Local influence diagnostics for generalized linear mixed models with overdispersion

Since the seminal paper by Cook and Weisberg [9 R.D. Cook and S. Weisberg, Residuals and Influence in Regression, Chapman &; Hall, London, 1982. [Google Scholar]], local influence, next to case deletion, has gained popularity as a tool to detect influential subjects and measurements for a variety of statistical models. For the linear mixed model the approach leads to easily interpretable and computationally convenient expressions, not only highlighting influential subjects, but also which aspect of their profile leads to undue influence on the model's fit [17 E. Lesaffre and G. Verbeke, Local influence in linear mixed models, Biometrics 54 (1998), pp. 570–582. doi: 10.2307/3109764[Crossref], [PubMed], [Web of Science ®] , [Google Scholar]]. Ouwens et al. [24 M.J.N.M. Ouwens, F.E.S. Tan, and M.P.F. Berger, Local influence to detect influential data structures for generalized linear mixed models, Biometrics 57 (2001), pp. 1166–1172. doi: 10.1111/j.0006-341X.2001.01166.x[Crossref], [PubMed], [Web of Science ®] , [Google Scholar]] applied the method to the Poisson-normal generalized linear mixed model (GLMM). Given the model's nonlinear structure, these authors did not derive interpretable components but rather focused on a graphical depiction of influence. In this paper, we consider GLMMs for binary, count, and time-to-event data, with the additional feature of accommodating overdispersion whenever necessary. For each situation, three approaches are considered, based on: (1) purely numerical derivations; (2) using a closed-form expression of the marginal likelihood function; and (3) using an integral representation of this likelihood. Unlike when case deletion is used, this leads to interpretable components, allowing not only to identify influential subjects, but also to study the cause thereof. The methodology is illustrated in case studies that range over the three data types mentioned.     

基于图方法的MA和ARMA模型系数检验

应用图模型方法来讨论传统的MA和ARMA模型,证明了MA和ARMA模型的系数为去掉其他时间序列分量线性效应的条件下的偏相关系数,且利用图模型推断算法提出了一种新的参数估计和检验方法。     

An extended random-effects approach to modeling repeated, overdispersed count data

Non-Gaussian outcomes are often modeled using members of the so-called exponential family. The Poisson model for count data falls within this tradition. The family in general, and the Poisson model in particular, are at the same time convenient since mathematically elegant, but in need of extension since often somewhat restrictive. Two of the main rationales for existing extensions are (1) the occurrence of overdispersion, in the sense that the variability in the data is not adequately captured by the model's prescribed mean-variance link, and (2) the accommodation of data hierarchies owing to, for example, repeatedly measuring the outcome on the same subject, recording information from various members of the same family, etc. There is a variety of overdispersion models for count data, such as, for example, the negative-binomial model. Hierarchies are often accommodated through the inclusion of subject-specific, random effects. Though not always, one conventionally assumes such random effects to be normally distributed. While both of these issues may occur simultaneously, models accommodating them at once are less than common. This paper proposes a generalized linear model, accommodating overdispersion and clustering through two separate sets of random effects, of gamma and normal type, respectively. This is in line with the proposal by Booth et al. (Stat Model 3:179-181, 2003). The model extends both classical overdispersion models for count data (Breslow, Appl Stat 33:38-44, 1984), in particular the negative binomial model, as well as the generalized linear mixed model (Breslow and Clayton, J Am Stat Assoc 88:9-25, 1993). Apart from model formulation, we briefly discuss several estimation options, and then settle for maximum likelihood estimation with both fully analytic integration as well as hybrid between analytic and numerical integration. The latter is implemented in the SAS procedure NLMIXED. The methodology is applied to data from a study in epileptic seizures.     

Determining the number of factors with potentially strong within-block correlations in error terms

ABSTRACT

We develop methods to estimate the number of factors when error terms have potentially strong correlations in the cross-sectional dimension. The information criteria proposed by Bai and Ng (2002 Bai, J., Ng, S. (2002). Determining the number of factors in approximate factor models. Econometrica 70:191–221.[Crossref], [Web of Science ®] , [Google Scholar]) require the cross-sectional correlations between the error terms to be weak. Violation of this weak correlation assumption may lead to inconsistent estimates of the number of factors. We establish two data-dependent estimators that are consistent whether the error terms are weakly or strongly correlated in the cross-sectional dimension. To handle potentially strong cross-sectional correlations between the error terms, we use a block structure in which the within-block correlation may either be weak or strong, but the between-block correlation is limited. Our estimators allow imperfect knowledge and a moderate misspecification of the block structure. Monte-Carlo simulation results show that our estimators perform similarly to existing methods for cases in which the conventional weak correlation assumption is satisfied. When the error terms have a strong cross-sectional correlation, our estimators outperform the existing methods.     

