Exploring the Potential for Predicting Project Dispute Resolution Satisfaction Using Logistic Regression

The success of a construction project depends on the coordinated efforts of project team members. This is especially crucial when a project is in dispute and hence the achievement of satisfactory project dispute resolution is critical to project success. This proposition has been empirically demonstrated a previous research that studied project dispute resolution satisfaction (DRS) using multivariate discriminant analysis (MDA). This paper reports on a study that builds on that research, with the specific aim of predicting project DRS through the use of logistic regression (LR). In this study, a LR model of project DRS (Model 1) is developed, and then compared with the MDA model. The findings suggest that the LR technique provides a higher hit rate and thus a higher proportion of correct classification. With the wider acceptance of the use of alternative dispute resolution (ADR) methods, the effect, on the LR model, of changing the demarcation between adverse and favorable project DRS is also examined. For this examination, another LR model (Model 2) was developed. It is believed that Model 2 may reflect the prevailing sentiment that ADR is viewed as an amicable way to resolve disputes. Both the MDA model and LR models (Model 1 and Model 2) indicated that "design changes" are the root cause of adverse project DRS. Within the scope of the project data, these findings suggest that design changes are not just disruptive to project progress but also a critical cause of construction disputes.