Моделювання та статистична оцінка впливу вищої освіти на формування показників злочинності в Україні

Abstract

The paper conducts an information-analytical study of correlations between the number of crimes committed on the territory of Ukraine and the total number of students of higher education institutions (HEIs). A detailed analysis of the corresponding correlation field, in the section of separate time periods, made it possible to identify three areas of points with a fairly close linear relationship between the specified indicators. In the first period (1990—1994), the rapid growth of the level of criminogenicity in the country with minor changes in the contingent of students of higher education institutions was due to the socio-economic crisis caused by the collapse of the USSR. In the second time period (1995—2015), a characteristic reverse cause-and-effect relationship is observed for the studied indicators. The constructed regression model proved to be adequate, with all its parameters being statistically significant at a confidence level of p = 0.95. Specifically, it was established that an average 1 % increase in the number of HEI students during this period is accompanied by a 0.43 % decrease in the total number of committed crimes. Similar statistical estimates were obtained considering a one-year lag in the values of the endogenous variable. In the third time period (2015—2021), there is a change in the direction of the correlation between the studied indicators. Namely, it was established that an average 1 % decrease in the number of HEI students is accompanied by a 2.23 % decrease in the number of committed crimes. This atypical behavior and violation of their causeand-effect relationship are explained by the systemic influence of several factors, among which the key ones are the hybrid military actions of the Russian Federation on the territory of Ukraine and the corresponding increase in migration processes due to the introduction of a visa-free regime with EU countries. The article attempts to combine the separate time periods (1995—2015 and 2015—2021) into one regression equation by introducing additional indicator variables. The generalized criminological model built as a result of the proposed approach has high accuracy (MAPE = 4.51 %), significant correlation indicators (R = –0.937), is adequate, and all its parameters are statistically significant.