In June 1999, the Idaho Transportation Department (ITD) in collaboration with the Department of Commerce and other agencies participated in a statewide travel survey, “1999-2000 Idaho Resident and Nonresident Motor Vehicle Travel Survey.” Using the highway intercept method and a mail-back questionnaire, the survey collected data on traveler characteristics, trip characteristics, and many other variables from resident and nonresident travelers in Idaho. This report describes the activities that have been completed for the Idaho Statewide Trip Generation Rates and Friction Factors project, the purpose of which was to use the survey data to develop Idaho-specific trip generation rates and friction factors.
The Idaho statewide travel survey was conducted between October 1999 and October 2000 at 56 sites based on a stratified random sampling method to make sure that nearly everyone who traveled in the state had an equal chance of being sampled. A total of 7284 questionnaires were returned. Because many returned questionnaires did not provide complete household and trip information, only 4285 questionnaires were useable for this project. An additional problem with the returned questionnaires was discovered in spring 2000: many respondents filled out the trip diary section of the questionnaire only for one person, instead of every household member. A one-page addendum was therefore developed and inserted in the questionnaire for the summer and part of the fall of 2000. A much simpler survey instrument, the addendum only asked for trip destinations and trip sequence made by other members of the household (i.e., besides the one who filled out the questionnaire). A total of 874 addendums were returned. In addition to the survey data, this project compiled a base year 1999 socioeconomic database using data from the U.S. Census Bureau, and the Idaho Division of Finance Management.
Cross-classification was the method for deriving trip production rates. Trip data from the questionnaire and the addendum were grouped by trip purpose into home-based work (HBW), home-based recreation (HBR), and home-based other (HBO). Household data were classified by income, household size, and number of vehicles owned. Three income brackets were used: low (less than $20,000), medium ($20,000-$50,000), and high (greater than $50,000). Both household size and number of vehicles owned were grouped into 1, 2, 3, and 4+. Trip production rates based on the questionnaire data appeared substantially lower than trip generation rates published in reports by NCHRP and other states. Trip production rates based on the addendum data were more reliable and could be used for Idaho statewide traffic demand modeling.
The development of trip attraction models was based on the trip purposes of HBW, HBO, HBR, and NHB (non-home-based). An initial experiment with regression models showed unsatisfactory results because of low R-Square values, collinearity between the independent variables, and negative regression coefficients. This project then developed a method, in which the model for each trip purpose was based on a set of activity indicators and the rates were calculated using the state-level data. These attraction models were later adjusted for unclassifiable trips in the survey data and persons under 14 years of age.
The ratios of attractions to productions by trip purpose ranged from 0.45 to 0.88 using rates derived from this project. These ratios fell outside the range of 0.90 to 1.10 recommended by FHWA. The imbalance could be the result of inadequate socioeconomic estimates, trip rates, or both. Because production rates are generally considered more trustworthy than attraction rates, attractions are usually adjusted to match productions by trip purpose at the end of the trip generation step in statewide travel demand modeling, thus resolving the problem of imbalance.
This project used travel time data from the survey and the Gamma impedance function recommended by the FHWA to develop the Idaho statewide friction factors. The Gamma function parameters were calibrated for each trip purpose of HBW, HBO, HBR, and NHB. Two time intervals, 15 and 30 minutes, were used to characterize the trip length frequency. The Gamma function was fit by log-linear transformation and linear regression, and the two time intervals were evaluated by comparing the respective root-mean-square errors. The curve with the lowest root-mean-square error was then chosen to represent the full trip length frequency distribution for each trip purpose.
The multi-purpose statewide travel survey did not entirely fulfill the data needs for this project. The questionnaire was too long and complicated for many respondents. This resulted in less reliable data and a lower response rate. Fortunately, some useful information was extracted from the questionnaire for determining friction factors and attraction rates. The addendum, although used only for part of the project, provided more useful data for estimating trip production rates and would be a desirable means for collecting trip data in future surveys.
This project has been beneficial to transportation planners in Idaho in several ways. First, experience from this project can certainly be used to improve the future survey process and its reliability and applicability to statewide transportation modeling. Second, the trip rates and friction factors derived from the project have provided for the first time Idaho-specific data that can be compared to data published by NCHRP and other states as well as data to be available from the 2001 National Household Travel Survey (NHTS) and the Census 2000 Journey to Work Survey (JTW). Third, the trip rates and friction factors derived from the project.