Empirical, Human-Centered Evaluation of Programming and Programming Language Constructs: Controlled Experiments

摘要

While the application of empirical methods has a long tradition in domains such as performance evaluation, the application of empirical methods with human subjects in order to evaluate the usability of programming techniques, programming language constructs or whole programming languages is relatively new (or, at least, running such studies is becoming more common). Despite the urgent need for such usability studies, few researchers are well-versed in such techniques, certainly when compared to the large number of researchers inventing new programming techniques or formal approaches. The main goal of this text is to introduce empirical methods for evaluating programming language constructs, with a strong focus on quantitative methods. The paper concludes with by explaining how and why a series of controlled experiments were gradually designed to study the usability of type systems.