A simulation study compared the performance of the Mann-Whitney "U" test, Student's "t" test, and the alternate (separate variance) "t" test for two mutually independent random samples from normal distributions, with both one-tailed and two-tailed alternatives. Type I error probabilities were computed with equal and…

An adaptive test can usually match or exceed the measurement precision of conventional tests several times its length. This increased efficiency is not without costs, however, as the models underlying adaptive testing make strong assumptions about examinees and items. Most troublesome is the assumption that item pools are unidimensional. Truly…

E.J. Bedrick recently derived the asymptotic distribution of the modified sample biserial correlation estimator of F.M. Lord and studied its efficacy for bivariate normal populations. A more detailed examination of the properties of Lord's estimator is provided, and the small sample bias and variance are studied. (SLD)

A model is proposed for investigating test validity as a predictor of a criterion variable when there are both missing and censored scores in the data set. Implications for maximum likelihood estimation are discussed, and the method is illustrated with hypothetical data sets. (SLD)

The psychometric objections to using essays and other subjective tests for measurement can be overcome by a many-facet Rasch model. This model enables judge-awarded grades to be transformed from their arbitrary, local, non-linear rating scale form into linear measures with explicit generalizable meaning of specifiable reliability (standard error)…

Single construct validity is examined, and appropriate statistical tests are described when criterion information is available for: (1) examinees selected at random; (2) examinees selected on the basis of the current test; and (3) examinees selected using other measures of the latent construct. Discriminant validity is also discussed. (SLD)

Many researchers have suggested that the main cause of item bias is the misspecification of the latent ability space. That is, items that measure multiple abilities are scored as though they are measuring a single ability. If two different groups of examinees have different underlying multidimensional ability distributions and the test items are…

Test validity is a concept that has often been ignored in the context of latent trait models and in modern test theory, particularly as it relates to computerized adaptive testing. Some considerations about the validity of a test and of a single item are proposed. This paper focuses on measures that are population-free and that will provide local…

Common approaches to examining the relationship between multitrait-multimethod (MTMM) data and variables outside the MTMM data are compared: averaging various means of each trait and estimating LISREL computer program models, and estimating only relationships between MTMM traits and the outside variables. Problems of correlational bias are…

Extensions of unidimensional appropriateness indices are developed for multiunidimensional tests (multidimensional tests composed of unidimensional subtests). Simulated and real data (scores of 2,978 students on the Armed Services Vocational Aptitude Battery) were used to evaluate the indices' effectiveness in determining individuals who are…

Effects of the following methods for linking metrics on detection of differential item functioning (DIF) were compared: (1) test characteristic curve method (TCC); (2) weighted mean and sigma method; and (3) minimum chi-square method. With large samples, results were essentially the same. With small samples, TCC was most accurate. (SLD)

This paper is the final report of a multi-year project sponsored by the Office of Naval Research (ONR) in 1987 through 1990. The main objectives of the research summarized were to: investigate the non-parametric approach to the estimation of the operating characteristics of discrete item responses; revise and strengthen the package computer…

A two-dimensional, compensatory item response model and a unidimensional model were fitted to the reading and mathematics items in the Woodcock-Johnson Psycho-Educational Battery-Revised for a sample of 1,000 adults aged 20-39 years. Multidimensional information theory predicts that if the unidimensional abilities can be represented as vectors in…

The difference between item bias and item impact and the way they relate to item validity are discussed from a multidimensional item response theory perspective. The Mantel-Haenszel procedure and the Simultaneous Item Bias strategy are used in a Monte Carlo study to illustrate detection of item bias. (SLD)