USF Libraries
USF Digital Collections

Access to public transportation

MISSING IMAGE

Material Information

Title:
Access to public transportation an exploration of the National Household Travel Survey appended data
Physical Description:
Book
Language:
English
Creator:
Maggio, Edward
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Demographic
Distribution
Mode
Ridership
Transit
Trip
Accessibility
Dissertations, Academic -- Engineering Science -- Masters -- USF   ( lcsh )
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: Understanding transit usage has become a critical transportation research interest and policy goal. This thesis presents results of an analysis of the 2001 NHTS data specifically focusing on the newly released appended variables that measure access or distance to public transportation. Statistically significant public transportation distance intervals from households and individuals were chosen for analysis in relation to other key variables in the original dataset. Actual relationships between public transportation and traditional household and person characteristics nationwide are explored, specifically focusing on both rail and bus transit modes for the work commute trip. Geographically, both inclusions and exclusions in analysis are conducted due to the widely accepted ubiquitous transit network present in the NY region.^ The analysis reveals strong differences in household and workplace access to transit as a function of race, income, auto ownership, and urban area size. Additionally, a very high sensitivity to access exists suggesting that the share of transit accessible trips is smaller than previously acknowledged. Approximately 53 percent of households are within aviimile of bus service and 40 percent within a quarter-mile. Approximately 10 percent of the population lives within one mile of rail. Over 50 percent of workplaces are within a quarter mile walk radius of a bus line. Not surprisingly, work is more closely concentrated near transit than are residences. Furthermore, mode share for transit declines approximately two thirds beyond the first interval beyond 0.15 miles from a bus route.^ These observations imply a high value to services in close proximity to residential areas.Historical work in this topic area include geographically specific data analysis obtained from surveys which potentially allow a degree of subjectivity in perceived responses whereas accessibility and distance data analyzed in this thesis are actual and spatially measured. Additionally, a regression model exploring the significance of actual access to transit upon mode choice is performed to explore the significance of influence by measured access variables. The analysis suggests that access is even more critical than might have previously been acknowledged by the transit planning profession.
Thesis:
Thesis (M.S.E.S.)--University of South Florida, 2006.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: World Wide Web browser and PDF reader.
System Details:
Mode of access: World Wide Web.
Statement of Responsibility:
by Edward Maggio.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 62 pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001920099
oclc - 187303350
usfldc doi - E14-SFE0001836
usfldc handle - e14.1836
System ID:
SFS0026154:00001


This item is only available as the following downloads:


Full Text

PAGE 1

Access to Public Transportation: An Exploration of the National Household Travel Survey Appended Data by Edward Maggio A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Engineering Science Department of Civil and Environmental Engineering College of Engineering University of South Florida Major Professor: Steven Polzin, Ph.D. Xuehao Chu, Ph.D. John Lu, Ph.D. Date of Approval: October 31, 2006 Keywords: demographic, distribution, mode, ridership, transit, trip, accessibility Copyright 2006, Edward Maggio

PAGE 2

ACKNOWLEDGEMENTS Were it not for the support of Dr. Steve Polz in and the confidence placed in me by the Center for Urban Transportati on Research (CUTR), this rese arch effort and subsequent milestone in my academic career would not have been possible. I am very grateful for this opportunity, especially for the Florida Department of Transportation (FDOT) and Tara Bartee for providing the opportunity to conduct this research. Additionally, I sincerely thank Dr. Xuehao Chu for the tr emendous assistance and guidance provided relative to his expertise and th e topic area explored in this work. I also would like to express my sincere appreciation to the thesis committee members, including Dr. Jian Lu for their continued support, and reviews during the researc h. Finally, I would like to thank my wife, Heather, for her unwavering support during this academic endeavor.

PAGE 3

i TABLE OF CONTENTS LIST OF TABLES.............................................................................................................iii LIST OF FIGURES ............................................................................................................iv ABSTRACT....................................................................................................................... vi CHAPTER 1 INTRODUCTION .........................................................................................1 Background .....................................................................................................................1 Problem Statement ..........................................................................................................2 Objectives ........................................................................................................................3 Methodology ...................................................................................................................4 CHAPTER 2 LITERATURE REVIEW ..............................................................................5 Background .....................................................................................................................5 Measurement of Access ...................................................................................................6 Transit Users ....................................................................................................................7 Considerations .................................................................................................................8 CHAPTER 3 NHTS DATA REVIEW ..............................................................................11 Background ...................................................................................................................11 Methodology .................................................................................................................11 Dataset ...........................................................................................................................12 New Data .......................................................................................................................13 CHAPTER 4 DISTRIBUTION OF ACCESS TO TRANSIT ..........................................15 Background ...................................................................................................................15 Minimum Access Concept ............................................................................................16 Access Measurement .....................................................................................................16 Access Distribution .......................................................................................................17 Access and Demographic Distribution ..........................................................................22 Access and Geographic Distribution .............................................................................28 Accessibility, Density, and Area Type ..........................................................................32 CHAPTER 5 TRANSIT USAGE AND ACCESS ............................................................37 Background ...................................................................................................................37 Mode Share ....................................................................................................................38 Matrix Mode Share ........................................................................................................39

PAGE 4

ii CHAPTER 6 ACCESS LOGISTIC REGRESSION MODEL ..........................................44 Introduction ...................................................................................................................44 Transit Mode Choice Regression Model .......................................................................45 Model Results ................................................................................................................52 CHAPTER 7 CONCLUSION ...........................................................................................54 Introduction ...................................................................................................................54 Transit Access ...............................................................................................................54 Transit Choice ...............................................................................................................55 Going Forward ..............................................................................................................55 REFERENCES ..................................................................................................................58 APPENDICES ...................................................................................................................60 Appendix A Geogrphaphic Household Access ............................................................61

PAGE 5

iii LIST OF TABLES Table 1 Household Distance to Tran sit by Population Density, 1995 NPTS...................33 Table 2 Household Distance to Tran sit by Population Density, 2001 NHTS..................34 Table 3 Household Distance to Transit by Area Type, 1995 NPTS................................36 Table 4 Household Distance to Transit by Area Type, 2001 NHTS................................36 Table 5 Model Results, Un-weighted Vari ables Not Including Measured Access..........48 Table 6 Model Results, Un-weighted Va riables Including Measured Access.................49 Table 7 Model Results, Weighted Vari ables Not Including Measured Access...............50 Table 8 Model Results, Weighted Va riables Including Measured Access......................51

PAGE 6

iv LIST OF FIGURES Figure 1 Measured versus Actual Walk Access to Bus..................................................17 Figure 2 Cumulative Distribution of Household Distance to Bus Line.........................18 Figure 3 Cumulative Distribution of H ousehold Distance to Rail Stop/Station............19 Figure 4 Cumulative Distribution of Person Distance to Bus Route..............................20 Figure 5 Cumulative Distribution of Distance from Work to Bus Route.......................21 Figure 6 Cumulative Distribution of Distance from Work to Rail Stop........................22 Figure 7 Distribution of Household A ccess Distance to a Bus Route by Income..........23 Figure 8 Distribution of Househol d Access Distance to a Bus Route by Area Type..................................................................................................................24 Figure 9 Distribution of Household Access Distance to a Rail Stop by Income...........25 Figure 10 Distribution of Household A ccess Distance to a Bus Route by Race..............26 Figure 11 Car Ownership Category, Per cent Households by Distance from Bus Route................................................................................................................27 Figure 12 Metropolitan Area Size Cate gory, Percent Households by Distance from Bus Route................................................................................................28 Figure 13 Rail Station Access by Trip End Distance, Nationwide, Excluding NY MSA..........................................................................................................29 Figure 14 Rail Station Access by Trip End Distance, Only NY MSA.............................30 Figure 15 Bus Station Access by Trip End Distance, Nationwide, Excluding NY MSA..........................................................................................................31 Figure 16 Bus Route Access by Trip End Distance, Only NY MSA...............................32 Figure 17 Bus Trip Mode Share by Household Distance.................................................38

PAGE 7

v Figure 18 Share of Bus Work Trips within Vehicle Availability Category by Bus Route Access Distance, Nationwide.........................................................39 Figure 19 Bus Transit Work Trip Mode Share by Trip End Distance to Bus Route, Nationwide...........................................................................................40 Figure 20 Rail Work Mode Share by Tr ip End Distance Interval to a Rail Station, Nationwide..........................................................................................41 Figure 21 Rail Work Mode Share by Trip End Distance to Rail Station, Only NY MSA..........................................................................................................43 Figure 22 Bus Work Mode Share by Trip End Distance to Bus Route, Only NY MSA..........................................................................................................43 Figure 23 Household Access to Transit by Density, 1995 NPTS (Percent Persons per Square Mile).................................................................................61 Figure 24 Household Access to Tran sit by Density, 2001 NHTS (Percent Persons per Square Mile).................................................................................61 Figure 25 Household Access to Tr ansit by Area Type, 1995 NPTS................................62 Figure 26 Household Access to Transit by Area Type, 2001 NHTS...............................62

PAGE 8

vi ACCESS TO PUBLIC TRANSPORTATI ON: AN EXPLORATION OF THE NATIONAL HOUSEHOLD TRAVEL SURVEY APPENDED DATA EDWARD MAGGIO ABSTRACT Understanding transit usage has become a critical transportation research interest and policy goal. This thesis presents resu lts of an analysis of the 2001 NHTS data specifically focusing on the newly released appended variables that measure access or distance to public tran sportation. Statis tically significant public transportation distance intervals from households and individuals were chosen for analysis in relation to other key variables in the original dataset. Actu al relationships between public transportation and traditional household and person characteri stics nationwide are explored, specifically focusing on both rail and bus transit modes fo r the work commute trip. Geographically, both inclusions and exclusions in analys is are conducted due to the widely accepted ubiquitous transit network present in the NY region. The analysis reveals strong differences in household and workplace access to transit as a function of race, income, auto ownership, and urban area size. Additionally, a very high sensitivity to access exists suggesting that th e share of transit accessible trips is smaller than previously acknowledged. Approximately 53 percent of house holds are within a

PAGE 9

vii mile of bus service and 40 pe rcent within a quarter-mile. Approximately 10 percent of the population lives within one mile of rail. Over 50 percent of workplaces are within a quarter mile walk radius of a bus line. Not surprisingly, work is more closely concentrated near transit than are residen ces. Furthermore, mode share for transit declines approximately two thirds beyond th e first interval beyond 0.15 miles from a bus route. These observations impl y a high value to services in close proximity to residential areas. Historical work in this topi c area include geographically spec ific data analysis obtained from surveys which potentially allow a degr ee of subjectivity in perceived responses whereas accessibility and distance data analyzed in this thesis are actual and spatially measured. Additionally, a regr ession model exploring the significance of actual access to transit upon mode choice is performed to explore the significance of influence by measured access variables. The analysis suggest s that access is even more critical than might have previously been acknowledge d by the transit planning profession.

