There are many different approaches to successful programs for improving and maintaining quality. But how do we know that a measurement program is successful?There are at least four key goals for any measurement program: understanding, control, forcasting, and business value. For each we must ask questions to help us decide if the measurement program is effective, see Table 14.2.
Not every use of measurement in practice is part of a large, organization-wide measurement program. More often, measurement is used to help answer a particular question on a given project or within a specific organization.Two examples are presented. It is stressed that a small degree of additional information, made more explicite, can result in a very large degree of additional management insight.
The goal was to evaluate reliabiliy and maintainability based on historical data. There was neither mean-time-between-failures nor operational usage information available, so reliability could not be evaluated directly. Size information that was missing from the original data was recovered. This revealed that one of the smallest system areas had the largest number of "faults".
It is rather amazing that measuring size was not part of the original measurement program.
The second example is on assessing the use of formal methods on the ultimate quality of a safety-critical system. By detailed analysis it was found, that code of higher quality likely was the result of the formal specification technique used, leading to components that were relatively simple and independent, making them relatively easy to unit-test.