Prediction of item psychometric indices from item characteristics automatically extracted from the stem and option text. (1st January 2011)

• Record Type: Journal Article
• Authors: Garcia, Carmen; Ponsoda, Vicente; Sierra, Alejandro
• Is Part Of: International journal of continuing engineering education and lifelong learning. Volume 21:Number 2/3(2011)
• Page Start: 210
• Page End: 221
• Publication Date: 2011-01-01
• Subjects: multiple choice items -- item writing guidelines -- item difficulty -- item discrimination -- item parameter prediction -- automatic item reading; Engineering -- Study and teaching -- Periodicals; 620.00715
• Languages: English
• ISSNs: 1560-4624

Abstract

This study has four main parts: first, some of the 31 guidelines proposed by Haladyna et al. (2002) were converted to 39 features that can be automatically extracted from a text file containing the test stem and options. Second, text files of a few multiple choice university exams were collected and the 39 features were extracted. Third, the difficulty and discrimination indices of each item were computed. Finally, linear regression was applied to find which features predict the items' psychometric indices. The number of special adverbs (always, never, …) in the distracters and the discrimination index show the strongest link.