00490nas a2200109 4500008004100000245011700041210006900158300001200227490000700239100001400246856012000260 2007 eng d00aThe comparison of maximum likelihood estimation and expected a posteriori in CAT using the graded response model0 acomparison of maximum likelihood estimation and expected a poste a339-3710 v191 aChen, S-K uhttp://iacat.org/content/comparison-maximum-likelihood-estimation-and-expected-posteriori-cat-using-graded-response02911nas a2200133 4500008004100000245016300041210006900204300000800273490000700281520232300288653003402611100001402645856011802659 1997 eng d00aA comparison of maximum likelihood estimation and expected a posteriori estimation in computerized adaptive testing using the generalized partial credit model0 acomparison of maximum likelihood estimation and expected a poste a4530 v583 aA simulation study was conducted to investigate the application of expected a posteriori (EAP) trait estimation in computerized adaptive tests (CAT) based on the generalized partial credit model (Muraki, 1992), and to compare the performance of EAP with maximum likelihood trait estimation (MLE). The performance of EAP was evaluated under different conditions: the number of quadrature points (10, 20, and 30), and the type of prior distribution (normal, uniform, negatively skewed, and positively skewed). The relative performance of the MLE and EAP estimation methods were assessed under two distributional forms of the latent trait, one normal and the other negatively skewed. Also, both the known item parameters and estimated item parameters were employed in the simulation study. Descriptive statistics, correlations, scattergrams, accuracy indices, and audit trails were used to compare the different methods of trait estimation in CAT. The results showed that, regardless of the latent trait distribution, MLE and EAP with a normal prior, a uniform prior, or the prior that matches the latent trait distribution using either 20 or 30 quadrature points provided relatively accurate estimation in CAT based on the generalized partial credit model. However, EAP using only 10 quadrature points did not work well in the generalized partial credit CAT. Also, the study found that increasing the number of quadrature points from 20 to 30 did not increase the accuracy of EAP estimation. Therefore, it appears 20 or more quadrature points are sufficient for accurate EAP estimation. The results also showed that EAP with a negatively skewed prior and positively skewed prior performed poorly for the normal data set, and EAP with positively skewed prior did not provide accurate estimates for the negatively skewed data set. Furthermore, trait estimation in CAT using estimated item parameters produced results similar to those obtained using known item parameters. In general, when at least 20 quadrature points are used, EAP estimation with a normal prior, a uniform prior or the prior that matches the latent trait distribution appears to be a good alternative to MLE in the application of polytomous CAT based on the generalized partial credit model. (PsycINFO Database Record (c) 2003 APA, all rights reserved).10acomputerized adaptive testing1 aChen, S-K uhttp://iacat.org/content/comparison-maximum-likelihood-estimation-and-expected-posteriori-estimation-computerized01795nas a2200169 4500008004100000245014100041210006900182300001200251490000700263520113700270653003401407100001401441700001301455700002101468700001401489856012201503 1997 eng d00aThe effect of population distribution and method of theta estimation on computerized adaptive testing (CAT) using the rating scale model0 aeffect of population distribution and method of theta estimation a422-4390 v573 aInvestigated the effect of population distribution on maximum likelihood estimation (MLE) and expected a posteriori estimation (EAP) in a simulation study of computerized adaptive testing (CAT) based on D. Andrich's (1978) rating scale model. Comparisons were made among MLE and EAP with a normal prior distribution and EAP with a uniform prior distribution within 2 data sets: one generated using a normal trait distribution and the other using a negatively skewed trait distribution. Descriptive statistics, correlations, scattergrams, and accuracy indices were used to compare the different methods of trait estimation. The EAP estimation with a normal prior or uniform prior yielded results similar to those obtained with MLE, even though the prior did not match the underlying trait distribution. An additional simulation study based on real data suggested that more work is needed to determine the optimal number of quadrature points for EAP in CAT based on the rating scale model. The choice between MLE and EAP for particular measurement situations is discussed. (PsycINFO Database Record (c) 2003 APA, all rights reserved).10acomputerized adaptive testing1 aChen, S-K1 aHou, L Y1 aFitzpatrick, S J1 aDodd, B G uhttp://iacat.org/content/effect-population-distribution-and-method-theta-estimation-computerized-adaptive-testing-cat