PAGE 10

1 CHAPTER 1 INTRODUCTION Background In the Transportation planning profession, including the realm of transportation research, there exists a common desire to understand the influences upon the decision processes involved in mode choice. Pa rticularly, striving for a co mprehensive understanding of transit usage has become of key importance especially in a society experiencing continuous growth in travel congestion. Resul ting from a wide range of research in the topic area, it has been widely accepted that many factors influence the decision to use the public transportation mode, and, that analyzi ng only one aspect of the influences upon travel behavior is not alone sufficient. Some of the factors that have been traditionally influential upon an individuals travel choi ces include those that may be directly controlled by local transit or government agencies such as level-of-service factors including frequency of service and similar tr aveler convenience factor s, route corridors, and fare structure. Other factors beyond the direct control of a responsible agency or government entity include variables such as geographical area population and density, land-use interaction, employment density or even petroleum price or some similar travel cost factor. Perhaps a link be tween both of these categories of factors is transit access and accessibility. Public transit serves va rious markets and is utilized by a diverse amount of individuals within various demogr aphics. Transit tends to capture a large

PAGE 11

2 portion of work commute trips in part due to location in central business districts (CBD) where concentrations of workplaces have t ypically been highest. Arguably, however, transit patronage levels are la rgely subject to the effects of changes in the economy and employment growth or decline, and perhaps foremost, are affected by varying levels of accessibility. Problem Statement Understanding transit usage has become a critical transportation research interest and policy goal because of the imp lications of high degrees of investment and longevity of infrastructure. There is a de sire in the transportation i ndustry to more fully understand the distribution of access to public transportati on as it relates to both individual segments of the population and the nationa l population as a whole in an effort to most accurately capture the transportation needs of the traveler Specifically, it is the desire of the Florida Department of Transportation (FDOT), and by the funding of this research effort to explore the relationship between transit acce ss and other geographic, demographic, and socioeconomic factors. This research effort presents an analysis and result of the 2001 NHTS data specifically focusing a set of newly released, appended va riables that spatially measure distances from households and workplaces (where applic able) to public transportation. Statistically significant public transporta tion access distance intervals that group residences and workplaces were chosen for analysis and correlated to other key demographic and geographic variables present in the complete (all add-on samples) NHTS dataset. Actual

PAGE 12

3 relationships between household distances fr om public transit locations and geographical and demographical characteristics nationwide are explored; the analysis specifically focuses on both rail and bus transit m odes for the work commute trip. Objectives The objective of this research effort is to obtain an improved understanding of the relationship between transit a ccess distances and population characteristics. This is accomplished by conducting analysis of the 2001 National Household Travel Survey (NHTS) database, including the appended variab le data sets which will be described in further detail. The resulting graphical relationships and conclusions can help professionals and policy makers make more informed decisions regarding the design and provision of transit services. Additionally, this research can successfully exhibit that planners would benefit from pursuing collecti on and analysis of meas ured data, with the help of advancements in technology, while re lying less on personal su rvey response data. This analysis will explore the land use va riables appended to the National Household Travel Survey data to further explore how land use characteristics influence transit use behavior using both aggregate national data, and New York metropolitan area specific data. While the 1995 Nationwide Personal Tr ansportation Survey (NPTS) and the 2001 National Household Travel Survey utilizes mo stly subjective or perceived measures of transportation characteristics, spatially measured proximity to transit for the household location is new for the 2001 survey. Also new for 2001 is spatially measured proximity to transit for the employment location for workers.

PAGE 13

4 Methodology SPSS and Microsoft Excel software are used to carryout the analysis Each are well suited to the task of organizing and graphically representing characteristics for a database of this size. The relationships betw een household access distances and the persontraveler characteristics are developed in tabula r and subsequently in graphical format in order to clearly visualize possi ble correlations in attributes The NHTS data set for the Household file, Person file, and Day Trip file all contain a ppended instances of the access distance variables, that is, each instance of a household, person, and trip is allotted an attribute for distance between the household and an attribute specifying the distance to the workplace where applicable. These comprehensive variable additions enable subsequent cross-tabulations while providing a means for desc riptive analysis and finally, conclusions. Due to the volume of data a nd enormous number of possible tabulations, the relationships deemed most feasible and relevant are analyzed. Additionally, access distances are categorized to the smallest or finest scale practicable, to the extent that adequate sample sizes allow. Specificall y, access distances are explored for possible existing relationships to key demographic va riables such as age, race, income, and vehicle ownership while evaluating in the desired geographical characteristics. To achieve a more appropriate representation of characteristics, both inclusions and exclusions in analysis are conducted because of the generally ubiquitous transit network present in the NY metropolitan region.

PAGE 14

5 CHAPTER 2 LITERATURE REVIEW Background As a vital part of the trans portation planning process, ofte n surveys or interviews are conducted to ascertain the reasoning pr ocesses behind the mode choice decision, especially for the work commuter. Very importantly, data is sought in order to numerically quantify and analy ze the many of the input factor s potentially influential on the mode choice process. Survey data has become a powerful asset to research professionals and the transit planning professi on. In some studies, interviews of transit agency managers have been conducted in order to learn their opinions or perceptions about what general factors affect transit ridership. (Taylor 2 002) As a prelude to further analysis of tangible survey da ta results, these interviews reveal a prevailing assertion among transit agency professiona ls that external factors, such as population change, new development, and regional economic conditions more than likely effect ridership to a greater extent than do agency or se rvice design initiatives. (Taylor 2002) Public Transportation can be conveniently categorized into two major components: rail and bus. Admittedly, various subcategories of public transportation exist. Alternatively, walk-ability plays a key role in that an indi vidual must walk from either a residence or business to their mode of choice. In a literature study conducte d by Robert Cervero,

PAGE 15

6 characteristics of rail stations, in particular, those adjacent to residences and commercial projects were examined. After hypothesi zing a relationship between proximity and usage, subsequent to review of several survey results and analysis, he concluded that rail transit usage varied significantly with di stance to rail lines and stations. Measurement of Access In what is considered a highly comprehens ive analysis of patr onage by a function of distance to key land development in the Washington D.C. area in 1887 and 1989, by JHK & Associates, conclusions indicated that transit trip mode share declined by approximately 0.65 percent for every one hundr ed foot increase in distance of a residential site from a Metrorail station. (C ervero 1993) Interestingly, this is somewhat more of a finely scaled documented instan ce of transit mode share variance by access distance. Additionally, transit work trip m ode share for rail ranged from 18 to 63 percent while the ridership experienced the highest pe rcentages for individuals from residences closest to rail stations. Furthermore, thr ough the course of that literature review, Mr. Cervero found that rail transit ridership declined steadily as distance between stations and employment offices increased. Not surprisingl y, the result implies that an increase in access causes a decrease in transi t use, and conversely, a decrease in distance in transit stops from residences and workplaces displa yed increases in ridership. (Taylor 2002) Another similar study which was related to tr ansit access wa s actually conducted prior to the Washington D.C. study; it focused on charac teristics in the geogr aphic locations of both Toronto and Edmonton, Canada. Notably, th is study revealed that individuals were

PAGE 16

7 willing to walk approximately 4,000 feet (almos t 0.8 miles) to a rail station. (Cervero 1993) However, this study focused on the imp acts of variances releva nt to the topic of Transit Oriented Development (TOD) whereby more pleasant and desirable urban spaces existed along the walking distan ces, more than catalogued by other surveys or data. As stated previously, many factors weigh into mode choice as a function of access distance, and walk-ability and related characteristics thereto may play a significant role. Transit Users One method in the exploration of the mode choice decision process involves analyzing characteristics of the individual. Arguabl y, there are many ways to classify a person, household, or trip in a context of transit or even automob ile usage, but in a paper by Beimborn et al in 2003 travelers are classified into two groups: choice users or captive users. This classification may play an impor tant role in the relationship between mode choice and accessibility. Th e key difference between these two groups is that option users have more than one choice available to them. As such, accessibility, or at least connectivity, relates to cap tivity and mode choice. The data used in this choice and captivity themed paper were obtained from a survey conducted in Portland, Oregon: the 1994 House hold Activity and Travel Diary Survey. It simultaneously explored viable individual characteristics from various other sources including but not limited to the Regional Land Information System for the Portland Area, and the U.S. Dept. of Energy. In essence, this paper compared traditional mode choice models with those that included a categorization of users to either captive or choice. The

PAGE 17

8 result of this work indicate d in part, that transportation models produce more accurate forecasts when captive users are asse ssed and catalogued based on individual accessibility. Thus, accurate assessment of acces sibility has been found to play a vital role in forecasting and transit model developm ent. Not surprisingly, out-of-vehicle travel time (OVTT) was found to be a highly influentia l factor in this anal ysis. Considerations for OVTT might prove very useful in the analys is of this research paper; however, data constraints and the lack of temporal variab les in the NHTS data set do not easily allow for such analysis. Finally, when compared w ith other factors, among choice transit users, variances in travel time was found to have li ttle influence upon mode choice, in direct contrast to the highly significan t effect of transit access upon mode choice. (Beimborn et al 2003) Considerations Two additional terms considered relevant and influential in the context of transit access include mode split and market share. Undeni ably, market share is a key aspect involved in transit planning or research since it is necessary to unde rstand the traveler and their needs. According to Beimborn et al, market share and mode split analysis prove difficult and could be captured inappropriately since data limitations exists. A true representative population or even an appropriate sample size for choice individuals or choice trips may be too illusive. When calcula ting a typical market share or ratio, the numerator may be known, for example trips or boardings; howev er, the denominator, the user group is somewhat less accurate or even arbitrary. (B eimborn et al 2003) Essentially, the number of individuals or group of individuals for whom transit is a viable option is not exact.

