Aetiology and Progression of Construction Disputes towards a Predictive Model

Peipei Wang, Lihan Zhang, Kun Wang, Peter Fenn

Research output: Contribution to journalArticlepeer-review

Abstract

Construction projects are at risk of dispute. Dispute prevention is thus expected to avoid that cost. This paper aims to establish a model predicting the occurrence of disputes and identifying strategies regarding resource allocation in dispute avoidance. Factors in the model were identified by a literature review and conceptualised into aetiology and progression of dispute formation based on an analogy with epidemiological investigations. The model structure was established upon this analogy, validated against Bradford Hill criteria, and then quantified by conducting a Bayesian network analysis with samples returned from a questionnaire survey. The model shows high accuracy rates with both complete and incomplete input data. Finally, a sensitivity analysis was conducted to simulate the results of potential management measures available to be taken. This predictive model not only serves the prediction function but also traces back to factors causing disputes, and hence can be used to assist decision making before and during a construction process. At the centre of the model sits the causal mechanism investigation, but the existing methods only shed lights on correlational research without causal consideration. This paper devised an epidemiological investigation which covers from causal proposition to causal examination and contributes greatly to model establishment.

Original languageEnglish
Pages (from-to)1131-1143
Number of pages13
JournalKSCE Journal of Civil Engineering
Volume25
Issue number4
Early online date12 Feb 2021
DOIs
Publication statusPublished - 16 Apr 2021

Keywords

  • Bayesian networks
  • Causation
  • Construction disputes
  • Epidemiology
  • Prediction

Fingerprint

Dive into the research topics of 'Aetiology and Progression of Construction Disputes towards a Predictive Model'. Together they form a unique fingerprint.

Cite this