A weak goodness-of-fit test for rank-frequency distributions

Abstract

We propose a goodness-of-fit test for cumulative rank-frequency distributions (Bradford curves). As the test is based on the correspondence of Gini indices and not on the curves themselves we lose information. Hence we refer to the procedure as a weak test. Confidence intervals for observed Gini indices are drawn using the jackknife estimator for the variance. Techniques are illustrated by testing the fit for 30 data sets. This clearly illustrates that the equation