Abstract:
The paper discusses estimation of normality of data collected from educational software.
It reviews tests for normality which have discrete approximations and may be directly implemented in program systems.
The paper discusses the necessity of these tests to educational data and their place in automation of education.
The results of the review are used a base of a test selection algorithm, which can be used in an educational measurement software system. (In Russian).
Key words and phrases:educational measurement, assessment, mathematical support, test for normality, measurement results distribution, program system, programming, software.