Strategies for computerized adaptive testing: golden section search, dichotomous search, and Z-score strategies

Xiao, Beiling

doi:10.31274/rtd-180813-9591

Cited by 1 publication

(3 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The finalθ is the MLE. Z-score strategies have also been shown to be very effective for a broad range of CATs (Xiao, 1993). ZGT was adapted from the Z-score strategies for CAT.…”

Section: Z-score Search Cat and Zgtmentioning

confidence: 99%

“…The obtained score, x, was compared to the expected score at the midpoint of the search region. If x was greater or less than, respectively, the upper or lower bound of the confidence interval of the expected score µ x|θ ± .7σ x|θ , that side of the search region was discarded [.7 is an arbitrary weight used in the present study, based on Xiao (1993) who showed that GSS strategies were most efficient at .7]. The size of successive golden search regions was reduced by a ratio t (see Equation 1).…”

Section: Testsmentioning

confidence: 99%

“…The finalθ is the maximum likelihood estimate (MLE). GSS strategies have been shown to be very effective for a broad range of CATs (Xiao, 1989(Xiao, , 1993. GGT differs from GSS strategies in that it adds a grade assignment evaluation after each item is administered.…”

mentioning

confidence: 99%

See 2 more Smart Citations

Strategies for Computerized Adaptive Grading Testing

Xiao

1999

Applied Psychological Measurement

Self Cite

View full text Add to dashboard Cite

IRT-based adaptive grading tests are designed to assign examinees to one of several grading categories. The item search algorithm in these tests can be based on either a golden section search, a Z-score, or an EAP-based search; these methods result, respectively, in the golden search grading test (GGT), the Z-score grading test (ZGT), and the EAP grading test (EGT). Grade assignments are evaluated after each item is administered and after the current trait estimate ([UNKNOWN]) has been determined. A test is terminated based on one of three conditions: (1) [UNKNOWN] is between two cutoff scores; (2) [UNKNOWN] is above or below the highest or lowest cutoff scores, respectively; or (3) a prespecified maximum number of items has been administered. Monte carlo studies using actual ACT Mathematics test item parameters showed that all three strategies effectively assigned examinees into multiple achievement grade levels. EGT had more correct classifications in the middle range of grade levels and more classification errors for extreme true grade levels than CGT and ZGT. However, when assuming simulees with a relatively larger variance priordistribution, or using maximum likelihood estimation for final [UNKNOWN] estimates, the systematic classification errors disappeared in EGT. GGT and ZGT required fewer items than EGT to terminate the test.

show abstract

“…The finalθ is the MLE. Z-score strategies have also been shown to be very effective for a broad range of CATs (Xiao, 1993). ZGT was adapted from the Z-score strategies for CAT.…”

Section: Z-score Search Cat and Zgtmentioning

confidence: 99%

Section: Testsmentioning

confidence: 99%