Personal webpage Hanneke van der Hoef
Cluster analysis in educational research: Best practice guidelines for finding groups
To find groups (clusters, profiles) in data researchers commonly use cluster analysis, a powerful statistical approach for unraveling patterns in complex data. Applying cluster analysis requires making various decisions, such as selecting a clustering method, choosing the variables, choosing the (dis-)similarity measure, and determining the number of clusters. To make cluster analysis more accessible for researchers, there is a need for guidelines on how specific requirements of the application can be connected with the available methods. Guidelines on cluster methodology and strategy should be developed given a specific application or domain. In this project, the focus is on the application of cluster analysis in educational research, where there is an emerging need to identify academic ability profiles of primary school students. The identification of such profiles may help improve appropriate school selection and facilitate tailored curricula.
The aim of this research project is to develop best practice guidelines for applying cluster analysis in educational research. First, a systematic review will be conducted to provide both a quantitative and qualitative overview of how cluster analysis has been applied to identify academic ability profiles in students. The subprojects hereafter will then focus on the major steps of the clustering procedure: selecting variables, determining the number of clusters, choosing a clustering method, and treating outliers. Project output will be shared with both academia and educational practice via several papers in peer-reviewed journals, conference contributions, R code, a white paper, and a workshop.
Prof. dr. M.E. Timmerman, dr. M.J. Warrens
University of Groningen
1 September 2018 – 31 August 2022