Vertical system analysis of students' psycho diagnostic data using the ‘decision tree' method

Introduction. Choosing forms and methods of psychological and educational support for students (individual and group-based) within the modern educational paradigm requires new means of data analysis. The purpose of this research is to investigate the possibilities of the ‘Decision Tree' method, which is a modern tool available for practical educational psychologists, for vertical system analysis of psycho diagnostic data and the choice of forms and methods of psychological and educational support for students. Materials and Methods. Based on a systematic approach in Psychology and Education, using one of the data mining tools - the "decision tree" method, the problem of classifying the results of students' psycho diagnostics is considered. On the example of a vertical system analysis of pre-adolescent schoolchildren tests, a hierarchical structure of connections of their multi-level psychological characteristics (inclinations, individual psychological and psychosocial characteristics) is constructed. Diagnostic tools were selected in such a way that the analyzed data conditionally present psychological characteristics of all levels. The authors used standardized, scaled methods of psycho diagnostics, which are quite widespread in psychological and educational practice. The two lower levels are represented by the type of nervous system (Ilyin's tapping test), and IQ (R. B. Cattell's Culture Fair Intelligence Test). Personality qualities in behavior and activity were analyzed on the basis of 12-factor Cattell's Personality Factor Questionnaire. The upper level of social and psychological relations is represented by such indicators as attitude to family, peers, to school, to oneself (V. Michal's "Unfinished Sentences" inventory). Motivational characteristics were investigated using ‘Need for Achievements' test. The results of psychological diagnostics of 83 schoolchildren (aged 11-12, fifth grade of secondary school) were processed (19 numerical test indicators were obtained). Results. When testing the ‘Decision Tree' method, it was revealed that the algorithm can be used by practical educational psychologists to analyze a relatively small sample of psycho diagnostic results - starting from several dozens of respondents. It is shown that the vertical system analysis of psychological characteristics can be clearly performed using a simplified procedure: comparing the significance of input attributes when classifying by different number of subsets of the target variable. The top-level indicators (motivation and relationship system) were used as variables for classifying data. The ‘Decision Tree' method allows analyzing and evaluating not only direct, but also latent (indirect, hidden) links of students' psychological data. For pre-adolescent age, the analysis of relations between different-level characteristics based on the results of classification shows a direct relationship between only some characteristics of the social level with the characteristics of the basic level (inclinations) and only an indirect relationship with communicative traits. Psychological interpretation of the revealed relationships of testing data allows the authors to clarify the age specificity of certain groups of students for subsequent psychological and educational support. The possibility of using the results for the analysis of transition problems from primary to secondary schools is discussed. Conclusions. The construction of hierarchical models of multi-level data links for students' psycho diagnostics proves to be an efficient tool for solving a wide range of problems within the fields of Education and Psychology.