OH--KY--IN CMSA 180 879,371 10.41 26.54 68.43 19.98 Cleveland--Akron, OH CMSA 319 1,439,342 9.51 29.30 60.00 22.47 Dallas--Fort Worth, TX CMSA 324 1,654,824 10.42 31.81 73.60 27.43 Denver--Boulder--Greeley, CO CMSA 287 1,256,429 10.20 28.34 81.71 22.48 Detroit--Ann Arbor--Flint, MI CM SA 433 1,910,668 10.13 31.11 68.84 23.43 Houston--Galveston--Brazoria, TX CMSA 300 1,737,975 10.44 31.73 71.70 27.06 Los Angeles--Riverside--Orange County, CA CMSA 1,037 6,297, 337 10.68 30.17 73.14 27.88 Miami--Fort Lauderdale, FL CM SA 225 1,664,172 10.49 32.95 75.35 30.07 Milwaukee--Waukesha, WI PM SA 184 713,697 10.21 27.39 90.00 19.08 New York--Nr.New Jersey--Long Island, NY--NJ --CT--PA CMSA 1,477 8, 161,875 9.85 32.66 78.50 31.72 Philadelphia--Wilmington--Atlantic City, PA--N J--DE--MD CMSA 500 2, 499,760 10.01 30.51 71.69 24.60 Portland--Salem, OR--WA CMSA 223 853,460 9.89 28.61 74.37 23.81 Sacramento--Yolo, CA CMSA 188 896,578 9.98 28.08 66.64 21.12 San Francisco--Oakland--San Jose, CA CMSA 612 3,370,225 9.87 31.87 79.04 25.90 Seattle--Tacoma--Breme rton, WA CMSA 382 1, 720,666 10.35 29.35 70.50 26.21 Washington--Baltimore, DC--MD--VA--W V CMSA 694 3,971,786 10.48 32.76 73.74 28.87 Average Commute Time 10.14 31.27 74.77 27.26

PAGE 68

Table 4.37 Average Commute Distance by CMSA (NHTS) Sample Size 3,401 (39.0 %) 4,484 (51.6) 813 (9.4%) 8,698 (100%) Weighted Population 17,706,776 (39.0%) 23,751,852 (38.6%) 4,374,181 (9.4%) 45,832,809 (100%) CMSA Name N Sample N Population Short Commuters Medium Commuters Long Commuters All Commuters Boston--Worcester--Lawrence, MA--NH--ME--CT CMSA 607 3,091,566 4.47 16.15 39.65 13.24 Chicago--Gary--Kenosha, IL--IN--WI CMSA 726 3,713,079 3.86 15.17 33.64 13.19 Cincinnati--Hamilton, OH--KY--IN CMSA 180 879,371 4.90 14.97 47.37 11.05 Cleveland--Akron, OH CMSA 319 1,439,342 4.92 17.66 26.54 12.85 Dallas--Fort Worth, TX CMSA 324 1,654,824 5.18 21.08 32.88 16.32 Denver--Boulder--Greeley, CO CMSA 287 1,256,429 4.11 15.47 37.01 10.88 Detroit--Ann Arbor--Flint, MI CM SA 433 1,910,668 4.52 18.71 41.07 13.26 Houston--Galveston--Brazoria, TX CMSA 300 1,737,975 5.24 20.58 32.36 15.92 Los Angeles--Riverside--Orange County, CA CMSA 1,037 6,297, 337 4.81 16.71 42.43 15.09 Miami--Fort Lauderdale, FL CM SA 225 1,664,172 4.52 17.22 41.98 15.22 Milwaukee--Waukesha, WI PM SA 184 713,697 4.85 15.89 70.00 10.66 New York--Nr.New Jersey--Long Island, NY--NJ --CT--PA CMSA 1,477 8, 161,875 3.47 14.58 35.40 13.55 Philadelphia--Wilmington--Atlantic City, PA--N J--DE--MD CMSA 500 2, 499,760 16.49 13.62 39.04 16.31 Portland--Salem, OR--WA CMSA 223 853,460 4.23 15.49 42.62 12.60 Sacramento--Yolo, CA CMSA 188 896,578 4.51 18.84 28.00 12.71 San Francisco--Oakland--San Jose, CA CMSA 612 3,370,225 4.18 16.41 44.59 12.99 Seattle--Tacoma--Breme rton, WA CMSA 382 1, 720,666 4.49 15.87 34.31 13.51 Washington--Baltimore, DC--MD--VA--W V CMSA 694 3,971,786 4.55 17.01 34.79 14.13 Average Commute Distance 5.05 16.42 37.57 13.88

PAGE 69

60 4.6 Summary of Person, Household and Area Characteristics The descriptive analysis of the data provided us with an idea of characteristics of the short and long commuters. The following Ta ble 4.38 shows some important person, household and area characteristics of an individu al that are significant in short and long commuters in NHTS 2001 and ACS 2000 Table 4.38 Summary of the Person and Household Characteristics NHTS 2001 ACS 2000 Characteristic Short Commuters Long Commuters Short Commuters Long Commuters Gender Male Male Age 16 to 20 years 25 to 64 years 16 to 20 years 25 to 64 years Race White Black or AfroAmerican White Black or AfroAmerican Education Bachelors, Graduate and Professional Bachelors, Graduate and Professional Income less than $15,000 $100,000 and above less than $19,999 greater than $ 25,000 Driver Status not a driver NA NA Household Size single person household single person household 5 or more household Family Structure One adult, no children Household Income $25,000-$49,999 $100,000 and above $25,000-$49,999 $100,000 and above Number of Workers single worker 3 or more workers single worker Number of Vehicles zero vehicle zero vehicle Number of Children no children 3 or more children no children 1 or more children MSA Population less than 250,000, not in MSA/CMSA 3 million or more NA NA Urban Area Type urban cluster Inside urban area, not in urban area NA NA

PAGE 70

61 4.7 Afro-American, Poor and Bus Users Analysis was done to know if the percenta ge of Afro-Americans (11.7), poor people (18.1) with personal annual income less th an $15,000 and bus tran sit users (11.00) of long commuters are same. The Figure 4.6 and Figure 4.7 reveal that Afro-American, Poor and Bus users do not belong to the same group in long commuters. Figure 4.6 Proportions of Combination of Afro-American, Poor and Bus user groups Afro-American Poor Bus Users Total Population (N) = 2,734,364 32.4% 8.3% 26.3% 5.3% 4.9% 7.3% 15.5%

PAGE 71

62 Figure 4.7 Percentage of Combination of Afro -American, Poor and Bus user groups in Long Commuters Afro-American Poor Bus Users Total Population (Long Commuters) (N) =8,941,813 9.9% 2.5% 8.0% 1.6% 1.5% 2.2% 4.7%

PAGE 72

63 4.8 Range of Short and Long Commuters The Figure 4.8 shows the dist ribution of commuters by ti me. The Figure 4.9 and Figure 4.10 shows how the percentage of short a nd long commuters changes with upper and lower limits of short and long commuters. Figure 4.8 Distribution of Commuters by Time 11.249 3.143 14.579 1.364 12.976 8.569 9.973 3.481 0.067 5.091 0.070 1.454 0.576 3.081 0.0170.005 2.579 17.417 3.851 0.314 0.142 0.000 2.000 4.000 6.000 8.000 10.000 12.000 14.000 16.000 18.000 20.0001-9 min 10 min 11-14 min 15 min 16-19 min 20 min 21-29 min 30 min 31-39 min 40 min 41-44 min 45 min 46-49 min 50 min 51-59 min 60 min 61-64 min 65 min 66-69 min 70 min >=71minTime (min)Percent of Commters

