A Group Decision-Making Aggregation Model for Contractor Selection in Large Scale Construction Projects Based on Two-Stage Partial Least Squares (PLS) Path Modeling

Modern projects are normally characterized by huge investments, long construction periods, and complex technology. Therefore, the selection of an appropriate contractor for smooth project delivery is challenging. Previous studies attempted to develop suitable frameworks for contractor selection. However, correlation among indicators, the subjectivity of indicator weights, and heterogeneity among experts' professional capabilities for selecting contractors were not successfully removed from the decision-making process. Typical partial least square (PLS) path modeling can solve these problems. However, it can only solve problems with the same direction correlation among the indicators (e.g., the correlation coefficients of the indicators are all positive). For indicators with different direction correlations, path modeling is helpless. Aiming to overcome this limitation, this research introduces a group decision model based on two-stage PLSs path modeling. This decision model can eliminate correlation among indicators and the impact of subjectivity and heterogeneity among experts on the reliability of weighting schemes; more importantly, it can effectively solve the problem of having correlation indicators with different directions. Through a literature review, we first established an indicator system of contractor selection on large scale construction projects. Second, a two-stage PLS path modeling combined with the maximization of deviations principle was proposed as an aggregation approach for performance evaluation. Finally, a comparison was made between the two-stage and typical PLS path modeling methods through a case study, which was conducted to validate the reliability of the new approach.