Empirical software engineering research can be organized in several ways, including experiments, cases studies, and surveys. Experiments sample over the variables, trying to represent all possible cases; cases studies sample from the variables, representing only the typical cases(s). Every case study or experiment should have a hypothesis to express the desired result. The experimental design is especially important because it identifies key variables and their relationships. The design uses balancing, blocking, and local control to help minimize error. Analysis techniques depend on the design, the distribution of the data, and the type of investigation being carried out. Different techniques allow us to look at variable interaction and to look at combinations of effects. Using a technique similar to a board game, we can determine when we have enough evidence to demonstrate clear relationships among variables.
原文链接
分享
引用
