Reliance on correlation data for complexity metric use and validation

Abstract

This paper reports the results of an experiment to illustrate the hazards of using correlation data as the sole determinant for software metric use and validation. Three widely cited complexity metrics have been examined in relation to the frequency of software development errors.