초록
영어
This study applied Testlet-based response theory (TRT) model to a set of reading comprehension data from a TOEIC practice test. The parameter estimates and test information were compared to those of the standard item response theory (IRT) model. The results show that the model choice could have a serious effect on item slope and guessing parameter estimates. However, the two models did not yield any significant difference in item difficulty and proficiency parameter estimates. Test information as a measure of global precision shows that the two models could yield a differential degree of test precision depending on the range of ability distribution. A further textual analysis was also performed to see why certain testlets have more dependence than others in the same test
목차
1. Introduction
2. Procedures
2.1. Data
2.2. Parameter Estimation
2.3. Test Information as a measure of test Precision
3. Results
4. Discussions and Conclusions
References