Search

The asymptotic posterior normality (APN) of the latent variable vector in an item response theory (IRT) model is a crucial argument in IRT modeling approaches. In case of a single latent trait and under general assumptions, Chang and Stout (Psychometrika, 58(1):37–52, 1993) proved the APN for a broad class of latent trait models for binary items. Under the same setup, they also showed the consistency of the latent trait’s maximum likelihood estimator (MLE). Since then, several modeling approaches have been developed that consider multivariate latent traits and assume their APN, a conjecture which has not been proved so far. We fill this theoretical gap by extending the results of Chang and Stout for multivariate latent traits. Further, we discuss the existence and consistency of MLEs, maximum a-posteriori and expected a-posteriori estimators for the latent traits under the same broad class of latent trait models.

The paper introduces a new class of row-column (RC) association models for contingency tables by allowing the user to select both the scale on which interactions are measured as in Kateri and Papaioannou (1994) and the type of logit (local, global, continuation) suitable for the row and column classification variables as in Bartolucci and Forcina (2002). These choices determine the matrix of interactions which is subjected to rank constraints. Examples are provided where the new models fit substantially better than traditional ones; an intuitive explanation for this behavior is outlined and supported by numerical investigations. An extension of the optimality property in Kateri and Papaioannou (1994) is derived, leading to a general representation theorem and reconstruction formulas for the joint probabilities. These results are the key to show that, given marginal logits, the generalized interactions introduced in this paper determine uniquely the bivariate distribution; in addition, for each pair of logit types, we establish which kind of positive association is implied by our extended interactions being non negative. Quick model selection within this wide class can be performed by an efficient algorithm for computing maximum likelihood estimates which allows for additional linear constraints both on marginal logits and interactions.

Summary We consider simple ordinal model-based probability effect measures for comparing distributions of two groups, adjusted for explanatory variables. An “ordinal superiority” measure summarizes the probability that an observation from one distribution falls above an independent observation from the other distribution, adjusted for explanatory variables in a model. The measure applies directly to normal linear models and to a normal latent variable model for ordinal response variables. It equals Φ(β/2) for the corresponding ordinal model that applies a probit link function to cumulative multinomial probabilities, for standard normal cdf Φ and effect β that is the coefficient of the group indicator variable. For the more general latent variable model for ordinal responses that corresponds to a linear model with other possible error distributions and corresponding link functions for cumulative multinomial probabilities, the ordinal superiority measure equals exp(β)/[1+exp(β)] with the log–log link and equals approximately exp(β/2)/[1+exp(β/2)] with the logit link, where β is the group effect. Another ordinal superiority measure generalizes the difference of proportions from binary to ordinal responses. We also present related measures directly for ordinal models for the observed response that need not assume corresponding latent response models. We present confidence intervals for the measures and illustrate with an example.

A step-stress accelerated life testing model is constructed that deals with type-I censored experiments for which a continuous monitoring of the tested items is infeasible and only their inspection at particular time points is possible, producing thus grouped data. A general scale family of distributions is considered for the underlying lifetimes, which allows for flexible modeling by permitting different lifetime distributions for different stress levels. The maximum likelihood estimators of its parameters and their density functions are derived explicitly only when the inspection points coincide with the points of stress-level change. In case of additional inspection points, the estimates are obtained numerically. Asymptotic, exact (whenever possible) and bootstrap confidence intervals (CIs) are considered. For the bootstrap CIs a smoothing-modification is introduced, accounting for the categorical nature of the data.

A multi-sample model for general step-stress experiments based on sequential order statistics is proposed and analyzed, where the numbers of observations under each stress level are pre-specified. Contrary to common step-stress approaches, an arbitrary absolutely continuous lifetime distribution can be chosen, and, due to the experimental design of repeated Type-II censoring, the existence of maximum likelihood estimators for the parameters of interest is always guaranteed. Estimation and testing of the model parameters associated with the stress levels is examined, and inference under a log-linear link function is considered. Simulation studies illustrate some of the statistical procedures.

In step-stress modeling with the cumulative exposure model it is well known that, for underlying exponential distributions, explicit expressions can be obtained for maximum likelihood estimators of parameters as well as for their conditional density functions or conditional moment generation functions, given their existence. Applying a failure rate based approach instead, similar results can be also obtained for underlying lifetime distributions out of a general scale family of distributions, which allows for a flexible modeling. Exemplarily, respective results are presented for Type-I and Type-II censored experiments.