A semiparametric extended hazard regression model with time-dependent covariates

We introduce a general class of semiparametric hazard regression models, called extended hazard (EH) models, that are designed to accommodate various survival schemes with time-dependent covariates. The EH model contains both the Cox model and the accelerated failure time (AFT) model as its subclasses so that we can use this nested structure to perform model selection between the Cox model and the AFT model. A class of estimating equations using counting process and martingale techniques is developed to estimate the regression parameters of the proposed model. The performance of the estimating procedure and the impact of model misspecification are assessed through simulation studies. Two data examples, Stanford heart transplant data and Mediterranean fruit flies, egg-laying data, are used to demonstrate the usefulness of the EH model.     

Statistical analysis of controlled calibration model with replicates

In this work, we generalize the controlled calibration model by assuming replication on both variables. Likelihood-based methodology is used to estimate the model parameters and the Fisher information matrix is used to construct confidence intervals for the unknown value of the regressor variable. Further, we study the local influence diagnostic method which is based on the conditional expectation of the complete-data log-likelihood function related to the EM algorithm. Some useful perturbation schemes are discussed. A simulation study is carried out to assess the effect of the measurement error on the estimation of the parameter of interest. This new approach is illustrated with a real data set.     

空间计量经济学理论体系的解析及其展望

随着空间经济理论研究出现重大突破,空间计量经济学也从边缘走向现代计量经济学的主流。重点对空间计量经济模型的设立包括空间横截面数据模型、空间面板数据模型以及空间离散数据模型进行讨论,对模型参数估计方法包括最大似然估计法、两阶段最小二乘法和矩估计法等进行分析,对模型检验方法包括Moran’s I方法和LM/RS方法等内容进行总结,最后展望了该理论研究未来的发展趋势。     

Comparison of naïve,Kenward–Roger,and parametric bootstrap interval approaches to small-sample inference in linear mixed models

In mixed models the mean square error (MSE) of empirical best linear unbiased estimators generally cannot be written in closed form. Unlike traditional methods of inference, parametric bootstrapping does not require approximation of this MSE or the test statistic distribution. Data were simulated to compare coverage rates for intervals based on the naïve MSE approximation and the method of Kenward and Roger, and parametric bootstrap intervals (Efron's percentile, Hall's percentile, bootstrap-t). The Kenward–Roger method performed best and the bootstrap-t almost as well. Intervals were also compared for a small set of real data. Implications for minimum sample size are discussed.     

计量经济学和统计学视角下的线性回归模型——再议线性回归模型的设定

从属性、构建方法及意义等方面,分析研究线性回归模型在计量经济学和统计学两学科视角下的差异,并根据这种差异进一步提出回归模型的基本设定思路。研究表明:识别这种差异是完成模型设定工作的基础性和必要性举措,有助于实现线性回归模型的正确设定。以经典例证对计量经济学和统计学回归模型在应用中的区别以及模型设定问题进行进一步展示和分析。     

Models Comparison for Step-Stress Accelerated Life Testing

In this article, four basic models for step-stress accelerated life testing are introduced and compared: cumulative exposure model (CEM), linear cumulative exposure model (LCEM), tampered random variable model (TRVM), and tampered failure rate model (TFRM). Limitations of the four models are also introduced for better use of the models.     

Estimation of rational transfer function models

In the estimation of rational transfer function models, it has been recommended that starting values of a transfer function component be assumed to be zero (or a constant) in the recursive computation of the transfer function response. It is demonstrated that such algorithms may lead to serious bias in the estimation of moving average parameters. This paper discusses several other algorithms that may rectify this problem. It is found that the starting-value-free (SVF) method is a more reliable algorithm. For computer programs using the traditional algorithm, i.e., the zero-starting-value (ZSV) method, the bias problem can be easily remedied using a short-cut method that omits appropriate number of values at the beginning of the residual series.     

