Multiattribute weighting models for best-value selection in public sector design-build projects

With the adoption of the design-build delivery method in the public sector, development of theoretical tools for selecting a design-builder has become imperative. The core of design-build is to select the most appropriate proposal. Best-value is a process used in design-build to select the most advantageous offer by evaluating and comparing factors in addition to cost or price. The best-value process can be represented as a linear additive formula. This is done by multiplying the score for each factor by the weight of that factor, and then adding all those weighted scores together. Weight assignment to evaluation factors has been based on two strong rationales. The first is that public agencies weigh the evaluation factors on the basis of direct subjective weight assignment. The second is that there is no reason to justify or check the weights because they are predetermined by experts. For these two reasons, there has been a lack of theoretical research interest in the domain of the assignment of weights. Because of the increase in the use of design-build delivery method in the public sector, this study of weight assignment is proper and timely. This thesis begins with questions in regard to the conventional way of assigning weights. Alternative weighting models were tested in a best-value selection process. An experimental case-based survey was carried out to compare and contrast five weighting models (ROC, Ratio Method, Swing Weighting, Indifference Weighting, and Pairwise in AHP) in a process of best-value selection. Three distinctive groups—United States expert group, South Korean expert group, and non-expert group—were involved in the test. As one of the conclusions, a framework for applying weighting models is proposed. This research project suggests four determinants: evaluation factors, scoring confidence, preferential certainty, and group decision. Three weighting models—Rank Order Centroid (ROC), Ratio, and Pairwise Comparison in Analytic Hierarchy Process (Pairwise in AHP)—are recommended in cooperation with the characteristics of the determinants. This project makes five principal contributions. First, it provides a basic analogy for evaluation of proposals, so that it can help public agencies to enhance efficiency in construction procurement. Second, this project extends the general study domain of weight assignment in MADA in a process of best-value selection, which is an impetus for researchers in construction industry. Third, this study can benefit researchers in decision science as a reference on weight assignment. Fourth, it offers two classification systems—project delivery methods and weighting models. Finally, this research develops surrogate measurement system to check the consistency of resulting weights by using statistical techniques.