. Explain why design metrics are by themselves an inadequate method of predicting design quality?

Answer 1

A) Lack of explicit and well defined goals The first major problem with metrics is inadequately defined goals. Metrics are not always defined in the context of some explicit and well defined measurement goal derived from an objective of industrial interest [4]. This is one of the major reasons due to which metrics proposed by various researchers are not accepted by industries. Therefore without well-defined goals, metric is useless. Following are the few causes of this problem

• Too generic a goal for example the goal “To reduce complexity of software” is not well defined goal.

• Sub goals are not defined.

• Goals are not derived from industrial interest.

• Goal specification is not formalized.

• Criteria of goal achievement are not specified.

B) Undefined context Metric specification may lack the precise context specification i.e. environment in which they will be used. When the context of a metric is not defined then it may be misinterpreted by its users. For example applying component oriented metrics in an object oriented environment.

C) Lack of validation

It is necessary to validate the metrics before implementing it. But most of the metrics suffers from lack of

validation. Without validation it not sure that metric is measuring the same for which it is developed and

correlated with existing metrics.

D) Inconsistency

Another major problem with metrics is inconsistency. Many metrics are not uniformly interpreted by all users.

Inconsistency is the major issue in software measurement right for the beginning i.e. Line of Code (LOC)

counting. Now when software engineering is so mature even then we are searching for a consistent terminology

of measurement. For instance, software measurement researchers and practitioners have not reached an agreement

on the precise meaning of some terms commonly used such as ‘measurement’, ‘measure’, ’metric’, ’measurable

attribute’ etc.