This paper explores the use of K. G. Joreskog's (1970) congeneric modeling approach to reliability using censored quantitative variables. Two Monte Carlo studies were conducted. The first explored the robustness of Normal Theory Generalized Least-Squares (NTGLS) estimates for a single-factor congeneric model across several sample sizes…

The purpose of this study is to determine (1) whether the previously observed vulnerability of WISC subtests to tester effects appeared under ordinary testing conditions, and (2) which subtests were most susceptible to these effects. Results support the presence of both general and differential vulnerability of subtests. (Author)

This study was designed to: (1) assess the influence of an instructional set given to an observer regarding the presence or absence of a predictable pattern in the observed interaction, (2) extend and replicate findings of a previous study of observer accuracy, and (3) identify some of the specific types of errors made by observers in coding…

Observer reliability and number of behaviors rated were compared following two types of observer training: (1) training of observers by one individual, and (2) self-training, or each pair of observers training itself. Results suggest that more adequate standards for the training of observers and the reporting of observer reliability be adopted by…

For classical, norm-referenced test reliability, Cronbach's alpha has been shown to be equal to the mean of all possible split-half Pearson product-moment correlation coefficients, adjusted by the Spearman-Brown prophecy formula. For criterion-referenced test reliability, in an analogous vein, this paper provides the rationale behind, the analysis…

Three successively simpler formulas for approximating the standard error of measurement were derived by applying successively more simplifying assumptions to the standard formula based on the standard deviation and the Kuder-Richardson formula 20 estimate of reliability. The accuracy of each of these three formulas, with respect to the standard…

The accuracy of the large sample standard error of weighted kappa appropriate to the non-null case was studied by computer simulation for the hypothesis that two independently derived estimates of weighted kappa are equal, and for setting confidence limits around a single value of weighted kappa. (Author/CTM)

Two designs for comparing a judge's ratings with a known standard are compared: (1) Subjects are categorized into classes, with no knowledge of the size of either group; and (2) the judge is told the actual number allowed in each class. The probability distribution of the total number of correct choices is developed for each case, and their power…

A formula is derived for ascertaining factor scores for the factor analytic method: Little Jiffy, Mark IV. This formula is then employed to derive a second formula giving an exact determination of the generalized Kuder-Richardson estimate of the reliability of scores on a Little Jiffy factor. (Author/JKS)

A comparison of three multivariate models (canonical reliability model, maximum generalizability model, canonical correlation model) for estimating test battery reliability indicated that the maximum generalizability model showed the least degree of bias, smallest errors in estimation, and the greatest relative efficiency across all experimental…

Current methods for obtaining reliability indices for mastery tests can be laborious. This paper offers practitioners tables from which agreement and kappa coefficients can be read directly and provides criterion for acceptable values of agreement and kappa coefficients. (TJH)