Events and Meetings of Italian Statistical Society, Statistics and Demography: the Legacy of Corrado Gini

Font Size: 
On the Construction of Composite Indices by Principal Components Analysis
Matteo Mazziotta, Adriano Pareto

Last modified: 2015-09-05

Abstract


Principal Components Analysis (PCA) is the most commonly used multivariate statistical technique in construction of composite indicators. However, PCA and related methods, such as Factor Analysis, are based on a reflective model where the individual indicators (manifest variables) are seen as functions of a latent variable (principal component or factor). When individual indicators are causes of the latent variable, rather than its effects, a formative model should be adopted. In this paper, we compare the two approaches, and we show an example of reflective and formative composite index calculation.

Full Text: Untitled