Statewide Rural-Urban Bus Travel Demand and Network Evaluation

This paper examines the characteristics of intercity bus riders within Tennessee and proposes methods to identify service gaps and prioritize network expansion, particularly focusing on rural-urban connections. Data were collected through an on-board survey and compared with intercity auto trips. Compared to personal auto users, intercity bus riders are more likely to be of minority races, unemployed, unable to drive, and from low-income households. Five demand levels were determined based on the population distribution with these characteristics. The service areas of existing bus stops were identified and compared with the high demand areas. The result shows that an insufficient number of stops are located in high demand area. Still, approximately 80 percent of stops connect to meaningful destinations such as hospitals. The results imply that bus stations are well-connected to destinations but poorly connected to potential riders. Changes to the current network could better cover high-demand areas. Introduction Scheduled intercity bus service declined by one-third between 1960 and 1980, and remaining service declined by half between 1980 and early 2006 in the United States (Schwieterman et al. 2007). With rising travel demand, escalating gasoline prices, Journal of Public Transportation, Vol. 15, No. 3, 2012 98 and Federal Transit Administration (FTA) 5311(f) funding for intercity bus agencies to provide or continue service, the industry is beginning to see more ridership. In recent years, fixed-route and scheduled intercity bus service have experienced a renaissance. It was rated as the fastest-growing mode of intercity transportation, outpacing air and rail transportation in 2010 (Schwieterman and Fischer 2010). In Tennessee, an Intercity Bus Demonstration Program was implemented in 2008 in response to growing public intercity travel needs, particularly focused on connecting rural and urban areas. There is a growing number of fixed-route, scheduled intercity bus services in the state. In the context of this study, we focus on intercity buses operating within the state of Tennessee and not directly connecting to different states, even though they do feed interstate bus and air terminals. In general, the intercity bus services described in this paper are short-haul buses that connect rural regions with urban centers. Approaches to modeling intercity and interstate travel have evolved over the decades. Several early papers applied econometric methods to investigate choice behavior (Ashiabor, Baik, and Trani 2007; Koppelman 1989; Morrison and Winston 1985; Stopher and Prashker 1976). Because of jurisdictional and funding boundaries, within-state rural transit is a special class of transit service that deserves special study. However, little contemporary research has been directed to within-state intercity bus services, particularly in the context of recent demographic changes and growth in demand. This paper investigates emerging rural-urban bus travel patterns. A framework is presented to supplement locally-collected bus rider data with general long-distance travel described in the National Household Travel Survey (NHTS) (representing very few intercity bus trips) and apply this method to the Tennessee bus network. The paper describes results of intercity bus rider surveys that explore rider and trip characteristics of intercity bus users and contrasts those results with intercity car travel from the NHTS. Those data were used to estimate and map high demand regions to provide a framework to analyze rural-urban intercity bus service and connectivity between potential riders and their destinations. The paper highlights previous relevant research in the next section, followed by a description of the survey methods. Previous Intercity Bus Studies Intercity Travel Demand Model Review A few papers have presented mode choice modeling approaches for intercity surface travel. Ashiabor et al. (2007) reviewed disaggregate nationwide travel demand Statewide Rural-Urban Bus Travel Demand and Network Evaluation 99 modes developed by Stopher and Prashker (1976), Grayson (1981), Morrison and Winston (1985), and Koppelman (1989) between 1976 and 1990. All four models used versions of National Travel Surveys (NTS), and they included bus as one of the transportation modes. The fifth attempt to model nationwide travel demand was carried out by Ashiabor et al. (2007), who developed a logit model based on 1995 American Travel Survey (ATS). However, both the 1977 NTS and 1995 ATS collected information only on trips of 100 miles or more, eliminating intercity trips shorter than 100 miles, which includes most in-state Tennessee intercity bus trips. Fravel et al. (2011), in “Toolkit For Estimating Demand For Rural Intercity Bus Service” (TCRP Report 147), introduced a toolkit to estimate demand for rural intercity bus corridors. The toolkit includes two demand estimation