Coefficient Kappa is generally defined in terms of procedures of computation rather than in terms of a population. Here a population definition is proposed. Factors influencing the magnitude of Kappa are identified. Strategies to improve reliability are proposed, including that of combining multiple unreliable diagnoses. (Author/CTM)

A Monte Carlo evaluation of 30 internal criteria for cluster analysis was conducted using four hierarchical clustering techniques. The results indicated that a subset of internal criteria was identified which appear to be valid indices of correct cluster recovery. (Author/JKS)

Several algorithms for computing the greatest lower bound to reliability or the constrained minimum-trace communality solution in factor analysis have been developed. The convergence properties of these methods are examined. A uniqueness proof for the desired solution is offered. (Author/JKS)

This study explores a strategy for improving the stability of variance component estimates when only small samples are available, using a series of small, less complex generalizability (G) study designs as a surrogate for a single large design. (Author/BW)

One argument against the G index is that, unlike phi, it is not a correlation coefficient; yet, G conforms to the Kendall and E-coefficient definitions. The G index is also equal to the Pearson product moment correlation coefficient obtained from double scoring. (Author/CP)

In at least two situations, use of the Spearman-Brown prophesy formula yields overestimates of the interobserver reliability of composite scores. The appropriate formulas for estimating the interobserver reliability of composite scores, as well as efficient means of estimating the elements in these formulas are presented. (Author)

Reliability generalization (RG) is a measurement meta-analytic method used to explore the variability in score reliability estimates and to characterize possible sources of the variance. Summarizes some RG considerations, and suggests how confidence intervals might be portrayed graphically. (SLD)

Used reliability generalization to study five versions of the Working Alliance Inventory (A. Horvath, 1981; WAI), analyzing 67 internal consistency estimates, 6 interrater reliability estimates, and 4 study characteristics. In general WAI scale scores appear to be robust. (SLD)

If two raters assign targets to categories, the ratings can be arranged in a two-dimensional contingency table. This article presents a model for the frequencies in such a contingency table for which Cohen's kappa is a parameter. Illustrates the model using data from a study of the psychobiology of depression. (Author/SLD)

Describes how the assumptions underlying the use of multiple regression are not satisfied in reliability generalization studies and introduces mixed effects modeling to overcome many shortcomings of traditional approaches. Provides an example using results from the Beck Depression Inventory. (SLD)

Average reliability of the coefficient is suggested as a multivariate extension of the classical reliability coefficient. It is shown that this coefficient is identical to the mean redundancy of the observed variables conditional on the true variables. (Author/SLD)

Examines psychometric reliability of Drug Use Screening Inventory (DUSI) utilizing adolescents with DSM-III-R diagnosis of Psychoactive Substance Use Disorder. Concludes that split-half, internal, and test-retest reliability is superior. Suggests that DUSI may be useful for identifying and quantifying substance use and related problems. Includes…

H. Kaiser (1992) has shown that the sum of coefficients alpha of a set of principal components does not change when the components are transformed by an orthogonal rotation. In this paper, the rotational invariance and the successive alpha-optimality are integrated and generalized in a simultaneous approach. (SLD)

Developed a graphical method of determining adequate sample size based on the power of L. Feldt's (1969) test of the difference between two values of Cronbach's alpha coefficient. Discusses assumptions on which this approach is based. (SLD)

Describes a study of judge leniency and consistency using a Rasch approach and involving 4,683 candidates and 53 judges. (SLD)