In simple step-stress experiments under Type-II censoring with the cumulative exposure model and exponentially distributed lifetimes, maximum likelihood estimates (MLE) of the expected lifetimes may not exist due to the absence of failure times either before or after the stress change point. For this reason, when planning a step-stress experiment, the change point could be chosen so as to minimize the probability of non-existence of the MLE. These non-existence probabilities are examined and compared in the one- as well as the two-sample situations. Moreover, the optimal allocations of the change points are discussed and the effects of the use of non-optimal choices for the change points are assessed.

In the model of sequential order statistics, prior distributions are considered for the model parameters, which, for example, describe increasing load put on remaining components. Gamma priors are examined as well as priors out of a class of extended truncated Erlang distributions (ETED), which is introduced along with some properties. The choice of independent priors in both set-ups leads to respective independent, conjugate posterior distributions for the model parameters of sequential order statistics. Since, in practical applications, the model parameters will often be increasingly ordered, a multivariate prior is applied being the joint distribution of common ETED-order statistics. Whatever baseline distribution of the sequential order statistics is chosen, the joint posterior distribution turns out to be a Weinman multivariate exponential distribution. Posterior moments are given explicitly, and HPD credible sets for the model parameters are stated. Journal of Multivariate Analysis

The step-stress model is a special case of accelerated life testing that allows for testing of units under different levels of stress with changes occurring at various intermediate stages of the experiment. Interest then lies on inference for the mean lifetime at each stress level. All step-stress models discussed so far in the literature are based on a single experiment. For the situation when data have been collected from different experiments wherein all the test units had been exposed to the same levels of stress but with possibly different points of change of stress, we introduce a model that combines the different experiments and facilitates a meta-analysis for the estimation of the mean lifetimes. We then discuss in detail the likelihood inference for the case of simple step-stress experiments under exponentially distributed lifetimes with Type-II censoring.

Square contingency tables with the same row and column classification occur frequently in a wide range of statistical applications, e.g. whenever the members of a matched pair are classified on the same scale, which is usually ordinal. Such tables are analysed by choosing an appropriate loglinear model. We focus on the models of symmetry, triangular, diagonal and ordinal quasi symmetry. The fit of a specific model is tested by the chi-squared test or the likelihood-ratio test, where p-values are calculated from the asymptotic chi-square distribution of the test statistic or, if this seems unjustified, from the exact conditional distribution. Since the calculation of exact p-values is often not feasible, we propose alternatives based on algebraic statistics combined with MCMC methods.

Association models constitute an attractive alternative to the usual log-linear models for modeling the dependence between classification variables. They impose special structure on the underlying association by assigning scores on the levels of each classification variable, which can be fixed or parametric. Under the general row-column (RC) association model, both row and column scores are unknown parameters without any restriction concerning their ordinality. However, when the classification variables are ordinal, order restrictions on the scores arise naturally. Under such restrictions, we adopt an alternative parameterization and draw inferences about the equality of adjacent scores using the Bayesian approach. To achieve that, we have constructed a reversible jump Markov chain Monte Carlo algorithm for moving across models of different dimension and estimate accurately the posterior model probabilities which can be used either for model comparison or for model averaging. The proposed methodology is evaluated through a simulation study and illustrated using actual datasets.

The likelihood ratio is used for measuring the strength of statistical evidence. The probability of observing strong misleading evidence along with that of observing weak evidence evaluate the performance of this measure. When the corresponding likelihood function is expressed in terms of a parametric statistical model that fails, the likelihood ratio retains its evidential value if the likelihood function is robust [Royall, R., Tsou, T.S., 2003. Interpreting statistical evidence by using imperfect models: robust adjusted likelihood functions. J. Roy. Statist. Soc. Ser. B 65, 391 404]. In this paper, we extend the theory of Royall and Tsou [2003. Interpreting statistical evidence by using imperfect models: robust adjusted likelihood functions. J. Roy. Statist. Soc., Ser. B 65, 391 404] to the case when the assumed working model is a characteristic model for two-way contingency tables (the model of independence, association and correlation models).We observe that association and correlation models are not equivalent in terms of statistical evidence. The association models are bounded by the maximum of the bump function while the correlation models are not. Journal of Statistical Planning and Inference

Describing the structure in a two-way contingency table in terms of an RC(m) association model, we are concerned with the computation of posterior distributions of the model parameters using prior distributions which take into account the nonlinear restrictions of the model. We are further involved with the determination of the order of association m, based on Bayesian arguments. Using projection methods, a prior distribution over the parameters of the simpler RC(m) model is induced from a prior of the parameters of the saturated model. The fit of the assumed RC(m) model is evaluated using the posterior distribution of its distance from the full model. Our methods are illustrated with a popular dataset.

