Data processing methods in the national testing system E. O. Bondarenko, Ph.D. student National aerospace university named after N. Ye. Zhukovsky “KhAI”
This work is devoted to the description of data processing methods in the national testing system. The calculated values of test quality in the context of threshold group test participants. With index p-value (significance level of performance is achieved) analyzed the success of tasks-choice threshold groups of participants.
The achieved significance level p(T) - is the probability under the null hypothesis to get the value of statistics, the same or even more extreme than T. Investigated a reliability test (б-Cronbach) for threshold groups of test participants. The reliability coefficient is an important criterion for evaluating the test results. It is a measure of the accuracy with which the testing of a character. For the implementation the above-described method is proposed to use the software SPSS Statistics («Statistical Package for the Social Sciences»). SPSS provides for this purpose a variety of methods, the default is the method of calculating Cronbach’s alpha with a value whose magnitude is between 0 and 1. References 1. ZeitlinN. The experience of analytical statistics. —;M.: Solar, 2006. -905 p. 2. Illustrated tutorial on SPSS [electronic resource]. - Access mode: http://www.leamspss.ru/hndbook/glava21 /cont2.htm. 3. Chelyshkova MB The theory and practice of constructing educational tests: a tutorial. - Moscow: Logos, 2002. - 432 p. 4. JM Neumann Introduction to modeling and parameterization of pedagogical tests / YM Neumann, VA Khlebnikov - New York: Prometheus, 2000. - 169 p. * * Scientific advisor PhD, Associate Professor M. Mazorchuk