PAGE 73

64 Figure 4.9 Share of Short Commuters by Upper 28.70% 46.40% 60.70% 0.00% 10.00% 20.00% 30.00% 40.00% 50.00% 60.00% 70.00% 1-10 min1-15 min1-20min Range (min)Percent in all commuters Figure 4.10 Share of Long Commuters by Lower Limit 8.16% 6.13% 3.04% 2.89% 0.00% 1.00% 2.00% 3.00% 4.00% 5.00% 6.00% 7.00% 8.00% 9.00% >=50 min>=60 min>=65 min>=70 min Range (min)Percent in all commuters

PAGE 74

65 CHAPTER 5 METHODOLOGY The descriptive analysis explained in the prev ious chapter highlighted the distributions of the individual, household and area related ch aracteristics of short, medium and long commuters and it helped in comparing th e characteristics between short and long commuters. But the descriptive analysis alone is not very effective in forming policies, as it does not reveal the sensitivity of the commute length to characteristics. More rigorous data analysis is necessary to explain the sensitivity of the commute length and in knowing the characteristics involved in the process of commuter type choice. Modeling the commuter behavior in this context would help in identifying the magnitude and type of effects of the aforementioned characteristi cs on the commute length and commuter type choice. The models developed in this study are cent ered on the commuter type choice and commute length. A Multinomial Logit Model (MNL) and a Structural Equations Model (SEM) was developed. MNL model was constructed to study the influence of the different characteristics on the choice of commuter ty pe. The SEM was constructed to measure the commute characteristics like commute time, commute distance and departure time. This chapter explains the theory related to the models developed in this study and the teststatistics used in evaluati ng the models. The following sect ions explain the theory and test-statistics related to Multinomial Logit Models and Structural Equations Models. 5.1 Theory of Multinomial Logit Models Modeling individual decision-ma king behavior is fundament al to predicting aggregate (population) behavior. Classical economic c onsumer choice theories offer convenient paradigms for modeling such individual choice behavior. These choice theories also consider the psychological processes underl ying decision-making behavior. A choice may be viewed as the result of a sequentia l decision-making process that includes the following steps: Definition of the choice Generation of the alternatives Evaluation of the attributes Choice Implementation

PAGE 75

66 A decision maker would collect information on the available alternatives, and then apply a decision rule to choose an alternative fo r the desired activity. So, any theory is a collection of procedures that define/describes the decision maker, choice set, attributes of the alternatives and decision rule. The multinomial logit models are based on the probabilistic choice theory with random utility functions. A basic assumption in discrete choice analysis is that each alternative in th e choice set of a decision maker is associated with a utility and that the decision maker choos es the alternative with the highest utility. Thus, the probability of choice i is equal to the probability that the utility of alternative i is greater than or equal to the utilities of all other alte rnatives in the choice set. P(i|C n ) = Pr [U in U jn all j C n ] Where, i = the choice alternative, j = other alternatives in the choice set not equal to i, Uin = Utility of an individual ‘n ’ choosing the alternative i, Ujn = Utility of an individual ‘n ’ choosing the alternative j, Cn = choice set, consists of all the alternatives feasible to the individual. It is a subset of Universal set of alternativ es represented as C, (Cn C) Utilities are not known to analyst with cert ainty and therefore treated as random variables. Choice probabilities are derived by assuming a joint probability distribution for the set of random utilities {U in i C n }. The utility is assumed to consist one part observable and one part not obs ervable by the analyst. The obs ervable part is called the systematic part of the utility function a nd the unobservable part as the random or stochastic part of the utility function. The utility function is represented as: Ui = Vi + i Where, Ui = total utility of the alternative i., Vi = observable part, and i = unobservable part

PAGE 76

67 Thus the probability of choice i is equal to, The unobservable part is assumed to be stoc hastic. This means that the alternative a decision-maker would actually choose cannot be predicted with certainty, but an assumption on the distribution of the random or stochastic part will allow one to predict the probability that it could be chosen. Thus for a population of decision-makers, the share of the population choosing each alternative may be pred icted. Since there are more than two choices or alternatives, deriva tion of multinomial choice models get more complicated. The most convenient way to express P n (i) is to reduce the multinomial choice problem to a binary one. To do this, we note that: U in U jn j C n j i is equivalent to In effect, we create a composite altern ative out of all the elements of C n other than i, and we use the utility of best alternative to represent the entire composite. If U in exceeds the utility of the composite alternative, then i is chosen; otherwise it is not. Thus Since U in is a random variable, max Ujn is also a random variable. The stochastic or random part of the utility func tion is assumed to be indepe ndent and identically Gumbel distributed. The properties of the Gumbel distribution the yields the following form for the probability of choice i. n jn inJ 1 j U U ne e (i) P Where, Pn(i) = Probability of individual n choosi ng alternative i, Ujn = Utility derived by individua l n from alternative j, Jn = Number of available alternative choices, ) i j C j U U Pr( ) i ( Pn jn in n ) i j C j V V Pr(n jn jn in in ) i j C j V V Pr(n in jn in jn jn C j inU max Ui j n )] V ( max V Pr[ ) i ( Pjn jn C j in in ni j n

PAGE 77

68 The utility derived by the individual n from a lternative j, may be modeled as a linear function of explanatory variables as follows: U jn = 0j + 1jX1nj + 2jX2nj +………….+ kjXknj + jn Where, 0j = alternative specific c onstant for alternative j 1j, 1j, 2j….kj = coefficients associated with explanatory variables X1nj, X2nj…..Xknj = explanatory variables for individual n jn = disturbance term k = number of explanatory vari ables included in the model The values reflect the sensitivity of the vari ables included in the model. The log of the denominator of the multinomial logit model equa tion also has useful property in that it can be interpreted as the expected maximum u tility of the alternatives in the choice set. 5.2 Test Statistics for Mu ltinomial Logit Models Multinomial logit models may be subjected to a series of statisti cal tests, which are briefly described here. The first test is the l og-likelihood ratio test ( LLR), which is similar in purpose to the F-test used with the linear regression models, and is used to test the overall significance of the model. The LLR test is used to test the null hypothesis that the coefficients of the demographic variables in the model are collectively zero. Under the null hypothesis that all the coefficients are zero, that is, 1 = 2 = … = k = 0, the statistic ) ( ) 0 ( 2 L L is 2 distributed with k degrees of freedom. More informative is to test the null hypothesis that all coeffi cients except for the alternative specific constants are zero. Test statistic is ) ( ) ( 2 L c L with K-J df, where J is the number of alternatives in the choice set and L(c) is the log likelihood of a model with only constants. L(c) can be obtained by estimating a m odel with J-1 alternative specific constants or Where, N i is the number of observations selecting alternative i and N is the total sample size. The 2 is an informal goodness of fit statistic th at measures the fracti on of an initial loglikelihood value explained by the model. J 1 i i iN N ln N ) c ( L

