FAULT GENERATION MODEL AND MENTAL STRESS EFFECT ANALYSIS

Various technologies for improving software reliability have been developed and widely used. These technologies, however, mainly concern how to eliminate faults generated in earlier phases. Our objective is to go one step further and determine how to avoid generating faults. This article first describes fault generation models based on a software development model. Those models are then shown to be consistent with the results of an analysis of data collected from fault reports and thorough interviews with software developers, that is, designers and programmers. The relationships among the various error factors and fault classes were also obtained through data analysis. The results clearly show that mental stress and human nature are the largest causes of developer-generated faults. Finally, the number of faults that could have been avoided is estimated.