Fan et al. (2012) demonstrated that inversely weighting the maximum information criterion (MIC) by the expected response time for each item effectively reduced the average completion time for a test of fixed length while maintaining efficiency of ability estimation. However, this also resulted in an extremely skewed distribution of item exposure rates. To counter this issue, they recommended the use of the a-stratification with b-blocking (ASB) method, which balanced item exposure very well but at the cost of compromising on the average test time reduction and increasing the error of ability estimation. Therefore, the current study investigated several alternative methods for item exposure control when using response time-informed MIC, of which the most promising technique was inversely weighting MIC by the expected response time centered around 1.3. Simulation results showed that this simple modification markedly improved performance over employing ASB when utilizing response time information for item selection in computerized adaptive testing (CAT).
【 预 览 】
附件列表
Files
Size
Format
View
Controlling item exposure for response time-informed item selection in computerized adaptive testing