USF Libraries
USF Digital Collections

An analysis of the travel patterns and preferences of the elderly

MISSING IMAGE

Material Information

Title:
An analysis of the travel patterns and preferences of the elderly
Physical Description:
Book
Language:
English
Creator:
Sikder, Sujan
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Mobility
NHTS
Socio-demographics
Multinomial logit model
Mixed-multinomial logit model
Dissertations, Academic -- Civil & Environmental Engineering -- Masters -- USF   ( lcsh )
Genre:
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: The number of elderly is increasing; to meet their transportation needs, it is important to clearly understand their travel patterns and preferences. Since travel patterns and preferences depend on socio-demographic and other factors, it is essential to identify these factors first to understand the travel behavior of the elderly. The main purpose of this thesis is to analyze the travel patterns and preferences of the elderly age 65 and above using 2009 National Household Travel Survey (NHTS) data. This thesis presents a detailed descriptive analysis of 2009 NHTS data to understand the travel patterns of the elderly. Along with a descriptive analysis, a multinomial logit model and a mixed- multinomial logit model are estimated to explore the factors associated with the overall travel preferences of the elderly and to identify individuals among the elderly who are the least mobile and at risk for social isolation. The analysis results indicate the differences in the trip characteristics between the elderly and non-elderly. Variation is found even among the different groups of the elderly. The model estimation results show the presence of different travel preferences among the elderly and identify those individuals among the elderly who are immobile for longer periods (e.g., a week) and at risk for social isolation. Elderly individuals with different travel preferences should be considered separately in research to determine the appropriate outcomes that can help transportation planners and policy makers improve planning and policy related to elderly individuals.
Thesis:
Thesis (MSCE)--University of South Florida, 2010.
Bibliography:
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Sujan Sikder.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains X 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:
usfldc doi - E14-SFE0004630
usfldc handle - e14.4630
System ID:
SFS0027945:00001


This item is only available as the following downloads:


Full Text

PAGE 1

An Analysis of the Travel Patterns and Preferences of the Elderly by Sujan Sikder A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Civil Engineering Department of Civil and Environmental E ngineering College of Engineering University of South Florida Major Professor: Abdul R. Pinjari, Ph.D. Xuehao Chu, Ph.D. Steven E. Polzin, Ph.D. Date of Approval: June 30, 2010 Keywords: mobility NHTS, socio de mographics multinomial logit model mixed multinomial logit model Copyright 2010, Sujan Sikder

PAGE 2

DEDICATION This thesis is dedicat ed to my father, Samiran Sikder and my mother Swapna Sikder for their love, affection and nurturing. I am greatly indebted to them for their supp ort and encouragement thorough my life. I would also like to dedicate this thesis to my brother for his warm companionship.

PAGE 3

ACKNOWLEDGEMENTS I express my sincere gratitude to my advisor, Dr. Abdul R. Pinjari, for his constant support and e ncouragement throughout the masters program. I thank him for his active encouragement, inspiration and invaluable ideas that gave me the opportunity to learn and experience research. I would like to thank Drs. Xuehao Chu and Steven E. Polzin for serving in my thesis committee and for providing valuable suggestions. I also thank the Department of Civil and Environmental Engineering for providing such good facilities and environment for research. I would also like to thank all of my friends and colleagues for their wonderful company and support throughout the research work.

PAGE 4

i TABLE OF CONTENTS LIST OF TABLES iii LIST OF FIGURES v ABSTRACT v i CHAPTER 1: INTRODUCTION 1 1.1 Backgroun d 1 1.2 Motivation 3 1.3 Objectives 6 1.4 Organization of the T hesis 6 C H APTER 2: LITERATURE REVIEW 8 2.1 Introduction 8 2.2 Effects of Socio Demographic Fact ors on Elderly Travel Behavior 8 2.3 Application of Modeling Technique s in Elderly Travel Behavior 16 CHAPTER 3: DATA DESCRIPTION 20 3.1 Introduc tion 20 3.2 Natio nal Household Travel Survey 20 3.3 Descriptive from the 2009 Nat ional Household Travel Survey 22 3. 3.1 Household Characteristics 22 3.3.2 Person Characteristics 27 3.3.3 Trip Characteristics 31 3.4 Conclusion 52 CHAPTER 4 : MODELING METHODOLOGY 54 4 .1 Introduction 54 4 .2 Data 54 4.2.1 Sample Preparation 54 4.2.2 Sample Description 56 4 .3 Multinomial Logit Model 60 4 .3.1 Random Utility Maximization Approach 61 4 .3.2 Estimation and Evaluation 63 4 .4 Multinomial Logit Model Results 64 4.5 Mixed Multinomial Logit Model 72 4.5.1 Error Components Specification 73

PAGE 5

ii 4.5.2 Estimation and Evaluation 74 4.6 Mixed Multinomial Logit Model Results 75 4.7 Conclusion 80 CHAPTER 5 : CONCLUSIONS AND FUTURE RESEARCH 83 5 .1 Conclusions 83 5 .2 Future Research 87 REFERENCES 8 8

PAGE 6

iii LIST OF TABLES Table 3.1 Household Ch aracteristics of the 2009 NHTS D ata 24 Table 3.2 Elderly Household (One Person and Two P erson) Characteristics 26 Table 3.3 Person Ch aracteristics of the 2009 NHTS D ata 28 Table 3.4 Person Characteristics (One Person and Two P erson Household s ) 30 Table 3.5 Trip Distribution by Age 32 Table 3.6 Trip Distribution by Age and Gender 32 Table 3.7 Trip Characteris tics by Age and Gender 33 Table 3.8 Distribution of Drivers and Non Drivers by Age and Gender 35 Table 3.9 Distribution of Person Trip s by Age, Driver Status and Mode 36 Table 3.10 Trip Characteristics by Age, Driver S tatus and Gender 37 Table 3.11 Distribution of Person Trip s by Age, Race and Mode of Transportatio n 38 Table 3.12 Trip Characteristics by Age and Race 40 Table 3.13 Distributio n of Person Trip s b y Age, Vehicle Ownership and Mode 41 Table 3.14 Trip Characteristics by Age and Vehicle Ownership 43 Table 3.15 Trip Characteristics by Age and Household Income 44 Table 3.16 Distribution of Person Trip s by Age, Mode and Trip Purpose 46 Tabl e 3.17 Trip Characteristics by Age and Trip Purpose 47 Table 3.18 Trip Characteristics by Age and Mode of Transportation 48 Table 3.19 Trip Characteristics by Age, Worker Status and Gender 49

PAGE 7

iv Table 3.20 Trip Characteristics by Age and Population Size of MSA 50 Table 4.1 Comparison of Mobility Patterns and Preferences by Age G roups 56 Table 4 .2 Sample Characteristics 58 Table 4 .3 Multinomial Logit Model Results 66 Table 4.4 Mixed Multinomial Logit Mode l Results 77

PAGE 8

v LIST OF FIGURES Figure 3.1 Average Daily Person Trips by Age and Gender 33 Figure 3.2 Distribution of Drivers and Non Drivers by Age and Gender 35 Figure 3.3 Dist ribution of Person Trip s by Age and Mode of Transportation 36 Figure 3.4 Average Daily Person Trip s by Age and Race 39 Figure 3. 5 Distribution of Trip s by Vehicle Ownership and Mode of Transportation 42 Figur e 3.6 Average Daily Person Trip s by Age and Vehicle Ownership 43 Figure 3.7 Average Daily Person Trip s by Age and Household Income 44 Fi gure 3.8 Distribution of Person Trip s by Age and Time of the Day 51

PAGE 9

vi An Analysis of the Travel Patterns and Preferences of the Elderly Sujan Sikder ABSTRACT The number of elderly is increasing; to meet their transportation needs, it is important to clearly understand their travel patterns and preferences. Since travel patterns and preferences depend on socio demographic and other factors, it is essential to identify these factors first to understand the travel behavior of the elderly. The main purpose of this thesis is to analyze the travel patterns and preferences of the elderly age 65 and above using 2009 National Household Travel Survey (NHTS) data. This thesis presents a detailed descriptive analysis of 2009 NHTS data to understand the travel patterns of the elderly. Along with a descriptive analysis, a multinomial logit model and a mixed multinomial logit model are estimated to explore the factors associated with the overall travel preferences of the elderly and to identify individuals among the elderly who are the least mobil e and at risk for social isolation. The analysis results indicate the differences in the trip characteristics between the elderly and non elderly. Variation is found even among the different groups of the elderly. The model estimation results show the pres ence of different travel preferences among the elderly and identify those individuals among the elderly who are immobile for longer periods (e.g., a week) and at risk for social isolation. Elderly individuals with

PAGE 10

vii different travel preferences should be con sidered separately in research to determine the appropriate outcomes that can help transportation planners and policy makers improve planning and policy related to elderly individuals.

PAGE 11

1 CHAPTER 1 INTRODUCTION 1.1 Background Transportation mobility is cr itically important to our li v e s Mobility is generally a derived demand for different needs in our daily life and it is the transportation system that gets us to work and other places for shopping, social interaction, personal errands etc. to fulfill th ose needs. A n accessible, affordable and reliable transportation system is desirable to us all Alt hough transportation mobility is critical in life no matter what the age, this issue is more important for the elderly due to their physical and mental conditions. T he total number of older people is increasing in almost all w estern European c ountries, North America, and Australia (Rosenbloom, 2001) and it is expected that the number of the elderly age 65 and over will be at least double by 2051 c ompared to 1999 (Alsnih and Hensher, 2003). Therefore, u nderstanding the travel patterns and preferences of the elderly is becoming increasingly important. Mobility is required not only for obtaining different commodities and goods in our daily li v e s it i s essentially important for participation in social relati ons and activities (Mollenkopf, 1997 ). Participation in such activities is important to our quality of life especially in the li v e s of older people because social activities involving mobility redu ce mortality in older people (Glass et al. 1999). Mobility provides som e psychological

PAGE 12

2 benefit as well (Metz, 2000). O lder people are likely to develop physical, sensory and cognitive limitations with the increase in age resulting in a decline in their mobility. While they can satisfy their needs for medical appointments and grocery shopping through some services, they face difficulties in conducting the social or recreational activities that are an important part of their lives. T hese difficulties and o ther factors such as physical impairments, the desire to stay home etc. decrease mobility among the elderly and ultimately, force them into social isolation. The immobile elderly are at risk for social isolation, which affects the ir quality of life and accelerates the decline in their personal health (Trilling and Eberhard, 2002). T he elderly who travel generally make fewer trips and the trip characteristics associated with those trips are different compared t o their younger counterparts (Collia et al ., 2003 ; Heaslip 2007). The characteristics of these trips depend on the goods and services they need and also on the ir desire for social interaction (Skinner and Stearns, 1999). The desire or preference to socialize is one of the factors affecting the tr ip characteristics of the elderly. Th is preference is the is difficult to assess. T he latest 2009 National Household Travel Survey (NHTS) provide d an opportunity to identify the fac tors associated with the mobility patt erns and preferences of the elderly by introducing some new questions on the 2009 questionnaire. a person did not make any trips on the given reporting day (travel day) of the survey and mentioned that he/she had stayed at the same place (for example, at home) all day, he/she is question was asked of only

PAGE 13

3 those persons who traveled more than a week ago from the given reporting day (travel day) of the survey Individuals who did not travel for long periods such as a week can be s e question s in directly distinguish the short term (less than a week) and long term (more than a week) immobil ity of individuals. Based on the questions mentioned above, a sample can be divided into four categories: individuals (1) who traveled on the travel day (2) w ho did not travel on the travel day but traveled in the past seven days (3) who did not travel in the past seven days but prefer going out more often and (4) who did not travel in the past seven days and do not prefer going out more often. The first cate the secon individuals and the last two categories can be combi individuals. consists of two separate groups prefer g but did not travel for longer periods due t o some constraints and do T hese two groups of individuals revealed their inherent travel prefe rences on the survey and thus provided an opportunity to identify the factors associated with these preferences. Considering the importance of the mobility issues of the elderly, this study focuses only on persons age 65 and above. The next sections of this chapter describe the motivation, objectives and organization of the thesis. 1.2 Motivation The elderly are likely to become transportation disadvantaged with the increase of age (Giuliano, 1999), and the travel behavior of this group increasingly is seen as an import

PAGE 14

4 into retirement age and the expectations of this group will be different from that of the current elderly because of their experience with affordable mobility and technology throughout their lives (Coughlin, 2009). The perceived differences in the mobility characteristics of the baby boomers and the current elderly warrant the need for in depth research on elderly travel behavior. The elderly are not homogeneous; differences exist in socio economic characteristics such as household structure, gender, lifestyle, and race (Kim and Ulfarsson, 2004). These differences affect the mobility and travel patterns of the different groups of elderly (Hilderbrand, 2 003).This diversity str ikes the researchers and transporta tion planners to find out the avenues to fulfill the special transportation needs of different groups of the elderly. It also is important to minimize the level of depression among the elderly that may result from the los s of driving ability (Alsnih and Hensher, 2003). Tacken (1998) emphasizes taking necessary steps to keep the elderly mobile rather than to reactivate their desire for mobility. To keep the elderly mobile, it is important to identify the factors affecting t heir travel preferences through an in depth analysis. Also, since some older persons do not travel but rather stay at home all day, it is necessary to understand the reasons for this behavior. The elderly are generally less likely to make trips, and the trips they make are usually shorter in distance compared to their young counterparts (Collia et al., 2003; Heaslip, 2007). The travel patterns and different characteristics associated with the travels such as travel distance and travel time are dependent u pon the travel preferences of the elderly. These travel preferences depend on several socio demographic factors as well as the physical and mental conditions of the elderly. Most of the research in the

PAGE 15

5 literature focuses only on the group that travels and the factors affecting their travel preferences But it is important to understand the motivations behind the immobile population of the elderly Since immobile persons are a share of the total elderly population and the success of any elderly mobility rela ted policy depends on the participation of both groups of elderly (travelers and non travelers), it is important to understand the underlying factors affecting the travel preferences of the elderly. To state it succinctly, it is necessary to understand cle arly those who are not traveling among the elderly and why they are not traveling. Essentially, it is important to identify and then distinguish the factors associated with the short and long term immobility of the elderly. These factors may help in takin g appropriate measures to retain the mobility desires of the travelers and to reactivate the non travelers. A multinomial and a mixed multinomial logit model are estimated in this study for analyzing the different travel preferences of the elderly. Most o f the elderly travel behavior related studies in the literature focus only on the travel patterns of the elderly, that is, how the trip characteristics of the elderly differ from their counterparts; the travel preferences of the elderly generally are not c onsidered in those studies. Even if considered, it is limited to only one day travel period data. However, the latest 2009 NHTS data provide an opportunity to analyze the travel preferences of the elderly for longer periods (such as a week) instead of focu sing only on one day travel period data. Many forms of special transportation services are provided for the elderly through support from the federal government and other sources (Trilling and Evarhard, 2002), and these services generate additional travel b y the elderly. However, to develop policy in this area for instance, to evaluate the efficient distribution of funds

PAGE 16

6 and the success of the services it would be valuable to identify the factors related to long term immobility among the elderly. 1.3 Object ives This study mainly aims to provide a detailed analysis of the travel patterns and preferences of the elderly in the context of socio demographic and other factors using data from the 2009 NHTS. Persons age 65 and older are considered to be elderly in this study. The specific objectives of this study are as follows: To understand the trip characteristics of the elderly in order to examine how their travel patterns differ from those of their younger counterparts. To examine the presence of different trav el patterns among the different groups of elderly. To identify the factors affecting the overall travel preferences of the elderly. To distinguish the socio demographic and other factors affecting the different travel prefer ences of the elderly ( a multinom ial logit model and a mixed multinomial logit model are developed to distinguish the factors) 1.4 Organization of the Thesis The remainder of this thesis is organized as follows. Chapter 2 provides an extensive review of the literature available related t o the thesis topic. Chapter 3 describes the 2009 NHTS data and gives a detailed description of household and personal characteristics of Americans. Trip characteristics of the elderly and how these characteristics differ among the elderly and non elderly a lso are provided in this chapter. Chapter 4 explains the modeling efforts undertaken in this study. In addition, sample characteristics, sample preparation for the model estimation, and model estimation results

PAGE 17

7 are provided. Finally, conclusions and the sc ope for further research are discussed in Chapter 5.

