Invariance levels across language versions of the PISA 2009 reading comprehension tests in Spain.

BACKGROUND The PISA project provides the basis for studying curriculum design and for comparing factors associated with school effectiveness. These studies are only valid if the different language versions are equivalent to each other. In Spain, the application of PISA in autonomous regions with their own languages means that equivalency must also be extended to the Spanish, Galician, Catalan and Basque versions of the test. The aim of this work was to analyse the equivalence among the four language versions of the Reading Comprehension Test (PISA 2009). METHOD After defining the testlet as the unit of analysis, equivalence among the language versions was analysed using two invariance testing procedures: multiple-group mean and covariance structure analyses for ordinal data and ordinal logistic regression. RESULTS The procedures yielded concordant results supporting metric equivalence across all four language versions: Spanish, Basque, Galician and Catalan. CONCLUSIONS The equivalence supports the estimated reading literacy score comparability among the language versions used in Spain.