Accelerated life regression modelling of dependent bivariate time‐to‐event data

To analyze bivariate time‐to‐event data from matched or naturally paired study designs, researchers frequently use a random effect called frailty to model the dependence between within‐pair response measurements. The authors propose a computational framework for fitting dependent bivariate time‐to‐event data that combines frailty distributions and accelerated life regression models. In this framework users can choose from several parametric options for frailties, as well as the conditional distributions for within‐pair responses. The authors illustrate the flexibility that their framework represents using paired data from a study of laser photocoagulation therapy for retinopathy in diabetic patients.     

A bias analysis of Weibull models under heaped data

Retrospectively collected duration data are often reported incorrectly. An important type of such an error is heaping—respondents tend to round-off or round-up the data according to some rule of thumb. For two special cases of the Weibull model we study the behaviour of the ‘naive estimators’, which simply ignore the measurement error due to heaping, and derive closed expressions for the asymptotic bias. These results give a formal justification of empirical evidence and simulation-based findings reported in the literature. Additionally, situations where a remarkable bias has to be expected can be identified, and an exact bias correction can be performed.     

Markov random-field models for estimating local labour markets

Summary. This work is motivated by data on daily travel-to-work flows observed between pairs of elemental territorial units of an Italian region. The data were collected during the 1991 population census. The aim of the analysis is to partition the region into local labour markets. We present a new method for this which is inspired by the Bayesian texture segmentation approach. We introduce a novel Markov random-field model for the distribution of the variables that label the local labour markets for each territorial unit. Inference is performed by means of Markov chain Monte Carlo methods. The issue of model hyperparameter estimation is also addressed. We compare the results with those obtained by applying a classical method. The methodology can be applied with minor modifications to other data sets.     

Are there causal relationships between the dimensions of the Maslach Burnout Inventory? A review and two longitudinal tests

Arising from interest concerning the possibility of causal relationships among the three components of the Maslach Burnout Inventory, several process models have been proposed for the development of burnout. The present paper first reviews the evidence in favour of the three most influential of these (Leiter and Maslach's model (1988); Golembiewski, Boudreau, Munzenrider, & Luo's (1996) phase model; and Lee and Ashforth's model (1993)). These three models, and our own model (which integrates of two of them, and includes feedback effects of depersonalization on emotional exhaustion) are then compared with each other using structural equation modelling, drawing on longitudinal data from two Dutch samples (total N=1185). The review revealed that none of the seven previous studies on this issue provided any convincing support for any particular causal order proposed so far. In contrast, our own study showed that high levels of exhaustion were associated with high levels of depersonalization over time across both samples. Further, higher levels of depersonalization led to higher levels of emotional exhaustion and lower levels of personal accomplishment. To our knowledge, the present research is the first to provide reliable longitudinal evidence for the conceptualization of burnout as a developmental process, although the effects are not large enough to be of practical use in the recognition of burnout.     

Survival Analysis For A Sequenced Clinical Trial With Covariates

In this paper we discuss the survival analysis for a clinical trial in which treatment categories and general prognostic data are realised at different stages during a patient's survival time. In the light of possible strategies for the parsimonious modelling of such data, a corresponding sequence of illustrative analyses is presented. Detailed results are given for a weighted least squares analysis and these generally agree with those obtained by maximum likelihood.     

Adaptive Gaussian Markov random fields with applications in human brain mapping

Summary.  Functional magnetic resonance imaging has become a standard technology in human brain mapping. Analyses of the massive spatiotemporal functional magnetic resonance imaging data sets often focus on parametric or non-parametric modelling of the temporal component, whereas spatial smoothing is based on Gaussian kernels or random fields. A weakness of Gaussian spatial smoothing is underestimation of activation peaks or blurring of high curvature transitions between activated and non-activated regions of the brain. To improve spatial adaptivity, we introduce a class of inhomogeneous Markov random fields with stochastic interaction weights in a space-varying coefficient model. For given weights, the random field is conditionally Gaussian, but marginally it is non-Gaussian. Fully Bayesian inference, including estimation of weights and variance parameters, can be carried out through efficient Markov chain Monte Carlo simulation. Although motivated by the analysis of functional magnetic resonance imaging data, the methodological development is general and can also be used for spatial smoothing and regression analysis of areal data on irregular lattices. An application to stylized artificial data and to real functional magnetic resonance imaging data from a visual stimulation experiment demonstrates the performance of our approach in comparison with Gaussian and robustified non-Gaussian Markov random-field models.     

