The Impact of Affect, Scenario and Task Characteristics on Developer Decision-Making

Abstract

Decision-making and strategy choice during software development are influenced by many factors, from the technologies we use to our own personal qualities. Better understanding these factors may help enable better understanding how and when development tools help developers. The goal of this thesis was to identify the relationship between contextual factors and the strategy selections that determine how developers approach problems. The factors that were examined were programming task, code scenario, and affect (mood). Task is the actual development task, such as debugging, implementing, or testing. Code scenario is a way to describe the type of code; our survey examines three scenarios: self-written code, framework code, and code containing callbacks. Affect describes psychological mood; it can be activated or deactivated (arousal), as well as positive or negative (valence). Our findings show that all three factors influenced the strategies participants chose. Participants with a highly positive viii affect were over twice as likely to ask for help. Participants were twice as likely to use developer tools for a debugging task, as in an implementation task. Participants rarely wished to create diagrams when dealing with framework code, as compared to their own code. These findings deepen our understanding of developer strategy choice; these insights can be used to improve recommendation systems, IDE’s, and other developer tools not only by improving the recommendations and content itself, but also by gaining more insight about when a developer might use those systems.