MODELOVÁNÍ VLIVU SOCIOEKONOMICKÉHO POZADÍ ŽÁKŮ A STUDENTŮ ŠKOL SR NA VÝKONNOST V MATEMATICE VYUŽITÍM HIERARCHICKÝCH LINEÁRNÍCH MODELŮ

Abstrakt

Over the past 20 years data from surveys of education is increasingly analysed using multilevel models. Simple linear regression models (without considering the concept of hierarchical modelling) did not take account the way the students or students enrolled in schools or classes when modelling the possible effects of addiction. Therefore simple linear regression models may not adequately represent the monitoring information on education systems, schools, students or pupils. Then simple linear regression models do not distinguish between different levels - between the level of students and the level of higher units - classes and schools that these students or students attending. In some educational systems, for example, schools can select students or pupils with a broad spectrum of socio-economic background, which is mirrored in the high variability of socio-economic background of pupils. Therefore it is necessary for quality analysis of hierarchical data use multilevel (hierarchical) linear regression models. This paper shows an example of using several types of multilevel linear regression models, which were modelled effects of socio-economic background of students on their performance in mathematics. The quality of the estimated models (how models match reality) was assessed according to the information criteria and the size of the residual variance.