An insight into technology diffusion of tractor through Weibull growth model

Most of the technological innovation diffusion follows an S-shaped curve. But, in many practical situations this may not hold true. To this end, Weibull model was proposed to capture the diffusion of new technological innovation, which does not follow any specific pattern. Nonlinear growth models play a very important role in getting an insight into the underlying mechanism. These models are generally ‘mechanistic’ as the parameters have meaningful interpretation. The nonlinear method of estimation of parameters of Weibull model fails to converge. Taking this problem into consideration, we propose the use of a powerful technique of genetic algorithm for parameter estimation. The methodology is also validated by simulation study to check whether parameter estimates are closer to the real value. For illustration purpose, we model the tractor density time-series data of India as a whole and some major states of India. It is seen that fitted Weibull model is able to capture the technology diffusion process in a reasonable manner. Further, comparison is also made with Logistic and Gompertz model; and is found to perform better for the data sets under consideration.     

A Simplified Treatment of the Estimation of Parameters and Tests of Hypotheses in Constrained Design Models with Unbalanced Data

A simple approach to the least squares estimation of parameters in the linear design model is given. An algebraically tractable example for unbalanced data is used to illustrate the approach.     

Bayes estimates in multivariate semiparametric linear models

Bayes estimates are derived in multivariate linear models with unknown distribution. The prior distribution is defined using a Dirichlet prior for the unknown error distribution and a normal-Wishart distribution for the parameters. The posterior distribution is determined and explicit expressions are given in the special cases of location-scale and two-sample models. The calculation of self-informative limits of Bayes estimates yields standard estimates.     

A refined parameter estimating approach for HIV dynamic model

HIV dynamic models, a set of ordinary differential equations (ODEs), have provided new understanding of the pathogenesis of HIV infection and the treatment effects of antiviral therapies. However, to estimate parameters for ODEs is very challenging due to the complexity of this nonlinear system. In this article, we propose a comprehensive procedure to deal with this issue. In the proposed procedure, a series of cutting-edge statistical methods and techniques are employed, including nonparametric mixed-effects smoothing-based methods for ODE models and stochastic approximation expectation–maximization (EM) approach for mixed-effects ODE models. A simulation study is performed to validate the proposed approach. An application example from a real HIV clinical trial study is used to illustrate the usefulness of the proposed method.     

PARTIALLY LINEAR MODEL SELECTION BY THE BOOTSTRAP

We propose a new approach to the selection of partially linear models based on the conditional expected prediction square loss function, which is estimated using the bootstrap. Because of the different speeds of convergence of the linear and the nonlinear parts, a key idea is to select each part separately. In the first step, we select the nonlinear components using an ' m -out-of- n ' residual bootstrap that ensures good properties for the nonparametric bootstrap estimator. The second step selects the linear components from the remaining explanatory variables, and the non-zero parameters are selected based on a two-level residual bootstrap. We show that the model selection procedure is consistent under some conditions, and our simulations suggest that it selects the true model most often than the other selection procedures considered.     

High dimensional multivariate mixed models for binary questionnaire data

Summary.  Questionnaires that are used to measure the effect of an intervention often consist of different sets of items, each set possibly measuring another concept. Mixed models with set-specific random effects are a flexible tool to model the different sets of items jointly. However, computational problems typically arise as the number of sets increases. This is especially true when the random-effects distribution cannot be integrated out analytically, as with mixed models for binary data. A pairwise modelling strategy, in which all possible bivariate mixed models are fitted and where inference follows from pseudolikelihood theory, has been proposed as a solution. This approach has been applied to assess the effect of physical activity on psychocognitive functioning, the latter measured by a battery of questionnaires.     

Bayesian model averaging in longitudinal regression models with AR(1) errors with application to a myopia data set

We propose a new iterative algorithm, called model walking algorithm, to the Bayesian model averaging method on the longitudinal regression models with AR(1) random errors within subjects. The Markov chain Monte Carlo method together with the model walking algorithm are employed. The proposed method is successfully applied to predict the progression rates on a myopia intervention trial in children.     

Book Reviews

Aggregate forecasts using McFadden's conditional logit model of discrete choice harbor the unrealistic implicit assumption of a random-utility distribution that is homogeneous across both alternatives and individuals. This article relaxes that assumption. A choice model is developed that describes the random-utility component as a probability-mixture model. Some numerical results illustrate that the derived model is not constrained by the independence-of-irrelevant-alternatives property. An experimental test of visual perceptions demonstrates the potential superiority of the model.     

Admissible variable-selection procedures when fitting misspecified regression models by least squares

For loss equal to squared error of prediction, Kempthorne(l984) has proved that all variable-selection procedures are admissible for choosing among least-squares fits of a normal linear regression model. We extend this result to the case of a normal linear regression model in which the form of the expected response vector is misspecified.