A study was undertaken to develop guidelines for the interpretation of the parameters of three multidimensional item response theory models and to determine the relationship between the parameters and traditional concepts of item difficulty and discrimination. The three models considered were multidimensional extensions of the one-, two-, and…

The appropriateness of the use of the standardized residual (SR) to assess congruence between sample test item responses and the one parameter latent trait (Rasch) item characteristic curve is investigated. Latent trait theory is reviewed, as well as theory of the SR, the apparent error in calculating the expected distribution of the SR, and…

A study was conducted to compare the quality of the item parameter estimates obtained from the ANCILLES and LOGIST estimation procedures using goodness of fit as a criterion. Statistics used to compare the fit included a chi-square statistic and a mean square deviation statistic. Other analyses performed included comparisons of the distributions…

To test for item bias, it must be determined whether an item fits the model. Two approaches to defining bias within the framework of the Rasch model are examined. One compares within-group fit mean squares and the other utilizes a between-group fit statistic. Results from both approaches overlap somewhat, but are distinct in many different but…

Two methods of item selection were used to select sets of 40 items from a 50-item verbal analogies test, and the resulting item sets were compared for relative efficiency. The BICAL program was used to select the 40 items having the best mean square fit to the one parameter logistic (Rasch) model. The LOGIST program was used to select the 40 items…

Believing that examinee-by-item interaction should be conceptualized as true score variability rather than as a result of errors of measurement, Lu proposed a modification of Hoyt's analysis of variance reliability procedure. Via a computer simulation study, it is shown that Lu's approach does not separate interaction from error. (Author/RL)

Some stochastic approximation procedures are considered in relation to the problem of choosing a sequence of test questions to accurately estimate a given examinee's standing on a psychological dimension. Illustrations are given evaluating certain procedures in a specific context. (Author/CK)

The Rasch model for test analysis is described and compared with two-parameter and three-parameter latent-trait models. Conditional maximum likelihood equations for estimating item parameters are derived, and estimates of person parameters are described together with their confidence intervals. Goodness of fit tests are discussed, including a…

Relationships between item bias, item difficulty invariance, Rasch tests of item fit, and item position in a speeded 72-item Michigan State University Vocabulary Placement Test were investigated using 143 black males, 289 black females, 2,953 white males and 3,271 white females. Measures of item bias and item difficulty invariance were determined…

Simulated tailored tests were used to investigate the relationships between characteristics of the item pool and the computer program, and the reliability and bias of the resulting ability estimates. The computer program was varied to provide for various step sizes (differences in difficulty between successive steps) and different acceptance…

TAILOR is a computer program that uses the implied orders concept as the basis for computerized adaptive testing. The basic characteristics of TAILOR, which does not involve pretesting, are reviewed here and two studies of it are reported. One is a Monte Carlo simulation based on the four-parameter Birnbaum model and the other uses a matrix of…

To improve the accuracy of college placement, Rasch scoring and person-fit statistics on the Comparative Guidance and Placement test (CGP) was compared to the traditional right-only scoring. Correlations were calculated between English and mathematics course grades and scores of 1,448 entering freshmen on the reading, writing, and mathematics…

Latent trait theory supposes that, in testing situations, examinee performance on a test can be predicted (or explained) by defining examinee characteristics, referred to as traits, estimating scores for examinees on these traits and using the scores to predict or explain test performance (Lord and Novick, 1968). In view of the breakthroughs in…