models, a regression model and a trip rate model, both of which give more accurate results for current rural intercity bus services than previous efforts to model intercity bus demand. However, this toolkit has limitations that cannot be used for regional transit, which includes much of the rural-urban bus service. A Minnesota intercity bus network study (KFH Group 2010) chose five transitdependent population characteristics to profile persons who rely on transit. Potential intercity bus needs were identified by comparing the locations served by the current network with the locations in Minnesota that have concentrations of persons more likely to rely on public transportation. The limitation with the Minnesota study is that characteristics of transit-dependent populations, which are mainly determined by urban public transit riders, could differ from intercity bus rider characteristics. This means that using transit-dependent population characteristics to determine areas of high intercity bus needs could introduce some bias. Also, identifying locations with high intercity bus need is not enough to evaluate an intercity bus network; how well the network connects to the destinations also should be studied. This paper extends the Minnesota study by comparing the characteristics of Tennessee rural-urban intercity bus riders to the characteristics of general travelers to obtain specific characterizations of potential riders. Intercity Bus Riders Characteristics A 1981 Tennessee intercity bus study (J. R. Wilburn and Associates 1981) developed a survey to ascertain a profile of passengers. It was conducted for a 24-hour period at several bus terminal locations. The survey results showed that a typical intercity bus passenger is age 16–25, uses the bus once a year to visit friend and relatives, travels over 10 miles by auto to get to and from the terminals, and come from households with total income of $7,501–$15,000 per year, which was lower Journal of Public Transportation, Vol. 15, No. 3, 2012 100 than the 1981 national median household income of $18,033. Findings indicated some variance in automobile ownership between cities. In Chattanooga, Memphis, and Nashville, most respondents indicated that they owned one automobile, while in Jackson and Knoxville, most respondents indicated that they did not own an automobile. The Bureau of Transportation Statistics (BTS) 1995 rider study (Bureau of Transportation Statistics 1997) concluded that intercity bus riders are more likely to be persons ages 65 years and older, female, minority, and less educated, who live in households with low income and no personal use vehicle available. Although the BTS study provides a good description of long distance intercity bus rider characteristics, the study parallels the scope of the ATS, focusing on people who travel more than 100 miles. Although these two studies have given a comprehensive view of intercity travel mode choice modeling and intercity bus rider characteristics, both are obsolete, and there is a gap in the literature on within-state long distance bus traveler characteristics, particularly trips linking rural areas with urban centers. Therefore, it is crucial to obtain information about intrastate long-distance travelers in order to determine their characteristics and identify the areas of potential demand. This paper addresses this gap and evaluates how rural-urban intercity bus rider population demographics are different from overall intercity traveling populations. In addition to identifying high intercity bus demand areas and assessing the connectivity of current network to those areas, methods to evaluate the connections of riders to destinations are introduced using a Geographic Information System (GIS) framework. This study proposes a framework to evaluate existing intercity bus network effectiveness at connecting probable intercity bus riders to destinations and introduces ways to improve these connections. Although this study observes only Tennessee rural-urban intercity trips and determined most to be less than 100 miles, it is reasonable to assume, because of similar state geography, that other states also have many intercity trips that are less than 100 miles. Survey Methods To gather information from intercity bus users, a questionnaire was developed for riders of each intercity bus route supported by the FTA 5311(f) program that funds fixed-route intercity bus service. This group included 5 rural transit companies (3 human resource agencies and 2 private service providers) that provide 756 Statewide Rural-Urban Bus Travel Demand and Network Evaluation 101 route-miles of service in Tennessee. The surveys were conducted between May 1 and August 21, 2010. Intercity bus passengers were asked about their trips and to provide personal information. The survey explored trip and demographic characteristics. Two survey methodologies were used. First, passengers are approached and interviewed by surveyors at different