PAGE 18

9 Although in this paper, acce ss considerations are certainly not limited to proximal distribution influences on market share, clearly, properly delinea ting and understanding user transit accessibility and distances is important to consid er in the planning process. Therefore, if market share factors are not properly calculated, the resulting conclusions and forecasts may contain errors that may have serious implications. Strong consideration of the characteristic s of public transit trips augments the understanding of public transit market share as a function of access distance relevant to the planning process. Previous research efforts have produced a focused analysis on several trip characteristics including trip dist ance, out of vehicle travel time, and travel speed to name a few. While these characteri stics are very important factors and deserve adequate consideration, real and obtainable statistics in these categories are mostly limited to personal survey response data. (P olzin and Chu 1995) These data are not actually measured; they are based on the percep tions of the survey re spondents involved. (Polzin and Chu 1995) Furthermore, historica lly, it has been widely accepted that trip makers are typically not accurate at reporti ng their own trip characteristics and may improperly estimate distances when responding to numerical-answer survey questions. (Polzin and Chu 2001) Perhaps no less important a consideration, tran sit service supply factors such as frequency, span, geographic connectivity, reliability, and cost, are usually not available nor are they all yet practic ally measurable for individual trips. Subsequently, statistical data of this nature may not often prove as reliable as might be desired by the transit re search community.

PAGE 19

10 An alternative approach to analyzing the effects of accessibility on transit includes examining the land-use interaction for a give n geographical area. One paper exploring these characteristics by Ross and Dunning in 1997 utilizes data from the 1995 Nationwide Personal Transportation Survey and examines the relationships between geographical land use variables and demographic characteristics. This prior work mainly focuses on area population densities and area type variables that were available in the 1995 NPTS in the context of access distance intervals. Additionally, demographic variables are examined by correlation of the land-use variables similar to the analysis performed in this research effort. Many of the results presented by Ross and Dunning, which utilize access data derived from user responses, are conveniently and directly compared to the actual measured 2001 NHTS access intervals to e xplore relationships with other geographic and demographic variable s. This comparison is carried out in a later chapter.

PAGE 20

11 CHAPTER 3 NHTS DATA REVIEW Background The 2001 NHTS is a sample survey of the natio ns daily personal travel and is generally considered the primary source for national personal travel behavior and related information. Although the data is not new by seve ral years, it is cons idered a tool that aids transportation planners and policy makers because of its uniqueness and relevance. The 2001 NHTS updates information gathered in prior Nationwide Personal Transportation Surveys (NPTS) and the Americ an Travel Survey (ATS). (NHTS 2001) These data include variables of information for all trips, modes, purposes, trip lengths, trip times, and geographical areas of the country. Methodology The 2001 NHTS was conducted from March 2001 through May 2002. Similar to prior surveys in the series, the procedure began with first obtaining a random sample of telephone numbers, then selecting only residential numbers from the sample. Exclusions from the pool of numbers included colle ge dormitory residents, nursing homes inhabitants, prisons, and military base reside nts. Next, a household member was queried over the phone for household and person characte ristics and traits in addition to some vehicle information and other administrative data. Perhaps of key importance to the

PAGE 21

12 survey, the household was assigned a travel-d ay for recording trip information. The respective respondent was asked to mail back a travel diary containing all pertinent travel information regarding the day and a subsequent follow-on interview was scheduled and conducted for eligible household members about their personal tr avel behavior. The NHTS data does not contain all of the information that the transportation planning profession might deem beneficial to transit planning and mode choice analysis. Some other possibly desirable information might include travel cost(s), travel routes, infrastructure type, and long-term tem poral variance in household activities. Additionally, actual house hold and workplace locations ar e not available to the public; however, a recent variable data set addition was derived containing measured distances from the household and employment location for workers to bus and rail transit. Dataset The data files utilized in the analysis in this study include the nominal rele ase of the 2001 NHTS dataset, including all subsequent geogr aphical area add-on samples to date. These files include Household File, Person File, a nd Travel Day File. The Household File contains data on household demographic, socio-economic, and residence location characteristics for 69,817 households. The Pers on File contains data about personal and household characteristics, att itudes about transportation, a nd general travel behavior characteristics such as usual modes of transportation to travel to work for 160,758 persons. The Travel Day File contains trip -based data on trip purposes, modes, trip lengths in terms of time a nd distance, and trip start times for 642,292 trips. Each

PAGE 22

13 comprehensive file has its own weighting vari able that approximates as accurately as practicable the national estimates for the H ousehold and Person Files, and annualized national estimates in the case of the Day Trip File. (NHTS 2001) New Data At the center and focus of analysis of this paper are four newly a ppended variables that were developed and released to the Center for Urban Transp ortation Research (CUTR) in 2006. These access variables augment the survey da ta file, for each of the data files, and accurately denote scalar distances from the household to transit and from the workplace (if applicable) to transit without revealing any privacy sensitive information or addresses. These new variables include: PTDISTHH ; Distance (in miles) from th e household location to the nearest bus line PTDISTWK; Distance (in miles) from the workplace location to the nearest bus line RRDISTHH; Distance (in miles) from the household to th e nearest rail stop (including light rail, comm uter rail, and subway) RRDISTWK; Distance (in miles) from th e workplace location to the nearest rail stop (including light rail, commuter rail, and subway) Bus route geographical location information cal culated for the new access variables was obtained from the 1995 Federal Transit Administ ration (FTA) database of transit routes for all reporting properties in the United States This route data is considered the most

PAGE 23

14 comprehensive available although it is expected that transit agencies have modified service routes and corridors since the time th e data was assembled. It is believed this data is still an appropriate representation of transit characteristics for 2001, which is conveniently and directly relate d to the time all other analysis variables were obtained. The location of the rail stops is known and current as of 2001. As stated, a realization and complete unders tanding of the dynamic relationship between transit accessibility and service planning a nd design would greatly benefit the transit profession. Numerous research initiatives have previously examined transit usage in relation to demographic vari ables such as age, race and ethnicity, income, auto availability, and gender. The National Hous ehold Travel Survey (NHTS) and similarly formulated regional or local survey data continue to provide a foundation for such analyses.

PAGE 24

15 CHAPTER 4 DISTRIBUTION OF ACCESS TO TRANSIT Background Figure 1 graphically illustrate s the mechanism of the access measured distances provided by the new NHTS data variables. Utilizing Geographical Information Systems (GIS) and related software, a straight-l ine distance is calculated betw een each residential address and the nearest bus route, perpendicular to the route. Bus stop information is present in the data set since a comprehensive and accurate database for nationwide bus stops is not available. Due to the availability of rail st ation information and evidently because of the permanency of a rail stop, they have been provided for in the data and allow for a stoplevel analysis for geographical areas with a rail system operating. Quite possibly, a less desired accurate representation of the distance may exist because a perpendicular distance may intersect the bus line halfway in betw een two stops. As displayed in Figure 1, measured perpendicular distances contained in the NHTS dataset for bus lines may be underestimated due to variances in the networ k walk path to the ac tual transit stop or station. As a result, the act ual walk path may be signi ficantly longer than actually presented in the data. Nonetheless, th e appended access distance dataset provides a unique opportunity to assess the extent of access to transit for the nation.

PAGE 25

16 Minimum Access Concept Generally, a larger percentage of the analysis content in this research effort considers access to bus route networks. Where appropriate to the analysis, both distances from the household to bus and rail are considered whereby a minimum access distance is generated for each household, person, or tr ip. Furthermore, in many cases, a new variable was constructed to denote minimu m access to transit, where transit included either bus or rail. Resulting from the generally higher availability and larger number of bus transit systems present nationally, th e distributions for minimum access would resemble the bus distribution in many cases, as it is determined that access to rail does not significantly affect overall transit acce ss nationally. Minimum access is utilized in the density and access analysis described later in this chapter for ease of comparison to preexisting analyses. Access Measurement From Figure 1, it can be seen that one may wi sh to compensate for an additional walking distance in order to capture a more accurate access distance to a bus stop. Generally, planners assume approximately 4 to 8 bus stops per mile for urban bus routes. Therefore, one might arguably and appropriately increase all the stated bus transit access distances by 0.1 miles to capture the variance in walk distance accounting for an additional one half the average bus stop distance per mile. (Polzin 2006)

PAGE 26

17 Access Distribution Figure 2 and subsequent graphics display th e national cumulative distributions for access to transit. Interval distances of one fift eenth of a mile were chosen to maximize the fineness of scale where statistical sample sizes mathematically allowed. Figure 2 illustrates that almost 50 percent of all individuals nationally live within half of a mile of a bus route. Additionally, a bout 65 percent of all households are located within five miles from a bus line. As illustrated, the slope of the line is a maximum at the close in short distances confirming the significance of the closer proximity, excluding the scale break at 5.11 miles. This of course supports the fact that bus lines are located in populated market areas where a higher population and household density is likely. The slope of the curve remains relatively flat beyond about the 1 mile distance interval. Measured straight-line shortest path distance to route Bus Route Bus Stop Actual walk distance is a function of walk network and path Theoretical maximum additional distance to bus stop = stop spacing Figure 1 Measured versus Actual Walk Access to Bus

