Przestrzenne zróżnicowanie wypadkowości przy pracy w Polsce

This study presents a classification of Polish voivodships according to selected data on accidents at work. Statistical inference has been applied to decide on the number of subgroups to be distinguished by means of Ward’s method. Critical values of the proposed test were obtained by two-step procedure, with rough critical values determined by computer simulation. They were then smoothed using the analytical function according to the number of objects (n = 16 Polish provinces) and attributes. In fitting the smoothed critical values, a model with a given loss function was applied. On the basis of this procedure two groups of provinces with different characteristics were designed.