Aktuelle Tendenzen / Current Trends
C-Tests consist of several short texts in which the missing parts of words have to be reconstructed. C-Tests have excellent psychometric properties and are used in many contexts as economical tests of language proficiency. This collection of papers illustrates the state of the art of C-Test research. It focuses on face validity and washback of C-Tests, C-Test performance and intelligence, C-Tests as screening devices for TestDaF and SIMTEST, ROC analyses for relating C-Test scores to the CEFR as well as use of C-tests in the classroom. A special feature of the volume is the editor’s comprehensive C-Test bibliography.
Do test takers with different language backgrounds take the same C-test? The effect of native language on the validity of C-tests
Monique Reichert, Martin Brunner & Romain Martin*
A great deal of research has empirically shown that test takers’ language background may affect the validity of language tests. Although C-Tests are widely used, this potential threat to their validity has not previously been thoroughly addressed. This study therefore rigorously examined the extent to which a German and a French C-Test developed for use in a large-scale study in Luxembourg measure a single latent trait – assumed to be general language proficiency – in an equivalent way across two language groups. Multi-group confirmatory factor modeling was used to test for measurement invariance across Luxembourgish- and German-speaking students as compared with students speaking a Romance language. It was hypothesized that, if the C-Tests measured a single construct in the same way across both language groups, then this construct would not only become apparent in both groups, but it would also affect the observed C-Test scores in the same way across the groups. In addition, cross-group differences in the means of the observed items would be attributable to differences in the means of the underlying construct. Finally, if the C-Tests were to be applied as high-stakes tests, then equivalence of the observed test variances (affected by the items’ residual variances) should also be given. The results supported the measurement invariance of the German C-Test. However, language background was found to have a stronger impact on the validity of the French C-Test. The results are discussed in terms of hypothesized differences in the...