Longitudinal Analysis for macthed series osf studies

Abstract

In the first phase of the STATIS methodology the information is contained in a series of studies is first condensed into a symmetric matrix. Models based in the spectral analysis of their mean matrices were developped,see Areia et al.(2008). When their first engenvalue is dominant a further condensation of the information into a structure vector, with the eigenvector associated to and a sum of square of residues is carried out. This enables the joint analysis of matched series of studies associated to the treatments of a base design. Then in the longitudinal analysis of such families of series we study the action of the factors in the base design (contrasts)taken for the components of the structure vectors. An application to elections in Portugal is presented.