This research study was conducted to test the accuracy and reliability of the function point software sizing metric. Basically, the study was designed around the hypothesis that estimate replicability is an indicator of estimate reliability. To test this hypothesis, the author measured estimate replicability both within (interrater) and between (intermethod) two groups. From a selected set of design documents and completed systems, one group estimated function points using the then-current IFPUG 3.0 standard. Using this same set, another group estimated function points using the entity-relationship approach.

Overall, this paper is excellent. The author has constructed a reasonable and easy-to-understand experiment. The development of the hypothesis and the results (the experiment did show both interrater and intermethod reliability) have definitely added to the body of knowledge of function points. For those of you considering the adoption of function points, this paper should help serve as justification. For those of you already using function points, this newer entity-relationship approach holds some promise of adding another dimension to function points.