Inter-item correlations among function points

Reports on an empirical investigation of Albrecht function points. They investigated the constituent elements of the function point counts to determine whether all the elements are required to provide a valid measure of size; whether the sum of all the elements is a better predictor of effort than the constituent elements; and whether the complexity adjustments are necessary. The study suggests that function points are not well-formed metrics because there is a correlation between their constituent elements. It also suggests that for the dataset under investigation, two of the constituent elements were as good at predicting effort as the raw function point count and that the unweighted counts can be reasonable predictors of effort