PAGE 27

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%<= .15 .31 .45 .61 .75 .91 1.05 1.21 1.35 1.51 1.65 1.81 1.95 2.11 2.25 2.41 2.55 2.71 2.85 3.01 3.15 3.31 3.45 3.61 3.75 3.91 4.05 4.21 4.35 4.51 4.65 4.81 4.95 4.96 5.10 5.11 13 >13 21 >21 29 >29 37 >37 45 >45 52 >52 60 >60 68 >68 76 >76 84 >84 92Distance in Miles (note scale break at 5.11 miles)Percent of Households Figure 2 Cumulative Distribution of Household Distance to Bus Line Figure 3 displays a national cumulative househol d distribution of distan ces to a rail stop. In contrast to the cumulative bus graphic in Figure 2, a significantly lower percentage of households are in proximity to a rail stop. A significantly less number of rail systems exist nationally which influences the shape and fl atter distribution in th is graphic. Figure 3 shows that approximately 10 percent of the national population lives within one mile of a rail station. 18

PAGE 28

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%<= .15 .31 .45 .61 .75 .91 1.05 1.21 1.35 1.51 1.65 1.81 1.95 2.11 2.25 2.41 2.55 2.71 2.85 3.01 3.15 3.31 3.45 3.61 3.75 3.91 4.05 4.21 4.35 4.51 4.65 4.81 4.95 4.96 5.10 5.11 88 >88 171 >171 254 >254 337 >420 500 >500 585Distance in Miles (note scale break at 5.11 miles)Percent of Households Figure 3 Cumulative Distribution of Ho usehold Distance to Rail Stop/Station 19 In the 1995 Nationwide Personal Transporta tion, respondents were asked about their perceived access to transit. As illustrate d in Figure 4, about 50 percent of households interviewed in the 1995 NPTS believed that they lived within one quarter of a mile from a public bus route. The figure compares the perceived acce ss distance by household respondents in the 1995 NPTS, to the measured sample in the 2001 NHTS. The comparison is not ideal due to the effects of systematic or behavioral changes over time; however, a similar service area can be assume d in which case interesting observations can be made. It appears th at over all household distances, the perceived household access distances to bus are consisten tly closer, which suggests that persons may tend to overstate their access to bus. This phenomenon is compounded by the effects of an already assumed greater access distance resulting fr om the probable walk access increase described in Figure 1. The relationship or differences between actual and perceived

PAGE 29

access to transit, as described by this graphi c may be of key importance to industry. As stated, transit access data is t ypically obtained by survey. C onsequently, transit planners have based decisions and planning principles on such research. The implication of a higher degree of accuracy of measurement may have serious implications to the decision making process going forward. 0% 20% 40% 60% 80% 100% <= .1.11 .24.25 .49.50 .99>1.0Distance in MilesPercent of Households Distance to Buslines in Miles from Household Distance to Transit Perceived in NPTS 1995 Figure 4 Cumulative Distribution of Person Distance to Bus Route As shown in Figure 5, a cumulative distributi on of distances to the workplace indicates that approximately 60 percent of workplaces are within a half of a mile to a bus line. The distribution is very similar in shape to the household distribution; however for workplaces, about 15 to 20 percent more workplaces are within the first three quarters of a mile than for households. This shows that a higher percentage of workplaces are in fact in close proximity to transit, which is exp ected as workplaces tend to be more densely and centrally located. (Polzin 2006) 20

PAGE 30

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%<= .15 .31 .45 .61 .75 .91 1.05 1.21 1.35 1.51 1.65 1.81 1.95 2.11 2.25 2.41 2.55 2.71 2.85 3.01 3.15 3.31 3.45 3.61 3.75 3.91 4.05 4.21 4.35 4.51 4.65 4.81 4.95 4.96 5.10 5.11 13 >13 21 >21 29 >29 37 >37 44 >44 52 >52 60 >60 68 >68 76 >7684 >84 92 >92 100Distance in Miles (note scale break at 5.11 miles)Percent of Workplaces Figure 5 Cumulative Distribution of Distance from Work to Bus Route Figure 6 illustrates that only about 10 percent of all wor kplaces are located within one half of a mile from a rail stop or stati on. The distribution indicates that workplace proximity to rail is bout 20 per cent higher within the first five miles than is the case for residences. The relative differences in geogr aphic availability be tween rail and bus in general play a large role in the dist ributions of these cumulative graphics. 21

PAGE 31

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%<= .15 .31 .45 .61 .75 .91 1.05 1.21 1.35 1.51 1.65 1.81 1.95 2.11 2.25 2.41 2.55 2.71 2.85 3.01 3.15 3.31 3.45 3.61 3.75 3.91 4.05 4.21 4.35 4.51 4.65 4.81 4.95 4.96 5.10 5.11 96 >96 188 >188 279 >279 370 >370 460Distance in Miles (note scale break at 5.11 miles)Percent of Workplaces Figure 6 Cumulative Distribution of Distance from Work to Rail Stop Access and Demographic Distribution In Figure 7, a household income bracket dist ribution is plotted against access distance intervals to a bus route. Income brackets were derived from the NHTS variable data; however, every two brackets in that dataset were combined to give $20,000 interval sizes for convenience and improved graphical representation. Several phenomena can be observed. Initially, the highest concentr ation of households for each income group occurs within the first access distance brack et of 0.15 miles. Approximately 37 percent of the under $20,000 income bracket resides with in the closest distance interval. These areas are likely more centralized and in higher density urban se rvice areas. (Polzin 2006) This is expected since historically, lower in come households have been concentrated in older, central urban areas. (Polzin 2006) Typically, these regions or closer distance intervals allow service to mo re patrons in general. 22

PAGE 32

0% 5% 10% 15% 20% 25% 30% 35% 40% 45%<= 0.15 .16 .30 .31 .45 .46 .60 .61 .75 .76 .90 .91 1.05 1.06 1.20 5.0+Distance in Miles (note scale break at 1.2 miles)Percent of Households <= 19,999 20k 39,999 40k 59,999 60k 79,999 80k 99,999 >= 100,000 Figure 7 Distribution of Household Access Distance to a Bus Route by IncomeThe highest income bracket, of greater than $100,000, displays the lowest concentration of households within th is first interval, about 22 percent. It is also evident from the graph that the highest income bracket has the lowest concentration compared to other brackets beyond 5 miles from bus transit. Interestingly, the highest income group has the highest concentration per centage consistently between distances of 0.15 and 5 miles. This observation could arguab ly indicate that a gr eater percentage of higher income persons choose to reside in ar eas likely considered suburban. These areas 23

PAGE 33

typically exhibit expanding access distances. Figure 8 provides some supplemental data relevant to this observation. Middle income brackets, according to Figure 7, appear to maintain their order of distri bution at least up to a 5 mile distance from a bus line. 0% 5% 10% 15% 20% 25% 30% 35% 40% 45%Urban Suburban Rural Second City TownArea TypePercent of Households <= 19,999 20k 39,999 40k 59,999 60k 79,999 80k 99,999 >= 100,000 Figure 8 Distribution of Household Access Distance to a Bus Route by Area Type As a supplement to the information provided by Figure 7, Figure 8 displays residential household area location by the same income br acket using 2001 NHTS data. It can be seen from the figure that the lowest income bracket displays its hi ghest concentration in urban regions in contrast to the highest bracket which experiences its highest concentration in suburban regions. Figure 9 displays the distribution by income by rail stop distances from the household. For rail access, income distribution is less obvi ously related according to the graphic. It 24

PAGE 34

can be seen that the effect of changing income is far less pronounced than for bus distances. Primarily, this is most likely the result of a lessened availability to rail in various markets throughout the country although servic e such as commuter rail may often serve the higher income suburban type markets. (Polzin 2006) 0.0% 0.5% 1.0% 1.5% 2.0% 2.5% 3.0%.01 .15 .16 .30 .31 .45 .46 .60 .61 .75 .76 .90 .91 1.05 1.06 1.20Distance in MilesPercent of Households <= 19,999 20k 39,999 40k 59,999 60k 79,999 80k 99,999 >= 100,000 Figure 9 Distribution of Household Access Distance to a Rail Stop by Income Figure 10 displays household access distance by r ace (not ethnicity). The concentration for White, within the first interval of 0.15 miles is the lowest for those shown, approximately 24 percent. African Ameri can, and Asian only and Hispanic Mexican only display the highest concentrations in the first distance interval to a bus route displaying 56 percent, and 47 percent respecti vely. The subsequent distance categories show a similar order of access concentration where an inverse relationship occurs beyond 25

PAGE 35

5 miles. The findings indicate that the minor ity populations may have the greatest access to transit by proximity. 0% 10% 20% 30% 40% 50% 60% <= 0.15.16 .30.31 .45.46 .605.0+Distance in MilesPercent of Persons White African American, Black Asian only Hispanic, Mexican only Figure 10 Distribution of Household A ccess Distance to a Bus Route by Race 26 Figure 11 illustrates the dist ribution for car ownership ca tegories by bus route access distance nationally. It is evid ent from the graph that zero car households display their highest concentrations within the first measured access dist ance category of 0.15 miles. Interestingly, the order of concentration mimics the number of cars owned per household as indicated by the NHTS data. However, onl y the zero and one car categories achieve their maximum within this first interval ca tegory. A very close concentration for all categories occurs through the next few dist ance categories, with a slightly decreasing percentage for each with rising distance. Notably, beyond a distance of 5 miles from a bus route, nationally, the zero car households exist in the lowest concentration of all categories. Three and four vehicle households are among the highest concentration when

PAGE 36

compared to other categories greater than 5 miles. Not surprisingly, a lower vehicle availability appears to be exactly inversely pr oportional to a bus transit access advantage. 0% 10% 20% 30% 40% 50% 60% 70% 80% <= 0.15.16 .30.31 .45.46 .605.0+Distance in MilesPercent of Households Zero Cars One Car Two Cars Three Cars Four Cars Figure 11 Car Ownership Category, Percent Households by Distance from Bus Route 27 In Figure 12, the Metropolit an Area Size (MSA) categories are displayed by bus line access intervals from national households. It can be seen from the graph that the concentration of households not within any MSA category show the lowest percentages in the first distance interval a nd the highest percentage in th e longer distance, greater than 5 mile interval. This is an expected resu lt from the existence of transit agency bus service that primarily exists in more populated areas consequently considered an MSA of a notable size. Also as expected, the more largely populated MSA categories generally exhibit higher concentrati ons at the closer proximity distance intervals.

