Correlations among Software Testability Metrics

• Francisco Gutenberg da Silva UFC
• Valeria Dantas UFC https://orcid.org/0000-0002-1210-7935
• Ismayle Santos UFC https://orcid.org/0000-0001-5580-643X
• Rossana Andrade UFC https://orcid.org/0000-0002-0186-2994

Resumo

Testability is a relevant characteristic to evaluate software quality and it is addressed in most quality models. This characteristic is directly related to software testing activities and defines how easily a program can be tested. However, testability is not an intrinsic characteristic of the software and, therefore, it is necessary to identify attributes and measures for evaluating the software testability that, respectively, may influence and assess it. Most studies on testability focus on evaluating it on white box perspective. Thus, the goal of this study is to establish the correlation between the testability metrics into two perspectives (white and black box) and based on this correlation, we can evaluate the testability on other software artifacts such as Test Specification. We inittially identified a set of attributes (13) and measures (30) related to testability. Next, we proposed a set of 10 metrics to evaluate Testability from a Test Specification. Then, we conducted the empirical study in two mobile systems from industry to evaluate 3 attributes and 25 measures regarding the testability in those perspectives. As a result, 8 metrics that we proposed to measure testability of a test specification (black box level) had a significant Pearson's correlation with 11 metrics commonly used to evaluate the testability in the white box level.

Palavras-chave: Testability, Software Measures, Software Testing