ERIC Number: ED362042
Record Type: Non-Journal
Publication Date: 1993-Jul-9
Construct Validation of Question Formats for Dutch Central Examinations in Foreign Language Reading Comprehension.
Krieken, Robert van
The research project reported in this paper had two aims: to demonstrate what standard multiple-choice questions and two new varieties (multiple-choice cloze and guided summary test and to assess whether the new formats test the same skills. To find out what the various question formats test, the study looked at a battery of tests, each of which was aimed at testing a different factor that might contribute to reading comprehension. Factors that are not intended to be measured by exam questions, such as non-verbal intelligence and topic knowledge were included as well as the known components of a model. In the first of two experiments, 500 pupils took the whole battery and 1 of the 2 multiple-choice formats. In a second experiment, a smaller group took a selection from the battery, multiple-choice items and a guided summary. Latent correlations between the tests in the battery demonstrated that there was more coherence among the tests of reading comprehension components than between these and the tests of topic knowledge and non-verbal intelligence. Differences in the content of the three question formats were not found. Each question format correlated to roughly the same amount with all the reading/comprehension abilities. The multiple-choice cloze, however, exhibited consistently lower correlations than the other question formats. (Author/JL)
Publication Type: Reports - Research
Education Level: N/A
Authoring Institution: Centraal Inst. voor Toetsonwikkeling, Arnhem (Netherlands).
Identifiers - Location: Netherlands