When comparing two tests that measure the same trait, separate comparisons should be made at different levels of the trait. A simple, practical, approximate formula is given for doing this. The adequacy of the approximation is illustrated using data comparing seven nationally known sixth-grade reading tests. (Author/RC)

A two-stage testing procedure, a routing test followed by one of several alternative second-stage tests, is studied in the situation where the purpose is measurement, not classification. Models are developed, examined, and compared with conventional tests and up-and-down procedures. (DG)

An upper-asymptote parameter was added to the three-parameter logistic item response model. This four-parameter model was compared to the three-parameter model on four data sets. The fourth parameter increased the likelihood in only two of the four sets. Ability estimates for the students were generally unchanged by the introduction of the fourth…

The stepped-up reliability coefficient does not have the same standard error as an ordinary correlation coefficient. Fisher's Z -transformation should not be applied to it. Appropriate procedures are suggested. (Author)

Two approaches for determining the optimal number of choices for a test item, presently in the literature, are compared with two new approaches. (Author)

A variety of practical applications of item characteristic curve test theory are discussed. Among these applications are tailored testing, two stage testing, determining whether two tests measure the same latent trait, and measuring item bias towards minority or other groups. (Author/JKS)

Given known item parameters for a test, unbiased estimators are derived for an examinee's ability parameter, his or her proportion correct true score for the variances of these parameters and for the parallel forms reliability of the maximum likelihood estimator of the ability parameter. (Author/JKS)

Under given conditions, conventional testing and computer-generated repeatable testing (CGRT) are equally effective for estimating examinee ability; CGRT is more effective for estimating the mean ability level of a group and less effective for estimating ability differences among individuals. These conclusion are drawn from domain-referenced test…