The generalized linear model (GLM) is reviewed and the log-linear models are integrated in this family. For GLMs, maximum likelihood estimation, model fit, and model selection are discussed. In the GLM framework the analysis of incomplete tables is more straightforward. The quasi-independence model is defined and illustrated in R. Furthermore, the family of generalized log-linear models (GLLMs) is briefly presented and a GLLM is illustrated with a representative example in R.

We present a one-parameter family of models for square contingency tables that interpolates between the classical quasisymmetry model and its Pearsonian analogue. Algebraically, this corresponds to deformations of toric ideals associated with graphs. Our discussion of the statistical issues centers around maximum likelihood estimation., Comment: 17 pages, 10 figures

For the quasi-symmetry (QS) model, applicable to square contingency tables with commensurable classification variables. it is proved that under certain conditions, it is the closest model to symmetry in terms of the KUllback-Leibler distance. Replacing the Kullback-Leibler distance by f-divergence we introduce a generalized quasi-symmetry model, the QS[f], and develop interpretational aspects for its parameters. QS is a special case of QS[f], whereas the most characteristic of the newly introduced QS-type models is the Pearsonian QS model. We compute maximum likelihood estimates of the parameters of the Pearsonian QS model and compare it, through examples and simulation studies, to the classical QS model in terms of goodness of fit and of the powers of the tests for marginal homogeneity conditional on the QS and Pearsonian PS models. Journal of the American Statistical Association

In this article, we propose a generalized estimating equations (GEE) approach for correlated ordinal or nominal multinomial responses using a local odds ratios parameterization. Our motivation lies upon observing that: (i) modeling the dependence between correlated multinomial responses via the local odds ratios is meaningful both for ordinal and nominal response scales and (ii) ordinary GEE methods might not ensure the joint existence of the estimates of the marginal regression parameters and of the dependence structure. To avoid (ii), we treat the so-called "working" association vector α as a "nuisance" parameter vector that defines the local odds ratios structure at the marginalized contingency tables after tabulating the responses without a covariate adjustment at each time pair. To estimate α and simultaneously approximate adequately possible underlying dependence structures, we employ the family of association models proposed by Goodman. In simulations, the parameter estimators with the proposed GEE method for a marginal cumulative probit model appear to be less biased and more efficient than those with the independence "working" model, especially for studies having time-varying covariates and strong correlation.

We generalize the classical conditional or triangular symmetry model for I x I contingency tables to three-way I x I x I tables with commensurable ordinal classification variables. The construction of the new family of models is such that the desirable property that connects conditional symmetry to complete symmetry and marginal homogeneity models in two-way tables is retained in three-way tables. Furthermore, connections between our proposed models obey a coherent structure. We provide maximum likelihood estimation for the new models which is illustrated with a real data example. (C) 2012 Elsevier B.V. All rights reserved. Journal of Statistical Planning and Inference

Responses of two groups, measured on the same ordinal scale, are compared through the column effect association model, applied on the corresponding 2 x J contingency table. Monotonic or umbrella shaped ordering for the scores of the model are related to stochastic or umbrella ordering of the underlying response distributions, respectively. An algorithm for testing all possible hypotheses of stochastic ordering and deciding on an appropriate one is proposed. Communications in Statistics-Theory and Methods

Bayesian analysis of correlated binary data when individual information is not available is considered. In particular, a binary outcome is measured on the same subjects of two independent groups at two separate occasions (usually time points). The groups are formulated through a binary exposure or a prognostic factor. Interest lies in estimating the association between exposure and outcome over time. Standard methods for this purpose apply on the individual item responses and are insufficient in case these are missing. Moreover it is assumed that the only available information is the marginal 2x2 cross-tabulations between the grouping variable and the response for each occasion. Assuming independent binomial distributions for the two groups, the success probabilities for each occasion as well as the associations between exposure and outcome, based on the corresponding odds ratios, are estimated. In order to deal with the missing information of each item's response and to estimate the corresponding transition probabilities, a Bayesian procedure is adopted. Computational Statistics & Data Analysis