PAGE 37

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% .01 .15.16 .30.31 .45.46 .605.0 +Distance in MilesPercent of Households <250,000 250,000 499,999 500,000 999,999 1,000,000 2,999,999 3,000,000+ Not In MSA Figure 12 Metropolitan Area Size Category, Percent Households by Distance from Bus Route Access and Geographic Distribution Figures 13 through 16 are three dimensional gr aphics which illustrate the access for workers to and from transit. Since access to transit in the New York Metropolitan area is unique, the graphics with and without the NY MSA data have been calculated. Access for both rail and bus transit has been delineated utilizing the aforementioned methodology with the stipulation that the sa mple size for the nation excluding the NY MSA is much larger, and that geographi cal areas around the nation are inclusive, particularly all areas that are rural or wher e transit systems are ge nerally not present. These figures assume connectivity among individual transit modes. 28

PAGE 38

.00 .15 .31 .45 .61 .75 9 1 1.05 1 21 1 3 5 0 1 1 5 3 1 4 5 6 1 7 5 9 1 1 0 5 1 2 1 1 3 5 0.0% 0.1% 0.2% 0.3% 0.4% 0.5% 0.6% 0.7% 0.8% Percent of Total Persons Miles from Wk Miles from Home Figure 13 Rail Station Access by Trip End Distance, Nationwide, Excluding NY MSA From Figure 13, one can infer that for the rest of the nation, rail acce ss for any subset of working individuals within a particular dist ance interval is marginal. In fact, those workers that reside in places that are with in 0.15 miles from a rail station, and where there workplace is located w ithin 0.15 miles from a rail stop make up the highest percentage off workers not inclusive in the New York MSA. It can be seen that for any given category in this graphic, the percentage of the tota l is modest resulting from a low overall national access to rail as shown in pr evious graphics. When considering an area where rail access is considered highly preval ent such as the New York Metropolitan area, a very different distribution emerges. 29

PAGE 39

.0 0 1 5 .31 45 6 1 .75 .91 1.05 1.21 1.35 0 1 1 5 3 1 4 5 6 1 7 5 9 1 1 0 5 1 2 1 1 3 5 0% 1% 2% 3% 4% 5% 6% Percent of Total Persons Miles from Wk Miles from Home Figure 14 Rail Station Access by Trip End Distance, Only NY MSA Figure 14 illustrates that within the first fe w distance intervals, a more gradual decrease in the overall percentage of working persons ex ists. Also from the figure, it can be seen that for those intervals where proximity to the workplace is closer than to the household, the percentages are generally hi gher. Notably, the closest a ccess interval for workplaces, not the closest interval for households exhibits the highest concentration of workers in the region. Generally, this agrees with the aforem entioned fact that wor kplaces are typically more centrally located than residences and th erefore are clustered mo re frequently around transit, especially in the New York area. 30

PAGE 40

.01 .15 .31 .45 6 1 .75 9 1 1 .0 5 1 .2 1 1. 3 5 0 1 1 5 3 1 4 5 6 1 7 5 9 1 1 0 5 1 2 1 1 3 5 0% 5% 10% 15% 20% 25% Percent of Total Persons Miles from Wk Miles from Home Figure 15 Bus Route Access by Tr ip End Distance, Nationwide, Excluding NY MSA Figures 15 and 16 each display a similar threedimensional analysis for access with the same geographical delineation but for access to a bus route instead of rail. Perhaps a key indication of these two graphi cs is that when consideri ng all areas in the nation, not including the New York MSA, approximately 20 percent of working travelers are within 0.15 miles from bus transportation for both thei r residence and workplace. In the New York area, more than double that percentage nearly 45 percent of workers in the region have very near access to bus transit. The result is somewhat expected since the bus transit network in New York is considered complex and uniquely dense in comparison to the rest of the nation with few exception. 31

PAGE 41

.01 .15 .31 .45 .61 75 9 1 1 .0 5 1.21 1. 3 5 01 1 5 3 1 4 5 6 1 7 5 9 1 1 0 5 1 2 1 1. 3 5 0% 10% 20% 30% 40% 50% Percent of Total Persons Miles from Wk Miles from Home Figure 16 Bus Route Access by Trip End Distance, Only NY MSA Accessibility, Density, and Area Type In a report by Ross and Dunning, the same report referenced earlier, the topic of land use interaction was explored by analyzing the 1995 Nationwide Personal Transportation Survey dataset. The geographic layout of va rious areas available in the NPTS and NHTS surveys may provide a further insight into the na ture of transit access. Of interest in this analysis, aspects of the relati onships between household distances to transit are correlated to variables for geographical area type and population density. The variable for area type present in both surveys is well suited for comparison since it utilizes the exact same categories for each. Similarly, the population dens ity data is very close in that only the very last interval was slightly modified in the latest survey. Thus, a unique opportunity exists to explore the data across both surv eys. Notably, the directly measured new 32

PAGE 42

appended household access variables for transit were banded slightly different than in the previous graphics so that they were identic al to the intervals de picted in the Ross and Dunning paper for a more appropriate comparison. Transit access data utilized in the 1995 data table was obtained from variables that were reported by householders in contrast to the 2001 dataset appe nded access dataset containing measured data. This comparison offers a unique insight to the differences between perception and measured data despite the fact that bo th surveys were taken some years apart. Access to transit considers th e minimum distance to either bus (commuter, transit) or rail (subway, li ght rail, commuter rail). Table 1 Household Distance to Tr ansit by Population Density, 1995 NPTS People per Square Mile Distance to Transit 0 to 249 250 to 999 1,000 to 3,999 4,000 to 9,999 10,000+ All < .1 mile 18.5% 20.1% 26.0% 38.4% 57.9% 36.0% .1 to .24 mile 2.4% 5.6% 13.0% 17.4% 18.3% 14.3% .25 to .49 mile 3.0% 6.5% 10.4% 13.3% 11.2% 10.8% .5 to .99 mile 18.7% 29.6% 35.1% 25.2% 11.3% 25.1% 1 mile+ 57.4% 38.2% 15.5% 5.7% 1.3% 13.8% Source: Ross and Dunning 33

PAGE 43

From Table 1, it can be observed that for highe r density areas, households within closer proximity intervals are more prevalent. Conversely, for lower density areas, access to transit is much less prevalent in that the highe st percentage of homes is not within close proximity to transit. Thus, in general, as population density in creases, transit access distance decreases. (Ross and Manning 1997) Interestingly, in the 1995 analysis, the closest distance interval of less than 0.1 mile s did not follow the trend exactly in that a significant concentration of households was pres ent in all density a ccess categories. In the 2001 dataset, this phenomenon did not occu r, and the trend was consistent ascending across all categories. Table 2 lists the same categorie s and displays the relationship utilizing the measured transit access data for the 2001 NHTS households. The percentages of lower density areas with l onger access distances and higher density areas with shorter transit distances were much higher in the later dataset. Table 2 Household Distance to Tran sit by Population Density, 2001 NHTS People per Square Mile Distance to Transit 0 to 249 250 to 999 1,000 to 3,999 4,000+ All < .1 mile 3.9% 14.7% 33.4% 53.4% 22.1% .1 to .24 mile 2.0% 10.6% 24.0% 28.5% 14.6% .25 to .49 mile 1.7% 9.2% 13.3% 8.4% 8.2% .5 to .99 mile 2.5% 9.9% 8.7% 4.2% 6.5% 1 mile+ 89.9% 55.7% 20.6% 5.4% 48.6% 34

PAGE 44

35 Table 3 and 4 compare both survey data sets in a similar manner as population density; however, they illustrate categories for geograp hical area type. For the 1995 data analysis, 52.5 percent of persons residing in an urban area are w ithin 0.1 miles from transit. (Ross and Manning 1997) Notably, the 2001 data anal ysis, illustrated in Table 4, shows a much lower percentage of households with access to transit within 0.1 mile than did the 1995 dataset listed in Table 3. As shown, nearly 60 percent of urban resi dences are within 0.1 miles from transit, an increase in percentage over the prior older survey result. This phenomenon agrees with analysis that sugge sts that respondents may tend to overstate their proximity to transit when asked for their perception. Additionally, it may be inferred that a shift of the share of total households has occurred. Several additional factors may contribute to this effect such as area development or redevelopment, service area sizes may have shifted or changed in size, and or geographical land use reclassification may have occurred. The m easured, 2001 data in Table 4 also illustrates the same circumstances for the Town category, and even the Urban category. For a visual comparison of the relative sizes of each category, Figures 23 through 26 are provided in Appendix A.