PAGE 18

8 CHAPTER 2 LITERATURE REVIEW 2.1 Introduction Mob ility issues related to elderly have been subject to extensive research by transportation planners and resear chers Understanding the travel pattern s of the elderly is becoming increasingly important due to their special trip making characteristics. It is of strong policy interest as well because the share of the elderly in the total population is significant. Interestingly, m ost of the studies on elderly travel behavior are based on descriptive analysis. But, it is always difficult to conclude confidently without considering the ef fects of all variables together, because the results obtained from the descriptive analysis might chang e when the effects of all possible variables are considered together (Kim and Ulfarsson, 2004). However, some modeling efforts ( Giuliano, 1999; Evans, 2001 ; Georggi and Pendyala 2001; Rosenbloom and Waldorf, 2001 ; Kim and Ulfarsson, 2004 ) were als o undert aken in this area This chapter provides a review of those modeling efforts and other research efforts in the direction of analyzing the travel patterns and preferences of the elderly. 2.2 Effects of Socio D emographics on Elderly Travel Behavior Several s tudies in the literature explored the effects of socio demographic characteristics on the travel patterns and preferences of the elderly. Among these,

PAGE 19

9 Lefrancois et al. (1998) measured the effect of age and other socio demographic factors such as gender, h ealth status, education and region on the activity involvement among the older adults aged 65 and above by analyzing a sample of 601 adults from Montreal and the eastern township, Canada. They tested a hypothesis that different factors (health status gend er, educational level, marital status rural and urban environment, social interaction) predict the reduction in activity among the older adults better than the age itself by using a canonical correlation approach. Travel activities (numb er of trips in the past year ) were included as one of the four major categories of activities considered in this study. The other three were: exercise and sport, social act ivities and outdoor recreations They found that health status played the most vi tal role in the reduc tion of activities among the elderly. Education was found to be positively related with the sport, travel and outdoor recreation activities participation. Coughlin ( 2001 ) explored the perceptions and preference of the persons age d 75 and above about their transportation options by using data from three focus groups and 17 one on one in person interviews conducted in Boston and Framingham, Massachusetts. The focus groups were formed in such a way that could reflect the characteristic s of the 75 plus age gro up and difference in various so cio demographic factors such as age, gender and driver status among the persons of this group. In addition, s through the variation in income of the partici pants. Coughlin observed that persons age d 75 and older were more inclined to the auto mobile based transport and the factors that drov e them towards this preference wer e reliability, convenience, personal security, and flexibility. In addition, the non dr iver older people were more likely to ask for rides to friends and/or

PAGE 20

10 family but the dependency and obligation result ed from requesting for rides were objectionable to them. Kim and & Ulfarsson (2004) showed the effects of personal, household, neighborhood and trip characteristics on the mode choice of retired elderly age d 65 and above by using the 2000 Puget Sound Transportation Panel (PSTP) data of the Puget Sound Regional Council (PSRC) in Washington State and 2000 census data. They found a negative re lationship between age and propensity to use privately owned vehicle. This result reveals the physical and cognitive deterioration of the elderly to drive and the dependency on transit. In addition, the negative relationship that they found between househo ld vehicle availability and the transit use clearly warrants the nee d of special transit system for zero vehicle elderly household s I ncome and d ista nce to the nearest bus stop were found to be negatively associated with the transit use of the elderly. Gag liard i et al.(2007) explored the effects of personal and environmental characteristics on the outdoor mobility and leisure activities of older people by using data from the interview of 3950 older adults ( age 55 and above) from fi ve different European coun tries : Germany, Finland, Hungary, The Netherlands and Italy. One important observation they made from their study was that non driver women of greater age with health problems were more likely to engag as comp ared to other groups of older adults. Henderson et al. (1998) and Mollenkopf et al. (1997) also found the similar results in their studies. By using the 1995 Nationwide Personal Transportation Survey (NPTS) data, Giuliano (1999) explored the relationship between land use and travel patterns among the elderly and other age groups. Though travel patterns were found to be affected by the

PAGE 21

11 land use, the effects were almost same across the three age groups considered in this study. In addition, a ge of the travel ers was found to be negatively associated with the trip making propensity and the travel distance. It was also found that the elderly were less likely to use transit even when it was accessible. Rosenbloom (1999) identified the basic travel patterns and trends among the elderly (age 65 and above) with special focus on drivers, non by using data from the 1995 nationwide personal transportation survey (NPTS) and from the office of Highway Information Management, FHWA. Auto mobil e was found to be the dominant mode for all elderly and a large gap was found in the total number of trips made by drivers and non drivers. Rosenbloom also observed that women were likely to make fewer trips and travel fewer miles as compared to men of al l age coho rts and this gap was found to increase with the age of the individuals. Besides this, elderly were found to be resided in the suburbs or in rural areas where automobile is necessity for mobility. Collia et al. (2003) analyzed the travel patterns of the eld erly (age 65 and above ) and compared wit h that of the young adults (age 19 to 65 years ) by using the 2001 National Household Travel Survey (NHTS) They found that older adults generally made fewer trips, traveled in mid day, traveled shorter dist ances and for shorter times. These patterns were more pronounced among the elderly women. In addition, old er drivers were found to make more percentage of trips as passenger as comp ared to the younger adults (age 19 to 65). They also found a lower percenta ge of alternative transportation mode uses on the travel day when compared to the share of the persons who had to give up driving due to medical condition. It was also found that only a small percentage (12%) of

PAGE 22

12 the people who identified the medical condit ion had affected their travel used special transportation service such as dial a ride on the travel day. The reasons behind this small percentage of alternative means and special transportation uses were not explored in this study. Since the percentage of the adults (24%) who reported medical condition ha d made their travel difficult was four times as compared to the younger adults (6%) (Collia et al. 2003), older adults were a good portion of the population for which special transportation systems were des igned. So, the success of these services is dependent on the uses of those services by the elderly people. The potential reasons for lower percentage of special transportation service users could be the serious medical condition that prevented them from ma king travel on the travel day or they used another mode of transportation on the travel day due to special circumstances or the special transportation systems were not accessible to the individuals. In addition, it might also happen that the individuals wa nted to travel on the day but some socio demographic conditions prevented them from doing go outside more often and liked to stay home. D ue to lack of data, it is not possible to explore each and every reasons mentioned above for not using special transportation service or alternative transpo rtation means on the travel day. But, the recently released 2009 NHTS provides a n opportunity to explore at least the effects of d ifferent factors on the last two issues i.e. the preferences of going outside of home Evans (2001) explored the personal and community characteristics associated with the trip making propensity among the non drivers aged 75 and above by using 1 day trave l period data from the 1995 Nationwide Personal Transportation Survey (NPTS) combined with community data prepared by Claritas, Inc. Though some contradictory

PAGE 23

13 findings such as low mobility in urban areas but high trip making propensity in areas with higher housing densities were found in this study, the overall results gave an idea of the personal and community characteristics associated with the trip making propensity of the non driver individuals aged 75 and above But the study was limited to 1 day trav e l periods and the author did not distinguish the short term and long term non travelers. More clearly, the difference be tween the individuals who did not travel on the travel day and those who did not travel for a longer time period such as a week were not differentiated in this study. The personal and community characteristics associated with these two different gr oups may vary and thus, warrant investigation. It is likely that some people may want to go outside and some people may not want to go outside m ore often. But, the possibility of the exis tence of two groups (who prefer go ing outsid e more often and who do not prefer go ing outside more often) within the long term non travelers were ignored in this study and the different types of non travelers were combined into a single due to the possibility of the existence of two groups mentioned above it is important to identify the factors affecting the travel preferences of the people who prefer go ing outside and who do not prefer go ing outside of home. This could be of great help to the transit industry and special transportation service providers to the elderly especially during taking decision on elderly prominen t area. If the situation is such that most of the elde rly in an area are not willing to go out more often, then the plan to provide special transportation service to that area will not be that much effective. So, from the policy perspective, the need of the individuals who want to go outside and who do not sh ould be identified first and then special transportation service should be provided accordingly.

PAGE 24

14 Heaslip (2007) analyzed the change of travel patterns among the older drivers (age 65 and above) by using the National Household Travel Survey data series from 1969 to 2001 and American Travel S urvey from 1977 to 1995. Heaslip found a dramatic increase in the total number of trips made by this group for medical, religious and social/recreat ional purposes from 1995 to 2001. In addition, strong inclination of the elderly towards the personal vehicles was found in this study. Besides this, older women were found to be less mobile in some studies (Collia et al., 2003; Rosenbloom, 1999) but they were more likely to use special transit services than the older men (Col lia et al, 2003). The reasons behind these issues were not explored clearly in those studies. One Australia based research (Alsnih and Hensher, 2003) urged elderly for t he better strategic plan of public transportation. They also focused on researching different groups (young and old) of elderly to understand the threshold of health change and the different needs of these two groups for the better implementation of the tr ansportation related plan and policy. On the other hand, driving characteristics of the older drivers are different from those of their younger counterparts and so, already address ed the attention of the researchers Chu (1994) investigated the mobility i ssue of the elderly (65 or more) driver by exploring the effects of age on six different driving habits: daily driving exposure, driving by time of the day and type of the roadway, vehicle size, number of passengers carried and driving speed by using the 1990 Nationwide Personal Transportation Survey (NPTS) He identified that the elderly made almost the same number of trips as their counterparts did but the total vehicle miles traveled (VMT) declined due to the shorter

PAGE 25

15 distances of those trips. He also found that older drivers showed a good self protection effort in their driving habits. Georggi and Pendy ala (2001) explored the long distance travel (trips greater than 100 mi ) behavior of the elderly and the low income group by utilizing the 1995 Americ an Travel Survey. They found that the elderly and the low income group made significantly fewer long distance trips than their counterparts. Also, when they traveled, they were more likely to travel by bus and the trip purposes were most likely to be socia l and personal business activities. They also identified the different travel patterns of the older elderly (aged 75 years and above) as compared to the individuals aged below 75 years and so, urged on further study on the travel pattern s of the elderly to explore the reasons behind their low mobility. Mallett (2001) also examined the long distance travel behavior of the low income households with special attention on elderly (65 or older) and children by using 1995 American Travel Survey. Mallett found a n egative relationship between the age of the persons and the number of per capita trips for the elderly group aged 65 and above. The effect was more pronounced among the older elderly aged 85 and above. Benekohal et al. (1994) examined the travel behavior of the elderly (65 or older) by using the data collected through a stateside survey of older drivers combined with the focus group meetings. They found that average vehicle trip length was negatively associated with the age of the elderly. Polzin et al. (2 001) explored the mode choice of minority population for non work travel by using data from the 1983, 1990, and 1995 National Personal Transportation Survey (NPTS) databases. They found that African American were more likely to use public transit for their non work travel as compared to

PAGE 26

16 other groups in the total population While Polzin et al. (2001) focused on whole minority population for non work travel mode choice, Rosenbloom and Waldorf (2001) considered only the elderly people in their mode choice mod el to explore the effects of race, ethnicity and residential location (Urban, Suburban, Second City and Town) on the mode choice of the elderly. It was found that minorities, non Hispanic and the residents in urban core areas were more likely to use public transportation as compared to Privately Owned Vehicle. 2.3 Application of Modeling T echniques in Elderly Travel Behavior Some of th e studies mentioned in section 2.2 used modeling techniques while analyzing the elderly travel pattern. This section provi des a brief overview of those modeling techniques. Evan s (2001 ) estimated four models (stepwise discriminant analysis) to explore the effects of personal and community characteristics associated with trip making propensity of the elderly non drivers aged 7 5and above Of these four models, first one identified the characteristics associated with mobility of the elderly and whether the person had gone out or not on the travel day The estimates indicated that the individuals wit h higher levels of education who owned home and who lived in an apartment in higher housing density areas were more likely to make at least one trip on the travel day. On the other hand, age of the individu als, household size, higher concentration of retail employment were found to be negatively associated with the trip making propensity of the non driving elderly aged 75 and above. In the second model, Evan s explored the factors associated with the transit availability. He found a positive relationship between the public transportation availability and some socio demographic

PAGE 27

17 factors such as age, female and African American. Also a negative correlation was found between the transit availability and the rural areas. The relationship between different factors and the transit use was explored in the third model. Only the individuals who had available public transportation were considered in this model. The results showed the positive effects of housing density, u rban areas, education, income and negative effects of age, household size, race (white) and detached house on the transit use. Finally a mode choice model was estimated to identify the factors associated with the m odal choice of the elderly aged 75 and ab ove It was found that this group of elderly was more likely to depend on transit for their daily trips though they were less likely to go out more often. Kim and Ulfarsson (2004) estimated a mode choice model of the retired elderly aged 65 and above with an aim to fill up the gap of application of modeling techniques in the literature related to elderly travel pattern. They used a multinomial logit model with four mode choice alternatives: private car or truck, carpool or vanpoo l, public transit and walk to identify the personal, household, neighborhood environment, trip characteristics and activity purp ose associated with the mode choice of the elderly. Some of the important variables in the mode choice model such as in vehicle and out of vehicle travel t ime, out of pocket cost were not considered in this model due to lack of data. In addition physical and mental abilities of the individuals which are important in the elderly related research were not included in the model. They found that mode choices of the elderly varied with the trip purposes and the distance to the nearest bus stop. Rosenbloom and Waldorf (2001) also estimated two logit models to explore the effects of race, ethnicity and residential location on the mode choice of the elderly. In this study, mode choice was limited to Privately Owned Vehicle (POV) and Public Transit. In the

PAGE 28

18 first logit model, they estimated the effects of race, ethnicity and residential location on the likelihood of choosing POV as a travel mode while in the second mod el, they estimated the effects of those socio demographics characteristics on the likelihood of choosing p ublic transportation as a mode during travel. They found that people of color, non Hispanic and the residents in urban area were more likely to use pu blic transportation as compared to privately Owned Vehicle. Georggi and Pendyala (2001) estimated linear regression models of trip generation for the elderly and the low income group people to explore the differences in the likelihood of making long distan ce trip among these two groups. They found that income, vehicle ownership, education level, employment and marital status positively affected trip generation and household size, single parent household types, African American were negatively associated wit h the trip generation. In addition, they also computed trip generation elasticity to explore the effe cts of vehicle ownership and income on the trip generation by different age groups. From this elasticity computation, it was found that vehicle ownership a nd income were positively associated with the trip making propensity of all the age groups except the older elderly (75 and above) and the possible reasons identified for this exception wa s the age related limitations of this older age group individuals. Giuliano (1999) estimated three models to explore the effects of land use on the travel patterns of the elderly and to examine whether these effects vary across the different age groups. Of these three, first one was a simple binary logit model to identify the effects of land use on trip making propensity on the travel day. It was found that age negatively affected the trip making propensity and this propensity was independent of the

PAGE 29

19 type of land use. Then the effects of land use on the daily travel distanc e were estimated by regression models. It was found that elderly were more likely to make shorter trips than their counterparts and the neighborhood characteristics had a little effect on the travel distance of all age groups. After this, a binary logit mo del was estimated to explore the effects of land use on the transit use by the elderly. It was found that elderly were less likely to use transit even when it was accessible and land use pattern supported it. In summary, the above reviews give an overall i dea about the importance of the elderly mobility issues in the transportation sector. But, most of the studies in the literature were based on descriptive statistics and emphasized on different trip characteristics such as number of trips, trip distance an d travel time associated with the trips of the elderly. Only a few studies (Evans, 2001; Giuliano, 1999) focused on the travel preferences of the elderly. But, these were limited to only 1 day travel period that means whether the indivi duals traveled on a given reporting day of the survey or not. It is likely that the travel preferences of the elderly may vary even within a week and there might be some individuals who want to be mobile and some who are completely inactive. Such variati on on the elderly trav el preferences when not considered, may affect the final outcome of the research. T his thesis aims to fill up the gap in the literature by considering the travel pre ferences of the e lderly for longer time periods such as a week

PAGE 30

20 CHAPTER 3 DATA DES CRIPTION 3.1 Introduction Travel surveys are the main sources of information that transportation planners and travel behavior analyst s need for their interests. Among these surveys, the NHTS is a national comprehensive survey of both daily and long dist ance travel that provides an opportunity for res earchers to analyze the travel pattern s of Americans. Since 1969, the addition the latest NHTS (2009), because of its larger sample size enables researchers to further analyze the different issues related to daily travel. The next sections of this chapter provide an overview of the 2009 NHTS data and the key household, person, and trip characteristics of Americans, with special focus on the elderly to explore the factors affecting the travel patterns and preferences of older Americans. 3.2 National Household Travel Survey The NHTS is a c omprehensive travel survey that collects information in the daily and long distance tr avel of Americans and began in 2001 Before 2001, th is travel information w as collected through two different surveys : the National Personal Transportation Survey (NPTS) and the American Travel Survey (ATS). The dataset used in this study wa s obtained from the 2009 NHTS conducted from April 2008 to May 2009

PAGE 31

21 Overall the 2009 NHTS sample consists of 150,147 households with 324 ,184 persons, 309,163 vehicles and 1,167,321 daily trips made by those individuals. These larger sample sizes of households and the corresponding persons, vehicles and trips provide a great opportunity for researchers to explore different issues that were not possible with the previous datasets of the NHTS series due to their smaller sample sizes. There are several different stages of NHTS data collection F irst a stratified random telephone number wa s obtained and screened to identify residential households. P eople living in college dormitories, nursing homes, other medical institutions, prisons and on military bases were excluded f rom the sample. Next, a member of each household was called and asked a series of questions about the number of persons and vehicles in the household. Following this interview, a travel diary wa s sent to the household to allow them to keep trip information of the travel day assigned for that household. Following the travel day, each eligible person in the household wa s interviewed for travel day trip information. One of the main aspects of the 2009 NHTS survey that makes it different f ro m the 2001 NHTS is t hat travel day trip information wa s collected on ly for persons age 5 or older ; t he 2001 NHTS contains travel information of the person s below 5 years of age The NHTS is the only data set available at the national level that provides information about the demographics of households, household members, vehicles owned by households and detailed trip information for household members. R esearchers in academics, consult ing, and government use this data set extensively for different purposes such as to explore the relationship between demographics and travel behavior, to quantify travel behavior and to analyze the change in travel characteristics over time.

