Missing data are a common practical problem for longitudinal designs. Time-series analysis is a longitudinal method that involves a large number of observations on a single unit. Four different missing-data methods (deletion, mean substitution, mean of adjacent observations, and maximum likelihood estimation) were evaluated. Computer-generated…

The similarity of item and ability parameter estimations was investigated using two numerical analysis techniques via marginal maximum likelihood estimation (MMLE) with a large simulated data set (n=1,000 examinees) and changing the number of quadrature points. MMLE estimation uses a numerical analysis technique to integrate examinees' abilities…

A partial least squares method is described for estimating parameters of linear structural relation models. This method is an extension of Herman Wold's proposal for estimation parameters without distributional assumptions, using some algorithms worked out by Paul Horst. The method (LISPLS) determines a different number of latent variables from…

In maximum likelihood factor analysis, there arises a situation whereby improper solutions occur. The causes of those improper solution are discussed and illustrated. (JKS)

Some extensions of the partial credit model are presented. A marginal maximum likelihood estimation procedure is developed to allow for incomplete data and linear restrictions on the item and population parameters. Two statistical tests for evaluating model fit are also presented. (SLD)

The present study expands the design of Warschauer (1996) surveying freshman foreign language students at a Turkish university. Motivating aspects of computer assisted instruction in terms of writing and e-mailing are explored through an exploratory factor analysis conducted on the survey developed by Warschauer (1996). Findings suggest that…

The linear logistic model with relaxed assumption is extended to designs with any number of time points or with different sets of items presented on different occasions, provided that one unidimensional subscale is available per latent trait. A sample application is presented. (SLD)

A modification index is presented to aid in reformulating hypothetical models rejected after analysis of empirical data. This index is an improvement over the one in the LISREL V computer program in that it takes into account changes in all parameters of the model when one parameter is freed. (SLD)

Functional and structural relationship alternatives to the standard "F"-test for analysis of covariance (ANCOVA) are discussed for cases when the covariate is measured with error. An approximate statistical test based on the functional relationship approach is preferred on the basis of Monte Carlo simulation results. (SLD)

For comparing nested covariance structure models, the standard procedure is the likelihood ratio test of the difference in fit, where the null hypothesis is that the models fit identically in the population. A procedure for determining statistical power of this test is presented where effect size is based on a specified difference in overall fit…

Simultaneous estimation of many parameters can often be improved, sometimes dramatically so, if it is reasonable to consider one or more subsets of parameters as exchangeable members of corresponding populations. While each observation may provide limited information about the parameters it is modeled directly in terms of, it also contributes…

The predictive analysis approach to adaptive testing originated in the idea of statistical predictive analysis suggested by J. Aitchison and I.R. Dunsmore (1975). The adaptive testing model proposed is based on parameter-free predictive distribution. Aitchison and Dunsmore define statistical prediction analysis as the use of data obtained from an…

The effect of prior information in Bayesian estimation is considered, mainly from the standpoint of objective testing. In the estimation of a parameter belonging to an individual, the prior information is, in most cases, the density function of the population to which the individual belongs. Bayesian estimation was compared with maximum likelihood…

This article describes a Monte Carlo study of 2 methods for treating incomplete nonnormal data. Skewed, kurtotic data sets conforming to a single structured model, but varying in sample size, percentage of data missing, and missing-data mechanism, were produced. An asymptotically distribution-free available-case (ADFAC) method and structured-model…

The development of new statistical methods of log-linear models by Leo Goodman has led to major advances in the statistical analysis of categorical data. Goodman's logit analysis, (the simplest form of log-linear models) can be applied in evaluation studies to estimate the "main effects" and "interaction effects" of categorical…