PAGE 78

69 It is defined as This statistic is similar to R2 in Linear regression models. Th ere are no general guidelines for when a 2 value is sufficiently high. For th e same estimation data set, the 2 of a model will always increase or at least stay the same whenever new variables are added (similar to R 2 in regression). For this reason, we also use the adjusted 2, It should be noted that if 2increases but 2decreases, then it means that the added variables do not provide suffici ent explanatory power to the model to compensate for the degrees of freedom utilized by the “larger” model specification. Another measure of goodness of fit is “percent predicted correctly” defined as Where, nyˆ is 1 if the highest predicted probability co rresponds to the chosen alternative and 0 otherwise. This should be used with considerable caution. The test statistics discussed above may be applied for overall model. The conventional tstatistic is used to test the significance of the coefficient of any given variable (as in linear regression model). The t-statistic is used to test whether i is equal to a certain value, say c (c = 0 in this case). That is we are interested in testing whether the population value of the coefficient, i, equals c H 0 : i = c Then, the statistic is: has a t-distribution with degrees of freedom [n-(p+1)] where n is the sample size and p is the number of explanatory variables included in the model. If the t-statistic has an extreme value, which will occur with only a small probability (say, 0.05), then we reject the null hypothesis and conclude that the popula tion value of i does not equal c. If this is ) 0 ( ) ˆ ( 12L L ) 0 ( ) ˆ ( 12L K L n ny Nˆ ) / 100 ( ii c s c i t ˆ

PAGE 79

70 true, it implies that variable Xi does not influence the utility function. In this case, the test statistic becomes: 5.3 Theory of Structural Equations Models A typical structural equations model (with ‘G’ number of endogenous variables) is defined by a matrix equation system as shown in Equation 5.1. Y Y YX BGG 11. . . (5.1) This equation can be rewritten as YBYX (5.2) (or) YIBX ()()1 (5.3) where Y is a column vector of endogenous variables, B is a matrix of parameters associated with right-hand-side endogenous variables, X is a column vector of exogenous variables, is a matrix of parameters associ ated with exogenous variables, and is a column vector of error terms a ssociated with the endogenous variables. Structural equations systems are estimated by covariances-based structural analysis, also called method of moments. In this approach of estimation, instead of minimizing sum of squared differences of observed and predic ted individual values, the difference between the sample covariances and the covariances predicted by the model is minimized. The observed covariances minus the predicted covariances form the residuals. The fundamental hypothesis for the covariances-bas ed estimation proced ures is that the covariance matrix of the observe d variables is a function of a set of parameters as shown in Equation 4: = () (5.4) where is the population covariance matrix of observed variables, is a vector that contains the model parameters, and () is the covariance matrix written as a function of ii c s i t ˆ

PAGE 80

71 Equation 5.4 implies that each element of th e covariance matrix is a function of one or more model parameters. The relation of to () is basic to an understanding of identification, estimation, and assessmen ts of model fit. The matrix () has three components: (a) the covariance matrix of Y (b) the covariance matrix of X with Y and (c) the covariance matrix of X Considering first YY () the implied covariance matrix of Y can be derived as: YY () = E (YY) = EIBXIBX [()()(()())] 11 = ()(()()()())() IBEXXEXEXEIB 1 1 = 1 1) )( ( ) ( B I B I (5.5) where = covariance matrix of X and = covariance matrix of The implied covariance matrix of X, XX () is equal to or XX () = E (XX) = (5.6) The final part of the implied covariance matrix is XY (), the implied covariance of X with Y : XY () = E (XY) = EXIBX [(()())] 1 = () IB1 (5.7) Now, assembling Equations 5.5 through 5.7, the implied covariance matrix of Y and X is () ()()()() () IBIBIB IB111 1 (5.8) Before estimating model parameters, it is firs t necessary to ensure that the model is identified. Model identification in simultaneous structural equations systems is concerned with the ability to obtain unique es timates of the structural parameters. The identification problem can be re solved if travel behavior th eory can be used to place restrictions on the set of simultaneous structur al equations. These restrictions may take a variety of forms such as the use of extran eous estimates of parameters, knowledge of exact relationships among parameters, knowl edge of the relative variances of

PAGE 81

72 disturbances, and knowledge of zero correla tion between disturbances in different equations. The restrictions usually employed ar e zero restrictions that take the form of specifying certain structural parameters to zero, i.e., cer tain endogenous variables and certain exogenous variables do not appear in certain equations. It has been shown that in the case of zero restrictions on structural parameters, each equation can be checked for identification by using either the rank conditi on or the order condition. If an equation is identified, it may be either 'exactly-identif ied' or 'over-identified'. An equation is 'exactly-identified' if the number of identif ying restrictions placed on the model is the minimum needed to identify the equation, and an equation is over-ide ntified if there are some additional restrictions beyond the minimu m necessary to identify the equation. In order to check for identification of a structural model, the commonly used identification rules are t-Rule, Null B Rule, and Recursive Rule. The t-Rule, the Null B Rule, and Recursive Rule are conditions for th e identification of the model as a whole. The first is only a necessary condition, but th e second and third are sufficient conditions. The t-Rule is the most general rule and applies to all of the models, whereas the Null B Rule is appropriate only when B=0, regardless of the form of The recursive rule is appropriate for models with tr iangular B matrices and diagonal matrices. Finally, the rank and order conditions establish the identifi cation status of equati ons. If each equation meets the rank rule, then the model as a whole is identified. Both, rank and order conditions allow any nonsingular (I-B) matr ix and assume no restrictions for the matrix. A detailed discussion on checks for id entification for structural equations models may be found in Bollen (1989) A summary is shown in the table below. Table 5.1 Identification Rules for Structur al Equations with Observed Variables Assuming No Measurement Error (y = By + x + ) Identification Rule Evaluates Requirements Necessary Condition Sufficient Condition t-Rule Model t () (p+q)(p+q+1) Yes No Null B Rule Model B=0 No Yes Recursive Rule Model B triangular diagonal No Yes Order Condition Equation Restrictions (p – 1) free Yes No Rank Condition Equation Rank (Ci) = p – 1 free Yes Yes p = number of endogenous variables; q = number of exogenous variables t = number of unknown parameters in For definition of Ci, see notes on identification in simultaneous equation systems (under rank condition) The unknown parameters in B, , and are estimated so that the implied covariance matrix, is as close as possible to the sample co variance matrix S. In order to achieve

PAGE 82

73 this, a fitting function F(S, ()) which is to be minimized is defined. The fitting function will have the following properties: F(S, ()) is a scalar; F(S, ()) 0; F(S, ()) = 0 if and only if () = S, and F(S, ()) is continuous in S and (). Available methods for parameter estim ation include maximum likelihood (ML), unweighted least squares (ULS), generalized least squares (G LS), scale free least squares (SLS), and asymptotically distribution-free ( ADF). Each of these methods minimizes the fitting function and leads to consistent estimators of Among these methods, the two most widely used estimation techniques are maximum likelihood (ML) and asymptotically distribution-free (ADF). The ML method of estimation is most appropr iate when all of the endogenous variables included in the model system are continuous variables. The fitting function that is minimized in the ML method of estimation of structural parameters is shown in Equation 5.9 FML = log | () | + tr (S -1 ()) log | S | (G + K) (5.9) where G = Number of excluded endogenous variables on RHS of the model, and K = Number of included exogenous variables on RHS of the model. The asymptotic covariance matrix for the ML estimator is given by, 1 ML 2F E 1N 2 (5.10) When is substituted for we have an estimated asymptotic covariance matrix that allows tests of statistical significance on parameters of The ML estimator provides a test of overall model fit for overiden tified models. The asymptotic distribution of (N-1) FML is a 2 distribution with ()(p+q)(p+q+1) t degrees of freedom, where t is the number of free parameters and FML is the value of the fitting function evaluated at the final estimates. The null hypothesis of the chi-square test is H0: = (). This implies that the overidentifying restrictions for the model are correct. Rejection of H0 suggests that at least one rest riction is in error so that (). In general, the suitability of the chi-squa re test depends on having a sufficiently large sample, on the multinormality of the observed variables, and on the validity of = ().

