«Technical Report Documentation Page 1. Project No. 2. Government Accession No. 3. Recipient's Catalog No. SWUTC/11/161127-1 4. Title and Subtitle 5. ...»
The 1995 ATS defined a long-distance trip as at least 100 miles one-way. The ATS (BTS, 1997) reveals that during the one year period, American households made 685 million long-distance trips (1 billion of personal trips), over 95% of which were to destinations within the United States. Of the trips with destinations inside the U.S., 45% were to destinations inside the traveler’ home state. Personal use vehicle (PUV) was heavily used for long-distance travel. The total longdistance PUV trips were about 505 million from 1995 to 1996, resulting in over 280 billion vehicle miles of travel (VMT) on the nation’s highways. The PUV were most used for shorter trips – 37% of the PUV were less than 300 miles round trips and 68% of the PUV trips were less than 500 round trips. On the contrary, commercial air was mostly chosen for longer trips – 72% of commercial airplane trips were 1,000 miles or more round trips.
Of all personal trips made during the survey year, 23% were for business, 30% were for leisure activities, one third were to visit friends or relatives, and 15% were for personal business such as attending wedding or funerals, or participating in school-related activities. The ATS also revealed the temporal pattern of long-distance trip making – the largest share of travel in 1995 occurred during the third quarter, July through September. It should be noted that household in Texas made more long-distance trips than the national average – the 1995 ATS data show that in Texas, about 87% of household took one or more long-distance trips to a destination 100 or more miles away, comparing to 80% of household in the nation.
The 1995 ATS has been used and analyzed by researchers for understanding intercity travel.
Many of the work were presented in the 1999 “Personal Travel: The Long and Short of It” conference held at Washington, D.C., and published in the Transportation Research E-Circular (E-C026). Several researchers concluded based on the ATS dataset that income is an important factor affecting intercity travel behavior including travel frequency and mode choice (Addante, 2001; Chin & Hwang, 2001; Georggi & Pendyaly, 2001; Hwang, et al., 2001; Mallett, 2001).
Mallett ((2001) compared the long-distance travel behavior of low-income people to the entire population. He found that Low-income people made much less long-distance travel than the entire population (1.6 per year vs. 3.9 per year) in 1995; for low-income people, one of the most important limiting factors is vehicle availability since air travel was beyond the means of most poor family. In addition, business and leisure trips are much more sensitive to income than those to visit friends and relatives or personal business. Georggi & Pendvaly (2001) conducted similar research analyzing long-distance travel (trip frequency, trip purpose, trip mode, trip distance, trip duration, and travel party size) across age groups and income groups, in considerable detail for the elderly and the low income. Their findings confirmed the conclusion that the low income (along with the elderly) had significantly lower log-distance mobility when compared to other segments of the population.
Based on the 1995 ATS data, Chin & Hwang (2001) assigned person vehicle trips and household vehicle trips between metropolitan areas to major highways to identify major passenger corridors in the U.S. They identified such corridors by different trip purposes, income levels, as well as private vehicles and buses. They found the most traveled highway corridors were concentrated on east and west coast and there was no significant east-west cross-country corridor. In addition, pleasure trips including recreation and visiting family and friends as well as trips made by the middle-income household group dominated the long-distance highway travel in 1995. They also indicated that business trips showed concentration between nearby metropolitan areas while nonbusiness travel showed distinct northeast and Pacific coastal corridors. In Chin & Hwang’s research, the most prominent corridor to Mexico was identified as starting from the Dallas and Houston areas passing through San Antonio and entering Mexico via the border of city of Laredo.
Moreover, Chin & Hwang pointed out that information on travel between cities that are less than 100 miles apart was missing from the ATS, which could lead to some incomplete understanding about long-distance travel.
The 1995 ATS data were also used to analyze intercity air travel. Addante (2001) studied the air travel market in the New England region. He revealed for air travel, the business and nonbusiness split was about the same, yet for all mode travel, the split was about 30% to 70%, and the most significant class of air travelers were well educated, high income, married male with children. Hwang et al. (2001) evaluated the relationship between the accessibility of the air transportation system and the demographic and socioeconomic status of travelers. They found that higher income households were more likely to choose air travel; more than half of all outbound air passenger drove to the airport and parked their vehicles at the airports, and the majority of travelers were either picked up by private vehicle or used a rental car to get from the airport to their final destination. Moreover, business travelers were more likely to drive to the airport and park and pleasure travelers were more likely to be dropped off at the airport.
The 1995 ATS data is a rich resource for intercity travel model development. O’Neil & Brown (2001) used the ATS data to develop a long-distance, non-business trip generation model using a cross-classification approach. The authors estimated the cross-classification trip rate for metropolitan and non-metropolitan areas separately based on household income and household type. The trip rates were defined by income level (low, medium, and high) and by household types (married with kids, married without kids, single parent with kids, family or non-family without kids, and non-family not living alone). It should be noted that the results of this research showed that whether there are kids or not would affect the long-distance travel decision. The authors also pointed out the ATS data have about half sample in metro areas and half in nonmetro areas. Then the nearly equal sample size of households in metropolitan and nonmetropolitan raised concerns that the metropolitan long-distance trips might not be well represented in the survey.
