Using multilevel random coefficient models to assess students’ spelling abilities

Abstract

This paper presents statistical models that analyze cross- sectional data related to student attainment in English and Maltese spelling. For each spelling test a random sample of 2040 students, whose age ranged from 6.5 to 16 years, was selected to examine the progression of spelling skills over time. The sample comprised equal numbers of male and female students attending state, church and private schools to investigate gender and school bias in students’ spelling abilities. This hierarchical nested data can be deemed as a type of two-level data, in which the students spelling scores are level-1 units and schools are the level- 2 units. This multilevel approach provides an adequate framework for modelling hierarchical data at several levels of nesting. To inspect the effect of age on student performance in English and Maltese spelling in different schools, a random coefficient model is fitted. This allows the school-specific coefficients describing individual trajectories to vary randomly when the spelling scores are regressed against the student age.