Time-cost relationship of building projects: statistical adequacy of categorization with respect to project location

Construction duration is one of the key elements to determine the success of a project. Several models have been developed to provide precise predictions of construction duration. The datasets in previous studies comprise observations from particular countries. This precludes examination of the role of project location at the country level for modelling construction duration. The aim, therefore, is to test the statistical adequacy of categorizing projects with respect to their locations for modelling construction duration. A formal procedure to evaluate the adequacy of categorizing observations is also presented. It is intended to contribute to the epistemology of modelling construction duration. The dataset was generated from a global client. It contains information on 246 projects from 25 different countries. This creates a categorical variable for location. Bromilow's time-cost model is chosen to assess the adequacy of this categorization. A full model with indicator and interaction variables is compared to a restricted one. Parameter estimation was performed according to the ordinary least squares method. F-statistics are calculated and the probability of the value is significant. The results show that two distinct groups with substantial changes in model parameters exist in the dataset. It is also noted that conclusions are valid over the selected population.