PAGE 83

745.4 Test Statistics for Stru ctural Equations Models The chi-square test of overall model fit is called the discrepancy in the model. The null hypothesis under test is that th e model fits the data, so one hopes to find a small, nonsignificant chi-square value for this test. This chi-square value is verified along with the degrees of freedom. ‘Degrees of freedom’ is the difference between the number of distinct sample moments and number of di stinct parameters to be estimated. The probability value tells us that the chi-square value obtained would be that large or larger with that chance if the null hypothesis that the model fits the data were true. If the probability value of the chi-squa re test is smaller than 0.05 level used by convention, one has to reject the null hypothesis that the model fits the data. Some descriptive statistics are RMR, GFI a nd AGFI, PGFI and RMSEA. The RMR is the root mean square residual. RMR is the squa re root of the mean squared amount by which the sample variances and covariances differ from the corresponding estimated variances and covariances, estimated on the assumption th at your model is correct. The smaller the RMR, the better the fit. GFI is the Goodness of Fit Inde x. GFI varies from 0 to 1, but theoretically can yield meaningless negative values. By convention, GF I should by equal to or greater than .90 to accept the model. By this criteri on the present model is accepted. AGFI is the Adjusted Goodness of Fit Index. AGF I is a variant of GFI, which uses mean squares instead of total sums of squares in the numerator and denominator of 1 GFI. It, too, varies from 0 to 1, but theoretically can yi eld meaningless negative values. AGFI should also be at least .90. By this criterion the present model is accepted. PGFI is the Parsimony Goodness of Fit Index. It is a variant of GFI, which penalizes GFI by multiplying it times the ratio formed by the degrees of freedom in your model and degrees of freedom in the independence model. RMSEA is the root mean square error of approximation, which incorporates the discrepancy function criterion (comparing observed and predicted covariance matrices) and the parsimony criterion. By convention, th ere is good model fit if RMSEA less than or equal to .05. There is adequate fit if RMSEA is less than or equal to .08. Other descriptive fit statistic to assess the overall fit a model to the data is comparative fit index (CFI). CFI compares the ab solute fit of the specified model to the absolute fit of the Independence model. The "independence model" is the model in which variables are assumed to be uncorrelated with the dependent(s ), so if the fit for "your model" is no better than for the "independence model," then the specified model should certainly be rejected. The greater the discrepancy between th e overall fit of the two models, the larger the values of the descriptive statistic. CFI vari es from 0 to 1. CFI close to 1 indicates a very good fit, and values above .90 an acceptabl e fit. There are many other fit measures.

PAGE 84

75 Each researcher has his or her favorite collect ion of fit statistics to report. The important fit measures considered are chi-square for a certain degrees of freedom and probability value, CFI and RMSEA. After the model has been eval uated for its goodness-of-fit, one would be interested in knowing the effect of explanatory vari ables on endogenous variables and more importantly the effects of endogenous variable s on other endogenous variables. There are two types of effects, direct an d indirect effects. A direct ef fect is one where a variable directly affects another variable as depicted by a direct arrow linki ng the two variables in the path diagram. On the other hand, an indir ect effect is one where a variable influences another variable through a mediating variable. The sum of direct and indirect effect is called the total effect. From the figure 5.1, A and B are the exogenous variables and C and D are endogenous variables. The e1 and e2 are the disturbances associated with C and D respectively. 1, 2, 3 and 4 are coefficients associated with the relation A and C, C and D, B and C, and B and D. The direct effect is th e effect of A on C, C on D, B on C, B on D. Indirect effect is the effect of A on D through D. It is important to note that A does not affect D directly but A influences D through C.

PAGE 85

76Figure 5.1 Direct and Indirect Effects 1 3 2 4 The direct effect of A on C is The indirect effect of A on D is given by Similarly, The direct effect of B on D is The indirect effect of B on D is The total effect of B on D is the su m of direct and indirect effects = B D C A e1 e2

PAGE 86

77 CHAPTER 6 MODEL ESTIMATION RESULTS 6.1 Commuter Type Choice Model A Multinomial Logit Model was developed to analyze the influence of individual, household and area related ch aracteristics on the commuter type choice. Multinomial Logit Model was developed using both the NHTS 2001 data set and ACS 2000 data set. In both the models, the utility function of medium commuter t ype has been set to zero as the base alternative. The util ity functions for short and long commuters were first defined by including all possible best combinati on of characteristics available in the corresponding datasets were included. The vari ables were tested for their significance at 95% level of confidence by running the mo dels in LIMDEP 3.0 (An Econometric Modeling Software Tool). All the significant variables were retained and the model was tested for good-of-fit using st andard test-statistics. The Table 6.1 and Table 6.2 show the results of MNL using the NHTS 2001 and ACS 2000 datase ts respectively. These models are estimated to determine potentially the maximum extent to which the demographic variables can explain the commut er type choice. In these models, typical individual characteristics lik e gender, age, race, income, education, driver and the household characteristics like household size, dr iver count, household income, number of children, property value and finally the area re lated characteristics like size of the area and urban area type are found to be signifi cant at 95% level of confidence. All the coefficients in the model have the expected signs. The constants in the MNL model shown in Tabl e 6.1 shows that there is general tendency for a commuter to be a short commuter. The gender variable (dummy =1 for male, 0 or else) shows a negative coefficient for short commuter utility function (SCUF) and positive for long commuter utility function (L CUF) indicating that females generally prefer to be short commuters or males have more tendency to be long commuters. The age variable shows a positive coefficient in SCUF indicating that as age increases the commuters tend to be short commuters. Th e dummy variable for middle age shows a negative coefficient in SCUF indicating that middle-aged commuters have more likely to be long commuters. The dummy variable for White American race group shows coefficient that is positively associated with short commuting and negatively associated with long commuting. The coefficient of ed ucation variable in SCUF is negatively associated and shows that highly educated pe ople have tendency to be long commuters. The coefficient of the dummy variable for driver status (driver = 1,0 or else) is negatively associated with both the SCUF and LCUF and shows that drivers are more likely to be

PAGE 87

78 short commuters than long commuters. The coefficient of the dummy variable representing the managerial or professional type of occupation in LCUF is positively associated with long commuting. The first of household characteristic, the household size variab le shows that its coefficient is negatively associated with SCUF indicating that as the household size increases the chances of being a short commut er decreases. The coefficient of driver count variable is positively asso ciated with SCUF indicating th at as the number of drivers in the household increases their chances of being a short commuter increases. The low household income coefficient is positively a ssociated with SCUF and the high household income coefficient is negatively associated with SCUF. Indicati ng that low household income increases the tendency of an individu al to be a long commuter. Also the high household income coefficient in LCUF is positively associated. The coefficient for number of children in the household is positively associated with SCUF indicating that presence of children restricts the individual to be a short commuter. The area related characteristic dummy for size of population of an area greater or equal to 3 million has a coefficient that is pos itively associated with LCUF. The dummy variable urban cluster and urban center has a co efficient that is negatively associated with LCUF. The MNL model using ACS 2000 data also reve aled the similar kind of results. There are some interesting variables in this model. The coefficient of personal income variable is negatively associated with SCUF. The coe fficient for the dummy variable for low personal income also reveals the same m eaning. The coefficient for dummy for the duration of residence at a pla ce for 10 years or more is posi tively associated with SCUF indirectly indicating that, as the commuting falls short the duration of status is as high as 10 years. The log-likelihood value at convergence for the MNL model using NHTS 2001 dataset is ) ( L= –19938.7. Therefore, the test statistics is ) ( ) ( 2 L c L = 1503.41 with 21 degrees of freedom. The criti cal (0.05 level) value of 2 with 21 degrees of freedom is 32.67. The log-likelihood value at converg ence for the MNL model using ACS 2000 dataset is ) ( L= –21299.9. Therefore, the test statistics is ) ( ) ( 2 L c L = 1208.2 with 21 degrees of freedom. The critical (0.05 level) value of 2 with 21 degrees of freedom is 32.67. Thus, the hypothesis that the coefficients of the individual, household and area related characteristics considered are zero is rejected. This confirms the importance of demographic variables in expl aining the commuter t ype choice behavior. The adjusted likelihood ratio index 2is 0.03586 for NHTS 2001 and 0.027 for ACS 2000. The values are low but this goodness-of-fit measure does not have the real statistical interpretation (R 2 in regression does have a statistical interpretation). The low value could be because of hidden effects of area related characteristics like urban growth, structure, congestion and the transportation facilities. The hidden effects could also be