PAGE 45

36 Table 3 Household Distance to Transit by Area Type, 1995 NPTS Area Type Distance to Transit City Rural Suburban Town Urban All < .1 mile 37.9% 21.4% 28.2% 22.1% 52.5% 36.0% .1 to .24 mile 16.0% 1.6% 13.4% 6.3% 19.6% 14.3% .25 to .49 mile 12.0% 4.9% 11.6% 5.7% 12.0% 10.8% .5 to .99 mile 24.3% 18.3% 34.4% 27.5% 14.3% 25.1% 1 mile+ 9.7% 53.8% 12.3% 38.4% 1.6% 13.8% Source: Ross and Dunning Table 4 Household Distance to Transit by Area Type, 2001 NHTS Area Type Distance to Transit City Rural Suburban Town Urban All < .1 mile 36.4% 1.7% 25.6% 7.2% 59.5% 24.6% .1 to .24 mile 21.0% 0.8% 21.6% 4.6% 27.8% 14.9% .25 to .49 mile 9.3% 0.9% 16.3% 4.2% 7.2% 7.9% .5 to .99 mile 6.5% 0.6% 13.4% 5.6% 2.9% 6.2% 1 mile+ 26.8% 96.0% 23.1% 78.3% 2.6% 46.3%

PAGE 46

37 CHAPTER 5 TRANSIT USAGE AND ACCESS Background Figure 17 illustrates a comparison between bus wo rk trips and the entire set of bus trips as a function of household access distance for the closest three intervals. A sharp decreasing slope is evident beyond the first inte rval which indicates that the work mode share for bus transit trips declines swiftly beyond 0.15 miles from a household. Beyond about a third of a mile distance from trans it, the all-trip mode share drops below 1 percent. For work trips, a 50 percent decrease in mode share occu rs beyond a third of a mile. For bus transit, the number of trips is comparatively low compared to automobile trips, therefore percentages alone do not capture the phenomenon. Notably, from Figure 17, it can be seen that the overall share of wo rk trips using bus tran sit is higher for each category thus illustrating the importance of th e work trip. (Polzin 2006) The decreases in share beyond 0.15 miles indicate that there is a distinct walk distance limit that travelers are willing to undertake. Historic ally, it has been accepted that individuals undoubtedly greatly value their time, and that walk trip dist ances beyond one quarter of a mile are generally undesirable. Some factor s influencing the propensity for shorter walk trip distances include but are not limited to weather conditions, physical conditioning, safety, and total allotted travel time.

PAGE 47

0% 1% 2% 3% 4% 5% 6% 7% <= .15 .16 .30 .31 .45MilesBus Mode Share Bus All Trips Bus Work Trips Figure 17 Bus Trip Mode Share by Household Distance Mode Share Figure 18 displays a bus work trip share of those persons within their own vehicle ownership category. The vehicle categories include those trips taken by persons who do not have access to a vehicle and all other pers ons taking trips, who ha ve access to at least one vehicle. The mode share is not typical in the sense that a disproportionate share of total work trips are taken by those with vehicl e access. Additionally, due to a diminished sample size and low percentage of trips within these subcategories, this data is presented by share within each own access distance interval. As displayed in the figure, over 55 percent of trips taken by those who live within 0.15 miles fr om a bus route and have no vehicle available make a bus transit trip. Conversely, for those person-trips made by individuals who live within 0.15 miles from a bu s route and have indicat ed that they have access to a car, only about 6 percent choose the bus transit mode. As expected, the figure 38

PAGE 48

indicates that a high propensity for bus transit use exists when no vehicle is available, but this propensity diminishes with distance from a bus route in favor of some other mode alternative. <=.1 5 1 6 .3 0 .3 1 .45 N o C a r A t l e a s t O n e C a r 0% 10% 20% 30% 40% 50% 60% Bus Work Mode Share Miles from Houehold Figure 18 Share of Bus Work Trips within Vehicle Availability Category by Bus Route Access Distance, Nationwide Matrix Mode Share For Figures 19 through 28, a three-dimensional analysis of mode share is developed to explore the percentage of tran sit trips (both bus and rail) th at were chosen within each particular access interval. The intervals rese mble a matrix of cells of individual work trips that fall into the specific access distance categories for both residences and 39

PAGE 49

workplaces. The percentages displayed repres ent mode share within each cell. This analysis of the data was developed to presen t the data by a visual method that relates the individual trip choices of workers to acces sibility on a finer scale than previously analyzed. < = 1 5 1 6 .30 < = 1 5 3 0 0% 1% 2% 3% 4% 5% 6% 7%Bus Work Trips Miles from Work Miles from Home Figure 19 Bus Transit Work Trip Mode Share by Trip End Distance to Bus Route, Nationwide Figure 19 exhibits a bus transit work trip m ode share by trip end distance for a national work trip. Illustrated is the access distance interval to transit from the household and from transit to the workplace for a given work trip. Two intervals are shown due to the lessening of market sample size beyond given distances. The highest mode share exists where the distance categories to and from a bus route are in the minimum categories. Thus, a higher percentage of individual bus transit trips are made where the proximity to transit is the closest on each trip end. 40

PAGE 50

<=.15 .16 .30 .31 .45 .46 6 0 61 7 5 < = 1 5 1 6 3 0 3 1 4 5 4 6 6 0 6 1 7 5 0% 10% 20% 30% 40% Rail Work Mode Share Miles from Wk Miles from Home Figure 20 Rail Work Mode Share by Trip End Distance Interval to Rail Station, Nationwide Similar to the previous graphic, Figure 20 i llustrates the work mode percentage for rail person-trips when correlated to rail accessibility for both the household and the workplace. It can be seen from the gra ph that a higher percentage within distance category occurs where the household distance is shortest, less than 0.15 miles and where the distance from the rail stat ion to individuals workplace is just under three quarters of a mile. Interestingly, the rail mode choice percentage for these proximity users is higher for rail than for bus nationwide, but the percen tage of users of tran sit bus declines with trip end distance. On the contrary, the access category mode share for rail tends to 41

PAGE 51

42 increase slightly with distances up to about three quarters of a mile and then again showing a decline. In contrast to national trends, New York ar ea rail trip percentages by access category far surpass that of bus transit. Many of the pr oximity distance categories from a rail stop to residences and workplaces for workers in the New York MSA exhibit approximately a 20 percent share for rail within each category, up to the half mile access distance intervals. Due to a less robust sample size available in the NHTS data for this market share, many intervals could not be shown fo r the same analysis for bus in Figure 21. However, from Figure 21, it can be seen that the local mode c hoice percentage for bus is less than that of rail for the New York MSA. Historically, in this region, ridership on rail has surpassed that of bus transit, especially for the work commute. (Source) This is as expected due to the usually higher overall speed of travel of th e heavy rail system in New York City. Vehicle speed of travel, stop intervals, and su rface traffic all play a role in the mode choice decision in New York City, in additi on to the obvious choice constraints resulting from available of desired origin-d estination pairs a nd transfers.

PAGE 52

<=.15 .16 .30 .31 .45 .46 .6 0 < = 1 5 1 6 3 0 3 1 4 5 4 6 6 0 6 1 7 5 0% 10% 20% 30% 40% 50% Rail Work Mode Share Miles from Wk Miles from Home Figure 21 Rail Work Mode Share by Trip End Distance Interval to Rail Station, Only NY MSA < =.15 16 .30 < = 1 5 1 6 3 0 3 1 4 5 0% 5% 10% 15% 20% 25% Bus Work Mode Share Miles from Wk Miles from Home Figure 22 Bus Work Mode Share by Trip End Distance Interval to Bus Route, Only NY MSA 43

PAGE 53

44 CHAPTER 6 ACCESS LOGISTIC REGRESSION MODEL Introduction Transportation forecasting usually begins with utilization of the traditional four step process. The four components of the forecas ting model consist of trip generation, trip distribution, mode choice, and r oute assignment. Notably, transit accessibility can play a large role in mode choice analysis and modeling. Afte r exhaustive usage of crosstabulation and correlation analys is of various contributing fact ors, it may be desirable to analyze the effects of many cont ributing factors at the aggregat e or disaggregate level. For instance, linear regression, or more appr opriately, logistical regression may be suitable to explore in mathematical models fo r possible predictability in mode choice. Binary choice logistic regression has been wi dely utilized in econometric analysis to investigate travel behavior (Racca and Ratledge 2004). The binary model is based on the following mathematical convention: Y=1 if Bx + u >=0, Y=0 otherwise. Where y is a choice outcome for behavioral re sponse such as mode c hoice, x is a vector of attribute variables, and B is a vector of parameters.

