The Impact of Aective States on Skills and Productivity in Software Development
Daniel Graziotin, Thesis Advisor, Pekka Abrahamsson · 2012
There is a constant need to improve the way the software is being developed due to growing needs of software in everyday life and devices. Even if there is an emerging, yet consolidated desire to consider software developers as individuals, software engineering (SE) research forgets that they are not industrial, unit-of-work machines. Software Developers are human beings. They solve problems in creative ways. For 20 years, human-related productivity factors in software development have been gaining attention. However, there are no factors that take into account the diversity of developers and their unique, complex minds. There is a well-established body of knowledge in Psychology and Cognitive Science that studies how the affective states of the individuals are related to their performance at work. Unfortunately, there are no studies to merge and test these theories in SE. In particular, the definition of the productivity of software developers is an open problem for more than 30 years. The mood of software developers is completely ignored. Affective states (feelings, emotions, and moods) are important. There are numerous studies not related to SE claiming that affective states have an impact on creativity, on the capacity of solving problems and on productivity. We claim that affective states can not be ignored by SE. Additionally, there is a recent call to adopt psychometrics in Empirical Software Engineering studies, to address the uniqueness of each individual participating in a Software Development Life-cycle. The discipline is tentatively called Individualized Software Engineering. In this study we pose a question that must be answered by SE via many present and future researches. The question is “What is the role of affective states for software developers?”. To provide first insights for answering the question, we conducted two empirical studies. The first study is related to how affective states are indicators of creativity and problem solving skills of software developers. In the second study we propose a linear mixed effects models that estimates the productivity of software developers by measuring their affective states related to the task they perform. We test the model with a repeated measures, longitudinal experiment in order to provide empirical results. We found significant evidence that software developers feeling positive affective states also possess higher problem solving skills. We believe that the same holds for the creativity, but we could not find significant evidence. Nevertheless, the analysis of the data encourages future research. We found significant, positive correlation with the affective states of software developers and their task productivity. We call for repetition of the experiment to further investigate our claims. The results of our studies are relevant for SE researchers, for Computer Science Education researchers and lecturers, as well for the Information Technology Industry. Information Technology firms may test the longterm affective states in the hiring phases of potential workers to predict their skills. Future studies in SE should research about how to raise positive affective states of developers and assess their skills and productivity. Future research may also lead to the creation of a feedback-based system for IT companies to assesses the productivity at different organizationallevels by measuring the mood of workers.