PROGNOSTIC VALIDITY OF THE COMPETITION SELECTION OF GRADUATES OF ECONOMIC COLLEGES

Abstract

One of the topical challenges facing higher educational establishments (HEE) is to build the most optimal model for competitive selection of entrants. Based on the results of the external independent assessment (EIA) and the average score of the entry certificate, the institution of higher education, by varying weights, tries to build such a rating list, which places in the first place exactly those applicants who are able to study better in the relevant specialty. Evaluation of the applied model of competitive selection is investigated by the indicator of the predictive validity of the competitive score. Predictive validity is the correlation coefficient between the indicator by which competitive selection is made and the student's performance during the first year of study. Therefore, by assessing the value of predictive validity, it is possible to examine the statistical relationships of EIA results in different subjects or their respective weights with student performance and to build optimal competitive selection models based on them. In this case, the effectiveness of the HEE entry system based on EIA is recognized as high, if the correlation coefficient (R) is greater than 0.5; sufficient, if the correlation coefficient is in the range [0,3, 0,5] and low, if the correlation coefficient is less than 0,3. Since 2018, analytical procedures for EIA have begun to be introduced gradually for college graduates, but only as a state final certification. Subsequently, the results of the EIA will be taken into account in the admission of junior specialists to the bachelor's degree. But now, enrollment in the HEE is based on the grade point average of the school certificate and the results of the university entrance examination. Therefore, the study of the prognostic validity index of competitive selection of college graduates is relevant both in terms of evaluating the existing model of admission to the HEE, and for the search of directions for its improvement. In order to evaluate the predictive value of competitive selection and its components, the results of the calculation of correlation indicators between the students' grades and the students' success in the subjects studied in the HEE during the first semester of the third year are presented. The data obtained indicate a sufficiently high prognostic validity (R>0,5) of the competition score. At the same time, a significant level of correlation was observed both with the average rates of students' education in the first semester and with success in almost every discipline (only for the discipline "Information Systems and Technologies of Accounting and Audit" the correlation coefficient is slightly less than 0.5). It should also be noted that the competition score is better correlated with the results of the current academic performance of students in the 1st semester (before the test session), than with the results of the overall success after the 1st session. As it turns out, this tendency to decrease correlation is due to the fact that several students did not participate in the exam and therefore their average learning outcomes are very different from the main group of observations. Also indicative for predicting student learning outcomes in the HEE are the data from the average score of the school certificate. This indicator, as shown by the correlation coefficient calculations, has high predictive validity to the average results of the first semester (R = 0.73 and R = 0.67) and to each individual discipline (R>0.5). Therefore, when introducing external independent assessment for college graduates, it is also advisable to enter the average score of the education document into the formula for calculating the competition score. On the other hand, a proficiency test of university applicants is a weak predictor of student performance during the first semester (R<0.5 for all HEE subjects without exception). The slight correlation between these criteria is related to the introduced professional test evaluation methodology and therefore requires further improvement. By analyzing the standardized values of deviations between the competition grade and the average success rate of a student in the first semester, the presence of abnormal levels of evaluation within the sample was investigated. It is shown that the Spirman rank correlation coefficient is less dependent on the presence of such anomalies and is more rational when estimating the prognostic validity of small samples.