Refinement of estimates: using logistic and multiple regression to predict cost growth

Assessing cost growth for major Department of Defense weapon systems can be a difficult task to accomplish. Although experience and subjective reasoning has its place, incorporating statistical analysis in this assessment is a valuable tool to consider. Not only are empirical data and historical trends important for statistical analysis, so is the methodology chosen. In this study, the authors highlight a two-step regression procedure for predicting the likelihood and expected percentage increase of cost growth. Using this methodology, they produce statistically significant models highlighting the viability of ordinary least squares regression in conjunction with logistic regression for cost analysts to consider and to adopt for future uses.