IDENTIFICATION OF SUB‐SKILLS OF READING COMPREHENSION BY MAXIMUM LIKELIHOOD FACTOR ANALYSIS1

Abstract

Differing interpretations have been made of the results obtained by Davis in a large‐scale study of the mental skills involved in reading comprehension among twelfth grade students. On the basis of a uniqueness analysis and a varimax‐rotated principal components analysis, Davis maintained that five of his eight postulated skills were experimentally distinguishable. A refactorization of Davis's data by R. L. Thorndike suggested that except for word knowledge, the reading skills were not separately distinguishable. The present study consists of a further refactorization of Davis's data, taking more advantage of their factor analytic design possibilities, and employing the more comprehensive procedures now available for maximum likelihood factor analysis. Word knowledge and three other skills were shown to be separately distinguishable, but the latter three skills were very highly correlated and thus could be predominantly measuring a single basic ability.