Clustering students by their subject choices in the Learning Curves Project

In the course of standard statistical analyses in the third and final phase of the Learning Curves project, questions were raised about the nature of students’ subject choices. We thought that if well-described patterns of subject choice could be found using our data, then we might also question the relationship of these patterns to available student demographics. 

A clustering technique was devised to isolate patterns of subject choice for each of year 11, 12 and 13 students in our sample, and the resulting clusters were cross-tabulated with students’ school, ethnic group, and gender. 

Some general technical background about clustering is presented, followed by the results of applying the technique to the Learning Curves data.

The paper is intended for those who wish to know more about the clustering technique referred to in Shaping our Futures: Meeting secondary students’ needs in a time of evolving qualifications (Hipkins et al. 2005).  No specialised statistical knowledge is required to understand the techniques described.