Open AccessAn Item‐Driven Adaptive Design for Calibrating Pretest Items
Author(s) -
Ali Usama S.,
Chang HuaHua
Publication year - 2014
Publication title -
ets research report series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.235
H-Index - 5
ISSN - 2330-8516
DOI - 10.1002/ets2.12044
Subject(s) - computerized adaptive testing , computer science , calibration , item response theory , adaptive design , sample (material) , item bank , machine learning , data mining , artificial intelligence , statistics , mathematics , psychometrics , medicine , chemistry , pathology , chromatography , clinical trial
Adaptive testing is advantageous in that it provides more efficient ability estimates with fewer items than linear testing does. Item‐driven adaptive pretesting may also offer similar advantages, and verification of such a hypothesis about item calibration was the main objective of this study. A suitability index (SI) was introduced to adaptively select pretest items, by which an easy‐to‐implement calibration methodology—adaptive design—can be used. A simulation study was conducted to evaluate the proposed adaptive design as compared to existing methodologies. Results indicate that the adaptive design has many desired features in item calibration, including less bias and more accurate parameter estimates, than the existing methods do. The SI is promising and flexible enough to apply additional constraints on the calibration sample and on the pretest items, for example, constraints on response time. It can also be used to try out individual item modules such as those used in multistage testing. Study limitations and future research are also covered in this report.