Impact of CS Programs on the Quality of Test Cases Generation: An Empirical Study

摘要

Background: Although most Computer Science (CS) programs offered by higher education institutions usually include a software engineering course, some works report a lack of formal training in software testing. Aim: With the aim of studying the possible impact of knowledge acquired from CS programs on software testing, this paper reports an investigation composed of four experiments. The experiments conducted in Spain, Mexico and Ecuador examine the quality of test cases (TC) generated using black-box and white-box methods. The subjects of the experiments were undergraduate and graduate students who were exposed to different levels of CS knowledge. Method: We pool together the data from the four experiments and apply logistic regression to investigate possible relations of the quality of test cases with students' level of exposure to CS knowledge. Results: The quality of test cases generated by students depend significantly on the amount of CS program studied. The odds of generating test cases that reveal failures against those that do not reveal decrease when students are exposed to a low level of CS knowledge. Conclusions: Software testing plays a key role in what is an increasingly complex process of developing and maintaining software products today. The results of our empirical study provide evidence in favor of greater formal training in software testing as part of CS programs.