PAGE 54

45 The field of travel demand modeling includes nu merous in depth research work into the exploration and predictability of an individuals travel behavior choices. In this paper, a general logistic regression m odel is conducted only to expl ore the possibility that the inclusion of a measured accessibility variable will improve a given model. It is hypothesized that the significance of such a model will improve, more than if the variable were a perceived access response variable. From extensive literature in the topic area, st udies indicate that many factors play a role in transit usage and mode choice. As previ ously mentioned, some of the variables that may be considered relevant and subsequently utilized in a predic tive regression model include, level of service vari ables, land use and geographic variables, socioeconomic and demographic variables, and accessibility or distance variables, although these are not exhaustive as arguably an infinite number of characteristics may be considered. Transit Mode Choice Regression Model Tables 5 through 8 list the results for a tr ansit model using the national NHTS sample variables. The variables were chosen based on traditional utilization in some classic mode choice models as annotated in the li terature. The Beta coefficients for each categorical variable are listed in the second column of the tabl e. In the third column, the standard error for each variable is listed. Si gnificance of a given variable in the model is determined by a ratio between the coefficient an d its standard error term, which is labeled the Z-Statistic in column 4 of both tables. SPSS provides th e resulting Wald statistic when calculating the model, which is the squa re of the aforementioned Z-ratio. Finally,

PAGE 55

46 the overall significance of each variable is li sted in the last column and provides an indication of how relevant the variable is when included in the equation and subsequent model. It should be noted that even though a va riable may be very si gnificant, it is not guaranteed to play a vital role in the overall equation. Highe r Wald statistics indicate stronger influences. Lower significance values or those close to zero, indicate a higher parameter relevance to the model. Table 5 and 6 list the coefficients and result s for the models with and without the access distance variables for the un-weighted sample of workers present in the NHTS dataset. Table 7 and 8 utilize exactly th e same variables but display the results of the model when the NHTS national person weighting factor is applied to the variables. That is, the total number of working persons in the models annotated by Tables 7 and 8, is expanded to include the entire population concerning workers. In a classic travel demand model, variables related to trip characteristics are typically included, but notably, are not ut ilized in this model. As mentioned, the NHTS dataset does not provide for service characteristics or measured temporal characteristics, therefore, the model is performed usi ng demographic and geographic variable information only while the objective of the varyin g models is to indica te the effects of the inclusion of the measured access variables on th e predictability of transit mode choice. Importantly, this model utilizes variables from the person file and relates them to the variable for an individuals usual mode choice for the prior week. Variables for

PAGE 56

47 household family income, respondent age, geog raphic area type, vehicle availability, and access distance to a bus route were utilize d. Arguably, geographic area type may be considered an exception to trad itional usage in this type of model, but was included because of the inclusion in the cross-tabula tion analysis earlier in this report. The variables were reclassified from the numer ous categories provided in the NHTS variable data set and grouped into less ca tegories of a more general nature before analyzing with SPSS. The mode choice variable, or usual-mode variable was recoded to indicate a one if bus transit was chosen as the primary mode, or zero if otherwise. Only workers were considered. Additionally, instances of missing or not available data were filtered from the set of utilized variable s. Essentially, the equation was modeled around a propensity to choose bus transit based on demographics while analyzing for bot h the inclusion and exclusion of the access distance component.

PAGE 57

48 Table 5 Model Results, Un-weighted Variables Not Including Measured Access B S.E. Z-stat Wald Sig. R_AGE_17 (Cat) 5.432 0.246 R_AGE_18 TO 29 (Cat) 0.353 0.354 0.997 0.994 0.319 R_AGE_30 TO 49 (Cat) 0.072 0.213 0.338 0.114 0.735 R_AGE_50 TO 64 (Cat) -0.055 0.203 -0.271 0.073 0.787 R_AGE_65 (Cat) 0.131 0.21 0.624 0.390 0.532 HHFAMINC_LOW (Cat) 51.487 0.000 HHFAMINC_MID (Cat) 0.906 0.135 6.711 45.047 0.000 HHFAMINC_HIGH (Cat) 0.439 0.087 5.046 25.712 0.000 HHVEHCNT_AVAIL (Cat) -2.244 0.095 -23.621 556.578 0.000 HBHUR_URBAN (Cat) 172.923 0.000 HBHUR_SUBURBAN (Cat) 1.067 0.09 11.856 139.584 0.000 HBHUR_RURAL (Cat) -1.612 0.385 -4.187 17.492 0.000 Constant -2.008 0.231 -8.693 75.834 0.000 Hosmer and Lemeshow 0.111

PAGE 58

49 Table 6 Model Results, Un-weighted Variables Including Measured Access B S.E. Z-stat Wald Sig. R_AGE_17 (Cat) 5.118 0.275 R_AGE_18 TO 29 (Cat) 0.371 0.367 1.011 1.022 0.312 R_AGE_30 TO 49 (Cat) 0.149 0.218 0.683 0.465 0.495 R_AGE_50 TO 64 (Cat) 0.029 0.209 0.139 0.019 0.891 R_AGE_65 (Cat) 0.216 0.215 1.005 1.005 0.316 HHFAMINC_LOW (Cat) 48.320 0.000 HHFAMINC_MID (Cat) 0.9 0.137 6.569 43.348 0.000 HHFAMINC_HIGH (Cat) 0.414 0.088 4.705 22.305 0.000 HHVEH_AVAIL (Cat) -2.214 0.096 -23.063 530.607 0.000 HBHUR_URBAN (Cat) 118.304 0.000 HBHUR_SUBURBAN (Cat) 0.977 0.095 10.284 106.016 0.000 HBHUR_RURAL (Cat) -1.15 0.399 -2.882 8.325 0.004 PTDISTHH (Continuous) -0.266 0.082 -3.244 10.539 0.001 PTDISTWK (Continuous) -0.035 0.024 -1.458 2.153 0.142 Constant -1.954 0.238 -8.210 67.423 0.000 Hosmer and Lemeshow 0.305

PAGE 59

50 Table 7 Model Results, Weighted Variables Not Including Measured Access B S.E. Z-stat Wald Sig. R_AGE_17 0.000 0.000 R_AGE_18 TO 29 0.827 0.007 118.514 0.000 0.000 R_AGE_30 TO 49 0.245 0.005 48.628 0.000 0.000 R_AGE_50 TO 64 0.054 0.005 10.775 0.000 0.000 R_AGE_65 0.092 0.005 17.664 0.000 0.000 HHFAMINC_LOW 0.000 0.000 HHFAMINC_MID 1.016 0.003 391.707 0.000 0.000 HHFAMINC_HIGH 0.795 0.002 444.226 0.000 0.000 HHVEHCNT_AVAIL -2.302 0.002 -1190.550 0.000 0.000 HBHUR_URBAN 0.000 0.000 HBHUR_SUBURBAN 0.601 0.002 341.853 0.000 0.000 HBHUR_RURAL -4.463 0.038 -116.924 0.000 0.000 Constant -1.931 0.005 -355.259 0.000 0.000 Hosmer and Lemeshow 0.000

PAGE 60

51 Table 8 Model Results, Weighted Variables Including Measured Access B S.E. Z-stat Wald Sig. R_AGE_17 15908.32 0.000 R_AGE_18 TO 29 0.467 0.008 57.411 3295.966 0.000 R_AGE_30 TO 49 0.358 0.005 68.485 4690.229 0.000 R_AGE_50 TO 64 0.147 0.005 28.491 811.7415 0.000 R_AGE_65 0.217 0.005 40.211 1616.922 0.000 HHFAMINC_LOW 249911.5 0.000 HHFAMINC_MID 1.127 0.003 426.065 181531.7 0.000 HHFAMINC_HIGH 0.794 0.002 430.211 185081.3 0.000 HHVEHCNT_AVAIL -2.217 0.002 -1110.73 1233726 0.000 HBHUR_URBAN 88189.29 0.000 HBHUR_SUBURBAN 0.533 0.002 273.791 74961.49 0.000 HBHUR_RURAL -4.279 0.038 -111.988 12541.42 0.000 PTDISTHH -0.024 0.001 -18.443 340.1576 0.000 PTDISTWK -0.002 0.000 -180.480 32573.03 0.000 Constant -1.979 0.006 -347.186 120538.2 0.000 Hosmer and Lemeshow 0.000

PAGE 61

52 Model Results In both sets of models, both with and without the access variable, it is evident that the vehicle availability variable with its relative ly high negative Beta value indicates a strong propensity not to use transit when a vehicle is available to the individual. This result is expected since a person with no vehicle ava ilable for use has a more limited choice set for their work trip. In fact, the vehicle vari able dominates the equation in each case. The income variable was categorized by low in come being less than $20,000, medium income between $20,000 and $50,000, and high income above $50,000. The medium and high income group shows a positive relationship for bus transit mode when compared to the low income group. This is an expected result as alternatives to transit tend to increase with income level. The variables with the lowest significance in the un-weighted model were the age groups. This lower value of significance is not unexpe cted, since the effects of age over the unweighted sample may be dynamically biased. Thus, this variable becomes a less appropriate predictor unless the sample size is expanded significan tly. Subsequently, when expanding the sample using the NHTS wei ghting variable factor in the second set of models, namely Table 7 and 8, the categoric al age variables increased in significance. The variables included in the analysis were m easured relative to the lowest age category, less than 17 years. All but one of the category coefficients was positive against the lowest in the un-weighted model, notably, the 50 to 64 year old age group, indicating a negative propensity for transit. Among the other three models, the age variables were all

PAGE 62

53 positive; however, the higher age groups do exhibit the lowest positive coefficient which may indicate a higher likelihood for tran sit than in the other age groups. The addition of the access distance variables from household to transit and from transit to the workplace for workers slightly increased the overall significan ce of the nationally unweighted model, as indicated by the Hosmer and Lemeshow goodness-of -fit test. In the weighted model, the Hosmer and Lemeshow test did not exhibit significance which may be a direct indication that the model is improved by the a ddition of other variables and warrants even further analysis. Perhaps most importantly in this analysis, the addition of the continuous distance variab les for the household and the workplace for individuals, resulted in the application of slight negativ e Beta coefficients thus indicating an overall negative propensity for transit use with distance as expected.

PAGE 63

54 CHAPTER 7 CONCLUSION Introduction Understanding transit usage has become a critical transportation research interest and policy goal. As stated, this research effort presents results of an analysis of the 2001 NHTS data specifically focusing on the newly released appended variables that measure access or distance to public transportation. Actual relationships between public transportation and traditi onal household and person char acteristics nationwide are explored by analyzing correlations between demographic and geographic variables. Notably, both inclusions and ex clusions in analysis are c onducted due to the widely accepted ubiquitous transit network present in the New York region. Additionally the contrasting distributions betw een New York and the rest of the nation are noteworthy and may be considered a more finely tuned analys is of transit access fo r planners. Overall, the observations imply a very high importance of close proximity transit to for travelers. Transit Access Access to transit is well served by an analysis of measured data versus perceived data due to the complexity of issues involved. The an alyses reveal strong differences in household and workplace access to transit as a function of race, income, auto ownership, urban area size, and population density. Additionally, a very high sensitivity to access is evident.

