Abstract:

This paper presents the specificities of using MANOVA and Discriminant function analysis in the example of research in social sciences. Prerequisites for the use of both analyses and limitations of their use have been elaborated. In essence, though very similar, they are essentially different, and their application is elaborated for example. In order to better understand the computation of multivariate analysis of the analysis of variance, id est MANOVA, the basics of linear algebra are presented (eigenvector, eigenvalues, residual matrices…). Multivariate tests, id est test statistics: Pillai`s trace, Wilks`lambda, Hotteling´s trace and Roy`s largest root are discussed (convenient for specific use). The assumption of MANOVA was compiled, such as multivariate normality and homogeneity of covariance matrices.

Additionally, are elaborated in the DFA standardized Canonical Discriminant functions coefficient and structure matrix, additional functions at group centroids (which are discriminated by a variation).

Keywords:

MANOVA, Discriminant function analysis, standardized Canonical Discriminant functions coefficient, structure matrix.