C Computational Statistics

data set and to identify the most inuential indicators and (b) compare proled performance of the consid-ered units to reveal what is driving the composite indi-cator results, and in particular whether the composite indicator is overly dominated by a small number of indicators. ● Links to other indicators identify the relationships between the composite indicator (or its dimensions) and other individual or composite indicators. ● Visualization of results should attract audience, pre-senting composite indicators in a clear and accu-rate way. Following the above-mentioned guidelines, the construc-tors of composite indicators should never forget that com-posite indicators should never be seen as a goal per se.ey should be seen, instead, as a starting point for initiating dis-cussion and attracting public interest and concern (Nardo et al. ). However, there is now general agreement about the usefulness of composite indicators:ere is a strong belief among the constructors of composite indicators that such summary measures are meaningful and that they can capture the main characteristic of the investigated phe-nomena. On the other side, there is a scepticism among the critics of this approach, who believe that there is no need to go beyond an appropriate set of individual indi-cators.eir criticism is focused on the “arbitrary nature of the weighting process ” (Sharpe ) in construction of the composite indicators