A two-step approach to modelling student performance: A methodology that accounts for measurement and structural error.

Year: 2001

Author: Webster, Beverley, Fisher, Darrell

Type of paper: Abstract refereed

This paper describes the methodology used to investigate influences on student mathematics achievement and addresses the techniques of secondary analysis and its associated limitations and includes a two-step approach to modelling. A total of 57 Australian schools were surveyed and data collected from 620 teachers and 4645 students. The two-step model building approach consisted of the analysis of two conceptually distinct models. The first was an analysis of the measurement model, which specifies the relationships between the observed variables and the latent variables. The results of this analysis identified the measurement properties (reliabilities and validities) of the observed and latent variables. The second involved a structural equation model, which specifies the relationships among the latent variables as posited by theory and previous research and also describes those links between the latent variables and student science and mathematics achievement. This two-step approach to modelling student performance allows for clarity of model fit and identifies whether any source of poor fit is due to the measurement or the structural model. This paper then presents an explanatory model of student performance incorporating the student home background, student attitudes towards mathematics, success attribution, instructional practices and school-level environment factors.