The polytomous unidimensional Rasch model with equidistant scoring (rating scale model) is extended so that two parameters are linearly decomposed into certain basic parameters. A conditional maximum likelihood estimation procedure and a likelihood ratio test are presented in the context of the extended model (linear rating scale model). (SLD)

A loglinear item response theory (IRT) model is proposed that relates polytomously scored item responses to a multidimensional latent space. The analyst may specify a response function for each response, and each item may have a different number of response categories. Conditional maximum likelihood estimates are derived. (SLD)

The statistically correct conditional maximum likelihood (CML) estimation method has not been used because of numerical problems. A solution is presented which allows a rapid computation of the CML esitmates also for long tests. CML has decisive advantages in the construction of statistical tests of goodness of fit. (Author/CP)

The problem of estimating item parameters of latent trait models in a multistage testing design is considered. Using the Rasch model and conditional maximum likelihood estimates does not lead to solvable estimation equations, but the use of marginal maximum likelihood estimation leads to solvable equations for both Rasch and Birnbaum models. (TJH)

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)

Testing the fit of the Rasch model is examined. Tests proposed are based on the comparison of expected and observed frequencies. Conditional maximum likelihood estimates (MLEs) and marginal MLEs are compared. A statistical testing procedure is proposed that is a diagnostic tool for identifying violations of the Rasch model. (SLD)

An extension to the partial credit model, the linear partial credit model, is considered under the assumption of a certain linear decomposition of the item x category parameters into basic parameters. A conditional maximum likelihood algorithm for estimating basic parameters is presented and illustrated with simulation and an empirical study. (SLD)

Combining Rasch and latent class models is presented as a way to overcome deficiencies and retain the positive features of both. An estimation algorithm is outlined, providing conditional maximum likelihood estimates of item parameters for each class. The model is illustrated with simulated data and real data (n=869 adults). (SLD)

Examined the accuracy of the Gibbs sampling Markov chain Monte Carlo procedure for estimating item and person (theta) parameters in the one-parameter logistic model. Analyzed four empirical datasets using the Gibbs sampling, conditional maximum likelihood, marginal maximum likelihood, and joint maximum likelihood methods. Discusses the conditions…

The unconditional--simultaneous--maximum likelihood (UML) estimation procedure for the one-parameter logistic model produces biased estimators. The UML method is inconsistent and is not a good alternative to conditional maximum likelihood method, at least with small numbers of items. The minimum Chi-square estimation procedure produces unbiased…

We consider two-level models where a continuous response R and continuous covariates C are assumed missing at random. Inferences based on maximum likelihood or Bayes are routinely made by estimating their joint normal distribution from observed data R[subscript obs] and C[subscript obs]. However, if the model for R given C includes random…

Rasch models are characterised by sufficient statistics for all parameters. In the Rasch unidimensional model for two ordered categories, the parameterisation of the person and item is symmetrical and it is readily established that the total scores of a person and item are sufficient statistics for their respective parameters. In contrast, in the…

Noncompliance to treatment assignment is widespread in randomized trials and presents challenges in causal inference. In the presence of noncompliance, the most commonly estimated effect of treatment assignment, also known as the intent-to-treat (ITT) effect, is biased. Of interest in this setting is the complier average causal effect (CACE), the…

The accuracy of the Markov chain Monte Carlo procedure, Gibbs sampling, was considered for estimation of item and ability parameters of the one-parameter logistic model. Four data sets were analyzed to evaluate the Gibbs sampling procedure. Data sets were also analyzed using methods of conditional maximum likelihood, marginal maximum likelihood,…

Necessary and sufficient conditions for the existence and uniqueness of a solution of the so-called "unconditional" and the "conditional" maximum-likelihood estimation equations in the dichotomous Rasch model are given. It is shown how to apply the results in practical uses of the Rasch model. (Author/JKS)