ABSTRACT
The topic of selecting the activities to be included in software project measurements is a difficult issue and cannot be taken lightly. Project-level measurements and phase-level measurements have been the most widely used for more than 50 years, but they are also the least accurate. Neither project level nor phase level data will be useful in exploring process improvements, or in carrying out multiple regression analysis to discover the impact of various tools, methods, and approaches. Collecting data only at the level of projects and phases correlates strongly with failed or canceled measurement programs, because the data cannot be used for serious process research. Measuring at the activity level does not imply that every project performs every activity. Many activities such as quality assurance and technical writing span multiple phases, so phase-level measurements are not effective for process improvement work. Worse, activities such as project management that span every phase are not broken out for separate cost analysis.
