Robust estimations of fault content with capture-recapture and detection profile estimators

Inspections are widely used in the software engineering community as effici ent contributors to reduced fault content and improved product understandin g. In order to measure and control the effect and use of inspections, the f ault content after an inspection must be estimated. The capture-recapture m ethod, with its origin in biological sciences, is a promising approach for estimation of the remaining fault content in software artefacts. However, a number of empirical studies show that the estimates are neither accurate n or robust. In order to find robust estimates, i.e., estimates with small bi as and variations, the adherence to the prerequisites for different estimat ion models is investigated. The basic hypothesis is that a model should pro vide better estimates the closer the actual sample distribution is to the m odel's theoretical distribution. Firstly, a distance measure is evaluated a nd secondly a chi(2-)based procedure is applied. Thirdly, smoothing algorit hms are tired out, e.g., mean ana meadian values of the estimates from a nu mber of estimation models. Based on two different inspection experiments, w e conclude that it is not possible to show a correlation between adherence to the models' theoretical distributions and the prediction capabilities of the models. This indicates that there are other factors that affect the es timation capabilities more than the prerequisites. Neither does the investi gation point out any specific model to be superior. On the contrary, the Mh -JK model, which has been shown as the best alternative in a prior study, i s inferior in this study. The most robust estimations are achieved by the s moothing algorithms. (C) 2000 Elsevier Science Inc. All rights reserved.