PAGE 32

22 In the 2009 NHTS questionnaire, several new questions we re included to aid researchers and other NHTS use rs in their respective interests. For example, if a person did not make any trips on the given reporting day of the survey and reported that he/she had stayed at another This study focuses on the question s mentioned above and persons age 6 5 or older to analyze the mobility patterns and preferences o f the elderly. 3.3 Descriptive from the 2009 National Household Travel Survey This section gives a brief overview of the 2009 NHTS sample used in the analysis of this study. As this study is intend ed to analyze the travel patterns and preferences of a sp ecific socioeconomic group the elderly this overview provide s descriptive statistics for the socio demographic characte ristics of the 2009 NHTS with special focus on that particular group All descriptive statistics presented in this study we re obtai ne d from the respectively. (Note: The Federal Highway Administration has decided to enhance the weigh ts of the 2009 NHTS, so these results may change with the new weights). The next subsections provide a descriptive analysis of household, person, and trip characteristics from the 2009 NHTS. 3.3.1 Househol d Characteristics Table 3.1 provides key descriptiv e information pertaining to the socio demographic characteristics of the households. The 1 st column of th e table shows the

PAGE 33

23 characteristics and the 2 nd column gives the share of the total household against those characteristics. The 3 rd and 4 th columns giv e the share of the households with elderly ( above 65 year s of age ) and without elderly (below 65 years of age ) corresponding to the characteristics mentioned in the 1 st column. Among the total 112 520 151 households, 25 583 764 households (22.74%) have at least one elderly person age 65 and above and the remaining 86 936 388 households (77.26%) do not have any elderly persons in their house Almost one quarter of the households in the United States have at least one elderly person in the house. The charact eristics of these households are considered separately from the non elderly households in this section. The average household size in the United States as a whole is 2.34, whereas for elderly and non elderly households, it is 1.83 and 2.70, respectively. T he reason for this lower average elderly household size is the higher percentages of one and two person households. Almost 42 percent of elderly households are one person households, and 47 percent are two person households. A higher percentage of the eld erly are living alone, which should be considered seriously in mitigating the mobility needs of the elderly. Even when they live with other non elderly, most of them live in two person households. Therefore, the composition o f two person households also sh ou ld be considered to determine how the age of the other household members affects the travel behavior of the elderly. The average number of children (<18 years) in households with elderly is 0.05, which is lower than that of the households without elderly (0.65). The life cycle composition of elderly households seems t o be different than that of non elderly households. When the number of elderly in t he household is considered, it wa s found that 69.2 percent are one elderly household and 30.7 percent are tw o or more elderly

PAGE 34

24 Table 3.1 Household Ch aracteristics of the 2009 NHTS D ata Characteristics All Households Household s with Elderly Household s w ithout Elderly Sample Size Weighted Households 150,147 112,520,151 62,405 25,583,764 87,742 86,936,388 H ousehold Size 1 Person 2 Person 3 Person 4 or more Persons 2.34 24.5% 36.3% 15.8% 23.4% 1.83 41.7% 47.0% 6.7% 4.7% 2.70 19.4% 33.1% 18.6% 28.9% No. of Children (Under 18) 0 Children 1 Children 2 Children 3+ Children 0.40 67.9% 14.6% 11.8% 5.8% 0.05 95.6% 2.6% 1.2% 0.6% 0.65 59.7% 18.1% 14.9% 7.3% No. of Elderly 0 Elderly 1 E lderly 2+ E lderly 77.3% 15.7% 7.0% NA 69.2% 30.7% NA NA NA No. of Workers 0 W orkers 1 W orkers 2 W orkers 3 or more W orkers 0.93 28.9% 40.4% 26.3% 4.5% 0.39 69.3% 23.3% 6.4% 1.0% 1.31 17.0% 45.4% 32.1% 5.5% No. of Drivers 0 D river 1 D river 2 D river 3 or more D river 1.80 4.9% 30.6% 51.6% 12.9% 1.56 11.3% 42.8% 40.3% 5.6% 1.96 3.0% 27.0% 54.9% 15.1% Annual Income < $ 25 K $ 25 K $ 50 K $ 5 1 K $7 5 K > $ 75K 25.4% 25.8% 16.6% 32.1% 40.5% 33.1% 12.2% 14.3% 21.3% 23.8% 17.9% 37.0% Vehicle Ownership 0 Vehicle 1 Vehicle 2 Vehicle 3 or more Vehicles 2.05 8.8% 28.9% 37.7% 24.6% 1.72 14.3% 41.7% 30 .7% 13.3% 2.28 7.2% 25.1% 39.8% 27.9% Dwelling Unit Type Detached Single House Duplex Row /Town House Other 66.2% 7.3% 22.0% 4.5% 63.6% 6.3% 25.0% 5.2% 66.9% 7.6% 21.1% 4.4% Residential Area Type Urban Rural 77.2% 22.8% 79.0% 20.9% 76.6% 23.4%

PAGE 35

25 households. The elderly in the second category (two or more elderly) ha ve the advantage of having the companionship of persons of the same age group; this may affect their travel behavior The average number of workers in households wi th elderly is significantly lower than that of households with non elderly (0.39 vs. 1.31 ). About one quarter of the $ 25 K at an aggregate l evel, which is lower than the households with elderly group. In this grou p, almost 40.5 percent of household s fall into the lower income category ( < $ 25K) whereas for the households without elderly it is 21.3 percent. A verage vehicle ownership is lower for households with elderly (1.72) compared to households without elderly ( 2.28) and h ouseholds in the U.S. as a whole (2.05). Average vehicle ownership and the income category variable indicate that most of Interestingly, the average number of drivers in the household is 1.80 which is less than the average number of vehicles per households (2.05) for U.S. as a whole. The same trend goes for both of the household groups: wit h elderly and without elderly. At an aggregate level i.e., for the U.S. as a whole about 66 per cent of households live in a detached single ho me, which is ve ry close to the percentage of households with elderly (63.6 percent ) and households without elderly (66.9 percent ) living in a detached s ingle home. The same goes for households living in a r o w h ouse/ t own h ouse So, there is no high variation among the households with elderly and without elderly by housing unit types. About 79 percent of households that have at least one elderly are in an urban residential location and the rest are in a rural l ocation

PAGE 36

26 As mentioned earlier, almost 42 percent of elderly households are one person households, and 47 percent are two person households. The characteristics of these households are shown in Table 3.2. Of the two person households, only 0.5 percent ha s children (below 18 years of age) and 58 percent has only elderly member (above 65 Table 3.2 Elderly Household (One Person and Two Person) Characteristics Characteristics One P erson HH Two P erson HH No. of Children (Under 18) 0 Children 1 Children 2 Children 3+Children 100% NA NA NA 99.5% 0.5% NA NA No. of Elderly 1 E lderly 2 E lderly 100% NA 42.0% 58.0% No. of Workers 0 W orkers 1 W orkers 2 W orkers 3 or more W orkers 88.0% 12.0% NA NA 64.1% 28.5% 7.4 NA No. of Drivers 0 D river 1 D river 2 D river 3 or more D river 22.8% 77.2% NA NA 3.5% 20.0% 76.5% NA Annual Income < $ 25 K $ 25 K $50 K $ 51 K $75 K > $ 75K 60.1% 28.1% 6.9% 4.9% 26.4% 38.1% 15.5% 19.9% Vehicle Ownership 0 Vehicle 1 Vehicle 2 Vehicle 3 or more Vehicles 26.8% 63.4% 8.2% 1.6 % 5.0% 28.7% 50.0% 16.3 % Dwelling Unit Type Detached Single House Duplex Row /Town House Other 47.3% 6.1% 40.2% 6.4% 75.4% 5.7% 14.6% 4.3% Residential Are a Type Urban Rural 83.3% 16.7% 75.4% 24.6%

PAGE 37

27 years of age) Moreover, it appears that the percentage s of 0 drivers, low income ( < $ 25 K ) and 0 vehicle households are significantly higher in the one person households. It indicates that the elderl y in one person households i.e. the elderly who live alone has several b arriers in making trips. I t is clear from this section that the household characteristics of the elderly are different from those of the non elderly. 3.3.2 Person Characteristics Table 3.3 gives an overview of the person socio demographic characteristics of the 2009 NHTS data. The 1 st column shows the characteristics and the 3 rd and 4 th columns show the share of elderly (age 65 and above) and the non elderly (age below 65 years) corres ponding to the characteristics in the 1 st column. Among the 299,801,601 persons in the United States as a whole 37,850,918 (12.63 % ) are elderly (age 65 years and above), and 261,950,683 (87.37 % ) are non elderly (age below 65 years). Since the elderly a re almost 13 percent of the total population and, in general, their personal characteristics are different from the non elderly, they are considered separately in this section. From T able 3.3 the percentage of female s is higher than the p ercentage of male s (57.8 % vs.42.2 % ) among the elderly, whereas for the non elderly these percentage s are almost equal (49.7 % vs.50.3 % ). This indicates the gender disparity among the elderly. From the age variable, it appears that a bout 53 percent of the elderly are in the 65 74 years of age. Although age is considered to be one of the barriers in the elderly mobility issue, it seems that the percentage of young (65 74years) and middle (75 84 years) elderly are higher than that of the older elderly (>= 85 years). T his lower percentage of older elderly is a good news in mitigating the mobility issues of the elderly because

PAGE 38

28 Table 3.3 Person Ch aracteristics of the 2009 NHTS D ata Characteristics All Persons Elderly Non Elderly Sample Size Weighted Population 324 1 84 299 801 601 86 113 37 850 918 238 071 261 950 683 Gender Male Female 49.2% 50.8% 42.2% 57.8% 50.3% 49.7% Age 0 5 years 6 15 years 16 25 y ears 26 64y ears 65 74 years 75 84 years Greater than 85 years 7.1% 13.8% 13.7% 52.8% 6.7% 4.5% 1.5% NA NA NA NA 52.7% 35.5% 11.8% 8.1% 15.8% 15.6% 60.5% NA NA NA Race White African American Other 72.5% 12.3% 15.2% 80.4% 11.8% 7.8% 71.4% 12.3% 16.3% Hispanic Status Hispanic Not Hispanic 15 .0 % 85 .0 % 7.5% 92.5% 16.1% 83.9% Worker Yes, a worker No, not a w orker 59.9% 40.1% 15.3% 84.7% 68.3% 31.7% Highest Education Level High S chool/Less Some College College Graduate Post Graduate 40.1% 28.1% 18.7% 13.1% 51.7% 24.0% 13.0% 1 1.3% 37.8% 28.9% 19.8% 13.4% Driver Status Driver Not a Driver 87.0% 13.0% 80.0% 20.0% 88.2% 11.8% Daily Travel Average Person Trips /day Average V ehicle Trips /day Average Person M ile s /day Average Vehicle M ile s /day 3.77 2.21 3 7.18 22.12 3.18 2.08 26.87 16.31 3.96 2.29 41.10 26.17

PAGE 39

29 the lower the age of the elderly person the easier to include him/her in different plans and policies. The percentage of W hite s (80.4%) among the elderly is higher than that among the non elderly ( 71.4%) and among the persons ( 72.5 %) at an aggregate level. The percentage of non workers among the elderly is 84.7, which is significantly higher than the percentages among the non eld erly (31.7%) and among the persons for the U.S. as a whole (40.1%) In addi tion, a little more than half of elderly people exhibit a lower level of education when compared with the other two groups. This should be seriously considered in mitigating the mobility needs of the elderly because lack of education may hamper the pu rsuit of many types of activities (Lefrancois et al., 1998). Older people are more likely to be dependent on the private car (Rosenbloom, 1999). The mobility of older persons is closely related to thei r driving status. From Table 3.3 it can be seen that t he percentage of non drivers is higher among the elderly as compared to the non elderly (20.0 % vs. 11.8 %). This requires special attention in elderly related research. On an average, each person makes 3.77 person trips and 2.21 vehicle trips per day at a n aggregate level whereas for the elderly these are 3.18 and 2.08 and for non elderly, 3.96 and 2.29, respectively. The same trend goes for average person miles and vehicle miles traveled per day by these groups. Th is indicate s that older people make fewe r trips and travel shorter distances as compared to their younger counterparts (Collia et al., 2003; Heaslip, 2007) Tabl e 3.4 shows the characteristics of the individuals from one person and two person households that have at least one elderly It seems that the percentage of female s is significantly higher than the percentage of male s among the elderly who live alone Moreover, when the elderly live with other non elderly member, they are more likely to

PAGE 40

30 Table 3.4 Person Characteristics (One Person and Two P erson Household s ) Characteristics One P erson Household s Two P erson Household s Elderly Elderly Non Elderly Gender Male Female 25.1 % 74.9 % 51.8 % 48.2 % 27.0 % 73.0 % Age 0 5 years 6 15 years 16 25 years 26 64years 65 74 years 75 84 years Greater than 85 years NA NA NA NA 40.4 % 43.2 % 16.4 % NA NA NA NA 59.4 % 33.5 % 7.1 % 0.1 % 0.6 % 2.2 % 97.1 % NA NA NA Worker Yes, a worker No, not a worker 12.2 % 87.8 % 16.2 % 83.8 % 49.1 % 50.9 % Highest Education Level High School/Less Some College College Graduate Post Graduate 46.9 % 28.9 % 13.3 % 10.9 % 43.2 % 25.5 % 17.0 % 14.3 % 38.4 % 29.5 % 17.8 % 14.3 % Driver Status Driver Not a Driver 85.1 % 14.9 % 90.2 % 9.8 % 94.1 % 5.9 % live with the persons age 26 65 years. This may be because the elderly are likely to depend on the middle age person for their daily needs (shopping, social meeting etc.). There is no high variation among the elderly by worker status. Among the non elderly in two person households the percentage of workers and non workers are almost equal. From the education variable, it seems that the e lderly in both types of households exhibit a lower educat ion level. Also, the percentage of non drivers is higher among the elderly especially who live alone. The next subsection provides a clear picture of the travel patterns and trip characteristics of Americans using 2009 NHTS data.

