A new approach for software testability analysis

Software testability analysis has been an important research direction since 1990s and becomes more pervasive when entering 21st century. In this paper, we summarize problems in existing research work. We propose to use beta distribution to indicate software testability. When incorporating testing effectiveness information, we theoretically prove that the distribution can express testing effort and test value at the same time. We conduct experiment and validate our results on Siemens programs. Future work concentrate on deducing a prior estimation of the distribution for given software and testing criterion pair from program slicing and semantic analysis.