Using an areal model to study the meaning of the normalisation and weighting of variables

Abstract

In this study a model of the area is prepared to study the usefulness of some statistical methods by simulating. Actual observation material is also used. The results indicate that normalisation of the variables was of no use, as neither was weighting by acreage, but weighting by square root of the size of the population of the areal unit may sometimes be. It is also apparent that the correlations are not independent of the areal unit system used and that in some cases the usual statistical tests can give misleading results when applied to areal data.