PAGE 41

31 3.3.3 Trip Characteristics Travel patterns vary among the elderly and the non elderly populations. Variation is f ound even among different groups of elderly. The main intent of this section is to give an overview of the travel patterns and trip characteristics of Americans with special focus on older adults age 65 and above. When a trip is made by a person, it can b e either as a passenger or as a driver of a vehicle. If the trip is made by a person as a driver of a privately operated vehicle, it is trips made by different modes of transportation. This section focuses mainly on person trips and the characteristics associated with those trips. Among the various characteristics of the trip, four characteristics average daily person trips, average daily person miles, average person trip length, and average person trip travel time are considered with special attention in this section. In addition, the total population is divided into three main categories: children (age 5 18 years), young adults (age 19 64 years) and older adul ts (age 65 years and above). To understand the travel behavior of the elderly more clearly, older adults are divided into three subcategories: young elderly (age 65 74 years), middle elderly (age 75 erms will be used in the remainder of the thesis. The total number of trips and the trip characteristics associated with those trips varies greatly with the age of the person making the trip. As shown in Table 3. 5 persons age 19 64 make almost 70.30 pe rcent of the total trips, and this trip percentage decreases significantly with the increase of age, especially after 64 years. When gender is cons idered along with age (Table 3.6 ), it is found that the percentage of trips by female is

PAGE 42

32 higher than those by male s in all age cohorts except children. It is important to note that the difference in the trip percentages between male and female is higher among the older elderly as compared to other age cohorts. This gender disparity in the trip characteristics of the elderly should be considered with special attention in elderly travel behavior Table 3.5 Trip Distribution by Age Table 3.6 Trip Distribution by Age and Gender research. Table 3.7 shows the variation in different trip characteristics such as average daily person trips, average daily person miles, average person trip length (miles) and average person trip travel time (minutes) by different age groups and gender. It was found that the average daily person trips of female older adults are lower than of male older adults, but the pattern i s the opposite for the children and young adul t group. This comparison of male and female average daily person trips by age is shown in Figure 3.1. Characteristics Number of Trips Percent 5 18 Years 71,461,997,508 18.40 19 64 Year s 273,500,704,534 70.30 65 74 Years 26,346 ,070,424 6.80 75 84 Years 14,351,731,207 3.70 85 Years 31,966,353,04 0.80 All Ages 388,857,138,977 100.00 Characteristics Male Female Total 5 18 Years 51.0 49.00 100.00 19 64 Year s 48.40 51.60 100.00 65 74 Years 46.90 53.10 100.00 75 84 Year s 46.60 53.40 100.00 85 Years 42.60 57.40 100.00 All Ages 48.60 51.40 100.00

PAGE 43

33 Table 3.7 Trip Characteristics by Age and Gender Figure 3 .1 Average Daily Person Trips by Age and Gender 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5 18 19 64 65 74 75 84 >=85 Average Daily Person Trips Age Category Male Female Characteristics Male Female Average Person Trips per day 5 18 Years 19 64 Years 65 74 Years 75 84 Years 85 Years 3.18 3.96 3.82 3.35 2.27 3.30 4.19 3.45 2.63 1.79 Average Person Miles Per day 5 18 Years 19 64 Years 65 74 Years 75 84 Years 85 Years 27.25 48.72 37.54 30.65 13.50 24.89 37.21 29.01 18.33 9.58 Average Person Trip Length (miles) 5 18 Years 19 64 Years 65 74 Years 75 84 Years >= 85 Years 8.83 12.52 9.97 9.26 6.06 7.80 9.16 8.62 7.32 5.75 Average Person Trip Travel Time(minutes) 5 18 Years 19 64 Years 65 74 Years 75 84 Years 85 Years 19.78 22.33 20.89 20.07 17.97 18.83 19.30 18.87 18.64 17.87

PAGE 44

34 It appears that as age increases, average daily person trips of older adults decreases, and the difference between male a nd female person trips is found to increase with the increase in age of older adult s. Interestingly, from Table 3.7 it was found that although females from two of the age groups (5 18 years and 19 64 years) have higher average person trips per day tha n males, the average trip characteristics such as person miles per day, person trip length, and person trip travel time of females are lower than those of males in all age cohorts. This reveals the shorter distance trip making tendency for females as compa red to males (Collia et al. 2003). In addition, all four trip charac teristics mentioned in Table 3.7 were found to decrease with increase in age among older adults, irrespective of gender. Therefore, trip characteristics of older adults vary from their yo unger counterparts. Trip making propensity and the trip characteristics associated with those trips depend on different personal characteristics such as driver status and worker status; household characteristics such as race, household income, household ve hicle ownership, etc.; and trip purposes and mode of transportation used for the trips. Table 3 .8 presents the percentage of drivers and non drivers by different age groups and gender, and Figure 3.2 provides a snapshot of the percentages of drivers and no n drivers among older adults. From the distribution of older adults, it was found that the percentage of non drivers is higher among the older elderly compared to other age cohorts. This pattern is more pronounced for the female older elderly. About 58 per cent of females in this group are non drivers, which requires special attention. This driver status of older women may explain the reason for the lower average daily person trips of the ol der women mentioned in Table 3.7 Figure 3.3 depicts the mode use pa ttern by the older adults from zero

PAGE 45

35 vehicle households. It seems that while the tendency to use a privately owned vehicle (POV) increases with the age of older adults, the propensity to walk and use public transportation decreases with age. Interestingly, individuals in the 65 74 age cohort and living in zero vehicle households are more likely to walk than use a POV for their daily Table 3.8 Distribution of Drivers and Non Drivers by Age and Gender Figure 3 2 Distribution of Drivers and Non Drivers by Age and Gender 0 10 20 30 40 50 60 70 80 90 100 65 74 75 84 65 74 75 84 Male Female Percent Age and Gender Category Driver Non Driver Characteristics Driver Non Driver Total Male 19 64 Years 65 74 Years 75 84 Years 85 Years All Ages 93.3 92.7 87.5 69.7 76.2 6.7 7.3 12.5 30.3 8.9 100.0 100.0 100.0 100.0 100.0 Female 19 64 Years 65 74 Years 75 84 Years 85 Years All Ages 90.0 84.3 70.3 42.2 72.8 9.9 15.7 29.6 57.8 13.3 100.0 100.0 100.0 100.0 100.0

PAGE 46

36 Figure 3 .3 Distribution of Person Trip s by Age and Mode of Transportation (For Zero Veh icle Household) trips. If the mode of travel is considered along with age and driver status (Table 3.9), it seems that POV is the dominant mode of travel for both the driver and non driver groups Table 3 9 Distribution of Person Trip s by Age, Driver Status and Mode for daily travel. As expected, the tendency of using a POV is higher among drivers compared to non drivers. It also appears that walking and using public transportation are the 2 nd and 3 rd dominant modes, res pectively, and the percentages of walk and public transportation trips made by non drivers are higher than those of drivers. Also, while the 0 10 20 30 40 50 60 POV Public Transportaiton Bike Walk Other Percent of Dailiy Trips Mode of Transportation 65 74 75 84 Characteristics POV Public Transportation Bike Walk Other Total Driver 19 64 Years 65 74 Years 75 84 Years 85 Years All ages 86.70 89.00 90.70 89.10 86.80 2.20 1.40 1.20 1.10 2.20 0.70 0.50 0.50 0.10 0.70 9.40 8.20 6.90 7.70 9.20 1.00 0.80 0.70 2.00 1.20 100.00 100.00 100.00 100.00 100.00 Non Driver 19 64 Years 65 74 Years 75 84 Years 85 Years All ages 54.60 59.40 68.80 73.50 59.80 5.40 5.00 2.20 0.60 4.00 2.10 0.80 0.10 0.10 1.80 31.00 23.90 19.50 15.50 25.20 6.80 11.00 9.40 10.30 9.20 100.00 100.00 100.00 100.00 100.00

PAGE 47

37 tendency to use a POV increases with the age, the tendency to use public transportation and to walk decreases with age. Although the same trend goes for both drivers and non driver the rate of increase or decrease is higher among non drivers. When elderly non drivers are considered, they are more likely to use a POV compared to their younger counterparts. The highest p ercentage using a POV (73.5%) is found for the older elderly. In other words, dependency on a POV increases with age among the older elderly. On the other hand, a substantial subset of the elderly is non drivers (Table 3.8). So it is important develop a special transportation system for the elderly that can reduce their dependency on POVs and keep them active. Table 3. 10 shows the variation in trip characteristics by age, gender, and driver status. It was found that all average trip characteristics are lower Table 3 10 Trip Characteristics by Age, Driver Status and Gender Characteristics Driver Non Driver Male Female Male Female Average Person Trips per D ay 19 64 Yea rs 65 74 Years 75 84 Years 85 Years 4.06 3.98 3.61 2.71 4.37 3.81 3.07 2.66 2.47 1.82 1.60 1.24 2.57 1.56 1.60 1.15 Average Person Miles p er D ay 19 64 Years 65 74 Years 75 84 Years 85 Years 51.02 39.69 33.21 16.39 39.19 32.85 22.17 14.11 16.64 10.18 12.84 6.86 19.50 8.40 9.22 6.26 Average Person Trip Length 19 64 Years 65 74 Years 75 84 Years 85 Years 12.76 10.12 9.30 6.15 9.20 8.81 7.51 5.55 7.03 5.83 8.48 5.63 8.59 5 .93 6.38 6.13 Average Person Trip Travel Time 19 64 Years 65 74 Years 75 84 Years 85 Years 22.33 20.11 19.91 17.23 19.00 18.73 18.26 16.73 22.42 42.61 22.55 21.73 23.85 20.83 20.39 19.83

PAGE 48

38 for non drivers compared to drivers, except for average trip travel time. Average trip travel time is higher for no n drivers. Also, the difference between the average person trip travel time of male elderly drivers and non drivers is higher than that of their younger counterparts. Table 3 .11 presents the distribution of daily person trips by age, race, and mode of tra nsportation, and Figure 3.4 depicts the variation in average daily person trips by age and race. From the distribution in Table3.11 it was found that although African are less likely to use a POV for their daily travel compared to Whites, the propensity to use walk and use public transportation is higher among this group compared to Whites. Table 3.11 Distribution of Person Trip s by Age, Race and Mode of Transportation Characteristics POV Public Transportation Bike Walk Other Total White 5 18 Yea rs 19 64 Years 65 74 Years 75 84 Years 85 Years All Ages 75.40 87.50 88.80 89.50 85.80 85.50 0.60 1.10 0.80 0.90 0.60 1.00 2.60 0.70 0.60 0.50 0.10 1.00 11.10 9.70 8.70 7.70 9.80 9.80 10.20 1.10 1.10 1.40 3.70 2.70 100.00 100.00 100.00 100.00 100.00 100.00 Black 5 18 Year s 19 64 Years 65 74 Years 75 84 Years 85 Years All Ages 58.40 75.60 82.00 76.10 76.40 72.20 4.90 8.20 4.60 5.70 3.10 7.20 3.70 0.80 0.30 0.10 0.00 1.40 21.00 12.80 9.70 12.80 6.60 14.40 12.00 2.60 3.40 5.40 14.00 4. 80 100.00 100.00 100.00 100.00 100.00 100.00 Other 5 18 Years 19 64 Years 65 74 Years 75 84 Years 85 Years All Ages 64.00 80.70 78.50 78.30 79.40 77.00 4.70 4.20 7.40 3.10 4.20 4.40 1.60 0.70 0.50 0. 10 0.30 0.90 20.30 12.70 11.80 16.10 12.70 14.40 9.40 1.60 1.90 2.40 3.40 3.40 100.00 100.00 100.00 100.00 100.00 100.00

PAGE 49

39 Figure 3 4 Average Daily Person Trips by Age and Race This may be one of the main reasons for higher average travel times for African Americans in the Table 3.12 In other words, it is their dependency on transit and walking that forces them to travel for a longer time. In addition, Table 3.12 shows that the average daily person trips and person miles of Whites are higher than th ose of African Americans, and the pattern is the same for all age groups If only older adults are considered, the difference in the average daily person trips of Whites and African Americans is higher than that of their younger counterparts. Vehicle owne rship plays an important role in the trip characteristics of household members. As shown in Table 3.13 walking and using public transportation are more likely to be used by persons from zero vehicle households compared to persons from one vehicle or two o r more vehicles households. Figure 3 .5 depicts this difference in trip distribution by vehicle ownership and mode of transportation. It appears that t he mode use patterns of persons from zero vehicle households are different from persons of other types of households In addition, within each type of households categorized by the 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5 18 years 19 64 65 74 75 84 >=85 Avearge Daily Person Trips Age Category white black Other

PAGE 50

40 number of vehicles as shown in Table 3.1 3 it appears that the likelihood to use a POV increases and the propensity to walk decreases with the increase in age. Even within zero vehi cle households, the propensity to use public transit and walk appears to decrease with age. Table 3.12 Trip C haracteristics by Age and Race Characteristics White Black Other Average Person Trips per D ay 5 18 Years 19 64 Years 65 74 Years 75 84 Years 85 Years 3.33 4.17 3.80 3.09 2.09 3.09 3.88 2.82 2.01 1.14 3.02 3.80 3.07 2.42 1.79 Average Person Miles p er D ay 5 18 Years 19 64 Years 65 74 Years 75 84 Years 85 Years 29.89 46.25 35.67 24.02 11.89 22.13 35.19 20.55 21.48 6.14 14.70 33.35 22.71 18.37 8.31 Average Person Trip Length (miles) 5 18 Years 19 64 Years 65 74 Years 75 84 Years 85 Years 9.11 11.24 9.50 7.95 5.90 7.55 9.49 7.65 11.66 6.58 5.34 9.47 8.1 2 8.62 5.04 Average Person Trip Travel Time (min) 5 18 Years 19 64 Years 65 74 Years 75 84 Years 85 Years 18.51 19.89 19.17 18.36 17.21 22.09 24.56 23.53 27.81 24.40 20.54 22.24 22.20 23.09 22.12

PAGE 51

41 Table 3.13 Distribution of Person Trip s by Age, Vehicle Ownership, and Mode Characteristics POV Public Transportation Bike Walk Other Total 0 Vehicle 5 18 Years 19 64 Years 65 74 Years 75 84 Years 85 Years All Ages 23.2 23.7 31.8 41.3 50.0 25.7 15.0 25.4 19.9 13.6 5.0 21.9 1.6 2.5 1.0 0.3 0.1 2.0 47.0 42.9 35.2 30.0 26.3 42.1 13.2 5.6 12.1 14.8 18.6 8.3 100.0 100.0 100.0 100.0 100.0 100.0 1 Vehicle 5 18 Years 19 64 Years 65 74 Years 75 84 Years 85 Years All Ages 63.8 80.3 87.1 90.2 89.0 79.4 3.2 3.5 1.4 0.5 0.3 3.0 4.9 0.7 0.7 0.7 0.2 1.3 15.8 13.8 9.7 7.6 7.9 13.1 12.3 1.7 1.1 1.0 2.7 3.2 100.0 100.0 100.0 100.0 100.0 100.0 2 Vehicles 5 18 Years 19 64 Years 65 74 Years 75 84 Years 85 Years All Ages 73.2 88.7 91.9 92.9 93.0 86.2 0.9 0.8 0.2 0.3 0.4 0.8 2.4 0.7 0.4 0.2 0.0 1.0 12.8 8.8 6.8 6.0 5.6 9.3 10.7 1.0 0.7 0.6 0.9 2.8 100.0 100.0 100.0 100.0 100.0 100.0 3 or more Ve hicles 5 18 Years 19 64 Years 65 74 Years 75 84 Years 85 Years All Ages 79.9 91.7 92.9 93.9 91.1 89.4 0.5 0.5 0.0 0.1 0.3 0.5 0.5 0.5 0.0 0.1 0.3 0.5 9.4 6.3 6.1 5.4 7.5 6.9 8.4 1.0 0.4 0.3 1.1 2.5 100.0 1 00.0 100.0 100.0 100.0 100.0

PAGE 52

42 Figure 3 5 Distribution of Trip s by Vehicle Ownership and Mode of Transportation Table 3.14 and Figure 3.6 show that average daily person trips for all age cohorts increase with vehicle ownership of the household except th ose with three or more vehicles. Average daily person miles and average person trip length also increase with the number of vehicles in a household, but average person trip travel time does not follow the same trend. As expected, this is higher for persons from zero vehicle households, and the differences among zero vehicle and other types of household are significant. As shown in Table 3.15 and Figure 3.7 a similar pattern of relationship is found with household income. While average daily person trips, p erson miles, and person trip length increase with the increase in household income, average person trip travel time decreases with the income of households, with a few exceptions in households with income > $75 K and in the elderly groups. It seems that, u p to a certain limit, income and age have a clear effect on average daily person trips. In addition, the effects of low income (< $25 K) on the trip characteristics of older adults are clear from their average daily person trips. These are significantly lo wer compared to other high income groups. Also, within the same income group, average daily person trips of older adults are found to decrease with the increase in 0 10 20 30 40 50 60 70 80 90 100 Percent of Daily Trips Mode of Transportation 0 vehicle 1 vehicle 2 vehicle 3 vehicle

