Page 240 - DCAP304_DCAP515_SOFTWARE_PROJECT_MANAGEMENT

Software Project Management

Notes The question of “what data to collect?” The answer is to collect all of the data required to provide
the metrics primitives and the additional qualifiers.
In most cases, the “owner” of the data is the best answer to the question of “who should collect
the data?” The data “owner” is the person with direct access to the source of the data and in many
cases is actually responsible for generating the data. Table 12.1 illustrates the owners of various
kinds of data.
Benefits of having the data owner collect the data include:
 Data is collected as it is being generated, which increases accuracy and completeness.

 Data owners are more likely to be able to detect anomalies in the data as it is being
collected, which increases accuracy.
 Human error caused by duplicate recording (once by data recorder and again by data
entry clerk) is eliminated, which increases accuracy.
Once the people who gather the data are identified, they must agree to do the work. They must
be convinced of the importance and usefulness of collecting the data. Management has to support
the program by giving these people the time and resources required to perform data collection
activities. A support staff must also be available to answer questions and to deal with data and
data collection problems and issues.

Table 12.1: Examples of Data Ownership

Owner Examples of Data Owned
Management  Schedules
 Budgets
Engineers  Time spent per task
 Inspection data including defects
 Root cause of defects
Testers  Test cases planned/executed/passed
 Problem reports from testing
 Test coverage
Configuration Management Specialists  Lines of code
 Modules changed
Users  Problem reports from operations
 Operational hours

A training program should be provided to help insure that the people collecting the data
understand what to do and when to do it. As part of the preparation for the training program,
suitable procedures must be established and documented. For simple collection mechanisms,
these courses can be as short as one hour. I have found that hands-on, interactive training, where
the group works actual data collection examples, provides the best results.

Without this training, hours of support staff time can be wasted answering the same questions
repeatedly. An additional benefit of training is that it promotes a common understanding about
when and how to collect the data. This reduces the risk of collecting invalid and inconsistent data.
If the right data is not collected accurately, then the objectives of the measurement program cannot
be accomplished. Data analysis is pointless without good data. Therefore, establishing a good data
collection plan is the cornerstone of any successful metrics program. Data collection must be:
 Objective: The same person will collect the data the same way each time.

 Unambiguous: Two different people, collecting the same measure for the same item will
collect the same data.

234 LOVELY PROFESSIONAL UNIVERSITY