Date of Award


Document Type


Degree Name

Master of Science


Department of Electrical and Computer Engineering

First Advisor

Mark A. Kanko, PhD


This thesis investigated several hypotheses that specific product measures could be used to predict later software lifecycle process or product measures. It collected software product and process measures from four consecutive major releases of a large Cobol legacy system (400K LOC). The types of product measures used were size and specific complexity measures. A statistical software package was used to calculate sample correlation coefficients between the measures. A 95% confidence interval was computed for each sample correlation coefficient that showed a strong or moderate linear correlation. The maintenance process measures provided were manhours used for each program changed or added, and defects detected during each change request. Sample correlation coefficients were derived to see if product measures such as size and complexity could reveal trends that could be used to estimate other software lifecycle measures such as effort or defects. The hypotheses to this research could neither be accepted nor rejected because the process measures collected by the system's owners were recorded at a level too high for sound statistical analysis. Weaknesses are identified in the way these process measures are collected, and suggestions are provided on how process measures can be better identified and recorded.

AFIT Designator


DTIC Accession Number