PRETEEN AGE: THE ANALYSIS OF THE MULTILEVEL PSYCHO-DIAGNOSTIC DATA BASED ON NEURAL NETWORK MODELS

Abstract

The use of the artificial neural network (ANN) models for vertical system analysis of psycho-diagnostic data is suggested. It is shown that the ANN training, as the problem of nonlinear multi-parameter optimization, allows to create effective algorithms for the psycho-diagnostic data processing when the results of psychological testing for the different level’s characteristics have different numerical scales. On the example of processing the author's data of psycho-diagnostics (preadolescent schoolchildren), it is shown that neural network models can be used to estimate latent (hidden) connections between psychological characteristics. The proposed algorithms are based on a statistical assessment of the quality of such models, do not require a large sample of respondents. The quantitative statistical criteria for evaluating the quality of the models are estimated. The approach is sufficiently clear for practical use by psychologists who do not have a special mathematical preparation.