Testing the demand-control-support model among health-care professionals: A structural equation model

This paper reports a further empirical validation of the Demand-Control-Support Model (DCS model), which was developed by Johnson and colleagues (1988, 1989). Data were collected from a heterogeneous group of health-care professionals (nurses and nurses' aides; n = 249). Three major refinements were made to the validation of the DCS Model. First, all relationships in the model were estimated simultaneously by means of covariance structure modelling (LISREL 8). Second, the control dimension was refined substantially, using a psychometrically more sound assessment of the workers' autonomy. Third, the model was applied to the work of health-care professionals. The data did not confirm the assumption that both job strain and motivation are multiplicative functions of job demands, autonomy and social support. First, the results suggested that high levels of autonomy attenuate the increase of emotional exhaustion due to job demands. These results partially supported Karasek's Job Demand-Control Model (Karasek 1979). Second, high levels of social support proved to attenuate the increase of emotional exhaustion due to autonomy. Finally, the main effect of autonomy on job challenge implied that an increase in autonomy is accompanied by an increase in job challenge (and, consequently, job involvement). In addition, low job demands and a high amount of work-related support seem to reduce feelings of exhaustion and, consequently, health complaints.     

Simulating from irregular data: Kernel Carlo Simulation

In this paper, we address the problem of simulating from a data-generating process for which the observed data do not follow a regular probability distribution. One existing method for doing this is bootstrapping, but it is incapable of interpolating between observed data. For univariate or bivariate data, in which a mixture structure can easily be identified, we could instead simulate from a Gaussian mixture model. In general, though, we would have the problem of identifying and estimating the mixture model. Instead of these, we introduce a non-parametric method for simulating datasets like this: Kernel Carlo Simulation. Our algorithm begins by using kernel density estimation to build a target probability distribution. Then, an envelope function that is guaranteed to be higher than the target distribution is created. We then use simple accept–reject sampling. Our approach is more flexible than others, can simulate intelligently across gaps in the data, and requires no subjective modelling decisions. With several univariate and multivariate examples, we show that our method returns simulated datasets that, compared with the observed data, retain the covariance structures and have distributional characteristics that are remarkably similar.     

Semiparametric score test for varying copula parameter in Markov time series

This article examines a semiparametric test for checking the constancy of serial dependence via copula models for Markov time series. A semiparametric score test is proposed for testing the constancy of the copula parameter against stochastically varying copula parameter. The asymptotic null distribution of the test is established. A semiparametric bootstrap procedure is employed for the estimation of the variance of the proposed score test. Illustrations are given based on simulated series and historic interest rate data.     

Modelling the travel time of transit vehicles in real‐time through a GTFS‐based road network using GPS vehicle locations

Predicting the arrival time of a transit vehicle involves not only knowledge of its current position and schedule adherence, but also traffic conditions along the remainder of the route. Road networks are dynamic and can quickly change from free‐flowing to highly congested, which impacts the arrival time of transit vehicles, particularly buses which often share the road with other vehicles, so reliable predictions need to account for real‐time and future traffic conditions. The first step in this process is to construct a framework with which road state (traffic conditions) can be estimated using real‐time transit vehicle position data. Our proposed framework implements a vehicle model using a particle filter to estimate road travel times, which are used in a second model to estimate real‐time traffic conditions. Although development and testing took place in Auckland, New Zealand, we generalised each component to make the framework compatible with other public transport systems around the world. We demonstrate the real‐time feasibility and performance of our approach in real‐time, where a combination of R and C++ was used to obtain the necessary performance results. Future work will use these estimated traffic conditions in combination with historical data to obtain reliable arrival time predictions of transit vehicles.