Two procedures are discussed for calculating cell weights to be applied to sample data to ensure that the expanded sample jointly represents the population on several background distributions. Deming's procedures minimize the squared deviation of cell weights from unity, whereas Johnson's procedure iteratively produces weights that match the marginal distributions, are all positive, and have low variance from unity. The theory and calculation procedure of these two approaches are developed and are illustrated by a simple example. Advantages and disadvantages of the procedures are described, and applications to various transportation problems are highlighted. The paper concludes that, although the Johnson technique is less well grounded in theory, it produces reasonable weights, prevents negative estimates, and converges quickly. Its use is recommended. /Author/

Media Info

  • Media Type: Print
  • Features: References; Tables;
  • Pagination: pp 73-76
  • Monograph Title: Transportation system analysis
  • Serial:

Subject/Index Terms

Filing Info

  • Accession Number: 00195975
  • Record Type: Publication
  • ISBN: 0309028221
  • Files: TRIS, TRB
  • Created Date: Sep 15 1979 12:00AM