A 1992 study by R. Gershon found discrepancies when comparing the theoretical Rasch item characteristic curve with the average empirical curve for 1,304 vocabulary items administered to 7,711 students. When person-item mismatches were deleted (for any person-item interaction where the ability of the person was much higher or much lower than the…

A previous simulation study of methods for assessing item functioning (DIF) in computer-adaptive tests (CATs) showed that modified versions of the Mantel-Haenszel and standardization methods work well with CAT data. In that study, data were generated using the three-parameter logistic (3PL) model, and this same model was assumed in obtaining item…

Computerized adaptive tests (CATs) make it possible to estimate the ability of each student during the testing process. The computer presents items to students at the appropriate level, and students take different versions of the same test. Computerized testing increases the flexibility of test management in that: (1) tests are given on demand and…

A nominal response model-based computerized adaptive testing procedure (nominal CAT) was implemented using simulated data. Ability estimates from the nominal CAT were compared to those from a CAT based upon the three-parameter logistic model (3PL CAT). Furthermore, estimates from both CAT procedures were compared with the known true abilities used…

The present monte carlo simulation study was designed to examine the effects of multidimensionality during the administration of computerized adaptive testing (CAT). It was assumed that multidimensionality existed in the individuals to whom test items were being administered, i.e., that the correct or incorrect responses given by an individual…

Major efforts of this project fall into four categories: (1) investigations were performed on the relationship between the dimensionality of a dataset and its underlying cognitive processes; (2) two approaches for diagnosing erroneous rules of operation were developed (an "error vector" system for constructing error diagnostic programs…

The use of item response theory (IRT) is a prerequisite to successful use of computerized test systems. In item response models, as opposed to classical test theory, the abilities of the examinees and the properties of the items are parameterized separately. Therefore, when measuring the abilities of examinees, the model implicitly corrects for…

This paper presents an explanation of the program design of the three separate Pascal-language programs which comprise UCLA's Diagnostic Testing Package "DX." The three parts of the DX test package are three closely interrelated programs--Editest, Runtest, and Summary. Editest serves the following functions: (1) the creation of a…

The development of a Spanish computerized adaptive placement test at Brigham Young University (Utah) is described. The test is based on an approach that locates an examinee's ability along a continuum of latent ability rather than determining the degree to which an individual's ability differs from the abilities of others taking the test, as in…

This paper has shown the advantages of tailored testing over traditional methods, the need for latent trait theory in the application of tailored testing, and some of the results available to show the quality of estimates obtained by tailored testing procedures. Since the tailored testing procedures overcome many of the problems with traditional…

This paper describes two procedures for making binary classification decisions using tailored testing: the sequential probability ratio test (SPRT) and a Bayesian decision procedure. The first procedure described, the SPRT, was developed by Wald for quality control work. It has not been widely applied for testing applications because the…

The consistency of student response patterns on a test of language arts was examined in a set of studies as a first step toward designing a computer adaptive test to diagnose errors. (Author/LMO)

This articles discusses ways to respond to the need for criterion referenced information using Bayes' theorem, a method coupled with criterion referenced testing in the early 1970s. The theorem determines the most likely classification for an examinee from a dichotomous choice or through placement on a categorical or interval scale. A discussion…

A considerable amount of data on computerized adaptive testing (CAT) has been conducted using simulated data. However, most researchers would agree that simulations may not fully reflect the reality of examinee performance on a test. This study used maximum likelihood procedures to investigate the accuracy and efficiency of examinee ability…

A small-scale study was conducted to compare test-taking strategies, problem-solving strategies, and general impressions about the test across computer and paper-and-pencil administration modes. Thirty-six examinees (high school students) participated in the study. Each examinee took a test in one of the content areas of English, Mathematics,…