Impact of outliers on inequality measures – a comparison between Polish Voivodeships

It is known that outlying (large) incomes strongly influence the results of inequality measuring. Thus, there is a question how to deal with such observations. In this paper the rule of excluding observations based on (Q1 – 1.5Q; Q3 + 1.5Q) interval is investigated, for data from household budget survey in 2011, for Polish voivodeships. It is shown that although including more observations obviously changes the values of inequality measures, the relative values of them are surprisingly quite stable, with the rank correlation coefficient never over 0.9.