Date of Award


Document Type


Degree Name

Master of Science in Cost Analysis


Department of Mathematics and Statistics

First Advisor

Edward D. White, PhD


This study sought to predict cost growth in major Department of Defense (DoD) acquisition programs using logistic and multiple regression. In recent years, the use of statistical regression has proven to be successful in predicting the relationships associated with cost growth. This research follows on the work of Sipple (2002) and Bielecki (2003) and further explores the possibilities of using statistical regression to accurately estimate the dollar value associated with risk and uncertainty early in a program's life cycle. In doing so, the author intends to reduce cost growth by increasing the accuracy of the original cost estimates subsequently used to compute cost growth. The author first used logistic regression to determine whether or not cost growth would occur in a program and, if so, continued with multiple regression to determine to what extent it would occur. Data were compiled from all DoD departments using the Selected Acquisition Reports published between 1990 and 2002. The study analyzes programs during the Engineering and Manufacturing Development phase in the Research and Development, Test and Evaluation (RDT&E) phase of acquisition. For the logistic regression portion of the research, the author produced a seven-variable model that accurately predicted 72 percent of the randomly selected validation data. For multiple regression, a six-variable model was produced that accurately predicted the amount of cost growth incurred for 91 percent of the programs that incurred cost growth. Results show that the two-step regression methodology offers a significant advantage over traditional methods by removing the data points that do not incur cost growth. The author concludes that there is no significant advantage gained by either isolating each cost variance category individually or combining them.

AFIT Designator


DTIC Accession Number