On the influence of Test Smells on Test Coverage

Software testing is a key practice in the software quality assurance process. Usually, the quality of a test is not analyzed before its execution, i.e., there are no tests to check the tests. When the quality of tests is not guaranteed, it may impair the quality of the software. Test Smells are an alternative to indicate problems in the test code that can affect test maintainability, more specifically readability and comprehension. This study investigates correlations between test coverage and test smells types. We also introduce the JNose Test, a tool to automate test smells detection. We analyzed 11 open source projects and detected 21 types of smells and 10 different test coverage metrics to each test class. We identified 63 out of 210 calculated correlations. Our results show that there is a relationship between test smells and test coverage, in which test smells may influence code coverage. Our findings might support software testers and help them understand the behavior and consequences of poorly written and designed tests.