Criteria for choosing the type of model and method of data normalization in the index approach of social process analysis

Abstract

Subject. We analyze the method of calculating the global knowledge index, results of ranking countries according to one of the components of the resulting index, calculated using additive and multiplicative models for data on a natural scale and normalized under various methods. Objectives. The aim is to identify alternative normalization methods to eliminate distortions in the ranking of countries to improve the accuracy of the results, as part of the analysis of the data normalization method used in the global knowledge index, to substantiate the expediency of replacing the currently used additive model with a multiplicative one. Methods. The study employs quantitative methods of statistical analysis. The information base of the study consists of official data from the World Bank and the UNESCO Institute of Statistics. Results. We compared three methods of normalization: the "minimum-maximum" method, which is proposed by the compilers of the index, the "distance to a reference" method, and the standardization method. The study proved that using the "distance to a reference" method, the ratios for data on a natural scale and normalized, as well as their distribution laws, are preserved. The preservation of the order of countries when calculating the resulting indicator on a natural scale and after normalization is observed when combining the proposed normalization method with a multiplicative model. Conclusions. When calculating the components of the global knowledge index, it is recommended to use a weighted geometric average, as well as to switch to the normalization method through "distance to a reference" to exclude violations of relationships between the initial and normalized data.