Proposed a maximum likelihood procedure for combining the standardized mean differences based on a noncentral-t-distribution and developed an EM algorithm. Simulation results favor the proposed procedure over the existing normal theory maximum likelihood procedure and the commonly used generalized least squares procedure. (SLD)

Using simulation, studied the influence of each of 12 imputation methods and 2 methods using the EM algorithm on the results of maximum likelihood factor analysis as compared with results from the complete data factor analysis (no missing scores). Discusses why EM methods recovered complete data factor loadings better than imputation methods. (SLD)

This article introduces a bootstrap generalization to the Modified Parallel Analysis (MPA) method of test dimensionality assessment using factor analysis. This methodology, based on the use of Marginal Maximum Likelihood nonlinear factor analysis, provides for the calculation of a test statistic based on a parametric bootstrap using the MPA…

In practical applications of item response theory (IRT), item parameters are usually estimated first from a calibration sample. After treating these estimates as fixed and known, ability parameters are then estimated. However, the statistical inferences based on the estimated abilities can be misleading if the uncertainty of the item parameter…

An approximation to the likelihood for the generalized linear models with random coefficients is derived and is the basis for an approximate Fisher scoring algorithm. The method is illustrated on the logistic regression model for one-way classification, but it has an extension to the class of generalized linear models and to more complex data…

A maximum likelihood method of estimating the parameters of the multiple factor model when data are missing from the sample is presented. A Monte Carlo study compares the method with five heuristic methods of dealing with the problem. The present method shows some advantage in accuracy of estimation. (Author/CTM)

Conventional wisdom in missing data research dictates adding variables to the missing data model when those variables are predictive of (a) missingness and (b) the variables containing missingness. However, it has recently been shown that adding variables that are correlated with variables containing missingness, whether or not they are related to…

The standard error of the maximum likelihood ability estimator is commonly estimated by evaluating the test information function at an examinee's current maximum likelihood estimate (a point estimate) of ability. Because the test information function evaluated at the point estimate may differ from the test information function evaluated at an…

Reexamined the nature of structural equation modeling (SEM) estimates of autoregressive moving average (ARMA) models, replicated the simulation experiments of P. Molenaar, and examined the behavior of the log-likelihood ratio test. Simulation studies indicate that estimates of ARMA parameters observed with SEM software are identical to those…

Investigated the data demands associated with the marginal maximum likelihood (MML) expected a posterior (EAP) methodology and the precision of the resulting parameter estimates when data fit the underlying model through simulation. Also studied the extent to which a misspecified prior distribution would affect the item and person parameter…

Simulated items from two test forms using joint maximum likelihood estimation (JMLE) and marginal maximum likelihood estimation (MML) in the vertical equating situation (using an anchor test) when data were nonrandomly missing. Under MML, when the different ability parameters of students were not taken into account, the item difficulty parameters…

Proposes an estimation method for the Rasch model that is based on the pseudolikelihood theory of B. Arnold and D. Strauss (1988). Simulation results show great similarity between estimates from this method with those from conditional maximum likelihood and unconditional maximum likelihood estimates for the item parameters of the Rasch model. (SLD)

Used Monte Carlo simulation to examine the performance of four missing data methods in structural equation models: (1)full information maximum likelihood (FIML); (2) listwise deletion; (3) pairwise deletion; and (4) similar response pattern imputation. Results show that FIML estimation is superior across all conditions of the design. (SLD)

The generalized graded unfolding model (GGUM) (J. Roberts, J. Donoghue, and J. Laughlin, 1998) is an item response theory model designed to analyze binary or graded responses that are based on a proximity relation. The purpose of this study was to assess conditions under which item parameter estimation accuracy increases or decreases, with special…

Developed two methods for estimating the individual level correlation coefficient that combines information from aggregate data with a small fraction of the individual level data. Results of a simulation study support the use of these methods. (SLD)