
148 ◾ Software Metrics
increases productivity compared with your existing tool and its produc-
tivity. Or, you may want to choose between two techniques, depending on
their eect on the quality of the resulting product.
Your experiment will consist of a series of tests of your methods or
tools, and the experimental design describes how these tests will be orga-
nized and run. In any individual test run, only one treatment is used. An
individual test of this sort is sometimes called a trial, and the experiment
is formally dened as the set of trials. Your experiment can involve more
than one treatment, and you will want to compare and cont