Another research done by Thakuriah et al. (2001) applied a gravity model to estimate patterns of demand between 50 metropolitan areas using the ATS data. The authors estimated the cost parameters by Maximum Likelihood and estimated the origin and destination parameters using a variant of Iterative Proportional Fitting. They defined different scenarios in which different cost variables including only distance, only time and combination of distance and time were used and concluded that model with the variable of combination of distance and time had slightly better result. The authors highlighted challenges associated with estimating inter-city demand from the ATS data – the need for better cost data since the cost data needed for model are not directly available from the ATS data set and need to be constructed from exogenous sources.
Richardson & Seethaler (2001) worked on the survey method for long-distance trips, specifically the selection of the period of observation for long-distance travel. The selection of the period of observation for long-distance travel has more difficulties than daily travel because of the infrequency and irregular characteristics of long-distance travel – too short period will not be able to catch enough long-distance travel while too long period will bring recall errors or put too much burden to respondents who make frequent long-distance travel. The typical long-distance travel survey defines a two-three month window and used retrospective recall methods while the 1995 ATS used prospective recording where respondents were given a diary in advance of the period and asked to record trips as they occurred. The authors proposed the "most recent trip survey method" for long-distance trip, which does not restrict a time period, but ask respondents the most recent long-distance trip they made no matter when. One of the most difficulties of using data obtained from such survey is how to calculating trip rate. The authors also proposed a probability method to calculate the trip.
Other than the 1995 ATS, the 1995 National Person Travel Survey (NPTS) also asked respondents questions about their long-distance travel. Hu & Young (2001) compared these two long-distance data sets. They summarized several major differences: The NPTS used residential telephone numbers as sample frame while the ATS used an address-based sampling frame (the ATS could capture more low-income household); The NPTS defined 75 miles (one way) as longdistance while the ATS defined 100 miles (one way) as long-distance; The NPTS had a 2-week window while the ATS had a 12-month window; The NPTS excluded college students and individuals younger than 5 years old while the ATS included all of them; ATS trip distances were calculated based on real route distance and from zip code centroid to zip code centroid while NPTS trip distances were calculated based on the great circle distance (22% shorter) and from MSA centroid to MSA centroid.
5.2. Census Transportation Planning Package (CTPP) and National Household Travel Survey (NHTS) The two other most comprehensive travel data currently available to public include the Census Transportation Planning Product (CTPP) and the NHTS. The CTPP is a set of special tabulations derived from the decennial Census long form questionnaire which was sent to approximately one in six households containing all of the questions on the short form plus additional detailed questions relating to the social, economic, housing characteristics of each individual and household. In the long form questionnaire, six questions related to the journey to work were asked to survey respondents who worked at least 1 hour during the "last week" before the survey date. The CTPP data include three parts containing residence end data, place of work data and journey-to-work flow data, respectively. The long form questionnaire has been stopped after the 2000 Census.
The NHTS is a nationwide travel survey sponsored by the Bureau of Transportation Statistics (BTS) and the Federal Highway Administration (FHWA) to collect data on daily travel by the American public. The NHTS survey obtained information of individual travel for a single day across all days of the survey year, and the people traveling, their household and their vehicles.
Compared to the decennial Census, the NHTS has a smaller number of samples; about 1 out of 1500 households in 2001 NHTS and 1 out of 750 households in 2009 NHTS were represented.
Yet, the NHTS data include the work trip distance reported by respondents, by which commuters can be easily grouped; the NHTS data also contain detailed personal and household information of individual travelers, which allows a thorough search of factors that influence long distance commuting decisions.
Base on their particular characteristics, the two datasets have been studied separately with different emphases. Since the CTPP has a larger sample coverage and contains journey-to-work flow data, in this proposed research, it was used to map the inter-metropolitan work flow in the Texas Triangle area. The NHTS data were then used to summarize the general characteristics of long distance commuters and long distance commuting trips, as well as to examine the relationship between socioeconomic characteristics of commuters and long distance commuting decisions. Due to the relatively small sample size of long distance commuters, the NHTS data were mainly analyzed at the national scale.
LDC/TRC commuting in this research should have two basic characteristics. First, the commute crosses the boundary of a metropolitan area; second, the length of the commuting trip should be at least 50 miles.
The criteria of long distance commuting vary among studies and there is no consensus on the minimum distance that constitutes long distance commuting. The US Census Bureau defines 90 minute one-way work trip as extreme commuting, and the US NHTS defines 50-mile one-way travel as long distance trip. Several European countries also defined long distance commuting based on survey results which varied between 15 kilometers and 100 kilometers (Sandow, 2011).
Commuting trips with one-way distance at least 50 miles are defined as long distance commuting in this research following the US NHTS's long distance trip concept. According to the NHTS, the average commuting distance in the US is about 12 miles, 50-mile commute is about four times of that distance.
The LDC/TRC commuting can be conducted daily or weekly when the distance between residences and work places exceeds the tolerable daily commuting length. Weekly commuting involves staying in a secondary residence near work place one or several nights each week.
Weekly commuting cannot be accurately distinguished from others in the NHTS data. For this analysis, 50-mile commute is considered as long-distance commuting. When the commute reaches 100 miles, approximately 2-hour drive, it is assumed that people will start weekly commuting. When the commute reaches 300 miles or more, it is assumed that weekly commuting is no longer feasible for driving, and people would prefer flight or reduce travel frequencies to residence.
5.3. The Analysis of the CTPP data
The 1990 and 2000 CTPP journey-to-work flow data were used to monitor the changes in the commuting flows between the four major metropolitan areas during the ten-year period.