Figural matrix items are widely used for measuring general intelligence. Because of its non-verbal character, this task type is considered to be relatively culture-fair, or at least not as culture-loaded as, for instance, test items with verbal content. However, recent investigations have shown that most non-verbal tests are as much culturally biased as verbal tests. In this study, automatic item generation algorithms were employed to construct matrix items. Test takers received a thorough instruction before the actual test in order to facilitate equal testing conditions. Data from a German and a Russian sample were investigated for the presence of Differential Item Functioning (DIF). It could be shown that there was no DIF in any of the items, leading to the conclusion that culture-fair matrix items can be generated using principles of automatic item generation and after creation of equal testing conditions.
|
