Using Surveys to Build-up Empirical Evidence on Test-Related Technical Debt

• Lucinéia Souza UFBA
• Sávio Freire UFBA / IFCE
• Verusca Rocha UFBA
• Nicolli Rios UFBA
• Rodrigo O. Spínola UNIFACS / UEBA
• Manoel Mendonça UFBA

Resumo

Background: Technical Debt (TD) refers to the problem of pending or incomplete tasks and artifacts that brings short term gain to a software project, but may have to be paid with interest later on in its life-cycle. We use the term test-related debt to refer to TD items faced by software testing practitioners. Goal: To investigate, from the point of view of software testing practitioners, TD types associated with test-related debt, the causes that lead to it, and the effects of its existence. Method: We perform a replication of a well-known TD survey with software testing practitioners and combine the obtained responses with the ones collected in the original study. In total, the study analyses, quantitatively and qualitatively, 54 responses from Brazilian testing practitioners. Results: Test-related debt items are associated with TD types such as documentation, code, and test debt itself. The survey identified 61 TD causes that are organized into eight categories. The main causes are deadline, outdated / incomplete documentation, and inappropriate planning. The survey identified 45 TD effects that are grouped into six categories. The main effects are delivery delay, rework, and low external quality. Conclusion: This paper organizes the identified causes and effects in probabilistic cause-effect diagrams, which provide a broad view of the state of the practice concerning test-related debt causes and effects.

Palavras-chave: test-related debt, technical debt causes, InsighTD, probabilistic cause-effect diagrams, Technical debt, software testing, technical debt effects