Traffic accidents result in heavy social and economic burdens on the society and the safety aspects ought thus to be an important factor upon priority rating of road investments. In order to take accident costs into account it is necessary to derive a method for calculating the number of accidents to be expected for alternative design standards of the different road projects. In this pilot study it is attempted through the use of multiple regression analyses, to develop mathematical models that describe the quantitative functional relationship between accident density (the number of traffic accidents per km) and number of road and traffic factors. Data for this pilot study have been available from the road accident (traffic) inventories as well as the roadway element inventories compiled by the Directorate of Roads. A total of 769 km main roads in the counties of Ostfold, Sor-Trondelag and Nordland have been investigated. The road-system was divided into 50 subsections, on which a total of 206 accidents were recorded by the police. The following parameters were tested as variables: roadway width, angle of deviation, type of road surfacing, volume/capacity ratio, ADT, number of driveways per km, number of intersections per km, sight conditions, gradient, and sudden changes in road standard. For the observation data available it has been shown, at a confidence level of 95%, that the accident density increases with: increasing ADT, increasing density of intersections, decreasing roadway width, decreasing volume/capacity ratio, decreasing sight distances, and decreasing gradient. The best statistical model based on the above mentioned 6 significant variables results in a coefficient of correlation of 0.964. The relative small number of data employed results however in the model being rather sensitive to changes of the parameter values. General use of the derived model can therefore not be recommended at the present stage. It is necessary to carry out more extensive investigation employing a large number of observation data and to improve certain parameter combinations. In addition to the 6 variables found significant, it is recommended that the other variables mentioned above be included in further studies.

  • Corporate Authors:

    Utvalg for Trafikksikkerhetsforskning

    Oslo,   Norway 
  • Authors:
    • Berg-Andreassen, T
  • Publication Date: 1971-10


  • Norwegian

Media Info

  • Features: References; Tables;
  • Pagination: 37 p.

Subject/Index Terms

Filing Info

  • Accession Number: 00098859
  • Record Type: Publication
  • Source Agency: National Safety Council Safety Research Info Serv
  • Report/Paper Numbers: No. 16
  • Files: TRIS
  • Created Date: Nov 5 2003 12:00AM