Search
Program Calendar
Browse By Day
Browse By Time
Browse By Person
Browse By Room
Browse By Unit
Browse By Session Type
Search Tips
Annual Meeting Registraion, Housing and Travel
Personal Schedule
Sign In
When pretesting test questions, maximizing the number of examinee responses can improve item parameter estimates; however, such responses are typically limited due to the time limits of the exam administration or because of the security risk posed by exposing pretest items. The problems raised by these competing goals can be ameliorated somewhat by supplementing examinee response data with auxiliary information based on incidental item features, which have been shown to improve estimation under small sample conditions. Using data from a large-scale high-stakes, licensure exam, this paper considers the relationships between 113 incidental item features that are linguistically-motivated and three outcomes that are often of interest when pretesting: item difficulty, response time, and item eligibility for scored use.
Janet Mee, National Board of Medical Examiners
Peter Baldwin, National Board of Medical Examiners
Le An Ha, University of Wolverhampton
Viktoriya Yneva, National Board of Medical Examiners