PAGE 88

79 due to the interrelationships that exist be tween mode-choice and commute type choice. This study merits particular atte ntion. This is include d in the further study in the research. Table 6.1 Commuter Type Choice Model (NHTS) Note: Medium Commuter is the base alternative ) 0 ( L= –26070.0 ) ( L= –19938.7 ) ( c L= -20690.0 2 = 1503.4; 2= 0.036 Short Commuters Long Commuters Variable Variable Type -Coeff t-stat -Coeff t-stat Constant 0.3708 3.214 -1.905 -12.277 Male (= 1 if male, 0 else) -0.1501 -5.335 0.4375 7.084 Age Continuous 0.0019 1.695 Middle Age (=1 if age 25-64, 0 else) -0.4387 -10.526 White (=1 if race is white, 0 el se) 0.1244 3.352 -0.1373 -1.918 Well educated (=1 if more than bachelors, 0 else) -0.1182 -3.781 Driver status (=1 if driver, 0 else) -0.2089 -2.333 -0.774 -5.749 Personal income Continuous Low personal income (=1 if income < 20,000; 0 else) High personal income (=1 if income >= 75,000, 0 else) Managerial/Profession al Occupation (=1 if such occupation, 0 else) -0.2398 -7.998 0.2891 4.374 Household size Continuous -0.1038 -3.261 Household income Continuous Low household income (=1 if income <15,000, 0 else) 0.2077 3.241 High household income (=1 if income >=75,000, 0 else) -0.1491 -3.725 0.1293 1.781 Number of children Continuous 0.1392 4.343 Number of workers Continuous Number of drivers Continuous 0.1399 4.517 High Property value (=1 if value >= 150,000, 0 else) Duration of stay for 10 years or more (=1 if stay >=10 years, 0 else) Area type (=1 if urban area/cluster, 0 else) 0.4672 13.984 -0.3311 -4.607 Area population greater than or equal to 3 million (=1 if size is >= 3 million, 0 else) -0.5198 -16.471 0.9112 14.221

PAGE 89

80 Note: Medium Commuter is the base alternative ) 0 ( L= –26732.5 ) ( L= –21299.9 ) ( c L= –21904.0 2 = -1208.2; 2= 0.027 The 2is low because of unobserved effects due to the missing variables related to network level of service variables, urban structure and job opportunities.

PAGE 90

81 Table 6.2 Commuter Type Choice Model (ACS) Note: Medium Commuter is the base alternative ) 0 ( L= –26732.5 ) ( L= –21299.9 ) ( c L= –21904.0 2 = -1208.2; 2= 0.027 Short Commuters Long Commuters Variable Variable Type -Coeff t-stat -Coeff t-stat Constant -0.274 -2.768 -1.70 -9.808 Male (= 1 if male, 0 else) -0.143 -4.869 0.39 6.776 Age Continuous 0.006 4.539 -0.01 -2.12 Middle Age (=1 if age 25-64, 0 else) -0.259 -6.002 White (=1 if race is white, 0 else) 0.297 7.3 -0.21 -2.963 Well educated (=1 if more than bachelors, 0 else) Driver status (=1 if driver, 0 else) Personal income Continuous 0.000 -4.61 Low personal income (=1 if income <= 20,000, 0 else) 0.466 12.747 -0.18 -2.403 High personal income (=1 if income >= 75,000, 0 else) 0.51 7.505 Managerial/Professional Occupation (=1 if such occupation, 0 else) Household size Continuous -0.130 -6.54 0.11 3.291 Household income Continuous 0.000 4.627 Low household income (=1 if income <15,000, 0 else) High household income (=1 if income >=75,000, 0 else) -0.134 -3.767 Number of children Continuous 0.163 7.167 -0.08 -2.037 Number of workers Continuous 0.117 4.409 -0.18 -4.047 Number of drivers Continuous High Property value (=1 if value >= 150,000, 0 else) -0.240 -8.029 Duration of stay for 10 years or more (=1 if stay >=10 years, 0 else) 0.171 5.992 Area type (=1 if urban area/cluster, 0 else) Area population greater than or equal to 3 million (=1 if size is >= 3 million, 0 else)

PAGE 91

82 6.2 Model of Commute Length The results of the SEM model of commute length are shown in Table 6.3. The discrepancy is 26.075 and the pr obability is 0.053 with 16 de grees of freedom. The Path diagram for the constructed model is show n in Figure 6.1. The model shows that the individual characteristics like gender, middle age, driver status, managerial or professional occupation and household char acteristics like numbe r of children, high household income and area related characte ristics like MSA population of over 3 million and urban area or urban cluster are significant. All the signs of the variables in the model are as expected. The variables gender, middle age, manage rial and professional occupation, and high household income, MSA with population of over 3 million and urban area or urban cluster are dummy variables and number of children is continuous variable. The gender has positive influence on distance this indicate s that males have tendency to travel long distances for jobs than females. The mode l shows that middle-aged people generally travel long distances than other age groups. The people working in managerial and professional occupational ar e also travel long commutes. The model shows that individuals living inside the urban area type or urban cluster have shorter commutes when all other characteristics are equal. The model shows that middle age, MSAs w ith population 3 million or more and urban areas or urban clusters ha ve positive influence on individual’s commute time. The commute distance has positive influence on commute time, which is expected. The model shows that number of children tends to decrease an individual’s commute time.

PAGE 92

83 Table 6.3 Structural Equations Model for Commute Length Chi-square: ( 2) = 26.075 Degrees of Freedom: (df) = 16 Probability Value: (P) = 0.053 Comparative Fit In dex: (CFI) = 1.000 Root Mean Square Error of Approximation: RMSEA = 0.0016 Intercept Effects Male Middle Aged Driver Status Managerial/ Professional Occupation Number of Children Household Income >= 75k Urban Area/Cluster MSA Size 3 million Commute distance Commute Distance 2.924 Total 0.189 0.393 0.000 0. 351 0.000 0.311 -0.749 0.483 0.000 Direct 0.189 0.393 0.000 0. 351 0.000 0.311 -0.749 0.483 0.000 Indirect 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 Commute Time 2.050 Total 0.167 0.503 -0.869 0. 310 -0.049 0.274 -0.398 0.741 0.000 Direct 0.000 0.156 -0.869 0.000 -0.049 0.000 0.263 0.315 0.881 Indirect 0.167 0.347 0.000 0. 310 0.000 0.274 -0.661 0.426 0.000

PAGE 93

84 Figure 6.1 Structural Equation s Model of Commute Length COMMUTE DISTANCE 1 COMMUTE TIME 2 MALE MIDDLE AGED DRIVER STATUS MANAGERIAL/PROFESSIONAL OCCUPATION NUMBER OF CHILDREN HOUSEHOLD INCOME >=75 K MSA POPULATION 3 MILLION INSIDE URBAN AREA/CLUSTER

