Background Most Western countries are undergoing considerable demographic changes through continuous international migration, with diverse language groups in the school teaching as an implication (D. Coleman, 2006; OECD, 2003). In order to analyse the covariation between socio-economic status, SES, contextual factors and mathematics achievements for the different groups of students, a model for description and analysing of SES considering the diversity is required. The aim of this study is to problematise the theoretical starting-points in how to measure SES for different groups of students, and to explore such a model. It is also investigated how this model correspond with data from the TIMSS study 2003. There is an empirically established relationship between student’s academic achievements and family socio-economic status, SES, (J. S. Coleman et al., 1966; Jencks et al., 1972). It will in this study be discussed in what way traditional models could be applied when measuring SES in classrooms portrayed by great diversity of student’s language- and cultural background. A traditional way to measure SES is by treating it as an one-dimensional single entity, dealing with parental education, occupation and family income. However, this is further challenged by more complex systems. Grounded in a theory by Bordieu (1986) which conceptualise the dimensions of family background in terms of different forms of capital, and in a theory by Coleman (1988) which expose the importance of family social networks, multi-dimensional models are explored (Yang, 2003; Yang & Gustafsson, 2004). It is in these studies revealed that dimensions and indicators varies between countries, but the significance of using one set of possession variables for all students within a country is not problematised. In another study it is however stated that in Sweden the correlation between performances and often used indicators for SES varies between students with foreign background and Swedish students (Elmeroth, 2006). Circumstances on the occasion of emigration or the economic situation in the new country are by Elmeroth mentioned as probable explanations for these differences. By using a multicultural model, student’s individual background factors could be given a substantial meaning. It will in the present study be investigated how conceptual differences between students and groups of students within a country could be applied to a model for description and analysing of SES. The study will be conducted by a two-level structural equation modelling technique (L. K. Muthén & B. O. Muthén, 1998).

The data source for the empirical study is obtained from the TIMSS 2003 study, focusing on mathematics for Swedish students in 8th grade. A latent variable analysis will be conducted in order to identify descriptive dimensions of socio-economic status, SES. Because of the design effect in survey research it will be necessary to account for the hierarchical structure of the population (Hox, 2002), hence a Multilevel Confirmatory Factor Analysis (MCFA) will be conducted. A two-level structural equation model approach with latent factors indicated by items from the student questionnaire will be adopted for the measurement model, and the Mplus and STREAMS software will be applied in the analyses. Theoretical starting-point will be a measurement model for SES explored by Yang and Gustafsson (2004). It is hypothesised, that in order to obtain significant effects of SES, documented values for possession items influenced by circumstances due to migration, has to be corrected.

Expected Outcomes The analysis of data is still under way so the final empirical results of the study are not yet available. However, the first steps of the analysis have been completed, with encouraging results. Results from this study do support the possibility to adopt this alternative multicultural perspective on the family background factor SES.