Using Logistic and Multiple Regression to Estimate Engineering Cost Risk

Abstract

This study explores a two-step regression procedure for assessing defense acquisition program cost growth using programmatic data from the Selected Acquisition Reports (SARs) between 1990 and 2000. We focus our analysis on cost growth in research and development dollars for the Engineering Manufacturing Development phase of the acquisition life cycle, specifically engineering cost growth. We illustrate the use of logistic regression in cost analysis to predict whether cost growth will occur. Given a program has a high likelihood of cost growth, we then use a log-transformed model to predict the amount of cost growth. Using this methodology, we produce statistically significant models highlighting the viability of this technique for cost analysts to consider and to adopt for future uses.