NON-RANDOM SAMPLING IN THE CALIBRATION OF DISAGGREGATE CHOICE MODELS

Most disaggregate travel demand models have been developed using random samples of households and trip-makers. However, it is also possible to use two types of non-random sampling in the calibration of demand models. The first, stratified sampling, uses different sampling rates based on the values of independent variables in the model. The second sampling process, termed choice-based sampling, is based on the observed choices made by travelers. Four basic conclusions are reached regarding alternative sampling procedures and the development of disaggregate choice models. (1) Disaggregate choice travel demand models calibrated on stratified samples require no special adjusting or weighting. (2) Disaggregate choice travel demand models calibrated on choice-based samples do require adjusting or weighting. (3) The appropriate weight for modifying each choice-based sample observation is simply: Probability of a randomly drawn sample making that choice; probability of a choice-based sample making that choice. (4) It is possible to update existing travel demand models using supplemental random, stratified or choice-based sample data using a formula based on Bayes' rule.

  • Corporate Authors:

    Cambridge Systematics, Incorporated

    100 Cambridge Park Drive, Suite 400
    Cambridge, MA  United States  02140

    Federal Highway Administration

    Office of Highway Planning, 1200 New Jersey Avenue, SE
    Washington, DC  United States  20590
  • Authors:
    • Lerman, S R
    • Manski, C F
    • Atherton, T J
  • Publication Date: 1976-2

Media Info

  • Pagination: 56 p.

Subject/Index Terms

Filing Info

  • Accession Number: 00094629
  • Record Type: Publication
  • Source Agency: National Technical Information Service
  • Report/Paper Numbers: Final Rpt., FHWA/PO-6-3-0021
  • Files: NTIS, TRIS, USDOT
  • Created Date: May 14 1976 12:00AM