A SIMPLICIAL DESIGN FOR THE ANALYSIS OF CORRELATIONAL LEARNING DATA
Benjamin Fruchter, Edwin A. Fleishman · Multivariate Behavioral Research · 1967
The usual matrix of intercorrelations among trials in a study of the learning process produces a superdiagonal matrix with resulting factors which are uninteresting: viz, a factor for the early trials and a factor for the later trials; with perhaps also a factor for the middle trials. By introducing several independent measures of learning for each trial a more meaningful factor structure can be obtained.