METHOD OF SIMULATION AND CHOICE OF FACTORS IN THE ANALYSIS OF PRINCIPAL COMPONENTS

Objective There are many methods to determine how many components shou ld be retained in principal components analysis. This choice can be ma de on the basis of arbitrary (Kaiser) or subjective (Interpretable fac tors) criteria. This work presents the simulation criteria of Lebart e Dreyfus. The method create a matrix of randomized numbers and a princ ipal component analysis is performed on the basis of this matrix. The components extracted from this data represent the cut off values. Thos e that exceed this cut off value should be retained. As an example, a principal component analysis is performed with the Hamilton depression rating scale (17 items) on a sample of 130 subjects. Results and Conc lusion The Simulation method is compared with the Kaiser method rind i s shown that the Simulation method maintains the components clinically significant.