PAGE 53

43 Table 3.14 Trip Characteristics by Age and Vehicle Ownership Figure 3.6 Average Daily Person Trips by Age and Vehicle Ownership 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5 18 19 64 65 74 75 84 >=85 Average Daily Person Trips Age Category 0 vehicle 1 vehicle 2 vehicle 3 + vehicle Characteristics 0 Vehicle 1 Ve hicle 2 Vehicle >= 3 vehicle Average Person Trips per D ay 5 18 Years 19 64 Years 65 74 Years 75 84 Years 85 Years 2.73 3.16 2.24 1.65 1.13 3.04 4.04 3.53 3.04 2.26 3.33 4.20 3.93 3.28 2.19 3.33 4.11 3.72 2.97 2.14 Average Person Miles p er D ay 5 18 Years 19 64 Years 65 74 Years 75 84 Years 85 Years 9.32 12.99 6.85 7.90 4.12 17.73 35.66 27.42 22.95 13.30 25.86 44.01 38.33 28.77 13.87 34.10 51.38 41.31 28.10 10.79 Average Per son Trip Length (mile) 5 18 Years 19 64 Years 65 74 Years 75 84 Years 85 Years 3.76 4.57 3.54 5.58 4.22 6.17 9.19 7.97 7.79 6.08 7.95 10.66 9.84 8.90 6.54 10.43 12.67 11.22 9.60 5.39 Average Person Trip Travel Time 5 18 Years 19 64 Years 65 74 Years 75 84 Years 85 Years 30.96 29.5 25.51 24.18 22.27 19.25 19.81 19.01 18.55 17.07 18.17 19.80 19.28 19.13 18.14 18.96 21.28 20.69 19.99 16.20

PAGE 54

44 Table 3.15 Trip Characteristics by Age and Household Income Figure 3.7 Average Daily Person Trips by Age and Household Income 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00 5 18 years 19 64 65 74 75 84 >=85 Average Daily Person Trips Age Category <25k 25 50k 51 74k >75k Total Characteristics < $ 25K $ 25 50k $ 51 75k > $ 75 k Average Person Trips per D ay 5 18 Years 19 64 Years 65 74 Years 75 84 Years 85 Years 2 .91 3.56 2.83 2.45 1.82 3.07 3.96 3.87 3.29 2.18 3.30 4.15 4.03 3.67 1.95 3.50 4.47 4.35 3.48 2.39 Average Person Miles p er D ay 5 18 Years 19 64 Years 65 74 Years 75 84 Years 85 Years 17.23 30.33 21.01 19.83 10.05 20.95 38.15 33.03 25.19 13.40 24.28 41.20 36.38 28.67 12.07 34.85 54.17 47.79 32.60 11.56 Average Person Trip Length (mile) 5 18 Years 19 64 Years 65 74 Years 75 84 Years 85 Years 6.44 9.10 7.78 8.60 5.77 7.03 9.83 8 .67 7.76 6.35 7.50 10.10 9.08 7.96 6.27 10.04 12.23 11.07 9.45 4.95 Average Person Trip Travel Time 5 18 Years 19 64 Years 65 74 Years 75 84 Years 85 Years 23.26 22.47 21.08 20.21 18.46 18.71 20.15 18.86 18.83 17.66 17.81 19.79 18.88 18.00 16.74 18.25 20.65 20.24 19.32 17.16

PAGE 55

45 age (Mallett [2001] reported such results for long distance travel). In short, Figures 3.6 and 3.7 indicate that older perso ns from zero vehicle and lower income (< $25K) population. So, while focusing on the mobility issues of the elderly, this group (with zero vehicle and income < $25 K) should be carefully considered Table 3.16 shows the distribution of trips by age, mode of transportation, and trip purposes. It appears that POV is the dominant mode for all age groups and for all trip purposes. Interestingly, the percentage of walking is highe st for social/recreational purposes and is applicable to all age groups, even older adults. Tables 3.17 and 3.18 show the variations in trip characteristics for different trip purposes and for different modes of transportation. It seems that older adults a re more likely to make trips for shopping purposes and their average daily person tri ps for this purpose decrease with the increase in age But average trip length and average trip travel time for this purpose are lower compared to other trip purposes. A lso, older elderly are found to have lower average daily person trips for so cial/recreational purposes (0.48 ) compared to other age cohorts. Since the benefit obtained from such trips can help the older elderly in many ways, proper care should be taken to increase the social/recreational trips among this group.

PAGE 56

46 Table 3.16 Distribution of Person Trip s by Age, Mode and Trip Purpose Characteristics To/From work Shopping Family/ Personal Social/ Recreational Other 5 18 Years POV Public Tra nsportation Bike Walk Other Transportation Total 87.10 4.40 0.60 6.90 1.00 100.00 86.50 2.30 0.60 9.70 0.90 100.00 83.90 0.70 1.30 10.80 3.30 100.00 73.40 1.50 6.00 16.30 2.80 100.00 61.90 2.20 1.90 1 5.10 19.00 100.00 19 64 Years POV Public Transportation Bike Walk Other Transportation Total 91.20 4.00 0.80 3.10 1.00 100.00 89.00 1.80 0.50 7.90 0.80 100.00 84.70 1.20 0.30 12.90 0.90 100.00 75.60 1.20 1.50 20.30 1.40 100.00 83.30 3.70 0.70 9.50 2.70 100.00 65 74 Years POV Public Transportation Bike Walk Other Transportation Total 91.80 5.10 0.20 2.40 0.60 100.00 91.50 1.40 0.50 5.80 0.90 100.00 88.90 1. 00 0.20 9.30 0.60 100.00 78.40 0.70 1.30 18.00 1.60 100.00 89.10 2.50 0.20 5.50 2.70 100.00 65 74 Years POV Public Transportation Bike Walk Other Transportation Total 93.80 1.90 0.00 3.20 1.20 100.00 92. 80 1.50 0.40 4.20 1.10 100.00 90.20 0.50 0.40 7.90 1.10 100.00 79.30 1.10 0.80 17.30 1.50 100.00 89.50 1.40 0.20 5.10 3.70 100.00 85 Years POV Public Transportation Bike Walk Other Transportation Total 92.10 0.10 0.00 7.80 0.00 100.00 90.70 0.70 0.00 5.90 2.60 100.00 86.20 0.70 0.10 8.40 4.60 100.00 72.80 1.10 0.30 20.70 5.00 100.00 88. 00 1.30 0.10 5.60 5.10 100.00

PAGE 57

47 Table 3.17 Trip C haracteristics by Age and Trip Purpose For all trip purposes, aver age person trip length is found to decrease with the increase in age among older adults. In addition, from Table 3.18, it was found that daily average Characteristics To/From work Shop ping Family/ Personal Med /DDS Social/ Recreatio n Other Total Average Person Trips per day 5 18 Years 19 64 Years 65 74 Years 75 84 Years 0.08 0.81 0.22 0.07 0.01 0.44 0.89 1.07 0.89 0.62 0.37 0.85 0.72 0.56 0.33 0.05 0.10 0.19 0.20 0.18 0.77 0.74 0.83 0.73 0.48 1.50 0.66 0.58 0.44 0.32 3.20 4.03 3.60 2.89 1.94 Average Person Miles Per day 5 18 Years 19 64 Years 65 74 Years 75 84 Years 0.59 9.77 2.03 0.57 0.09 2.99 5.99 6.84 5.03 2.73 2.54 5.55 5.44 4.15 2.04 0.39 0.94 1.84 1.89 1.11 6.40 6.38 7.05 4.93 2.62 12.29 12.70 8.75 6.03 2.21 25.20 41.33 31.93 22.59 10.80 Average Person Trip Length (miles) 5 18 Years 19 64 Years 65 74 Years 75 84 Years 7.31 12.38 9.65 8.69 7.24 7.13 6.88 6.52 5.78 4.58 7.14 6.69 7.65 7.53 6.25 9.03 10.1 10.1 9.81 6.84 8.53 8.84 8.66 6.98 5.83 8.43 19.89 15.67 14.14 7.14 8.32 10.8 9.26 8.24 5.89 Average Person Trip Travel Time (minutes) 5 18 Years 19 64 Years 65 74 Years 75 84 Years ars 16.57 24.86 22.10 20.7 0 25.77 17.54 15.88 15.96 16.08 15.18 16.12 15.46 16.96 17.24 17.27 21.7 23.4 25.1 24.1 20.6 19.52 19.60 21.48 19.82 19.39 20.26 28.67 24.22 23.73 19.28 19.3 20.8 19.8 19.3 17.9

PAGE 58

48 Table 3.18 Trip C haracteristics by Age and Mode of Transportation person trips by public transportation are lowest among the older elderly (0.02 ). Also, public transportation is found to have the longest overall average travel time. Table 3.19 shows that the average daily person miles of male workers and non workers are higher than workers and non workers in the female cohor t. It indicates that worker status does not change the shorter trip making propensity of females. In addition, the difference Characteristics POV Public Transporta tion Bike Walk Other Average Person Trips per day 5 18 Years 19 64 Years 65 74 Years 75 84 Years 85 Year s 2.30 3.46 3.16 2.57 1.66 0.06 0.10 0.06 0.04 0.02 0.08 0.03 0.02 0.01 0.00 0.45 0.43 0.33 0.25 0.19 0.33 0.05 0.05 0.05 0.09 Average Person Miles Per day 5 18 Years 19 64 Years 65 74 Years 75 84 Years 85 Year s 21.42 37.97 29.84 21.66 10.51 0.32 0.84 0.24 0.14 0.06 0.07 0.09 0.06 0.03 0.00 0.29 0.32 0.23 0.14 0.10 3.99 3.67 2.40 1.35 0.34 Average Person Trip Length (miles) 5 18 Years 19 64 Years 65 74 Years 75 84 Years 85 Years 9.55 11.19 9.58 8.66 6.56 6.51 10.6 6.18 4.65 4.07 0.87 3.23 3.02 2.28 2.17 0.65 0.76 0.71 0.58 0.54 12.61 72.83 53.48 30.73 5.17 Average Person Trip Travel Time 5 18 Years 19 64 Years 65 74 Year s 75 84 Years 85 Years 18.14 20.19 19.29 19.04 17.47 44.43 50.67 43.50 47.24 41.03 14.42 23.74 22.85 19.49 27.12 15.61 15.75 18.44 15.25 16.03 29.17 41.93 34.35 32.59 25.05

PAGE 59

49 Table 3.19 Trip C haracteristics by Age, Worker Status and Gender in daily person mile traveled between worker males and females is higher than that between non worker males and females. The same trend is true for average person trip length as well. Another important obser vation is that the average daily person miles and average person trip travel time decreases with the increase in age. This is applicable to all age groups, irrespective of their gender and worker status. In addition, as expected, the average person miles p compared to other age groups. Characteristics Worker Non Worker Male Female Male Female Av erage Person Trips per day 19 64 Years 65 74 Years 75 84 Years 85 Years 3.93 4.10 4.43 4.34 4.41 4.23 3.65 3.39 3.40 3.60 3.24 2.33 3.81 3.26 2.57 1.76 Average Person Miles Per day 19 64 Years 65 74 Years 75 84 Years 85 Years 51.44 47.15 42.77 9.96 38.40 34.12 23.88 16.32 35.87 33.39 28.80 13.46 32.27 26.83 16.97 8.43 Average Person Trip Length (mile) 19 64 Years 65 74 Years 75 84 Years 85 Years 12.87 11.07 10.13 7.51 9.13 8.38 6.76 4.85 10.92 9.49 9.11 5.99 9.23 8.69 7.36 5.79 Average Person Trip Travel Time 19 64 Years 65 74 Years 75 84 Years 85 Years 22.43 20.62 21.10 19.87 19.16 18.75 19.05 15.51 21.8 8 21.01 19.89 17.87 19.59 18.91 18.61 17.95

PAGE 60

50 Table 3.20 Trip Characteristics by Age and Population Size of MSA As shown in Table 3.20, when the population size of a Metropolitan Statistical Area (MSA) is consid ered, the older elderly have the lowest daily average person trips in all categories. Travel distance varies more than travel time. Also, people living outside the MSA were found to travel the most miles compared to other groups (Giuliano, 1999) but the tr avel distance is significantly lower for the older elderly. Figure 3.8 shows the variations in trip start time among the different groups of travelers. It appears that daily Characteristics Not in MSA <1 million 1 3 million >3 million Average Person Trips per day 5 18 Years 19 64 Years 65 74 Years 75 84 Years 85 Years 3.25 3.95 3.51 2.93 2.02 3.35 4.11 3.90 3.05 2.00 3.26 4.15 3.63 2.94 1.86 3.15 4.06 3.48 2.82 1.99 Average Person Miles Per day 5 18 Years 19 64 Year s 65 74 Years 75 84 Years 85 Years 30.77 46.51 41.19 28.82 13.72 27.91 42.66 33.00 21.76 11.76 24.96 41.61 31.77 22.26 10.62 23.66 42.33 27.96 21.85 9.49 Average Person Trip Length 5 18 Years 19 64 Years 65 74 Y ears 75 84 Years 85 Years 9.62 11.91 11.87 10.02 6.89 8.61 10.62 8.65 7.36 6.13 7.82 10.23 8.87 7.77 5.92 7.84 10.74 8.29 8.10 5.18 Average Person Trip Travel Time (minutes) 5 18 Years 19 64 Years 65 74 Years 75 84 Years 85 Years 19.08 19.96 20.25 19.41 16.32 20.47 19.65 19.68 18.12 17.91 17.57 19.86 18.28 18.49 17.48 19.71 22.34 20.60 20.65 18.99

PAGE 61

51 Figure 3.8 Distribution of Person Trip s by Age and Time of the Day trips of the different travelers are not distributed evenly throughout the day, and older a dults show different travel patterns compared to other groups of travelers. They are more likely to start their trips in the late morning (10 a.m 12 a.m ) and mid day per iods (until 3 pm). It seems that older people tend to avoid the morning and after work peak traffic times by choosing a different time for traveling (Collia et al., 2003). Interestingly, it appears that the older the traveler the higher the tendency to st art travel in the late morning and mid day periods The timing of travel is always important from the Congestion is a common occurrence in most large cities and metropolita n areas. When the elderly are considered, this is more important to consider because of their special transportation needs and trip characteristics. The timing of elderly travel, as shown in Figure 3.8, can help planners and special transportation service providers to mitigate the mobility problems of the elderly. 0.00% 2.00% 4.00% 6.00% 8.00% 10.00% 12.00% 14.00% Midnight 1am 1am 2am 2am 3am 3am 4am 4am 5am 5am 6am 6am 7am 7am 8am 8am 9am 9am 10am 10am 11am 11am 12am Noon 1pm 1pm 2pm 2pm 3pm 3pm 4pm 4pm 5pm 5pm 6pm 6pm 7 pm 7pm 8 pm 8pm 9pm 9pm 10pm 10pm 11pm 11pm midnight Percent of Daily Trips Time of the Day 19 64 65 74 75 84 >=85

PAGE 62

52 3.4 Conclusion In summary, this chapter provides an overview of household and person characteristics of the 2009 NHTS. In addition, different trip characteristics of those persons are discussed i n this chapter. From the discussion, it is clear that the trip characteristics of older persons are different from their younger counterparts. For example, older adults make fewer trips, and these trips are shorter in distance (Collia et al., 2003; Heaslip 2007). An interesting observation is the variation in trip characteristics of the female elderly. In the child and young adults group, female s were found to make more average daily trips compared to males, but in the older adults group, the situation is the opposite: average daily person trips are higher for males compared to females in this cohort. In addition, females are more likely to make short distance trips, and this pattern is the same among the working population. Also, the elderly are more likel y to depend on the POV for their daily travel, but a substantial number of them is non drivers. Proper measures should be taken to keep the elderly active. For instance, special transportation services for the elderly should be provided in such way as to reduce their dependency on a POV. It also will reduce the frustration among older adults that develops from giving up driving. In addition, older adults make more trips for shopping purposes, and the older elderly have the lowest trip rate (0.48 ) for socia l and recreational purposes among all age cohorts. Appropriate measures should be taken to increase the social/recreational trips of the older elderly because social activities involving mobility can reduce mortality in older people (Glass et al. 1999) Variation is observed even among the different groups of elderly. It was found that the difference between the average daily person trips of males and females increase

PAGE 63

53 with the age of the older adults. In addition, female older adults are found to travel s horter distances compared to male. This gender variation should be carefully considered in elderly travel behavior. Young elderly living in zero vehicle households are more likely to walk than using a POV for their daily travel. The tendency to use public transit and walk appears to decrease with age. This pattern is observed even among the older adults from zero vehicle households. In addition, it is found that older elderly living in portation The variation in the mode use patterns of the elderly due to the differences in vehicle ownership and household income provides a better picture of elderly travel behavior. In addition, the variation in trip start times of the different groups of older adults can help transportation planners and policy makers to develop appropriate strategies for the elderly. It would also help special transportation providers to meet the special transportation needs of the different groups of elderly at a particular time of day. However, several cautions should be exercised before directly using the results obtained from descriptive analysis. All of the tables and figures in this chapter are based on a descriptive ana lysis. The travel patterns of the elderly obtained from this chapter may vary with the presence of other factors or when all the factors are considered together. To understand the travel patterns and preferences of the elderly more clearly, these variables should be taken into account through a modeling effort. Only then can the results be helpful for transportation planners and policy makers in developing a comprehensive plan and other strategies for older Americans.

