| Popis: |
This thesis aims to conduct a review and survey to establish the appropriateness of the types of risk management techniques currently used in the construction industry, to investigate risk perception in construction and its impact on project performance, and to develop a procedural model for the elicitation of expert opinions about risks that minimises the adverse effects of risk perception on individual estimates of risk, and provides these opinions as input variables to the rigorous and probabilistic analysis of contractual risks. The work is a cross-cultural study, applying mail questionnaire surveys, interviews, Delphi and Vignette techniques, and analysing risk management approaches and applications of the elicitation model developed by the study in both United Kingdom and Ghana. The data generated by the elicitation model are analysed using relative likelihood methods to develop subjective prior probability distributions for use as input variables in the Bayesian analysis of contractual risks in construction. The study concludes that although relative frequency data are often unavailable for contractual risks, existing predominant practices for contractual risk analysis are inappropriate for the nature of contractual risks. Furthermore, individual perceptions about risks significantly affect expert judgements about risks (and consequently project performance) in spite of their expertise. Using the expert elicitation model developed by the study and the analytical approaches applied, it is possible to capture, encode and aggregate the knowledge and experiences of a group of relevant experts to derive probability distribution functions of contractual risks to be applied as input variables to a Bayesian analysis of contractual risks, and thereby achieve a more appropriate, systematic and rigorous approach to contractual risk analysis. Evidence from the study also indicates that this approach need not involve any significantly high costs as the analysis can be done using standard spreadsheet software and add-in programmes that companies already have on their computer systems. |
