As good engineers, we use metrics to make decisions on quality, testing. My view is that measurements have inherent assumptions that we have to be cognizant to, otherwise decisions can be dangerous. For example, we use defect arrival rate to make decisions on quality, with the inherent assumption that test cases are relevant and complete. We use the metric of coverage to make decisions on adequacy of test cases, with the assumption that all system behaviors have indeed been coded.
I feel that measurements like test case immunity, test case growth, and quality growth could be possibly interesting indicators. These ensure that that we stay focused on the goal of effective testing. Find these thoughts outlined in this short presentation.