Test suite coverage is a commonly used metric to evaluate the test suite's ability to detect faults.   Inozemtseva et al found that there is only a low to moderate correlation between coverage and test suite effectiveness, however.   Other literature on the topic that vouches for coverage as a good metric seems to have been flawed in one way or another, such as using a small code base or synthetic programs.  This can lead to skewed results, as  Inozemtseva  et al showed by studying five large Java programs.  The results showed that there is a moderate to high correlation between coverage and test suite effectiveness when suite size is ignored, and a low to moderate correlation when size is controlled.  This shows that coverage is not necessarily a good metric for test suite effectiveness, as the correlation is more-so related to the fact that test suites with high coverage have a greater number of test cases.