Finding and interpreting lower bounds for reliability coefficients for tests with nonhomogenous items has been a problem for psychometricians. This paper presents a mathematical formula for finding the greatest lower bound for such a coefficient. (Author/JKS)

This discussion illustrates the application of generalizability theory to a design commonly employed in the collection of evaluation data and provides a detailed analysis of the dependability of student evaluations of college teaching. (RC)

This paper provides analytic evaluations of expected (marginal) true-score measures for binary items given their item response theory (IRT) calibration. Under the assumption of normal trait distributions, marginalized true scores, error variance, true score variance, and reliability for norm-referenced and criterion-references interpretations are…

This paper reviews various procedures for constructing an interval for an individual's true score given the assumption that errors of measurement are distributed as binomial. This paper also presents two general interval estimation procedures (i.e., normal approximation and endpoints conversion methods) for an individual's true scale score;…

Exact formulas for classical error variance are provided for Rasch measurement with logistic distributions. An approximation formula with the normal ability distribution is also provided. With the proposed formulas, the additive contribution of individual items to the population error variance can be determined without knowledge of the other test…

Classical test theory is based on the concept of a true score for each examinee, defined as the expected or average score across an infinite number of repeated parallel tests. In most cases, there is only a score from a single administration of the test in question. The difference between this single observed score and the underlying true score is…

Article commented on a study by Harris, who presented formulas for the variance of errors of estimation (of a true score from an observed score) and the variance of errors of prediction (of an observed score from an observed score on a parallel test). (Author/RK)

This reply refers to PS 502 345 which in turn is a criticism of EJ 045 083. (CB)

The problem of whether a precise connection exists between the stochastic processes considered in mathematical learning theory and the Guttman simplex is investigated. The approach used is to derive a set of conditions which a probabilistic model must satisfy in order to generate inter-trial correlations with the perfect simplex property.…

Compares four closed-formula estimators (Burkett, Claudy, Rozeboom, Browne) and the omit-one method for estimating TRS, the true shrunken correlation (not to be confused with TR, the true multiple correlation). Recommendations are based on artificial populations with known TRS. (Author/SV)

It is demonstrated that, in the presence of population information, a linear combination of true scores can be estimated more efficiently than by the same linear combination of the observed scores. Three criteria for optimality are discussed, but they yield the same solution, described as a multivariate shrinkage estimator. (Author/SLD)

Stability among equating procedures was studied by comparing item response theory (IRT) true-score equating with IRT observed-score equating, IRT true-score equating with equipercentile equating, and IRT observed-score equating with equipercentile equating. On average, IRT true-score equating more frequently produced more stable conversions. (SLD)

Two independent replications of a sequence of simulations were conducted to aid in the diagnosis and interpretation of equating differences found between representative (random) and matched (nonrandom) samples for three commonly used conventional observed-score equating procedures and one item-response-theory-based equating procedure. (SLD)

The procedure of M.L. Stocking and F.M. Lord (1983) for computing equating coefficients for tests having dichotomously scored items is extended to the case of graded response items. A system of equations for obtaining the equating coefficients under the graded response model is derived. (SLD)