This paper discusses the application of maximum likelihood analysis to the prediction of long-term accident rates at road/rail grade crossings. Theoretical concepts of the method are developed, including significance testing, and it is shown that traditional regression techniques are not usually applicable to sparse data of this type since accidents cannot be assumed normally distributed. Various model forms are developed and discussed from both a theoretical and practical viewpoint. Finally, the application of maximum likelihood methods to a fairly large set of accident data is described and some general conclusions given. /TRRL/

  • Record URL:
  • Availability:
  • Corporate Authors:

    ARRB Group Limited

    500 Burwood Highway
    Vermont South, Victoria    3133
  • Authors:
    • HERBERT, A J
    • Smith, NMH
  • Publication Date: 1976-9

Media Info

  • Features: References; Tables;
  • Pagination: p. 24-33
  • Serial:

Subject/Index Terms

Filing Info

  • Accession Number: 00147872
  • Record Type: Publication
  • Source Agency: Transport and Road Research Laboratory (TRRL)
  • Files: ITRD, TRIS, ATRI
  • Created Date: Feb 16 1977 12:00AM