PAGE 94

85 CHAPTER 7 CONCLUSIONS AND FURTHER RESEARCH 7.1 Conclusions The individual, household, trip and area related characteristics of short, medium and long commuters are discussed and models were developed to measure the commute time and propensity for a commuter type. The desc riptive analysis using NHTS 2001 and ACS 2000 revealed that the characteri stics of short and long commuters are different in nature. The descriptive analysis of da ta provides better u nderstanding about th e variation in the aforementioned characteristics influencing the commuting pattern. The influences of these characteristics are examined in detail for each of the commuter types. The analysis provides better understanding about the behavior al nature of the commuters in making short and long trips to work. This information can be used to formulate adequate commuter type choice models. The commuter type choice models developed ba sed on probabilistic th eory with random utility function provides a way in order to de velop choice models for commute type. The commuter type choice models develope d using the NHTS 2001 and ACS 2000 have expected signs for all the coefficients. The models confirmed the importance of demographic variables in explaining the co mmuter type choice behavior. The commute length measurement model developed using th e structural equations framework captures the simultaneous effect of the demographic ch aracteristics on commut e distance and time. The models reveal that demographic characte ristics should be includ ed in explaining the commuter type choice behavior and commute length that in way reflects the choice for residential and workplace location. There are some deficiencies in the models that can be attributed to the limitation of the datasets. This study provides analysis for policy makers who are concerned about the job accessibility and mobility options of the poor. Alan E. Pisarski had done a lot of research in this direction. In his studies on commuting to work, he focused on transportation’s role in providing mobility options, policy impact s on poor, impacts of urban sprawl on inter city travel, and impacts of congestion on co mmute time. His studies include public policy planning, travel behavior pla nning and statistical analysis Our study provided in-depth analysis of different commuter types. This st udy can be used for policy planning in the direction of providing mobility options for the poor as it provi des insights into whether an

PAGE 95

86 individual of particular type is because of his own preference or other external constraints. 7.2 Further Research The analysis of trip charac teristics of short, medium and long commuters revealed interesting trends about trip rates and trav el time expenditures. Study of trip chaining behavior, travel time expenditure, activity dur ations and vehicle u tilization patterns of short and long commuters would be an intere sting area to explore. More research is needed to find the direction of the relati onship between commuter type choice and mode choice of an individual. Models that cons ider combined choice of commuter type and mode choice would capture: The effects of demographic characteris tics on commuter type choice and mode choice simultaneously. The influence of mode on the residential location choice, which is reflected in commuter type choice or the commute time. For this type of research to be done the data has to be available rega rding the availability of public transit (NHTS 2001 provides information of availability of only rail transit to a household). However, even if the data is available the mo de choice modeling considers only those who have access to both auto and transit and this will exclude the zero-vehicle households as a result the low-income people ma y not be considered for the study. In this the case the research would not be useful for policy planning in the direction of transportation equity.

PAGE 96

87 REFERENCES Abraham, J.E. and Hunt, J.D., 1997. Specificatio n and estimation of a nested logit model of home, workplace and commuter c hoice by multiple worker households. Transportation Research Record 1606 (1997), pp. 17–24. Adjibolosoo, S., 2000. A critical anatomy of pe rformance effectiveness of the structural adjustment program in Tanzania: a huma n factor assessment. Balance, spring. Alonso, W., 1964. In: Location and Landuse Harvard University Press, Cambridge, Massachusetts. AMOS User’s Guide, Version 4. Anas, A. and Kim, I., 1996. General equilibr ium models of polycentric urban land use with endogenous congestion and job agglomer ation. Journal of Urban Economics 40, pp. 44-47. Baker, J.L., 2001. Social Exclusion in Urban Uruguay The World Bank, Washington, DC. http://www.worldbank.org/lac/. Bajari, P. and Khan M., 2001. Why Do Blacks Li ve in The Cities and Whites Live in the Suburbs?Stanford and Tufts Universities. http://wwwecon.stanford. edu/faculty/workp/swp00007.pdf. Bell, D.A., 1991.Office location––city or suburbs?. Transportation 18 (1991), pp. 239– 259. Ben-Akiva, M. and Bowman, J., 1998. Integrat ion of an activity-bas ed model system and a residential location model. Urban Studies 35 7, pp. 1131–1154. Ben-Akiva, M., Lerman, S. (Eds.), 1985. Discrete Choice Analysis: Theory and Application to Predict Travel De mand. MIT Press, Cambridge, MA. Berkovec, J. and Rust, J., 1985. A nested lo git of automobile holding for one-vehicle households. Transportation Research 19B 1, pp. 275–285. Bhat, C.R., 1997. Work travel mode choice and number of non-work commute stops. Transportation Research B 31 1, pp. 41–54.

PAGE 97

88 Bhat., C.R., 1997. Work travel mode choice and number of non-work commute stops. Transportation Research B 31 1, pp. 41–54. Bieber, A., Massot, Marie, H., Orfeuil a nd Jean, P., 1994. Prospects for Daily Urban Mobility. Transport Reviews 14(4),pp. 321-339. Boarnet, M. and Sarmiento, S., 1998. Can land-us e policy really affect travel behaviour? A study of the link between non-work travel and land-use characteristics. Urban Studies 35 7, pp. 1155–1170. Bollen, K. A. Structural Equations with Latent Variables Wiley, New York, 1989. Brown, H.,1975. Changes in workplace and residential locations. Journal of the American Institute of Planners 41, pp. 32–39. Cervero, R. and Kang-Li, Wu., 1998. Sub-centr ing and commuting: evidence from the San Francisco bay area. Urban Studies 35 7, pp. 1059–1076. Cervero, R. and Wu, K.L., 1997. Polycentrism, commuting, and residential location in the San Francisco Bay Area. Environment and Planning A 29, pp. 865–886. Cervero, R. and Wu, K.L., 1998. Sub-centering and commuting: evidence from the San Francisco Bay Area. Urban Studies 35, pp. 1059–1076. Cervero, R. and Landis,. J., 1992. Suburbanization of jobs and the journey to work: a submarket analysis of commuting in the San Francisco Bay Area. Journal of Advanced Transportation 26, pp. 3. Clark, W.A.V., Huang, Y. and Withers, S., 2003. Does commuting distance matter? Commuting tolerance and residential change. Journal of Regional Science and Urban Economics. Volume 33, Issue 2, pp. 199-221. Dijst and Vidakovic, 2000. M. Dijst and V. Vi dakovic Travel time ratio: the key factor in spatial reach. Transportation 27, pp. 179–199. Doorn P.K. and Rietbergen, A.V., 1990. Lifetime mobility: In terrelationships of labor mobility, residential mobility and household cycle. The Canadian Geographer 1, pp. 33– 48. Ellwood, D., 1986. The spatial mismatch hypothesis: Are there jobs missing in the ghetto?, in “The Black Youth Employment Crisis” (R. Freeman and H. Holzer, Eds.),Univ. of Chicago Press, Chicago.

