Beyond academic tracking: using cluster analysis and self-organizing maps to investigate secondary students' chemistry self-concept

Abstract

Academic tracking, placing students in different classes based on past performance, is a common feature of the American secondary school system. A longitudinal study of secondary students' chemistry self-concept scores was conducted, and one feature of the study was the presence of academic tracking. Though academic tracking is one way to group data for analysis, since students are naturally grouped in their classes, we aimed to uncover other groupings characterized by their self-concept scores. A cluster analysis was performed using scores from the chemistry and mathematics subscales of the chemistry self-concept inventory. The analysis yielded five clusters, four of which demonstrate a positive relationship between chemistry and mathematics self-concept scores. One cluster, however, was composed of students with low chemistry self-concept scores and high mathematics self-concept scores. Self-Organizing Maps (SOMs), new to chemistry education research (CER), were used to identify smaller groupings of students within the clusters to better understand students' self-concept. This technique was also used to explore longitudinal trends in students' chemistry self-concept data. This work has important implications for tracking in chemistry, the value of considering the affective characteristics of chemistry students, and the prospect of SOMs as a new CER tool.