Some considerations to carry out a composite indicator for ordinal data

Composite indicators (CIs) are important and useful tools in many elds\r\nto assess, compare and rank performances, development stage, quality and\r\nmany other dierent targets. CIs are an overall measure of a multidimen-\r\nsional, not directly observable, concept and are obtained by means of a set of\r\nmanifest variables (elementary indicators) that contribute to dene the over-\r\nall measure. In this paper, some matters regarding methods to build CIs are\r\nreviewed, assuming elementary indicators are ordinal and quantication is\r\nnecessary to convert observed data into a numerical form. Scoring methods,\r\naggregating functions and weighting systems are considered. In particular,\r\na scoring method based on the observed distribution or the use of dissim-\r\nilarity indices for quantication together with the Kendall- association or\r\na heterogeneity measure for weighting are suggested. Some of the reviewed\r\nprocedures are compared using students' satisfaction data.