Shlomo Yitzhaki's book The Gini Methodology: A Primer on a Statistical Methodology offers a compendium of suggestions regarding the use of Gini's Mean Difference (GMD) as a measure of variability whenever an analyst is not ready to impose, without questioning, the convenient, well-developed world of linear, symmetric and normally distributed information. As he notes, one of the advantages of the GMD is that classic variance estimation is a subset class, in other words, when information is linear, symmetric and normally distributed the GMD offers no new information. Yitzhaki proposes substituting GMD estimation into ANOVA, regression, and so on, basically any multivariate technique which has classic covariance structures at its core. My question is this: does SAS offer any statistical procedures leveraging the GMD...that is beyond use of, e.g., the Gini coefficient and concentration curves regarding income inequality?
... View more