Test-Driven Development: a systematic review

Abstract

Test-Driven Development (TDD) is a software development practice that became famous when Kent Beck defined it as an essential part of Extreme Programming (XP). The present study analyzed, previously published, experiments and study conclusions, related to the effects of TDD on the developers' productivity and on the quality of the software produced, contrasting TDD with Test-Last Development (TLD). Then, a systematic bibliographic review was conducted considering articles published between 2003 and 2020. At the end of the review process, approximately 73\% of the studies analyzed, consisted of experiments with TDD and in 27\% of them, the main theme was TDD itself. The analysis carried out shows that 43\% of the studies pointed to a considerable increase in software quality, while no article pointed to a decrease in quality. Regarding productivity, 28\% of studies pointed to a drop in productivity and 47\% were inconclusive. However, studies did not show significant improvements in productivity when TDD was used. According to the analysis, TDD promotes higher quality, even though some studies indicate the opposite. Regarding productivity, TDD analysis is inconclusive. Therefore, according to the papers analyzed there is no final position regarding the cost-benefit involved in this practice, we discuss some of the possible reasons for this conclusion.