EXPERIMENTAL MODELS FOR VALIDATING COMPUTER TECHNOLOGY

Experimentation is important within science for determining the effectiveness of proposed theories and methods. However, computer science has not developed a concise taxonomy of methods applicable for demonstrating the validity of a new technique. In this paper we discuss the methods generally employed to validate an experiment and propose a taxonomy consisting of 12 techniques that can be used to show that a new technology achieves its hypothesized goals. An evaluation of over 600 papers published from 1985 through 1995 shows that the 12 methods can be effectively applied to research papers, and we provide some observations of how well the research community validates its claims in these papers.