Cluster Analysis – A Standard Setting Technique in Measurement and Testing
Cluster analysis, Standard Setting, K-Means Clustering, Hierarchical Clustering
Standard setting plays an important role in educational and psychological testing. This paper is focused on standard setting using ‘cluster analysis’ technique. Cluster analysis is a statistical procedure for forming homogenous groups of subjects (examinees). It explores the process of doing cluster analysis and its types are – K-Means and Hierarchical clustering. In the hierarchical cluster analysis, all objects are initially being considered to be a unique cluster. The analysis proceeds sequentially by merging clusters together one step at a time until all objects are merged into a single cluster. In the K-Means cluster analysis, the number of clusters into which the objects which will be portioned is specified initially. The K-means algorithm then establishes the centers of each cluster which are represented by a vector of means (called the cluster centroid) corresponding to the variables used to cluster subjects. The procedure was applied to an achievement test in science. A five cluster solution best separated the examinees according to their proficiency skills. The study concludes that cluster analysis has an edge over other techniques in regard to reducing subjectivity based on expert ratings of items and applicability to performance-based assessments. It does not remove subjectivity from the standard setting process, but does provide subject-matter experts and test developers with a quantitative method for determining different groups of test takers.