Mark Agranovich - Candidate of Sciences, Head of Center for Monitoring and Statistics of Federal Institute of Development of Education. E-mail: magran@firo.ru

The article discusses the possibilities of different tests results comparisons. The issue of correct compare of different tests results raises in relation of analysis of tests results for different years and in a number of other cases. These results could not be compared directly, without some adjustments. The article presents the approach to make tests results with different parameters of distribution comparable. The suggested approach based on the assumption that the level of students’ knowledge in different subjects are more or less equal and differences in parameters of tests results distributions arise due to instruments and differences in testing scaling. Suggested scheme of using statistical methods allow to eliminate these differences in testing instruments and make results of different tests comparable. Applications of the suggested method is shown on an example of comparison of results of Unified State Exam (upper secondary graduation tests) in different subjects and for different years, and also comparison of Unified State Exam and State Final Attestation (low  secondary graduation tests) results. The sphere of application for the suggested method is limited by the following conditions: the number of objects should be sufficient to use mathematical statistics tools applied to evaluate the closeness of relationship; the paper compares the sets of values obtained in different tests for the same objects.