Decomposable models, which allow for the interdependence of structure among observable variables, are described. Each model is fully characterized by a set of conditional interdependence restrictions and can be visualized with an undirected as well as a special type of directed graph. (Author/JKS)

Conceptual rationales from five research contexts are presented, which all posit unmeasured variables that interact with observed independent variables to produce a complete model of the dependent variable. Strategies for overcoming related difficulties are outlined, including increased longitudinal assessment, oversampling of levels of…

A model was developed for estimating individuals' true changes on test scores using the additional information provided by an auxiliary variable correlated with a trait being measured. This provides an improved and more precise estimate when compared with simple difference scores. It involves the reliability of "X" scores only. (SLD)

This Monte Carlo study explored relationships among standard and unstandardized regression coefficients, structural coefficients, multiple R_ squared, and significance level of predictors for a variety of linear regression scenarios. Ten regression models with three predictors were included, and four conditions were varied that were expected to…

Discusses the interdependencies between parameter estimation and properties of probabilistic models, such as dependency assumptions, binary vs. nonbinary features, and estimation sample selection. An optimum estimation for binary features applicable to information retrieval is defined, a method for computing this estimation using empirical data is…

Argues that a major difficulty in using Lotka's law in information science arises from the misuse of goodness of fit tests in parameter estimation. Three approaches for studying Lotka's law are presented: an index approach, a time series approach, and a generating mechanism incorporating these two influential variables to derive an equilibrium…

Using an infinite item test framework, it is argued that the usual assumption of local independence should be replaced by a weaker assumption--essential independence. The usual assumption of unidimensionality is replaced by a weaker and more appropriate statistically testable assumption of essential unidimensionality. (TJH)

Direct and indirect effects in decomposed zero-order correlations among variables in causal models are considered. Under certain circumstances, the components of the decompositions could be interpreted as direct, indirect, and spurious causal effects, plus a component called joint associations. The sum of the direct and indirect effects is the…

A method is suggested to estimate the relationship between a latent trait and one or more manifest predictors without estimating subject parameters. The method, developed for the Rasch model, can be generalized to two-parameter and three-parameter logistic latent trait models. The model is illustrated with simulated and empirical data. (SLD)

In research, data sets often occur in which the variance of the distribution of the dependent variable at given levels of the predictors is a function of the values of the predictors. In this situation, the use of weighted least-squares (WLS) or techniques is required. Weights suitable for use in a WLS regression analysis must be estimated. A…

This paper is part of a series of applied statistics monographs intended to provide supplementary reading for applied statistics students. In the present paper, derivations of the unbiased standard error of estimate for both the raw score and standard score linear models are presented. The derivations for raw score linear models are presented in…

Variance formulas of H. E. Daniels and M. G. Kendall (1947) are generalized to allow for the presence of ties and variance of the sample tau correlation. Applications of these generalized formulas are discussed and illustrated using data from a 1965 study of contraceptive use in 15 developing countries. (SLD)

An evaluation of the rules-of-thumb used to determine the minimum number of subjects required to conduct multiple regression analyses suggests that researchers who use a rule of thumb rather than power analyses trade simplicity of use for accuracy and specificity of response. Insufficient power is likely to result. (SLD)

Methods for predicting specific college course grades, based on small numbers of observations, were investigated. These methods use collateral information across potentially diverse institutions to obtain refined within-group parameter estimates. One method, referred to as pooled least squares with adjusted intercepts, assumes that slopes and…

This study examined the accuracy of predictions of college grades in English, mathematics, social studies, and natural science courses, and the accuracy of predictions of overall freshman grade point average (GPA) based on American College Testing Program (ACT) assessment test scores and on high school course work and grade information from the…