Clustering Effects in Timss

Analyses of TIMSS data have, naturally enough, largely concentrated on comparisons of overall attainment between and within countries. There is however a huge range of data collected, around an educational framework, and there is substantial scope for secondary analyses on other themes. This presentation describes the result of one such investigation, on the degree of within-school clustering of variables, both outcomes and background factors. The degree of similarity among members of a group, such as a school or classroom, is a very important factor in the design and analysis of studies in education. It is measured by two statistics, the intracluster correlation and the design effect. These help determine the total size and structure of the sample of respondents used. They also determine the extent to which it is possible to discover school or class effects on behaviours such as learning. This paper, using data from TIMSS, classifies outcomes into a number of categories and estimates intracluster correlation and design effect. The relevance of the results to the design and analysis of international studies is discussed. We also discuss briefly what these findings tell us about a shared youth culture.