PAGE 98

89 Gabriel, S. and Rosenthal, S.,S., 1996. Comm utes neighborhood effects, and earnings: An analysis of racial discrimination and compensating differentials, Journal of Urban Economics 40,pp. 61–83. Getis, A.,1969. Residential location and the journey to work. Proceedings, Association of American Geographers 1, pp. 55–59. Godard, X. and Olvera, L.D., 2000. Pauvret et transports ur bains. Exprience franaise et villes en dveloppementRapport pour la Banque Mondiale dans le cadre de l'Urban Transport Strategy Review, SITRASS, Lyon Golob, T.F., 2000. A simultaneous model of hous ehold activity part icipation and trip generation. Transportation Research B 34, pp. 355–376. Golob, T.F. and McNally, M.G., 1997, A model of activity particip ation and travel interactions between household heads. Transportation Research B 31, pp. 177–194. Golob, T.F., Bradley, M.A. and Polak, J.M., 1995. Travel and activity participation as influenced by car availability and use. Wo rking Paper, No. 286, University of California Transportation Center, Berkeley. Golob, Beckmann, T.F., Martin J., Zahavi and Yacov, 1981. A Utility-Theory Travel Demand Model Incorporating Travel Budgets. Transportation Research B 15(6), pp. 375389. Gordon, I., and Vickerman, R., 1982. Opportuni ty, preference and constraint: an approach to the analysis of metropolitan migration. Urban Studies 19, pp. 247–261. Gordon, P., Kumar, A. and Richardson, H. W., 1989. Congestion, changing metropolitan structure and city size in the United States. International Regi onal Science Review 12, pp. 45–56. Grieco, M., Apt, N. and Turner, J., 1996. At Christmas and on Rainy Days. Transport, Travel and the Female Traders of Accra. Avebury, Aldershot. Gunn, Hugh F. 1981. Travel Budgets – A Review of Evidence and Modeling Implications. Transportation Research 15A, pp. 7-23. Hamed and Mannering, 1993. M.M. Modeling trav elers' postwork activity involvement: toward a new methodology. Tran sportation Science 27 1993, pp. 381–394. Hensher, D. and Johnson, N., Editors, 1982. Applied Discrete Choice Modeling, Wiley, New York.

PAGE 99

90 Hensher, D., Barnard, P., Smith, N. and Milthorpe, F., 1991. An empirical model of household automobile holdings. Applied Economics. Holzer, H., J., Ihlanfeldt, K.R., and Sjoquist D.L., Work, search, and travel among white and black youth, Journal of Urban Economics 35, pp. 320–345. Hupkes and Geurt, 1982. The Law of Constant Travel Time and Trip-Rates. Futures February,pp. 38-46. Ihlanfeldt, K.R., 1988. Intrametropolitan va riation in earnings and labor market discrimination: An economic anal ysis of Atlanta labor market, Southern Economic Journal 55, pp. 123–140. Ihlanfeldt, K.R., 1992.“Housing Segregation and the Wages and Commutes of Urban Blacks, The Case of Atlanta Fast-Food Rest aurant Workers,” Policy Research Center Paper 30, Georgia State University. Ihlanfeldt, K.R., 1993. Intra-urban job accessibili ty and hispanic youth employment rates, Journal of Urban Economics 33, pp. 254–271. Ihlanfeldt, K.R. and Sjoquist, D. L., 1990. Job accessibility and raci al differences in youth employment rates, American Economic Review 80, pp. 267–276. Ihlanfeldt, K.R. and Sjoquist, D. L., 1991. The effect of job access and black youth employment: A cross-sectional analysis, Urban Studies 28,pp. 255–265. Ihlanfeldt, K.R. and Sjoquist, D. L., 1998. The spatial mismatch hypothesis: A review of recent studies and their implications for welfare reform, Housing Policy Debate 9,pp. 849–892. Ingram, G., 1998. Patterns of me tropolitan development: what have we learned?. Urban Studies 35 7, pp. 1019–1036. Kain, J.F., 1968. Housing segregation, negro employment, and metropolitan decentralization, Quarterly Journal of Economics 82,pp. 32–59. Kain, J.F., 1992. The spatial mismatch hypothesis: Three decades later, Housing Policy Debate 3,pp. 371–460. Kain, J.F.,1962. The journey to work as a determinant of residential location. Papers and Proceedings, Regional Science Association 9, pp. 137–161. Kenworthy J.R. and Laube F.B., 1999. Patterns of automobile dependence in cities: an international overview of key physical and ec onomic dimensions with some implications for urban policy. Transportation Research A 33.

PAGE 100

91 Khattak, A.J, Amerlynck and V., Quercia, R.G., 1999. Are Travel Times And Distances To Work Greater For Residents Of P oor Urban Neighborhoods?. Transportation Research Board, Washington D.C. h ttp://www.unc.edu/~khattak/wktt.html. Kitamura, R., Chen, C. and Narayanan, R., 1998. Traveler, destination choice behavior: effects of time of day, activity duration, and home location. Transportation Research Record 1645, pp. 76–81. Kitamura, R., Nishii, K. and Goulias, K., 1990. Trip chaining behaviour by central city commuters: a causal analysis of time-space constraints. In: P. Jones, Editor, Developments in dynamic and activitybased approaches to travel analysis Avebury, Aldershot (1990), pp. 145–170. Leete, L., Bania, N. and Coulton, C., 1996. Measuring Commute Times and Job Access. Center On Urban Poverty And Soci al Change. Briefing Report No. 9911. Leete, L., Bania, N. and Coulton, C., 1996 Distance and Commute Times to Work For Welfare Exiters. Briefing Report No. 9908. Leete, L., Bania, N. and Coulton, C., 1996 City or Suburbs? Job and Residential Locations of Welfare Exiters in 1996. Cent er On Urban Poverty And Social Change. Briefing Report No. 9907. Levinson, D.M., 1998. Accessibility and the journey to work. Journal of Transport Geography 6 (1998), pp. 11–21. Levinson, D.M., 1999. Space, money, life-stage, and the allocation of time. Transportation 26 (1999), pp. 141–171. Levinson, M., 1998. Accessibility and the journey to work. Journal of Transport Geography 6 (1998), pp. 11–21. LIMDEP Reference Guide, Version 3. Linneman, P. and Graves, P., 1983. Migration and job change: a multinomial logit approach. Journal of Urban Economics 14 (1983), pp. 263–279. Lu, X. and Pas, E.I., 1999. Socio-demogra phics, activity participation and travel behavior. Transportation Research A 33 (1999), pp. 1–18. Macek, M., Khattak, A. and Quercia, R.G., 2001. Transportation Research Board, Washington D.C. http://www.msu.edu/user/maceknat/resource/mp.pdf. Mannering, F. and Winston, C., 1985. Dynamical em pirical analysis of household vehicle ownership and utilisation. Rand Journal of Economics 16 2, pp. 215–236.

