When I first heard of test suite coverage, I thought it seemed like a faulty metric.  I appreciate the time  Inozemtseva  et al put in to their research to show that this widely accepted method can be misleading.  Their results were shown with numerous graphs depicting normalized effectiveness scores plotted against size and coverage.  Their work set out to be different than other work by using larger, more realistic code bases, which was a good choice for proving the impracticality of the test suite coverage metric.