bus stops or onboard. This type of survey has a high response rate, a high quality of data collected, and allows surveyors to collect open-ended observations from riders. However, a considerable drawback of this method is that it has high cost for interviewers, owing to low bus service frequency, dispersed bus stop locations, and relatively few riders. A pilot intercept survey was performed to test the method. During the two-day pilot, 27 riders were interviewed. Another survey method distributed questionnaires to bus riders with the help of the driver. Survey packages were distributed to the transit agencies, and drivers gave the surveys to boarding riders along with a pencil and mail-back envelope. This survey method had a relatively low response rate, but it greatly increased the cost-effectiveness of the data collection. Using this method, 446 questionnaires were sent out and 92 were returned (21% response rate). The true response rate is somewhat uncertain because we were unable to confirm that all surveys were actually distributed to passengers. Also, because of lack of supervision, some surveys were returned incomplete. Survey Analysis and Comparison Results Considering the low number of intercity bus trips recorded in the 2009 NHTS (i.e., nationwide, only 48 trips made by intercity bus out of 62,968 trips of greater than 30 miles), it is difficult to model intercity bus travel from this dataset. Indeed, the NHTS does not record any trips in Tennessee made by intercity bus, making it impossible to follow traditional mode-choice modeling strategies. Therefore, we adopted an alternative approach to estimate potential intercity bus rider demand. The data for all intercity trips made in Tennessee by all modes were extracted from NHTS. We supplemented the NHTS data with our on-board survey data. Comparing data in our survey to the dataset extracted from NHTS illustrates the diverging characteristics of intercity bus riders and their trips from car-based transportation. Furthermore, intercity bus rider attributes can serve as a reference to determine the number of potential intercity bus riders in each census tract in Tennessee; this Journal of Public Transportation, Vol. 15, No. 3, 2012 102 was converted to estimate likely intercity bus rider population densities. Identifying these areas of population density helped to determine the areas with higher potential intercity bus demand. The characteristics of intercity bus riders and trips were summarized from the survey responses and compared with those of intercity car trips of the same range of travel distance, extracted from the 2010 NHTS. For our survey, recorded trip lengths range from 6 miles to 162.5 miles (2 trips were recorded at less than 30 miles, which may be the result of misunderstanding the survey questions or a writing error). We defined the shortest length of an intercity trip as 30 miles, while the upper limit was rounded to 170 miles. The intercity bus trip distance includes distance from rider origin to boarding bus stop, travel distance on the bus, and distance from alighting stop to destination. A geographic criterion was used to filter the data from NHTS so that only trips made within Tennessee were selected. This was done to ensure consistency with the scope and administrative boundaries of the study. Because the trip origin and destination are unknown in the NHTS data, the state in which a survey responder’s household is located was adopted as an alternative means to select the trips made in Tennessee. The filtered NHTS dataset included 1,116 intercity trips distributed among all modes in Tennessee. Figure 1 shows that 1,075 trips were made by non-public transportation and no trips were made by intercity bus. Of these trips, 129 were made by private vans, which could include commuter vanpools. Figure 1. Transportation Mode Choice Percentage of NHTS Intercity Trips Statewide Rural-Urban Bus Travel Demand and Network Evaluation 103 In addition to transportation mode choice, 11 trip and rider characteristics were compared, including race, gender, age, employment status, ability to drive, household annual income, household size, number of vehicles available in household, education level, trip purpose, and trip distance. We assumed that these 11 characteristics influence a traveler’s mode choice and, therefore, are included in both the NHTS and our survey. Comparing the NHTS intercity trips (mostly car) in Tennessee with our dataset revealed significant differences in all variables with the exception of gender, shown in Table 1. Table 1. Demographic Comparison between NHTS and Intercity Bus Trips Variable Name Category NHTS Percentage Survey Percentage P-value Race White 94 86 0.0079 Sex Male 56 53 0.6357 Employment status Employed 63 46 0.0014 Capability to drive Able 97 91 0.0036 Household income Under $15,000 12 49 <0.0001 $15,000-$27,499 9 20 $27,500-$52,499 27 8 $52,500-$89,999 32 8 $90,000 and over 20 14 Household size 1 6 31 <0.0001