Methodological Blind Spots in Contextual Analyses: The Example of School’s Cultural Capital and Academic Achievement

• In Event: 3429 - Section on Sociology of Education Refereed Roundtables (1 hour)
  In Refereed Roundtable 60min: 4. Diversity, Inequality and Education: Good Schools/Bad Schools

Abstract

Many contextual analyses which address the question whether (socio-)spatial contexts like neighborhoods, schools, regions, or countries affect individual outcomes have two methodological problems: first, contextual effects are nearly universally operationalized of by using mean values to measure a context’s characteristics, and second, they do not consider conditional contextual effects. This paper addresses both problems using the example of schools’ aggregate cultural capital on students’ academic achievements. In a first step, alternative measures to the mean value are compared with respect to their explanatory power in multilevel regression models. In the second step, cross-level-interaction effects are added to the regression model fitting the data best. The results show that choosing the mean value for characterizing a school’s aggregate cultural capital is not wrong, but additionally, the proportion of students with high cultural capital resources affects academic achievement. More precisely: while with increasing mean cultural capital all students’ competences increase, an increasing proportion of high cultural capital students leads to decreasing competences for students with low individual cultural capital. With respect to educational research the results indicate that schools and teachers tend to be oriented towards average or high cultural capital students when planning lessons, providing explanations, and giving feedback and support, and that students compare themselves with the most resourceful schoolmates. In result, students with low cultural capital resources are at risk to be left behind in high cultural capital schools. With respect to methodological problems of contextual analyses, the results reveal the general need of a more elaborate theoretical base to be able to derive specific hypotheses and the respective operationalization.

Authors

• Alexandra Nonnenmacher, University of Siegen

• Alexandra Wicht, GESIS, Leibniz Institute for the Social Sciences