Assessing different components of reading comprehension: fact or fiction?

Abstract

The starting point of this study was the debate about the structure of L1 reading comprehension. Kalb, Rabenstein and Rost (1979a; 1979b) claim to assess several different subskills of reading comprehension with the test battery Lesen und Verstehen - Diagnose (Reading and Comprehension- Assessment) (RuC-A) they designed specially for this purpose. To check this, 220 second-grade German elementary-school pupils were tested with the RuC-A. The results show medium-to-high correlations for the eight subtests (0.56 ≤ r ≤ 0.85), and some correlation coefficients are nearly as high as the corresponding subtest reliabilities. Corrected for attenuation, they reach 0.66 ≤ rk ≤ 0.99. Depending on the theoretical perspective, factor analyses (with the reliabilities of the subtests as communalities) yield either one broad factor, 'General reading competence' (accounting for 85% of the nonchance variance) or, at most, two factors, 'inferential reading comprehension' and 'vocabulary' (accounting unrotated for 85% and 6%, rotated for 55% and 36% of the reliable variance). As in other comparable L1 reading compre hension tests, RuC-A apparently cannot measure several clearly distinguish able components of reading comprehension. A reliable and valid diagnosis of typical L1 reading comprehension profiles is not possible.