Analysing classroom interaction using two-step cluster analysis.

Objectives: Research in education recognises that group work within classrooms can benefit pupils¿ socially and academically. Studies within primary education have identified that groups are a feature of classrooms; however comparative research within secondary education remains scarce. This research aims to address this disparity focusing on two teaching years and teaching subjects, who frequently rely upon group work.


Design: Naturalistic systematic observation took place within first and third year Science and English classrooms and was recorded onto a grid, based upon an approach detailed within previous research (Howe et al., 2007; Thurston et al., 2008). Systematic observation allows an accurate record of behaviour to be developed and is suitable for use within classrooms

Methods: Over 600 observations of pupils were recorded within group work and conventional lessons as classroom interaction is not restricted to group work lessons. Inclusion of both group work and conventional lessons allowed variables associated with these teaching methods to be identified. The observation grid contained four sections, which allowed the precise interactive nature of a lesson to be evaluated. Three complementary sections identified whether pupils were instructed to work as groups or otherwise, whether pupils completed tasks as individuals, groups or as part of a whole class, and whether pupils were actively focused on completing tasks or not. The final section focused on the dialogue produced by pupils during lessons.


Results: Two step cluster analysis produced two clusters. This form of cluster analysis is unique as it can automatically detect the appropriate number of clusters. Cluster profiles were computed using the cluster membership of each case in the data set, and supported the distinction between group work and conventional lessons. Group work and conventional lessons shared similarities in their task and organisation structure but had important differences. Within both types of lesson although similar tasks were most frequent, pupils varied their approach to such tasks, according to whether they were situated within group work or conventional lessons. Specific forms of dialogue were a feature of pupil interaction observed within group work lessons.


Conclusions: Two step cluster analysis can be used effectively with observational data; its features allowed clear interpretation and comparison of the two clusters. Clearly classroom interaction is not constrained to group work lessons, however specific forms of dialogue featured within group work lessons and were observed whilst pupils actively completed tasks. Whether the change in pupils¿ behaviour within these two types of lessons stems from the classroom environment or follows from teachers¿ behaviour needs to be addressed. These findings will directly contribute to a group work initiative aimed at secondary classrooms.