PAGE 64

54 CHAPTER 4 MODELING METHODOLOGY 4.1 Introduction This chapter begins with a data (sample) description and is followed by the Multinomial Logit Model (MNL) methodology, results, and, finally, the Mixed Multinomial Logit (M MNL) model for analyzing the travel patterns and preferences of the el derly. 4.2 Data This section describes the sample preparation that was required for modeling purposes and the sample characteristics of the elderly from the 2009 NHTS. 4.2.1 Sample Preparation This subsection presents the structure of the 2009 NHTS data files and the main file used for the estimation of the MNL and M MNL models to analyze the travel patterns and preferences of the elderly. The 2009 NHTS has four different files: a household file, a person file, a travel day file, and a vehicle file. The household file contains different household level information such as total number of workers, total number of household vehicles, and household income, all obtained from a household interview. The person file contains demographic information such as age, gender, and race of the interviewed persons from

PAGE 65

55 each selected household for the 2009 NHTS. The travel day file contains travel related information such as starting and ending time of trips, means of transportation, trip distance, and travel time for each trip made by household persons on a particular day of the week. The vehicle file contains information about each of the vehicles of the household. This study mainly focuses on the first three file household, person, and travel day to analyze the travel patterns and preferences of the elderly. reveals the information of those who traveled more than a week ago from the given reporting day of the survey (travel day) and whether the i ndividuals prefer going out of home more often or not. More specifically, if the individuals did not make any trips on the travel day and reported that they had stayed at the same place (for example, at home) g ago did you take a trip to another the next mobility patterns and preferences were divided into four categories: (1) traveled on travel day, (2) did not travel on travel day but traveled in past seven days, (3) did not travel in past seven days but prefer going out of home more often, and (4) did not travel in the past seven days and do not prefer g oing out of home. In other words, individuals who made at least one trip in a week were included in the first two groups, and the last two groups include only the persons who did not make any Past seven Table 4.1 shows the variation in mobility patterns and preferences by age group. It appears that the older adults (age 65 and above) are more likely to be immobile than their younger counterparts. The pattern

PAGE 66

56 Table 4.1 Comparison of Mobility P atterns and Preferences by Age G roups is similar for both short and long term immobility. A new file of persons aged 65 and above was created fr om the original person file to analyze the travel patterns and preferences of older Americans. In addition, different household level characteristics such as total number of trips of household members were imported into the new segmented person file from t he household file. After this, several screening and consistency checks were performed, and records with missing and inconsistent data were eliminated. Finally, dummy variables were created in the segmented file based on the literature and an understanding of what factors potentially could affect the travel behavior of the elderly for the model estimation. 4.2.2 Sample Description Table 4.2 presents a brief summary of the socio demographic characteristics of the individuals in the sample. As shown in the t able, there are a total of 71,261 elderly persons age 65 and above. Among them, 3,946 (1,962 + 1,984) individuals did not make a trip in the travel week and, of these 3,946 individuals, 1,962 individuals did not prefer to go out of home more often and 1,98 4 individuals did prefer to go out of home more often. A total of 54,783 individuals traveled on the travel day, and 12,532 individuals did Age Traveled in Past Seven Days Did Not Travel in Past Seven Days On Travel Day Not on Travel Day Prefer Going O ut of Home Do Not Prefer Going O ut of Home 19 64 Years 90.1% 8. 6% 0.9% 0.4% 65 74 Years 80.8% 15.3% 2.1% 1.9% 75 84 Years 72.4% 20.2% 3.7% 3.8% 55.2% 27.8% 7.8% 9.2% Total 87.4% 10.3% 1.3% 1.0%

PAGE 67

57 not travel on the travel day but traveled during the travel week. The percentage of elderly female in the groups wh o did not travel during the travel week is approximately 67 percent, which is higher than that of the U nited S tates as a whole (55.3%). Elderly females are less likely to go outside of home compared to elderly males. As expected, the percentage of individu als age 85 and above in the groups who did not travel in the travel week is higher than that in the groups who traveled at least once during the travel week. In addition, from the percentages of race, education, and employment status variables, it appears that African Americans, individuals with lower education, and non workers are more likely to stay home compared to individuals from other races, with higher education, and workers. As expected, medical condition made travel difficult for 71 percent of the people who prefer going out of home more often, and when length of medical condition is considered, it is longer among the individuals who did not travel compared to those who traveled at least once during the travel week. The percentages of non drivers a nd individuals living in zero vehicle households are also higher in the groups who did not travel compared to those who traveled in the travel week. Percentage of Internet use (88.4) and household size (30.2) variables indicate that persons who never use t he Internet and live alone are not likely to travel more frequently. Among the individuals who do not prefer going out of home more often, about 62 percent live in households with income less than $25K, and 18 percent live in a rented house, whereas for th e whole United States, these percentages are 29.8 percent and 8.8 percent, respectively. Finally, the distributions of birth status, dwelling unit type, residential area type, life cycle classification, MSA size, and presence of multiple

PAGE 68

58 Table 4.2 Sample Characteristics Variables Traveled in Past Seven Days Did Not Travel in Past Seven Days Total On Travel Day (1) Not on Travel Day (2) Prefer Going O ut of Home (3) Do Not Prefer Going Out of Home (4) Number of Persons 54,783 12,532 1,984 1,9 62 71,261 Gender Male Female 47.5 52.5 36.5 63.5 33.0 67.0 32.7 67.3 44.7 55.3 Age Young Elderly (65 74 years) Middle Elderly (75 84 years) Older Elderly ( >= 85 years ) 58.3 34.2 7.5 46.8 38.4 14.8 33.6 41.7 24.7 27.0 43 .0 30.0 54.7 35.4 9.8 Hispanic Status Hispanic Not Hispanic 4.1 95.9 4.5 95.5 6.4 93.6 9.5 90.5 4.4 95.6 Race White only Black only Other 91.4 4.6 4.1 89.7 5.8 4.5 82.9 10.4 6.7 80.1 10.6 9.3 90.5 5.1 4.4 Highest Level of Education High School Graduate or Lower 42.0 43.4 14.5 54.9 37.0 8.0 67.4 27.3 5.3 71.6 24.4 4.0 45.8 41.3 12.9 Employment Status Full Time Part Time Not employed 8.3 10.0 81 .7 2.8 3.8 93.4 0.7 0.8 98.6 0.8 0.7 98.6 6.9 8.4 84.7 Medical Condition Yes No 16.3 83.7 35.7 64.3 70.5 29.5 52.0 48.0 22.2 77.8 Length of Medical Condition 0 4 years 5 10 years 10 years or more 7.3 3.5 5.4 16. 7 8.1 11.0 36.7 15.7 18.0 24.9 11.7 15.4 10.2 4.9 7.0 Driver Status Driver Not a Driver 93.0 7.0 76.3 23.7 46.5 53.5 45.5 54.5 87.5 12.5 Birth Status Born in U.S. Not born in U.S. 93.3 6.7 93.3 6.7 91.1 8.9 89.1 10.9 93.1 6 .9

PAGE 69

59 Table 4.2 (Continued) Variables Traveled in Past Seven Days Did Not Travel in Past Seven Days Total On Travel Day (1) Not on Travel Day (2) Prefer Going Out of Home (3) Do Not Prefer Going Out of Home (4) Internet Use Almost Every Day Sometimes Never Never 37.1 16.3 46.6 23.2 12.7 64.2 8.2 6.8 85.1 7.5 4.1 88.4 33.1 15.1 51.9 Household Size 1 person 2 persons 3 persons >= 4 persons 23.9 65.9 7.0 3.1 25.6 59.4 9.6 5.5 28.0 53.4 11.7 6.9 30.2 48.9 13.7 7 .2 24.5 63.9 7.8 3.8 Count of Household Vehicle 0 vehicle 1 vehicle 2 vehicles 3 and above vehicles 3.2 33.9 44.2 18.8 7.9 37.0 37.4 17.8 19.2 38.9 28.5 13.4 20.0 40.0 27.1 12.9 4.9 34.7 42.1 18.3 Household Income < $25 K $25K $50K $51K $75K > $75K 25.8 37.0 17.2 19.9 37.5 35.4 12.7 14.4 59.1 26.4 7.6 7.0 62.0 22.6 8.1 7.3 29.8 36.1 15.9 18.3 Housing Unit Own Rent 92.3 7.7 89.1 10.9 82.9 17.1 82.4 17.6 91.2 8.8 Dwelling Unit Type Single De tached House Other (Duplex, Condo, Mobile etc.) 79.6 20.4 77.4 22.6 72.0 28.0 72.2 27.8 78.8 21.2 Residential Area Type (Urban/Rural ) Urban Rural 71.9 28.1 67.7 32.3 72.0 28.0 72.2 27.8 71.2 28.8 Life Cycle Classification 1 adult, no children/youngest child 0 21 years 1 adult, retired, no children 2+ adults, no children/ youngest child 0 21 years 2+ adults, retired, no children 5.1 19.0 11.5 64.3 4.0 21.7 10.1 64.2 4.3 24.2 9.7 61.7 4.9 25.8 10.5 58.9 4.9 19.8 11.3 64.1 MSA size MSA less than 1 million MSA 1 to 3 million MSA more than 3 million Not in MSA 31.8 21.6 24.6 22.0 31.8 20.5 23.9 23.8 29.8 19.5 25.7 25.1 29.1 21.0 24.7 25.1 31.7 21.3 24.5 22.5

PAGE 70

60 Table 4.2 (Continued) job holder variables appear to be the same across the groups with different mobility patterns and preferences. From the household structure and number of vehicle variables, it appear s that elderly persons living with two or more household members and in zero vehicle households are more likely to stay home. 4.3 Multinomial Logit Model This section describes the methodology of the Multinomial Logit Model (MNL), which is one of the most widely used models in choice analysis. A random utility based MNL model is specified in this study to analyze the travel patterns and preferences of the elderly. The following subsections describe this random utility approach, followed by the estimation a nd evaluation techniques of the MNL. Variables Traveled in Past Seven Days Did Not Travel in Past Seven Days Total On Travel Day (1) Not on Travel Day (2) Prefer Going O ut of Home (3) Do Not Prefer Going Out of Home (4) Household Structure Only elderly One other member Two or more other members 73.4 21.0 5.6 69.2 21.6 9.2 60.3 27.0 12.7 60.1 26.1 13.8 71.9 21.4 6.6 Worker Status 0 worker 1 worker 2+ workers 68.2 24.1 7.8 74.0 19.8 6.2 78.1 17.2 4.7 77.2 17.2 5.6 69.7 22.9 7.3 Presence of Multiple Job Holders Yes No 1.5 98.5 1.9 98.1 2.0 98.0 2.2 97.8 1.6 98.4

PAGE 71

61 4.3.1 Random Utility Maximization Approach Discrete choice models are based on a random utility maximization approach. This approach assumes that an individual will select the alternative from the choice set that prov ides him/her the maximum utility. A choice set generally contains all the alternatives available to the individual. In other words, if a decision maker has C choice ty of choice set. This can be expressed mathematically as: w here, is the utility function are the vectors of attributes describing alternatives i and j, respectively is a vector of the characteristics of individual n If the situation is such that there is no uncertainty in the indiv i.e., if he/she always selects the alternative with highest utility from the choice set, a However, the presence of different type s of errors in the utility functions drives the analyst to use a random utility or probabilistic choice model instead of a deterministic of these errors are: (1) lack of complete information to the decision maker about all the alternatives available to him/her, (2) incomplete information to the analyst about the alternatives available to the decision makers, and (3) the specific circumstances that the decision makers face in real life are completely unknown to the analyst (Koppelman and

PAGE 72

62 Bhat, 2006). To capture all of this absent information in the choice prediction, a random utility approach is used. In this approach, the utility function is decomposed into two components and the u ncertainty that results due to la ck of information is included as one of those two components. This can be expressed as : w here, is the true utility of the alternatives i to the decision maker n i s the obs ervable portion of the utility is the error portion of the utility unknown to the analyst The observable or the deterministic portion of the utility can be expressed as w here, is the observable portion of the utility of alternative i to decision maker n is the portion of the utility associated with the characteristics of alternativ e i is the portion of utility associated with the characteristics of decision maker n is the portion of the utility that results from interaction between the attributes of alternative i and characte ristics of decision maker n Again, these observable portions of the true utility function can be expressed as a linear function of the explanatory variables : w here, are the parameters associated with the attributes of alternatives and individual character istics

PAGE 73

63 The error terms mentioned in the utility equation are assumed to be identically and independently distributed type 1 extreme values (the Gumbel distribution) Also, the error terms are assumed to be independent from the irrelevant alternative, which is well known as the independent from the irrelevant alternative (IIA) property of the MNL model. All of these assumptions lead to the MNL model for the following probability expression: OR w here, is the probability of choosin g alternative i of individual n, and is a vector of coefficients of the observed characteristics of decision makers and the attributes of alternatives. 4.3.2 Estimation and Evaluation of the alternatives are estimated using the Maximum Likelihood Method. The mathematical expression of this maximum likelihood method is given below : Let N denote the sample size and define = 1 if decision maker n choose altern ative i = 0 otherwise The likelihood function for a general multinomial logit model becomes

PAGE 74

64 w here, for the linear in parameters logit model: Taking the logarithm provides the log likel ihood function as Generally t statistic is used to make the decision about the single variable in the model. However, when there is a need to make a decision about the whole model, the likelihood ratio test is used. In other wor ds, the superiority of the model with respect to the base line model is determined by the likelihood ration test. It helps the researchers to make a decision during the hypothesis testing. The statistics are given by : is distributed with degrees of freedom. w here, is the Log likelihood at market share is the Log likelihood at convergence of the specified m odel is the number of parameters corresponding to K variables in the model is the number of alternatives 4.4 Multinomial Logit Model Results This section provides the results of t he Multinomial Logit model. Table 4.3 presents the parameter estimates and the corresponding t statistics of the variables used to

PAGE 75

65 analyze the mobility patterns and preferences of the elderly. These results offer reasonable hints that are consistent with t he expectation. As shown in Table 4.3, the choice set of the elderly mobility patterns and preferences is composed of four alternatives: individuals (1) who traveled on the travel day, (2) who did not travel on the travel day but traveled in the past seven days, (3) who did not travel in the past seven days but prefer going out of home more often, and (4) who did not travel in the past seven days and do not prefer going out of home more often. As mentioned in subsection 4.2.1, the past seven days are descri individuals who traveled at least once in the travel week other than the travel day, i.e., they travel but not as frequent as the frequent travelers from the first category. On the other hand, the individuals who did not travel at least once in the travel week but prefer going out of home more often are des viduals in the rest of the thesis. The last column in the model result table shows the difference in the effects of variables on the last two preferences. In this column, while some coefficients are positive, some are negative and some are blank. The posi tive sign of a coefficient in the last column against a variable indicates that the effect of that variable is higher on the preference of staying home compared to the preference of going out. The opposite is true for the negative coefficient. When it is b lank, it indicates that the effects of that variable

