Premium
Measurement Characteristics of the lEA Reading Literacy Scales for 9‐ and 10‐Year‐Olds at Country and Individual Levels
Author(s) -
Gustafsson JanEric
Publication year - 1997
Publication title -
journal of educational measurement
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.917
H-Index - 47
eISSN - 1745-3984
pISSN - 0022-0655
DOI - 10.1111/j.1745-3984.1997.tb00517.x
Subject(s) - variance (accounting) , narrative , literacy , reading (process) , dimension (graph theory) , confirmatory factor analysis , psychology , latent variable , factor (programming language) , mathematics education , econometrics , linguistics , statistics , computer science , structural equation modeling , mathematics , pedagogy , philosophy , accounting , pure mathematics , business , programming language
This article presents an approach for studying measurement characteristics of instruments designed for comparative studies of educational achievement. It presents a reanalysis of data from 22 countries and 73,818 9‐ and 10‐year‐old students from the IEA reading literacy study. Using 2‐level confirmatory factor analysis procedures derived by Muthén (1990, 1994), it is demonstrated how the overall variance in performance on items from 3 domains—‐Documents, Narrative Prose, and Expository Text—may be decomposed into multiple latent sources of variance at country and individual levels. For the Documents items, a 1‐factor model was fitted at both levels. For Narrative Prose and Expository Text items, 2 latent variables (a general dimension of reading performance and a factor capturing high versus low performance on passages towards the end of the tests) were fitted at both country and individual levels. For the Documents dimension, estimates of country scores were close to those of the original analysis; however, for performance in the Narrative Prose and Expository Text domains, differences between the original analysis and the reanalysis were noted for many countries.