Enhancing construction productivity through continuous process improvements

Abstract

This dissertation presents a series of regression models that relate job site productivity to process improvement initiatives executed both before and during construction. Applied during early project stages, these models help predict the expected value of productivity based on certain inputs related to pre-construction planning and construction execution. These models demonstrate the strong relationship of project performance to a variety of process improvement initiatives including design completeness, definition of a project vision statement, testing oversight, project team stability, and project manager experience and dedication. The correlational research methodology targeted seventy-five projects representing approximately $274.53 million in civil construction. The data collection effort considered 47 process improvement initiatives (independent variables) using 106 quantitative and qualitative measures. The modeling technique involved the use of multiple linear regression, a method that exploits available data from multiple, independent sources to focus on specific outcomes. Models were developed from both owner and contractor information and were subjected to rigorous statistical analysis. The models provide project managers with a deliberate yet practical approach to project management and productivity enhancement. The dissertation results include verification analysis of the models using project data excluded from the data base that supported model development, sensitivity analyses aiding in beneficial application and interpretation, and an extensive discussion of the models' usefulness and limitations.