A strategy for the pairwise assessment which may be used to evaluate the nature of both "prerequisite" and "transference" relations existing among a set of traits is presented. Both confirmatory and exploratory strategies are presented. An example of how the strategy would be used is presented in the exploratory context.…

Goodness of fit tests for the Rasch model are typically large-sample, global measures. This paper offers suggestions for small-sample exploratory techniques for examining the fit of item data to the Rasch model. (Author/JKS)

In the study based on scores from 221 mentally retarded residents, the individual items in the first 13 Part Two domains of the American Association on Mental Deficiency Adaptive Behavior Scale were weighted for the perceived severity of the behavior they represent. (Author/SW)

A methodology of computing prescribed point biserial correlation coefficients and then evaluating the coefficients with respect to three criteria was developed. It can be used to assess the item convergent validity of nonobjectively scored test items. (Author/PN)

Mathematical and statistical relationships between factor analysis and smallest space analysis are discussed. As spatial analysis of correlation matrices, factor analysis is a special case of smallest space analysis. The two differ in six ways: Shepard diagram, dimensionality, correction for communality, similarity coefficients, regions versus…

Two algorithms for marginal maximum likelihood estimation for the Rasch model are provided. The more efficient of the two algorithms is extended to estimation for the linear logistic model. Numerical examples of both procedures are presented. (Author/JKS)

Most psychological tests and inventories, particularly in personality and attitude measurement, contain common items. The present report specifically considered common-item contributions to test internal consistency reliability using extant data from the California Psychological Inventory. No negative contribution of item overlap was found.…

A model is proposed for analyzing ability tests which utilizes a cognitive approach to understanding test items and latent trait analysis for quantifying subjects responses to test items. The model adopts a multicomponent perspective on the processes involved in answering test items. (JKS)

A new method--similar to Christoffersson's--for the factor analysis of dichotomous variables uses information from the first and second order proportions to fit a multiple factor model. The estimation is considerably simplified. Items from Rotter's locus of control inventory are included. (Author/JKS)

Deciding whether sets of test data are consistent with any of a large class of item response models is considered. The assumption of local independence is weakened to a new condition, local nonnegative dependence (LND). Necessary and sufficient conditions are derived for a LND item response model. (Author/JKS)

Using Monte Carlo generated item response data, this research sought to determine the effectiveness, sufficiency and similarity of selected techniques for detecting item bias. The three-parameter latent-trait test model was used to generate the simulated data. (Author/JKS)

A chain of lower bound inequalities leading to the greatest lower bound to reliability is established for the internal consistency of a composite of unit-weighted scores (such as a test). Algorithms for obtaining various reliability coefficients are presented. (Author/JKS)

Investigations of item bias provide an empirical basis for the identification and elimination of items appearing to measure different traits across population/culture groups. This paper reviews the psychometric rationales of six types of approaches to biased item identification. (Author/JKS)

Computer simulations were employed to investigate sample sizes needed for item analysis when one's primary purpose is to rank order items in terms of indices of quality or to compare indices of quality against a fixed criterion. Necessary sample sizes were much smaller than those indicated by the literature. (Author/JKS)

The negative bias of biserial correlations in the presence of guessing is demonstrated by Monte Carlo studies, and another estimator is described that is free of such bias. The usual biserial-correlation assumptions and assumptions about guessing are discussed. Brogden's coefficient of selective efficiency and the triserial correlation are…