A Method to Evaluate Estimates Produced by the Capture-Recapture Model

The capture-recapture techniques could be used for post-inspection estimation of residual defects in the inspected software artifacts. However, adoption this approach in real world software inspections may still face several challenges. On one hand, most of the capture-recapture estimators could easily produce extreme estimates on residual defects, on the other, there are very limited studies on the evaluation on the estimates so as to identify these extreme estimates. As a result, quality management decisions(e.g., whether a re-inspection is required) rely on the estimates produced by the capture-recapture techniques may possess high risk of mistakes. In this paper, we propose and verify an approach to evaluate the capture-recapture estimates, which could help practitioners identify extreme estimates to a certain degree.