Elicitation of multivariate prior distributions: A nonparametric Bayesian approach

In the context of Bayesian statistical analysis, elicitation is the process of formulating a prior density f(·)$f(·)$ about one or more uncertain quantities to represent a person's knowledge and beliefs. Several different methods of eliciting prior distributions for one unknown parameter have been proposed. However, there are relatively few methods for specifying a multivariate prior distribution and most are just applicable to specific classes of problems and/or based on restrictive conditions, such as independence of variables. Besides, many of these procedures require the elicitation of variances and correlations, and sometimes elicitation of hyperparameters which are difficult for experts to specify in practice. Garthwaite et al. (2005) discuss the different methods proposed in the literature and the difficulties of eliciting multivariate prior distributions. We describe a flexible method of eliciting multivariate prior distributions applicable to a wide class of practical problems. Our approach does not assume a parametric form for the unknown prior density f(·)$f(·)$, instead we use nonparametric Bayesian inference, modelling f(·)$f(·)$ by a Gaussian process prior distribution. The expert is then asked to specify certain summaries of his/her distribution, such as the mean, mode, marginal quantiles and a small number of joint probabilities. The analyst receives that information, treating it as a data set D   with which to update his/her prior beliefs to obtain the posterior distribution for f(·)$f(·)$. Theoretical properties of joint and marginal priors are derived and numerical illustrations to demonstrate our approach are given.     

Improved likelihood-based inference for the stationary AR(2) model

An improved likelihood-based method based on Fraser et al. (1999) is proposed in this paper to test the significance of the second lag of the stationary AR(2) model. Compared with the test proposed by Fan and Yao (2003) and the signed log-likelihood ratio test, the proposed method has remarkable accuracy. Simulation studies are performed to illustrate the accuracy of the proposed method. Application of the proposed method on historical data is presented to demonstrate the implementation of this method. Furthermore, the method can be extended to the general AR(p) model.     

A frailty model for (interval) censored family survival data,applied to the age at onset of non-physical problems

Family survival data can be used to estimate the degree of genetic and environmental contributions to the age at onset of a disease or of a specific event in life. The data can be modeled with a correlated frailty model in which the frailty variable accounts for the degree of kinship within the family. The heritability (degree of heredity) of the age at a specific event in life (or the onset of a disease) is usually defined as the proportion of variance of the survival age that is associated with genetic effects. If the survival age is (interval) censored, heritability as usually defined cannot be estimated. Instead, it is defined as the proportion of variance of the frailty associated with genetic effects. In this paper we describe a correlated frailty model to estimate the heritability and the degree of environmental effects on the age at which individuals contact a social worker for the first time and to test whether there is a difference between the survival functions of this age for twins and non-twins.     

An alternate version of the conceptual predictive statistic based on a symmetrized discrepancy measure

The conceptual predictive statistic, C_p, is a widely used criterion for model selection in linear regression. C_p serves as an estimator of a discrepancy, a measure that reflects the disparity between the generating model and a fitted candidate model. This discrepancy, based on scaled squared error loss, is asymmetric: an alternate measure is obtained by reversing the roles of the two models in the definition of the measure. We propose a variant of the C_p statistic based on estimating a symmetrized version of the discrepancy targeted by C_p. We claim that the resulting criterion provides better protection against overfitting than C_p, since the symmetric discrepancy is more sensitive towards detecting overspecification than its asymmetric counterpart. We illustrate our claim by presenting simulation results. Finally, we demonstrate the practical utility of the new criterion by discussing a modeling application based on data collected in a cardiac rehabilitation program at University of Iowa Hospitals and Clinics.     

On the advantages of the non-concave penalized likelihood model selection method with minimum prediction errors in large-scale medical studies

Variable and model selection problems are fundamental to high-dimensional statistical modeling in diverse fields of sciences. Especially in health studies, many potential factors are usually introduced to determine an outcome variable. This paper deals with the problem of high-dimensional statistical modeling through the analysis of the trauma annual data in Greece for 2005. The data set is divided into the experiment and control sets and consists of 6334 observations and 112 factors that include demographic, transport and intrahospital data used to detect possible risk factors of death. In our study, different model selection techniques are applied to the experiment set and the notion of deviance is used on the control set to assess the fit of the overall selected model. The statistical methods employed in this work were the non-concave penalized likelihood methods, smoothly clipped absolute deviation, least absolute shrinkage and selection operator, and Hard, the generalized linear logistic regression, and the best subset variable selection.The way of identifying the significant variables in large medical data sets along with the performance and the pros and cons of the various statistical techniques used are discussed. The performed analysis reveals the distinct advantages of the non-concave penalized likelihood methods over the traditional model selection techniques.     

Bayesian discrete latent spatial modeling of crack initiation in orthopaedic hip replacement bone cement

In this paper, we propose a spatial model for the initiation of cracks in the bone cement of hip replacement specimens. The failure of hip replacements can be attributed mainly to damage accumulation, consisting of crack initiation and growth, occurring in the cement mantle that interlocks the hip prosthesis and the femur bone. Since crack initiation is an important factor in determining the lifetime of a replacement, the understanding of the reasons for crack initiation is vital in attempting to prolong the life of the hip replacement. The data consist of crack location coordinates from five laboratory experimental models, together with stress measurements. It is known that stress plays a major role in the initiation of cracks, and it is also known that other unmeasurable factors such as air bubbles (pores) in the cement mantle are also influential. We propose an identity-link spatial Poisson regression model for the counts of cracks in discrete regions of the cement, incorporating both the measured (stress), and through a latent process, any unmeasured factors (possibly pores) that may be influential. All analysis is carried out in a Bayesian framework, allowing for the inclusion of prior information obtained from engineers, and parameter estimation for the model is done via Markov chain Monte Carlo techniques.     

Correlational biases in mean response latency differences

Multifarious psychological constructs are indexed by the mean latency difference (MLD), the within-subject difference between mean response latency on two tasks. Two associations consistently emerge in mean latency data. Firstly, across subjects, mean latencies on distinct tasks are positively correlated. This correlation arises from individual differences in general rates of information processing that are a shared influence on response latency in diverse tasks. Secondly, across tasks, the mean and variance of mean latency are positively correlated. Compared to a simple task, a complex task has both a larger average mean latency and a larger variance of mean latency, across subjects. Taken together, these associations make the interpretation of the MLD problematic by biasing correlations between the MLD and (a) task mean latencies, (b) the average of the mean latencies, (c) external criteria, and (d) other MLDs. A variety of mean latency transformations were evaluated and, while they differed in their effectiveness, they did not satisfactorily rectify MLD biases. An alternative approach, focusing on scale invariant contrasts of within-subject response latency distributions, is introduced in the conclusion.