Frequency of change orders in highway construction using alternate count-data modeling methods

A common problem at state transportation agencies is the inability to complete projects within the original scope of work. Change orders, which are contractual documents issued to accommodate the additional work in a contract, are generally due to root causes such as design errors, unexpected site conditions, and weather conditions, and intermediate causes such as bidding characteristics. At the preaward phase of project management, an improved understanding of the factors that are associated with change orders will be of value and also can serve as a basis for taking steps to reduce concomitant contractual aberrations such as time delay and cost overruns. Recognizing that the occurrence of change orders is a count variable, this paper analyzes the frequency of change orders using a variety of count-modeling methods including the negative binomial, Poisson, zero-inflated negative binomial, and zero-inflated Poisson. Using 5 years of contract data from Indiana highway projects, appropriate models are estimated to assess the influence of project type, contract type, project duration, and project cost on the frequency of change orders.