Keyword: «latent variable»

The features of measuring latent variables are discussed and compared with classical testing theory and other systems. This article is part of the structure of the Final qualification work on the topic «Measurement and analysis of the level of creativity of students in secondary schools».

The developed software for measuring latent variables is structured as follows: programs for measuring of latent variables, categorization of continuous indicators for measurement of latent variables, modeling experimental situations, investigation the precision of measurement of latent variables, analysis the quality of tests and questionnaires as measuring tools.

Within the framework of the theory of latent variables based on the Rasch model, the influence of the test shift on the accuracy of measuring the level of preparedness of the subjects is investigated. Situations are considered when the difficulty of the test corresponds to the proficiency level of subjects, as well as when the test is shifted relative to the level of subjects’ proficiency by one and two logits. It is shown that increasing shift, decreases the measurement precision of the latent variable.

The aim of the work is to determine the precision of measurement of latent variables. A comparative analysis of the classical theory of testing and the theory of latent variables is given. The study is based on a simulation experiment within the framework of the theory of latent variables.

The procedure for modeling test results based on the Rasch model is considered. The simulation results are processed within the framework of the theory of latent variables. Model values of the level of testees’ performance, their estimates, as well as estimates errors of the level of performance are approximated by linear models. It is shown that poorly prepared testees are evaluated with less precision than more prepared ones.