A simulation study to determine appropriate linking methods for adaptive testing items was designed. Three basic data sets for responses were created. These were randomly sampled, systematically sampled, and selected data sets. The evaluative criteria used were fidelity of parameter estimation, asymptotic ability estimates, root-mean-square error…

This paper explores the effects of item choice on ability estimation when using a tailored testing procedure based on the Rasch simple logistic model. Most studies of the simple logistic model imply that ability estimates are totally independent of the items used, regardless of the testing procedure. This paper shows that the ability estimate is…

The purpose of this study was to compare and evaluate three online pretest item calibration/scaling methods in terms of item parameter recovery when the item responses to the pretest items in the pool would be sparse. The three methods considered were the marginal maximum likelihood estimate with one EM cycle (OEM) method, the marginal maximum…

In a previous study (1998), how to evaluate whether adaptive testing data used for online calibration sufficiently fit the item response model used by C. Glas was studied. Three approaches were suggested, based on a Lagrange multiplier (LM) statistic, a Wald statistic, and a cumulative sum (CUMSUM) statistic respectively. For all these methods,…

Mantel-Haenszel and SIBTEST, which have known difficulty in detecting non-unidirectional differential item functioning (DIF), have been adapted with some success for computerized adaptive testing (CAT). This study adapts logistic regression (LR) and the item-response-theory-likelihood-ratio test (IRT-LRT), capable of detecting both unidirectional…

The concept of testing for partial knowledge is considered with the concept of tailored testing. Following the special usage of latent trait theory, the word valdity is used to mean the correlation of a test with the construct the test measures. The concept of a method factor in the test is also considered as a part of the validity. The possible…

Maximum likelihood and Bayesian procedures are presented for item selection and scoring of multidimensional adaptive tests. A demonstration with simulated response data illustrates that multidimensional adaptive testing can provide equal or higher reliabilities with fewer items than are required in one-dimensional adaptive testing. (SLD)

A computerized flexilevel test was implemented and its ability estimates were compared with those of a Bayesian estimation based computerized adaptive test (CAT) as well as with known true ability estimates. Results showed that when the flexilevel test was terminated according to Lord's criterion, its ability estimates were highly and…

Monte Carlo simulation was used to investigate score bias and information characteristics of Owen's Bayesian adaptive testing strategy, and to examine possible causes of score bias. Factors investigated in three related studies included effects of item discrimination, effects of fixed vs. variable test length, and effects of an accurate prior…

Measures of the robustness of disease class-specific diagnostic concepts could play a central role in training programs designed to assure the development of diagnostic competence. In the pilot study, the authors used disease/sign-symptom conditional probability estimates, Monte Carlo procedures, and artificial intelligence (AI) tools to create…

Responses to previously calibrated items administered in a computerized adaptive testing (CAT) mode may be used to recalibrate the items. This live-data simulation study investigated the possibility, and limitations, of on-line adaptive recalibration of precalibrated items. Responses to items of a Rasch-based paper-and-pencil licensure examination…

This report examines how selection fairness is influenced by the characteristics of a selection instrument in terms of its distribution of item difficulties, level of item discrimination, degree of item bias, and testing strategy. Computer simulation was used in the administration of either a conventional or Bayesian adaptive ability test to a…

Monte Carlo simulation procedures were used to study the psychometric characteristics of two two-stage adaptive tests and a conventional "peaked" ability test. Results showed that scores yielded by both two-stage tests better reflected the normal distribution of underlying ability. Ability estimates yielded by one of the two stage tests…

The accuracy of estimation of the subjects' latent ability maintained by tailoring for each testee the order of item presentation and the border of item dichotomization was compared to the information provided by the original graded test items. (RC)

An adaptive testing strategy is described for achievement tests covering multiple content areas. The strategy combines adaptive item selection both within and between the subtests in the multiple-subtest battery. A real-data simulation was conducted to compare the results from adaptive testing and from conventional testing, in terms of test…