PAGE 76

66 Table 4.3 Multinomial Logit Model Results Variables Traveled in Past Seven Days Did not Travel in Past Seven Days Difference in the Effects of Variables on (3) and (4) Preferences On Travel Day (1) Not on Travel Day (2) Prefer Going Out of Home (3) Do Not Prefer Going Out of Home (4) Par. t stat. Par. t stat. Par. t stat. Par. t stat. Par. t stat. Constant 1.97 65.19 4.56 48.18 4.52 48.90 Gender (Male is base ) Female 0.28 12.80 0.16 4.03 0.16 4.03 Age (< =75 years is base) Middle Elderly (75 84 years) Older Elderly (>= 85 years) 0.06 0.27 2.41 7.60 0.11 0.33 1.96 4.88 0.42 0.87 7.12 12.59 0.32 0.54 4.07 6.08 Race ( White and Others are base) Black only 0.26 4.33 0.26 4.33 Driver Status (Driver is base) Non Driver 0.68 19.35 0.95 15.16 1.34 21.13 0.39 5.08 Worker Status (Not worker is base) Worker 1.61 11.96 0.8 4 6.03 Education (High school graduate through higher degree) High School Graduate or Lower 0.22 9.82 0.37 7.09 0.50 9.43 0.13 1.86 Household Size (single elderly household is base) 2 + elderly household (no other member) 1 other member household 2+ other member household 0.19 0.34 7.19 8.53 0.07 0.50 0.50 1.30 8.79 5.44 0.07 0.50 0.63 1.30 8.79 6.98 0.14 1.38 Household Income (>$ 25k is base) <$25k 0.19 8.03 0.31 6.02 0.24 4.73 0.07 1.02

PAGE 77

67 Table 4.3 (Continued) Variables Traveled in Past Seven Days Did Not Travel in Past Seven Days Difference in the Effects of Variables on (3) and (4) Preferences On Tra vel Day (1) Not on Travel Day (2) Prefer Going Out of Home (3) Do Not Prefer Going Out of Home (4) Par. t stat. Par. t stat. Par. t stat. Par. t stat. Par. t stat. Household Location (Urban/Rural) Urban 0.09 2.12 0.09 2.12 Number of HH Vehicle (2+ is base) 0 vehicle 1 vehicle 0.45 0.09 6.46 1.65 0.45 0.17 6.46 3.02 0.08 1.14 Internet Use ( Never is base ) Almost Everyday Sometimes 0.97 0.83 14.81 10.9 8 0.71 0.64 10.40 8.12 Medical Condition made travel difficult ( No is base) Yes 0.59 21.67 1.43 22.32 0.50 7.32 0.93 10.35 Medical Condition results giving up driving (No is base) Yes 0.20 4.40 0.69 9.99 0.51 6.40 0.18 1.91 Log likelihood at convergence 44346.86 Number of Cases 71261

PAGE 78

68 on the last two preferences are same. It is important to note here that there are two types of base variables in the model results. One is mentioned in parenthesis with the variable name in the first column of the table and another is indicated by the blank sign in the row against a variable. Variables in the model syst ems are different socio demographic factors such as age, gender, race, education, driving and working status, number of household members, number of household vehicles, geographic location of the household (urban/rural), housing unit owned or rented, frequency of Internet use, household income, and different medical con ditions related information of the elderly. These variables were selected based on the literature review and the judgment on what factors potentially could The model estimates show that females are mo re likely to stay home compared to males, and this inclination increases with age. Age related variable effects more specifically indicate that the elderly of age greater than 75 years are less likely to travel frequently compared to the young elderly of a ge 65 to 75 years. This pattern is more pronounced for the senior elderly of age greater than or equal to 85 years. Also, the positive coefficients in the last column against the age variables show the preferences of eater than 75 and indicate that this preference increases with the age. This may be because the elderly women generally do not have sufficient funds to support their mobility due to lower earning and pensions in their late age when compared to males and, u ltimately, some of them need to give up driving too early (Skinners and Stearns, 1999). The positive coefficient of the race related variable indicates that the African Americans elderly are not likely to travel Since people of color are more likely to us e the walking mode and use public transit for their daily travel

PAGE 79

69 (Rosenbl oom, 1995; Rosenbloom and Waldorf 2001), this result may be because their older age that is preventing them from walking and using public transit and ultimately forced them to reduce traveling. At the same time, it may also be due to lack of a special transit system in a minority dominant area. In addition, mobility of the elderly depends largely on the driving status of the individuals, especially in the areas where public transport ation is not available. In this study, working and driving status of the elderly were found to significantly contribute to more likely to travel frequently compared to non drivers and non workers. The positive parameter estimates in the last column against the driver variable show that non drivers are more likely to stay home. In other words, they are not likely to go outside of home frequently. At the same time, the parameter estimates for the worker variable indicate that they are more likely to travel compared to non workers. This may be because workers need to travel to their work place more frequently compared to their counterparts. As expected, individuals with low education are less likely to travel frequently compared to their counterparts. effects on the preferences of staying home and going out i.e. individuals with low and high education have the same preferences. But, the parame ter estimate in the last column rejects the hypothesis and indicates that the higher the education level, the higher the desire for mobility. Besides this, number of members in households is likely to affect the travel preferenc es of the elderly. When only two or more elderly (no other member in the household) live in a household, they are more likely to travel compared to single elderly households because of having the companionship of a person of same mentality and

PAGE 80

70 physical con dition. On the other hand, when the elderly live with non elderly in the same house, they are not likely to travel frequently because they tend to depend on the non elderly person for their different needs such as shopping, social meeting etc. In other wor ds, non elderly persons are likely to do these things on behalf of the elderly persons in the household (Evans, 2001; Sommers and Rowell, 1992). From the results, it seems that the tendency to depend on non elderly persons increases with the increase of th e number of household members. This may be because the higher the number of non elderly persons in the household, the more the options the elderly person has to rely on other members for their different needs instead of traveling outside But the differenc e in the effects on the last two preferences is not statistically significant. In addition, it was found that individuals living in low income households are less likely to travel frequently compared to their counterparts. The parameter in the last colum n for the income variable indicates that the preference of staying home increases with the increase in income. It appears that although income plays a role in the overall mobility patterns and preferences of the elderly, but when it comes to the preference of s taying home, it does not dominate in the last column is not statistically significant. The geographical location variable indicates that individuals living in urban areas are more likely to tra vel. This result may be due to the characteristics of the urban area because urban areas are characterized as residential areas with high density and of mixed land use where public transportation is available and enables older people to take their daily tr ips independently. The estimated results also show the difference in the effects of the number of household vehicles on the travel preferences of the elderly. Individuals

PAGE 81

71 from zero and one vehicle households are less likely to travel compared to individual s from two or more vehicle households. Also, it appears that individuals living in households with one vehicle are less likely to stay home without making any trips compared to households with zero vehicles, but they do not travel as frequently as the elde rly from households with two or more vehicles. Results related to income and the number of household vehicles indicates that individuals living in low income and zero ation disadvantaged The Internet use variab le effects indicate that individuals who use the Internet almost every day or often in a week or month are more mobile than those who never use it. In other words, Internet use is positively associ ated with the mobility of older Amer icans. The elderly use the I nternet generally for e mail or for researching different topics such as health, investing, and entertainment. Ford et al. (2009) fo und that use of the Internet reduces depression among the elderly by 20 percent. The positive relationship of Internet use and being mobile in this model estimate indicates that the elderly who are using the Internet are more likely to be active and, hence, happy compared to th eir counterparts who never use I nternet. Medical condition related information is important in t ravel related research, especially when the elderly are considered. One might argue about the significant relationship between age and medical condition of the elderly, but the possibility of having different mental and physical conditions at the same age should not be ignored. In addition to the age of the elderly, medical information should be included in research to reflect the true effect of medical condition (Kim and Ulfarsson 2004). In this study, as expected, it appears that medical condition affect s the travel preferences of the elderly.

PAGE 82

72 The significant positive co medical conditions results in giving up driving are positively associated with not traveling frequently. Interestingly, wh en the difference in the effects of medical condition related variables on the last two preferences are considered, it is hypothesized that individuals with medical condition are more likely to stay home compared to go out. But, the significant negative co efficient in the last column reject s the hypothesis and indicates that individuals with medical condition are more likely to go out. So, medical condition does not dominate th e preferences of staying home when compared to the preferences of going out. In other words, i other factors that drive him/ her towards the decision of staying home 4.5 Mixed Multinomial Logit Model The mixed logit model is widely used for discrete choice modeling in transportatio n, economics, marketing, and many other fields because of its ability to capture the random taste variation, unrestricted substitution patterns, and the correlations in unobserved factors over time, which are the limitations of the standard logit model (Tr ain, 2009). The mixed logit model offers high flexibility in terms of capturing correlations between the unobserved factors affecting choice alternatives. In this study, a mixed multinomial logit (M MNL) model is estimated to capture the correlations among the unobserved factors affecting the travel preferences of the elderly. The following subsections provide an overview of the mixed logit model methodology and briefly compare the estimated M MNL results with the MNL results.

PAGE 83

73 4.5.1 Error Components Spe cification As mentioned in subsection 4 .3.1, the probability expression for standard MNL is : w here is the probability of choosing alternative i of the individual n. The integral of this multinomial logit model probability expression over the density of parameters gives the probability expression for mixed logit model, as shown in the following equation : (1) w here, is the logit probability evaluated at parameters is a density function Among the different ways of deriving the probability expression of the mixed logi t model, random coefficients and error components based specifications are widely used. The error components specifications are used in this study to capture the correlations among the utilities for different alternatives. The utility equation is specified as : w here, is the vector of observed variables relating to alternative j is a vector of fixed coefficients is a ve ctor of random terms with zero mean and are the error terms denoting the stochastic portion of utility

PAGE 84

74 In short, the random part of the utility is If = 0, the utility expression is converted into the standard logit model. If is not equal to zero i.e. if the utility is correlated over alternatives, then w here, is the covariance of In the mixed logit model with error components specifications, various correlations can be captured through the appropriate choice of variables as error components. For instance, a dummy variable was introduced in the model specification of this study that equals 1for the alternatives, among which there is a correlation and 0 for other alternatives 4.5.2 Estimation and Evaluation A simulation method is generally used for the mixed logit model estimation. As shown in equation (1), the probability expression for the mixed logit model is : w here, For the estimation the researchers specify the functional form f( .) and estimate the mean and covariance of the error terms. In simulation, first, a value of is drawn and then the

PAGE 85

75 probability is calculated using that value in the logit formula. T he same procedure is repeated many times, and the results are averaged. The average simulated probability is w here, R is the number of draws is the unbiased estimator of represents the serial draws This simulated probability then is used in the log likelihood function to obtain a simulated log likelihood (SLL), as shown in the following equation: w here, = 1 if n chooses j and zero otherwise. Then, the model with the best specifications is chosen based on the parameter estimate of the covariance. 4.6 Mixed Multinomial Logit Model Results This section presents the resu lts of the Mixed Multinomial Logit Model. Table 4.4 shows the parameter estimates and corresponding t statistics for all four alternatives used in the Multinomial Logit model. The last column in the table presents the difference in the effects of variables on 3 rd and 4 th alternatives. More specifically, the parameter estimates and the corresponding t statistics in the last column show the existence of two groups with different travel preferences within the elderly. The main objective of this

PAGE 86

76 model estimatio n is to capture the correlation among the unobserved factors affecting the different travel preferences of the elderly. In this model, different types of correlations were explored, and only the specification that induced correlations in a parsimonious fas hion is presented in this section. This specification captures the correlations between the unobserved factors affecting the last two travel preferences of the elderly. The assumption that there is no correlation in utility over the alternatives gives th e IIA property to the MNL model. Such assumption may result in distorted estimates of the influence of various factors on the travel preferences of the elderly. This can be observed by comparing the two model estimates as shown in Table 4.3 and Table 4.4. The significant coefficient of the standard deviation presented in Table 4.4 represents the level of correlation between the last two travel preferences, which is non negligible. In addition, after capturing the correlations between the last two travel pre ferences, some significant changes are found in the t statistics and parameter estimates of the variables in the M MNL model results, although the overall effects remain same. Among these, the t st ats of the constants (represent the average effects of all unaccounted factors) for the last two travel preferences reduce greatly from the MNL ( 48.18 & 48.90) to M MNL model ( 7.22 & 7.17). On the other hand, the parameter estimates of these constants increase from the MNL model to M MNL model. The patterns ar e similar for the gender age and race related variables, although the reductions of t stats for these variables are not that high like constants of the M MNL model. The new parameter estimates of these variables for the last two travel preferences in t he M MNL model indicates the higher preferences of staying home compared to the preferences found in the MNL model. But the relative differences in the effects of these variable s on the last two alternatives,

PAGE 87

77 Table 4.4 Mixed Multinomial Logit Model Resul t s Variables Traveled in Past Seven Days Did not Travel in Past Seven Days Difference in the Effects of Variables on (3) and (4) Preferences On Travel Day (1) Not on Travel Day (2) Prefer Going Out of Home (3) Do Not Prefer Going out of Ho me (4) Par. t stat. Par. t stat. Par. t stat. Par. t stat. Par. t stat. Constant 1.97 64.93 6.63 7.22 6.57 7.17 Gender (Male is base ) Female 0.28 12.70 0.21 3.55 0.21 3.55 Age(< =75 years is base) Middle Elde rly (75 84 years) Older Elderly (>= 85 years) 0.06 0.26 2.36 7.47 0.22 0.60 2.48 4.16 0.51 1.11 5.96 7.82 0.30 0.51 3.79 5.70 Race ( White and Others are base) Black only 0.42 3.81 0.42 3.81 Driver Status (Driver is base) Non Driver 0.68 19.36 1.41 6.45 1.73 7.89 0.32 4.05 Worker Status (Not worker is base) Worker 1.88 7.94 1.11 4.61 Education (High school graduate through higher degree) High School Gradu ate or Lower 0.22 9.72 0.53 5.27 0.66 6.61 0.13 1.76 Household Size (single elderly household is base) 2 + elderly household (no other member) 1 other member household 2+ other member household 0.18 0.34 7.04 8.51 0.09 0.70 0.69 1.21 5.65 4.44 0.09 0.70 0.84 1.21 5.65 5.37 0.14 1.42 Household Income (>$25k is base) <$25k 0.18 7.90 0.42 5.18 0.35 4.36 0.07 1.03

PAGE 88

78 Table 4.4 (Continued) Var iables Traveled in Past Seven Days Did not Travel in Past Seven Days Difference in the Effects of Variables on (3) and (4) Preferences On Travel Day (1) Not on Travel Day (2) Prefer Going Out of Home (3) Do Not Prefer Going Out of Home (4) Par. t stat. Par. t stat. Par. t stat. Par. t stat. Par. t stat. Household Location (Urban/Rural) Urban 0.12 2.01 0.12 2.01 Number of HH Vehicle (2+ is base) 0 vehicle 1 vehicle 0 .70 0.14 4.59 1.83 0.70 0.22 4.59 2.81 0.08 1.17 Internet Use ( Never is base ) Almost Every day Sometimes 1.24 1.10 7.30 6.60 0.98 0.91 5.72 5.41 Medical Condition made travel difficult ( No is base) Yes 0.59 21.68 1.74 10.08 0.81 4.65 0.93 10.27 Medical Condition results giving up driving (No is base) Yes 0.19 4.06 1.01 6.11 0.91 5.31 0.10 1.05 Standard Deviation 2.16 3.76 2.16 3. 76 Log likelihood at convergence 44322.20 Number of Cases 71261

PAGE 89

79 those who prefer going out of home and those who do not prefer going out of home, are almost same between the two models. The effects (estimated in the M MNL model) of the driver s tatus variable indicates that the true preference of staying home among the non drivers is higher than the preference estimated in the MNL model. The t statistics also decrease in the M MNL model after accounting for correlations between the unobserved fac tors of the last two alternatives. The parameter estimates of the worker, household location, and Internet use variables also indicate the higher tendency to travel than the tendency estimated in the MNL model. In the same way, for other variables in the m odel, it is found that the true effects (estimated in M MNL model) are higher than the effects estimated in the MNL m odel without considering the correlations between the unobserved factors of the alternatives. Interestingly, the relative differences in t he effects of variables on the last two preferences are almost same as estimated in the MNL model, except the last variable, which indicates whether the medical condition resulted in the person giving up driving or not. For this variable, the parameter est imates of the differences ( 0.10) slightly decrease as compared to the MNL model ( 0.18), although the patterns of overall effects on the alternatives remain the same. In short, the M MNL model explores the true effects of the variables on the travel prefe rences, and ignoring them may lead to poor model fit and biased estimates of variable coefficients. In the current empirical context, the log likelihood value deteriorates from 44,322 (M MNL) to 44,347 (MNL) when the correlations between the unobserved f actors affecting the last two alternatives are not considered. This log likelihood difference is equivalent to a log likelihood ratio 50 which is greater than the

