Application Of Generalized Linear Latent Variable Modeling For Multi-Group In Education

The problem discussed in this paper is the relation of students’ achievement andtheir age, sex. Student education research has become a focus of social concern atpresent, our country has spent much time on this research and achieved great success.It’s obvious that student achievement is related with their age and gender. It’smeaningful to prove that in theory.People always analysed the relationship between students’ score and their age,sex by factor analysis model, there are great limitations that factor analysis model isoften assumed that each factor is independent. In fact, the latent factor that weassumed is not independent. So we take advantage of linear latent variable to solvethis problem.In this paper, six subjects of each student divided into two categories, a categoryincluding chinese, english and history which are affected by liberal arts factor; and theother including mathematics, physics, chemistry which are affected by science factor.We make a model of the six subjects, each subject is affected by both liberal artsfactor and science factor, and these two latent factors also subject to the normaldistribution of a binary. We will discuss the influence of each factor on each subjectby estimating every parameter.We also make a model of student subjects with different age and gender,and theneatimating each parameter belong to different group. Finally, the t-test explain thedifferences between science and liberal arts factor in the different groups, in order toexplain the differences between the different groups.