PAGE 101

92 Mannering, F. and Winston, C., 1985. Dynamical em pirical analysis of household vehicle ownership and utilisation. Rand Journal of Economics 16 2, pp. 215–236. Mannering, F., Kim, S., Barfield, W. and Ng, L, 1994. Statistical anal ysis of commuters' route, mode and departure time flexibili ty. Transportation Research 2C 1, pp. 35–47. Marchetti, C., 1994. Anthropological I nvariants in Travel Behavior. Technological Forecasting and Social Change 47, pp. 75-88. Mokhatarian, P.L. and Chen, C., 2002. TTB or Not TTB, that is the Question: A Review and Analysis of the Empirical Literatu re on Travel Time (and Money) Budgets. Department of Civil and Environmental Engi neering, University of California, Davis. http://www.its.berkeley.edu/publications/I TSReviewonline/spri ng2003/trb2003/Mokhtari an-ttb.pdf. Muth, R., 1969 In: Cities and Housing: The Spatial Pa ttern of Urban Residential Land Use University of Chicago Press, Chicago, IL (1969). Guckin, N. and Srinivasan, N., 2001. Journey to Work Trends in the United States and its Major Metropolitan Areas 1960 – 2000. Olvera, L.D.O., Plat, d. and Poschet, P., 2003. Transportation conditions and access to services in a context of urban sprawl a nd deregulation. The case of Dar es Salaam. Transportation Policy Volume 10, Issue 4, pp. 287-298. Ommeren, V.J.N., Rietveld, P. and Nijkamp, P., 1997. Commuting in search of jobs and residences. Journal of Urban Economics 42 (1997), pp. 402–421. Palma, A.D. and Rochat, D., 2000. Mode choi ces for trips to work in Geneva: an empirical analysis. Journal of Transpor tation Geography. Volume 8, Issue 1, pp. 43-51. Pas, Eric I., 1998. Time in Travel Choice Mode ling: From Relative Obscurity to Center Stage. In: Theoretical Foundations of Travel Choice Modeling. Pinto, M.S., 2002. Residential choice, mobilit y, and the labor market. Department of Economics, University of Urbana-Champaign Journal of Urban Economics 51, pp. 469–496. Pisarski, A.E., 2002. Testimony of “Mobility, Congestion and Intermodalism”. Committee on Environment and Public Wo rks. Washington D.C. Paper No. 012946.http://alanpisarski.com/Testimonybefore theSenateCommitteeonEnvironmentandTra nsportation2002.pdf Pisarski, A.E., 2002. The Democratization of Mobility in America: Cars, Women and Moniorities. Automobity and Freedom Pr oject. http://www.cei.org/PDFs/pisarski.pdf.

PAGE 102

93 Reid, C.E., 1985. The effect of residential location on the wages of black women and white women, Journal of Urban Economics 18,pp. 350–363. Rouwendal, J., 1999. Spatial job search and commuting distances. Regional Science and Urban Economics 29 (1999), pp. 491–517. Schafer and Andreas, 1998. The Global Demand for Motorized Mobility. Transportation Research A 32(6), 455-477. Schafer and Andreas, 2000. Regularities in Trav el Demand: An International Perspective. Journal of Transportation and Statistics 3(3), December. Schafer, Andreas and David G. Victor ( 2000). The Future Mob ility of the World Population. Transportation Research A 34(3), pp.171-205. Schafer, Andreas and David G. Victor, 2000. The Future Mobility of the World Population. Transportation Research A 34(3),pp. 171-205. Schwanen, T. and Dijst, M., 2002. Travel-tim e ratios for visits to the workplace: the relationship between commuting time and wo rk duration. Journal of Transportation Research, Part A. Volume 36, Issue 7, pp. 573-592. Sermons, N.W. and Koppelman, F.W., 1999. U nderstanding the differences between female and male commute behavi or in two-worker households. SPSS 11.5 for Windows, User’s Guide. Straszheim, M.R., 1980. Discrimination and the sp atial characteristics of the urban labor market for black workers, Journal of Urban Economics 7,pp. 119–140. Thobani, M., 1984. A nested logit model of tr avel mode to work and auto ownership. Journal of Urban Economics 15, pp. 287–301. Tommy Garling, Thomas Laitila, and Kerstin Westin (Eds.). El sevier, New York, 231250. Train, K., 1980. A structured logit model of au to ownership and mode choice. Review of Economic Studies 157 2, pp. 357–370. Train, K., Editor, 1986. Qualitative Choice An alysis. Theory, Econometrics, and an Application to Automobile De mand, MIT Press, Cambridge, MA. Turner and Niemeier, 1997. T. Turner and D. Niemeier Travel to work and household responsibility: new evidence. Transportation 24 (1997), pp. 397–419.

PAGE 103

94 Turner, J. and Kwakye, E., 1996. Transport a nd survival strategies in a developing economy: case evidence from Accra, Ghana. Journal of Transport Geography 3 4, pp. 161–168. Vasconcellos., E.A, 2001. Urban Transport, Environment and Equity. The case for developing countries Earthscan, London. Vilhelmson, Bertil, 1999. Daily Mobility and the Use of Time for Different Activities: The Case of Sweden. GeoJournal 48, pp.177-185. Vrooman, J. and Green.eld, S., 1980. Are bl acks making it in the suburbs? Some new evidence on intrametropolit an spatial segmentation, Journal of Urban Economics 7,pp. 155–167. Wachs, M., Taylor, B., Levine, N. a nd Onh, P., 1993. The changing commute: A casestudy of jobs-housing re lationship over time. Urban Studies 21 (1993), pp. 15–29. Waddell, P. Exogenous workplace choice in residential location models: Is the assumption valid?. Geographical Analysis 25 (1993), pp. 65–75. Werlin, H., 1999. The slum upgrading myth. Urban Studies 36 9, pp. 1523–1534. Wingo, L., 1961.In: Transportation and Urban Land Resources for the Future, Washington, DC (1961). Zahavi, Yacov and Ryan, James M., 1980. Stab ility of Travel Co mponents over Time. Transportation Research Record 750, pp.19-26. Zahavi, Yacov and Talvitie, Antti, 1980. Regularities in Travel Time and Money Expenditures. Transportation Research Record 750,pp. 13-19. Zahavi and Yacov, 1979. “UMOT” Project Prepared for U.S. Department of Transportation, Washington, D. C. and Ministry of Trans port, Federal Republic of Germany, Bonn. Rept. DOT RSPA-DPB-20-79-3. Zax, J.S. and Kain, J.F., 1996. Moving to th e suburbs: Do relocating companies leave their black employees behind?, Journal of Labor Economics 14,472–504.


xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001469423
003 fts
005 20051108063305.0
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 040524s2004 flua sbm s000|0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0000324
035
(OCoLC)55731559
9
AJR1177
b SE
SFE0000324
040
FHM
c FHM
090
TA145
1 100
Vaddepalli, Srikanth.
3 245
An analysis of characteristics of long and short commuters in the United States
h [electronic resource] /
by Srikanth Vaddepalli.
260
[Tampa, Fla.] :
University of South Florida,
2004.
502
Thesis (M.S.C.E.)--University of South Florida, 2004.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
538
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
500
Title from PDF of title page.
Document formatted into pages; contains 103 pages.
520
ABSTRACT: An in-depth-analysis was carried out on short, medium and long commuters using the National Household Travel Survey (NHTS) of 2001 and American Community Survey (ACS) of 2000 to determine the role of individual,household, trip and area related characteristics on commute length. The individuals with commute time less than or equal to 15 min were considered as short commuters and individuals with commute time greater than 15 min but less than 60 min were considered as medium commuters and the individuals with commute time 60 min or more were considered as long commuters. The commute time is considered as a link joining the residence and workplace locations. The availability of the desired mode used is considered as flexibility in moving the location of these points in the area. As the jobs get dispersed the lower income people face more and more transportation problems in linking the residence and workplace. There is a potential threat in their social, physical and economic isolation in the society. The individual, household, and area related characteristics are assumed to influence both the commute time and location of these points. The descriptive analysis using NHTS 2001 and ACS 2000 revealed that the characteristics of short and long commuters are different in nature. A commuter type choice model and commute length measurement models were used to estimate the influence of socio-demographic characteristics on the residential and workplace separation. Multinomial Logit Model (MNL) methodology was adopted to develop the commuter type choice model and Structural Equations Model methodology (SEM) was adopted with commute time and commute distance as endogenous variables to estimate the commute length on a continuous scale. The models confirmed the importance of demographic variables in explaining commuter length.
590
Adviser: Pendyala, Ram M.
653
commuter behavior.
socio-demographic characteristics.
job access.
residence and workplace location.
transportation equity.
social isolation.
0 690
Dissertations, Academic
z USF
x Civil Engineering
Masters.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.324