Development and validation of a short form for the Raven's Coloured Progressive Matrices using a machine learning approach

Abstract

Raven's Coloured Progressive Matrices (CPM) is a widely used assessment tool for measuring general cognitive ability in developmental and educational research, particularly in studies involving young children. However, administering the full set of the 36-item CPM can be burdensome for young participants, hindering its practicality in large-scale studies and reducing research efficiency. In the current study, a short form of the CPM was developed based on a sample of preschoolers (n = 336, mean age = 5.8 years) using penalised regression, a machine learning approach that allows for variable selection. The resulting 12-item CPM short form demonstrated a very strong correlation with the total score of the 36-item full form (r = .94). Further investigations into the short form's item stability, content validity, and concurrent validity collectively supported its psychometric properties as a reliable and valid alternative to the full form. The significance of the CPM short form is also discussed.