PAGE 90

80 95% critical chi square v alue for 2 degrees of freedom ( mean and standard deviation parame ters ), indicating the superiority of the M MNL model. 4.7 Conclusion This chapter employs a multinomial logit model and a mixed multinomial logit model to explore the mobility patterns and preferences of the elderly. Although mobility patterns of the elde rly were considered in some of the earlier studies, these were limited to only one day travel period data in other words, whether the elderly made at least one trip on the travel day or not were considered in those studies; long term (such as a week) mob ility preferences were ignored in those studies. This study addresses two long term mobility preferences of the elderly (1) prefer going out of home more often, and (2) do not prefer going out of home more often using 2009 NHTS data. The multinomia l logit model results identify the factors affecting the overall mobility patterns and preferences of the elderly. Some of the important factors, such as medical condition, household income, and household vehicle ownership, were ignored or found insignific ant in the models used in earlier studies (based on one day travel period data) to analyze the mobility patterns of the elderly. When these factors are considered in this study along with long term mobility preferences, they are found to affect significant ly the overall mobility patterns of the elderly. In addition, some contradictory results such as mobility appear reduced in urban areas, opposite to the conventional wisdom found in earlier studies based on one day travel period data. In this study, urban areas are found to increase the mobility of the elderly, which is more intuitive. These model estimation results also distinguish the effects of different factors on the mobility preferences of the elderly. Several important findings were obtained from

PAGE 91

81 th is analysis. For example, female elderly are more likely to stay home compared to male, and this propensity increases with the age of older adults. The composition of elderly households, especially the presence of another elderly person in the same househo ld, was found to affect the travel behavior of the elderly. When two or more elderly (no other member) live in a household, they are more likely to travel compared to single elderly households because of having the company of a person of the same mentality and physical condition. In addition, household income and medical condition of low income and medical condition are not the driving factors on the preferences of stayi ng home when compared to the preferences of going out preference or the effects of other factors that compel them to stay home. decrease in mobility amo ng the elderly an autonomous choice or is it the result of reduced physical abilities or of increasing psychological and 1997 ) that planners and policy makers might have about the elderly mobility issue. Moreover, the results i dentify the individuals who are inactive and at risk for social isolation. Necessary steps should be taken to make them active and mobile because isolation can affect the quality of life. The mixed multinomial logit model results capture the correlation b etween the unobserved factors affecting the travel preferences of the elderly. In short, it explores the true effects of the variables on travel preferences after capturing correlation; ignoring this may lead to poor model fit and biased estimates of varia ble coefficients. From the results, it seems that the true effects of the variables give the same pictures of elderly travel

PAGE 92

82 behavior obtained from the multinomial logit model results. In short, these model estimates provide a clear idea about the different travel preferences of the elderly and distinguish the factors that d rive them towards these different travel preferences.

PAGE 93

83 CHAPTER 5 CONCLUSIONS AND FUTURE RESEARCH 5 .1 Conclusions This thesis used the recently released 2009 NHTS to analyze the travel patterns and preferences of the elderly, persons age 65 or older. The four types of travel choices considered in this thesis are traveling frequently (traveled on the travel day), traveling less frequently (traveled in the travel week but not on the travel day), prefer going out of home more often (did not travel in the travel week but want to go outside of home more often), and prefer to stay home (did not travel in the travel week and do not want to go outside of home more often). In other words, this study considers both the short term (less than a we ek) and long term (more than a week) immobility among older Americans. Also, the last two choices indicate the inherent travel preferences of the elderly, which were not considered in earlier studies. The earlier studies were limited to only one day travel decisions of the elderly, and their mobility was defined based on whether they made at least one trip on the travel day or not. It was recognized that the limitation of one day travel period data may restrict the reliability of the final outcome. The resu lts obtained from this thesis after considering long term mobility provide greater insight into elderly travel behavior and contribute to the literature.

PAGE 94

84 The detailed descriptive analysis provided in this thesis shows the difference in the trip characteris tics of the elderly and non elderly. An interesting observation is the variation in the trip characteristics of the female elderly. In the children and young adults groups, females were found to make more average daily trips compared to male, but in the ol der adult groups, the situation is the opposite: average daily person trips are higher for males compared to females. In addition, females are more likely to make short distance trips, and this pattern is the same among the working population. Also, the el derly are more likely to depend on the POV for their daily travel, but a substantial number of them are non drivers. Proper measures should be taken to keep the elderly active. For instance, special transportation service for the elderly should be designe d and provided in such a way as to reduce their dependency on the POV. This would also reduce the frustration among older adults that develops from giving up driving. In addition, older adults are found to make more trips for shopping purposes, and the old er elderly have the lowest trip rate (0.48 ) for social and recreational purposes among all age cohorts. Appropriate measures should be taken to increase the social/recreational trips of the older elderly because social activities involving mobility reduce mortality in older people (Glass et al. 1999) One of the important questions in elderly mobility patterns, it is essential to understand the travel behavior of the different groups of elderly to get the answer to this question. In this thesis, the elderly are divided into three groups young elderly (age 65 74 years), middle elderly (age 75 84 years), and 85 years) to identify the variations in the travel patterns of older

PAGE 95

85 Americans. The descriptive analysis provided in this thesis shows that the difference between the average daily person trips of males and females increase with age. In addition, female older adults are found to travel shorter distance compared to male. This gender variation should be carefully considered in elderly travel behavior. Young elderly living in zero vehicle households are more likely to walk than use a POV for their daily tra vel. The tendency to use public transit and walk appears to decrease with age. This pattern is observed even among older adults from zero vehicle households. In addition, it was found that older elderly living in zero vehicles and low income (<$25 K) house holds in trip start time of the different groups of older adults found in this study can help transportation planners and policy makers to develop appropriate strateg ies for the elderly. It would also help special transportation providers to meet the special transportation needs of the different groups of elderly at a particular time of the day. In short, understanding the different needs of these sub groups among the elderly will help transportation planners and policy makers identify the services and facilities suitable for these heterogeneous groups. The travel preferences of the elderly were examined through a multinomial logit model framework that provides insights into the effects of different factors on the inherent travel preferences of the elderly. Several important findings were obtained from this analysis. For example, female elderly are more likely to stay home compared to males, and this propensity increases with age. The composition of elderly households, especially the presence of another elderly in the same household, was found to affect the travel behavior of the elderly. When only two or more elderly persons (no other member)

PAGE 96

86 live in a household, they ar e more likely to travel compared to single elderly households because of having the company of a person of the same mentality and physical condition. preference to stay home. In other words, low income and medical condition are not the driving factors of the preference to stay home when compared to the preference to go out stay home Moreover, mixed multinomial logit model results show the presence of the correlation between the unobserved factors affecting the travel preferences of the elderly and thus illustrated the necessity of considering such relationships through the appropri ate modeling effort while analyzing the travel patterns and preferences of the elderly. In addition, this model result explores the true effects of the variables on the travel preferences of older adults after capturing the correlations between the unobser ved factors. Overall, this study summarizes the travel patterns and preferences of the elderly, which can help transportation planners and policy makers to develop or improve the planning and policy related to elderly mob ility issues. The general idea for the decrease in mobility among the elderly is a limited or ineffective transpor tation system. This study showed the existence of different travel preferences among the long term immobile elderly. Some of them prefer going out of the home often, but certai n constraints do not allow them to do so, while some of them prefer staying home, which may affect their overall travel decision. If this is case, necessary steps should b e taken to increase the mobility of these individuals who are at risk for social isol ation because isolation can

PAGE 97

87 affect their quality of life. Transportation planners and policy makers must take into consideration the existence of different travel preferences among the elderly to accommodate their mobility needs, including planning for var ious special transportation services. However, several cautions should be exercised before directly using the results obtained from descriptive analysis and model estimates. The e lderly might have several other characteristics that are not considered in th is study. For example, past travel behavior of the elderly are not considered in this study. In addition, the Federal Highway Administration has decided to enhance the weights of the 2009 NHTS and so, these results may change with the new weights. In short results obtained from this study should be used with caution. The next section provides some scopes for future research 5 .2 Future Research This thesis analyzed the travel patterns and preferences of the elderly using 2009 NHTS cross sectional data. How ever, travel patterns and preferences change over time. The analysis could be carried out using a panel dataset to get better insights into elderly travel patterns and preferences. Besides this, the discrete choice components of travel choices used in this study could be modeled in a nested structure. Accessibility to preferable transportation is closely related to mobility, which is not considered in this study due to the limitations of the dataset. This could be addressed in future research for potential policy implications. Since the elderly are not homogeneous in nature, thorough research could be undertaken to address the diverse needs of the heterogeneous older acc essibility, acceptability, affordability, and adaptability.

PAGE 98

88 REFERENCES Alsnih, R. and D.A. Hensher. The mobility and accessibility expectations of seniors in an aging population. Tran sportation Research Part A, Vol. 37, 2003 pp.903 916 Benekohal, R H., R.M. Michaels, E. Shim, and P.T.V. Resende. Effects of Aging on Presented at the 73 rd Annual Meeting of the Transportation Research Board Washington, D.C. 1994 Chu, X. The Effects of Age on the Driving Habit s of the Elderly: Evidence from 1990 National Personal Transportation Survey. Research and Special Programs Administration Report No. DOT T 95 12, USDOT, 1994. Collia, D. V., J. Sharp and L. Giesbrecht. The 2001 n ational household travel survey: A look i nto the travel patterns of older Americans Journal of Safety Research Vol 34, 2003 pp.461 470 Coughlin, J. Transportation and older persons: Perceptions and Preferences A Report on Focus groups, 2001 Coughlin, J. F. Longevity, Lifestyle, and Antici pating the New Demands of Aging on the Transportation System. Pubilc Works Management & Policy Vol. 13, No. 4, 2009. pp. 301 311 Evans, E. L. Influence on Mobility Among Non Driving Older Americans Transportation Research Circular E C026 Personal Tra vel: The long and Short of It 2001, pp. 151 1 68 Ford, S. G., and S. G. Ford. Internet Use and Depression Among the Elderly. Phoenix Center Policy Paper Series No. 38 2009 Gagliardi C., L. Spazzafumo, F. Marcellini, H. Mol lenkopf, I. Ruoppilla, M.T acken, and Z. Szemann. The outdoor mobility and leisure activities of older people in five European c ountries. Ageing & Society No. 27, 2007 pp 683 700 Georggi, N.L. and R. M. Pendyala. Analysis of Lon g Distance Travel Behavior of the Elderly and Low income. Transportation Research E Circular E C026, 2001 pp. 121 150.

PAGE 99

89 Giuliano, G. Land Use and Travel Patterns among the Elderly. Transportation in an aging society: A decade of experience 1999 pp. 192 210 Glass, T. A., C. Mendes de Leon, R. A. Marot olli, and L.F. Berkman. Population based study of social and productive activities predictors of survival among elderly Americans. British Medical Journal No 319, 1999, pp. 478 483. Heaslip, K. Are Travel Patterns of Older Drivers are C hanging ? Submitte d to the 29 th International Association for Time Research Conference 2007 Henderson, K.A., D. Stalnaker, and G. Taylor. The relationship between barriers to recreation and gender role personality traits for women. Journal of Leisure Research 1998 p p. 69 80 Hilderbrand, E. Dimensions in Elderly Travel Behavior: A simplified Activity Based Model Using Lifestyle Clusters. Transportation Vol.30, 2003, pp. 285 306. Kim, S., and G. F. Ulfarsson. Travel Mode C hoice of the Elderly: Eff ects of Personal, Household, Neighborhood and Trip Characteristics. Transportation Research Board No. 1894, 2004, pp 117 126 Koppelman, F.S., and C. Bhat. A Self Instructing Course in Mode Choice Modeling: Multinomial and Nested Logit Models 2006 Lefrancois, R., G. Leclerc, and N. Poulin. Predictors of Activity Involvement Among Older Adults. Activities Adaption & Aging Vol.22, No. 4, 1998, pp. 15 29 Mallett, W. J. Long Distance Travel by Low Income Households. Transportation Research E Circular E C026: Persona l Travel: The Long and Short of It TRB, National Research Council, Washington D.C., 2001 pp.169 177. Metz, D.H. Mobility of older pe ople and their quality of life. Transport Policy Vol. 7, 2000, pp. 149 152. Mollenkopf, H., F. Marcellini, I. Ruoppil la, P. Flaschentrager, C. Gagliardi, and L. Spazzafumo. Outdoor mobility and social relationships of elderly people. Archives of Gerontology and Geriatrics Vol. 24, No. 3, 1997 pp. 295 310 Polzin, S.E., X. Chu, and J. R. Rey. Mobility and Mode Choice of People of Color for Non Work Travel. Transportation Research E Circular E C026: Personal Travel: The Long and Short of It TRB, National Research Council, Washington D.C., 2001 pp. 391 412. Rosenbloom, S. Travel by Elderly. Nationwide Personal Transp ortation Survey: Demographic Special Reports FHWA, U.S. Department of Transportation, 1995

PAGE 100

90 Rosenbloom, S. Mobility of the Elderly: Good News and Bad News, Transportation in an aging society: A decade of Experience, 1999 pp. 3 21 Rosenbloom, S. and B Waldorf. Older Travelers: Does Place or Race Make a Difference? Transportation Research E Circular E C026 : Personal Travel: The Long and Short of It, TRB, National Research Council, Washington D.C., 2001 pp. 103 120. Skinner, D., and M. D Stearns. Saf e Mobility in an Aging W orld. P repared for P resentation at the Annual Meeting of the Transportation Research Board Washington, D.C. 1999 Sommers, D., and K. Rowell. Factors Differentiating Elderly Residential Movers and Non Movers: A longitudinal Analys is. Research and Policy Reviews Vol. 11, No. 3, 1992. Tacken, M. Mobility of the elderly in tie and space and Netherlands: An analysis of the Dutch national travel survey. Transporta t i on, Vol.25, 1998, pp. 379 393 Train, K. E. Discrete Choice Methods with Simulation Cambridge University Press, second edition, 2009. Trilling, D. and J. Eberhard. Safe Mobility for a Maturing Society: A National Agenda .Transportation in an aging society: A decade of experience 2002, pp. 313 318


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 22 Ka 4500
controlfield tag 007 cr-bnu---uuuuu
008 s2010 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0004630
035
(OCoLC)
040
FHM
c FHM
049
FHMM
090
XX9999 (Online)
1 100
Sikder, Sujan.
0 245
An analysis of the travel patterns and preferences of the elderly
h [electronic resource] /
by Sujan Sikder.
260
[Tampa, Fla] :
b University of South Florida,
2010.
500
Title from PDF of title page.
Document formatted into pages; contains X pages.
502
Thesis (MSCE)--University of South Florida, 2010.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
3 520
ABSTRACT: The number of elderly is increasing; to meet their transportation needs, it is important to clearly understand their travel patterns and preferences. Since travel patterns and preferences depend on socio-demographic and other factors, it is essential to identify these factors first to understand the travel behavior of the elderly. The main purpose of this thesis is to analyze the travel patterns and preferences of the elderly age 65 and above using 2009 National Household Travel Survey (NHTS) data. This thesis presents a detailed descriptive analysis of 2009 NHTS data to understand the travel patterns of the elderly. Along with a descriptive analysis, a multinomial logit model and a mixed- multinomial logit model are estimated to explore the factors associated with the overall travel preferences of the elderly and to identify individuals among the elderly who are the least mobile and at risk for social isolation. The analysis results indicate the differences in the trip characteristics between the elderly and non-elderly. Variation is found even among the different groups of the elderly. The model estimation results show the presence of different travel preferences among the elderly and identify those individuals among the elderly who are immobile for longer periods (e.g., a week) and at risk for social isolation. Elderly individuals with different travel preferences should be considered separately in research to determine the appropriate outcomes that can help transportation planners and policy makers improve planning and policy related to elderly individuals.
590
Advisor: Dr. Abdul R. Pinjari
653
Mobility
NHTS
Socio-demographics
Multinomial logit model
Mixed-multinomial logit model
690
Dissertations, Academic
z USF
x Civil & Environmental Engineering
Masters.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.4630