Using multivariate techniques for developing contractor classification models

Abstract

Today’s growing numbers of contractor selection methodologies reflect the increasing awareness of the construction industry for improving its procurement process and performance. This paper investigates contractor classification methods that link clients’ selection aspirations and contractor performance. Multivariate techniques were used to study the intrinsic link between clients’ selection preferences, i.e. project‐specific criteria (PSC) and their respective levels of importance assigned (LIA), during tender evaluation for modelling contractor classification models in a data set of 68 case studies of UK construction projects. The logistic regression (LR) and multivariate discriminant analysis (MDA) were used. Results revealed that both techniques produced a good prediction on contractor performance and indicated that suitability of the equipment, past performance in cost and time on similar projects, contractor relationship with local authority, and contractor reputation/image are the most predominant PSC in the LR and MDA models among the 34 PSC. Suggests contractor classification models using multivariate techniques could be developed further.

Citation

Wong, C.H., Nicholas, J. and Holt, G.D. (2003), "Using multivariate techniques for developing contractor classification models", Engineering, Construction and Architectural Management, Vol. 10 No. 2, pp. 99-116. https://doi.org/10.1108/09699980310466587