PAGE 64

55 Approximately 53 percent of households are w ithin a mile of bus service and 40 percent within a quarter-mile. Approximately 10 per cent of the population liv es within one mile of rail. Nearly 60 percent of urban residences are with in 0.1 miles from transit, a significant increase from prior survey analysis Over 50 percent of workplaces are within a quarter mile walk radius of a bus line. Not surprisingly, work is more closely concentrated near transit than are residences. Transit Choice Mode choice analysis in relation to tr ansit access distance ov erall suggests a high preference for users to be very near transit services. Mode share for transit declines approximately two thirds beyond the first in terval beyond 0.15 miles fr om a bus route. The more urban an area, the better transit access is. It has been shown that typically, some transit dependent groups such as zero vehicle householders ha ve an advantage in greater access to transit, as expected. One explanation for the differences in measured versus actual usage may be attributable to non -user segments of transit not being aware of transit proximity or service t hus accounting for deviances from prior survey. This may in fact be quite useful to future planning due to the higher degree of accuracy for access data, and the lessening of uncertainties. Going Forward This analysis may still be considered the tip of the iceberg in regard to planning tools. Many factors weigh into the planning and ulti mate success of transit systems, and this analysis of measured access contributes only a fraction to a comprehensive

PAGE 65

56 understanding. Increased ridership is one of the key metrics for success for these systems. It is recognized th at this research effort main tains focus on essentially one aspect of transit service supply. Importantly, some of the analysis of this research effort can be continued in the future as advances in technology and data collection techniques allow for a more accurate and measured databa se in aspects such as accessibility and other service supply variables re lated to frequency or span. Perhaps future studies will in clude or append to the dataset a higher number of measured variables that may likely include such variables or information as agency service area size, service frequency. Additionally, comprehe nsive stop-level and ro ute data, or actual origins and destinations at the trip level coul d be captured in the dataset and as measured variables for analysis. Going forward, it is expe cted that there will be an increase in the reliance and usage of Geographic Informati on Systems (GIS) in constructing future databases, as some of the technology alrea dy exists to analyze and manipulate the large geographical databases as those addressed by the appended dataset. Even beyond that technology which is current utilized by the Federal Transit Administration (FTA) and Federal Highway Administration (FHWA), the near future may reveal newer, better techniques. As an improvement upon this work, the curren t analysis technique could be enriched by the future addition of more accurate data and information in the dataset. For example, the four newly appended NHTS distance variable s may be calculated with an increased accuracy. The actual distance measurement could be calculated as a shortest path

PAGE 66

57 distance instead of a Euclidean or straightline distance. The stre et and roadway path network is known, and the current technology may allow for such calculation. Thus, the variables could be appended with the improved distance calculation more accurately capturing access distance for a household. As a first step related to this analysis, it is recommended to construct a bus stop level database to more accurately describe bus transit access. For the bus route access calculation (see Figure 1), a newer bus route dataset could be constructed, improved over the older dataset, if stop-level data was appe nded to the data. It is conceded that bus stops and service changes more dynamically th an rail and are usua lly shorter-distance spaced, however, the accuracy of the access distance measurement would undoubtedly be significantly improved. As evidenced from the analysis, there exists a high sensitivity to short access distances for choice transit users. It is understandable that this influences, in part, a traditionally lower mode share for transit. Of key importa nce is the ability to relate access to mode choice more closely than previously believe d. Quite possibly a key contributor to the success of future transit networks may be pl anning for a higher threshold level of transit access to the population for both rail and bus.

PAGE 67

58 REFERENCES Beimborn, Edward A., Michael J. Greenwal d, Xia Jin. 2003. Transit Accessibility and Connectivity Impacts on Transit Choice and Captivity. Center for Urban Transportation Studies and Department of Urban Planning. University of WisconsinMilwaukee, Transportation Research Board. Cervero, Robert. 1993. Ridership Impacts of Tr ansit-Focused Development in California. UCTC No 176, Institute of Ur ban and Regional Development. University of California at Berkeley, Berkeley, CA. Chu, X. and S. Polzin. 1998. Public Tran sit in America: Findings from the 1995 Nationwide Personal Transportation Surve y, Center for Urban Transportation Research, Tampa, FL. Federal Transit Administration, National Transit Database. 2006. http://www.ntdprogram.com/ntdprogram. Polzin, S. and X. Chu. 2005. Public Transit in America: Results from the 2001 National Household Travel Survey, Center for Ur ban Transportation Research, Tampa, FL. Polzin, S. and E. Maggio. 2006. Working Pape r. Exploring the Ava ilability of Public Transportation Services Through Analysis of the National Household Travel Survey Appended Data. Center for Urban Tran sportation Research. Tampa, FL. Pucher, J. 2002. Renaissance of Public Tran sport in the United States? Transportation Quarterly Vol. 56, No. 1. Washington D.C. Pucher, J. and J. L. Renne. 2003. Socioeconom ics of Urban Travel: Evidence from the 2001 National Household Travel Survey. Tr ansportation Quarterly Vol. 57, No. 3. Washington D.C. Pucher, J. and J. L. Renne. 2004. Urban-Rural Differences in Mobility and Mode Choice: Evidence from the 2001 National Household Travel Survey.

PAGE 68

59 Racca, David P. and Edward C. Ratledge. 2004. Factors That Affect and/or Can Alter Mode Choice. Prepared for Delaware Transporta tion Institute and The State of Delaware Department of Transportation. University of Delaware, Newark, DE. Ross, C. L. and A. E. Dunning. 1997. Land Use Transportation Interaction: An Examination of the 1995 NPTS Data. Atlanta, GA, Federal Highway Administration, U.S. Department of Transportation. Taylor, Brian. 2002. Increasing Transit Riders hip: Lessons from the Most Successful Transit Systems in the 1990s. MTI Report-0122. The Mineta Transportation Institute. San Jose State University, San Jose, CA. U.S. Department of Transportation, National Household Travel Survey, 2001. http://nhts.ornl.gov/2001/index.shtml 2001 NHTS Users Guide. U.S. Department of Transportation, National Household Travel Survey, 2001. http://nhts.ornl.gov/2001/usersguide/index.shtm

PAGE 69

60 APPENDICES

PAGE 70

Appendix A Geographic Household Access 0% 20% 40% 60% 80% 100%Percent of Persons < .1.1 to .24 .25 to .49 .5 to .99 1+Distance in Miles 0 to 249 250 to 999 1,000 to 3,999 4,000 to 9,999 10,000+ Figure 23 Household Access to Transit by Density 1995 NPTS (Percent Persons per Square Mile) 0% 20% 40% 60% 80% 100%Percent of Persons < .1.1 to .24 .25 to .49 .5 to .99 1+Distance in Miles 0 to 249 250 to 999 1,000 to 3,999 4,000+ Figure 24 Household Access to Transit by Density 2001 NHTS (Percent Persons per Square Mile) 61

PAGE 71

Appendix A (Continued) 0% 20% 40% 60% 80% 100%Percent of Persons < .1.1 to .24 .25 to .49 .5 to .99 1+Distance in Miles Second City Rural Suburban Town Urban Figure 25 Household Access to Transit by Area Type 1995 NPTS 0% 20% 40% 60% 80% 100%Percent of Persons < .1.1 to .24 .25 to .49 .5 to .99 1+Distance in Miles Second City Rural Suburban Town Urban Figure 26 Household Access to Transit by Area Type 2001 NHTS 62


xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001920099
003 fts
005 20080104095633.0
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 080104s2006 flu sbm 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001836
035
(OCoLC)187303350
040
FHM
c FHM
049
FHMM
090
TA145 (ONLINE)
1 100
Maggio, Edward.
0 245
Access to public transportation :
b an exploration of the National Household Travel Survey appended data
h [electronic resource] /
by Edward Maggio.
260
[Tampa, Fla] :
University of South Florida,
2006.
3 520
ABSTRACT: Understanding transit usage has become a critical transportation research interest and policy goal. This thesis presents results of an analysis of the 2001 NHTS data specifically focusing on the newly released appended variables that measure access or distance to public transportation. Statistically significant public transportation distance intervals from households and individuals were chosen for analysis in relation to other key variables in the original dataset. Actual relationships between public transportation and traditional household and person characteristics nationwide are explored, specifically focusing on both rail and bus transit modes for the work commute trip. Geographically, both inclusions and exclusions in analysis are conducted due to the widely accepted ubiquitous transit network present in the NY region.^ The analysis reveals strong differences in household and workplace access to transit as a function of race, income, auto ownership, and urban area size. Additionally, a very high sensitivity to access exists suggesting that the share of transit accessible trips is smaller than previously acknowledged. Approximately 53 percent of households are within aviimile of bus service and 40 percent within a quarter-mile. Approximately 10 percent of the population lives within one mile of rail. Over 50 percent of workplaces are within a quarter mile walk radius of a bus line. Not surprisingly, work is more closely concentrated near transit than are residences. Furthermore, mode share for transit declines approximately two thirds beyond the first interval beyond 0.15 miles from a bus route.^ These observations imply a high value to services in close proximity to residential areas.Historical work in this topic area include geographically specific data analysis obtained from surveys which potentially allow a degree of subjectivity in perceived responses whereas accessibility and distance data analyzed in this thesis are actual and spatially measured. Additionally, a regression model exploring the significance of actual access to transit upon mode choice is performed to explore the significance of influence by measured access variables. The analysis suggests that access is even more critical than might have previously been acknowledged by the transit planning profession.
502
Thesis (M.S.E.S.)--University of South Florida, 2006.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
538
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
500
Title from PDF of title page.
Document formatted into pages; contains 62 pages.
590
Adviser: Steven Polzin, Ph.D.
653
Demographic.
Distribution.
Mode.
Ridership.
Transit.
Trip.
Accessibility.
690
Dissertations, Academic
z USF
x Engineering Science
Masters.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.1836