Cognitive biases in software development life cycle and strategies to mitigate them

Abstract

Cognitive biases are systematic errors in human judgement that affect decision-making throughout software engineering, from effort estimation and design to testing and code review. Although their influence is being recognised increasingly, the literature remains scattered regarding where these biases occur and ways to reduce them. This thesis explores how cognitive biases influence decision making across SDLC and identifies strategies for mitigating their impact. A systematic literature review was conducted following the guidelines of Kitchenham and Charters (2007). A search of IEEE Xplore, Scopus, and ACM Digital Library, supplemented by Google Scholar and snowball sampling, yielded 31 primary sources published between 1995 and 2025, which were analysed by bias type and software development life cycle stage. The review found that anchoring and confirmation bias are the most frequently studied and empirically confirmed biases, and that they persist across the life cycle regardless of developer experience. Biases were shown to propagate across development stages, producing inaccurate estimates, weaker designs, and undetected defects. Researchers found that the state of being aware of bias is not enough to change outcomes. Structural interventions like Planning Poker, decision criteria that people define in advance and tools for structured review are the methods that lower bias most effectively — those results show that developers cannot remove cognitive bias from software engineering in its entirety — but teams can reduce it significantly through structured practices applied consistently over time.