Allowing examinee choice in educational testing

Cankar, Gašper (author)

Achievement tests sometimes entertain examinee choice - a situation where an examinee is presented with a set of items among which s/he has to choose one (or few) to answer that will be scored in her/his total test score. Basic assumption is that choice items are equivalent regarding both content and psychometric characteristics and therefore it doesn't matter which particular item examinee selects. Choice items often also share same maximum number of points and examinee score on a test is usually achieved by summing scores from all items taken by the examinee regardless of their combination. Choice items present conceptual problems like why enabling choice when items should be equivalent in the first place and methodological ones like how item scores from different combinations of items should contribute to comparable total score on test. Author used Rasch's model within Item Response Theory framework to test the assumption of equivalence of choice items by scaling all item difficulties on same scale. Physics 2008 and 2009 tests from Slovenian General Matura examination are analyzed as an example to explore equivalence of choice items. Differences in difficulty of choice items in those tests are presented and discussed. It seems that examinee choice still doesn't work in educational testing and should be avoided when possible.

• Language: English
• Work type: Not categorized
• Typology: 1.01 - Original Scientific Article
• Organization: FDV - Faculty of Social Sciences
• Year: 2010
• Number of pages: Str. 151-166
• Numbering: Vol. 7, no. 2
• PID: 20.500.12556/RUL-23134
• UDC: 311.1:373.5.091.27(497.4)
• ISSN on article: 1854-0023
• COBISS.SI-ID: 29643357
• Publication date in RUL: 11.07.2014

Record is a part of a journal

• Title: Advances in methodology and statistics
• Shortened title: Metodol. zv.
• Publisher: Fakulteta za družbene vede
• ISSN: 1854-0023
• COBISS.SI-ID: 215795712