具有随机效应和AR(1)误差的非线性纵向数据模型中组间方差和自相关系数的齐性检验

组间方差和自相关系数的齐性是纵向数据分析的基本假设之一,然而这种假设需要进行统计检验. Zhang \&; Weiss$^{[15]}$ 讨论了线性随机效应模型的组间和组内方差齐性的检验问题;林金官 \&; 韦博成$^{[10]}$ 研究了具有AR(1)误差但没有随机效应的非线性模型的自相关系数的齐性检验.该文研究具有随机效应和AR(1)误差的非线性模型的组间方差和自相关系数的齐性检验问题,构造了几个score检验统计量, 并通过Monte Carlo模拟方法研究了检验统计量的性质.最后利用该文的方法分析一组实际数据和一组模拟数据.     

具有一致相关的纵向数据模型中方差和相关系数的齐性检验

在纵向数据分析中, 模型方差的齐性是一个基本假定, 但是该假定未必正确. 林金官、韦博成^[1]讨论了具有AR(1)误差的非线性纵向数据模型中方差和相关系数的齐性检验. 本文对具有一致相关协方差结构的纵向数据模型, 研究了方差齐性和相关系数齐性的检验, 得到了检验的score统计量, 并应用于葡萄糖数据. 最后, 本文还给出了模拟结果.     

Inference in binomial AR(1) models

This paper studies inference methods for stationary time series with binomial distributions. Such series describe, for example, the number of rainy days in consecutive weeks. First, we formulate the renewal sequence version of the model that seemingly generates a new class of stationary binomial series. The model is shown to obey an AR(1) recursion in cases where the renewal lifetime has a constant hazard rate past lag one. Explicit asymptotic variances of the parameter estimators in the AR(1) case are derived from conditional least squares methods; likelihood techniques are also considered.     

Detecting changes in the ar parameters of a nonstationary arma process

We present a method for detecting changes in the AR parameters of an ARMA process with arbitrarily time varying MA parameters. Assuming that a collection of observations and a set of nominal time invariant AR parameters are given, we test if the observations are generated by the nominal AR parameters or by a different set of time invariant AR parameters. The detection method is derived by using a local asymptotic approach and it is based on an estimation procedure which was shown to be consistent under nonstationarities.     

General projection-pursuit estimators for the common principal components model: influence functions and Monte Carlo study

The common principal components (CPC) model for several groups of multivariate observations assumes equal principal axes but possibly different variances along these axes among the groups. Under a CPCs model, generalized projection-pursuit estimators are defined by using score functions on the dispersion measure considered. Their partial influence functions are obtained and asymptotic variances are derived from them. When the score function is taken equal to the logarithm, it is shown that, under a proportionality model, the eigenvector estimators are optimal in the sense of minimizing the asymptotic variance of the eigenvectors, for a given scale measure.     

Simultaneous estimation of means of classified normal observations

Simultaneous estimation of normal means is considered for observations which are classified into several groups. In a one-way classification case, it is shown that an adaptive shrinkage estimator dominates a Stein-type estimator which shrinks observations towards individual class averages as Stein's (1966,Festschrift for J. Neyman, (ed. F. N. David), 351–366, Wiley, New York) does, and is minimax even if class sizes are small. Simulation results under quadratic loss show that it is slightly better than Stein's (1966) if between variances are larger than within ones. Further this estimator is shown to improve on Stein's (1966) with respect to the Bayes risk. Our estimator is derived by assuming the means to have a one-way classification structure, consisting of three random terms of grand mean, class mean and residual. This technique can be applied to the case where observations are classified into a two-stage hierarchy.     

Inferring missing links in partially observed social networks

Determining the pattern of links within a large social network is often problematic due to the labour-intensive nature of the data collection and analysis process. With constrained data collection capabilities it is often only possible to either make detailed observations of a limited number of individuals in the network, or to make fewer observations of a larger number of people. Previously we have shown how detailed observation of a small network can be used, which infer where in the network previously unconnected individuals are likely to fit, thereby attempting to predict network growth as new people are considered for inclusion. Here, by contrast, we show how social network topology can be inferred following a limited observation of a large network. Essentially the issue is one of inferring the presence of links that are missed during a constrained data collection campaign on the network. It is particularly difficult to infer network structures for those organizations that actively seek to remain covert and undetected. However, it is often very useful to know if two given individuals are likely to be connected even though limited surveillance effort yields no evidence of a link. Specifically, we show how a statistical inference technique can be used to successfully predict the existence of links that are missed during network sampling. The procedure is demonstrated using network data obtained from open source publications.     

非线性纵向数据模型中方差和自相关系数的齐性检验

刻画纵向数据的协方差结构有三个可能因素：随机效应、序列相关和随机误差．在纵向数据分析中，模型方差的齐性是一个基本假定．但是，该假设未必正确．Zhang和、Weiss^[1]研究了具有随机效应的线性模型的异方差检验．林金官和韦博成^[2]将Zhang和、Weiss^[1]的结果推广到非线性情形．本文对具有自相关误差的非线性纵向数据模型，研究了方差齐性和相关系数的齐性检验，得到了检验的score统计量并应用于血浆渗透数据(见Davidian和Giltian^[3])．最后，本文还给出了模拟结果．     

Degeneracy in Heteroscedastic Regression Models

The maximum likelihood estimation in a regression model with heteroscedastic errors is considered. When the design matrices in the model are inappropriately specified, the maximum likelihood estimates of the variances of certain observations are found to be zero irrespective of the observed values, resulting in degeneracy. Necessary and sufficient conditions for degeneracy are given and used for its avoidance.     

Density Estimation with Replicate Heteroscedastic Measurements

We present a deconvolution estimator for the density function of a random variable from a set of independent replicate measurements. We assume that measurements are made with normally distributed errors having unknown and possibly heterogeneous variances. The estimator generalizes well-known deconvoluting kernel density estimators, with error variances estimated from the replicate observations. We derive expressions for the integrated mean squared error and examine its rate of convergence as n → ∞ and the number of replicates is fixed. We investigate the finite-sample performance of the estimator through a simulation study and an application to real data.     

����ģ�ͱ������ı�Ҷ˹����

本文主要讨论了变点的先验分布为beta-binomial分布 和Ibrahim等(2003)提出的幂型先验的条件下, 有一个变点的线性模型的贝叶斯统计推断问题, 并且我们假定变点两边的观测值的方差是相等的. 我们得到变点、回归系数、共同方差的后要分布的显示表达式. 本论文不仅把Ferrira(1975)论文从变点先验分布服从离散均匀分布推广到了更好描述变点 的形状的beta-binomial分布, 而且进一步将变点的先验分布推广到包含的历史信息的幂型先验. 当变点的先验分布为beta-binomial分布和幂型先验时, 模拟结果显示了贝叶斯方法具有更高的准确性.     

Computing Bounds on the Expected Maximum of Correlated Normal Variables

We compute upper and lower bounds on the expected maximum of correlated normal variables (up to a few hundred in number) with arbitrary means, variances, and correlations. Two types of bounding processes are used: perfectly dependent normal variables, and independent normal variables, both with arbitrary mean values. The expected maximum for the perfectly dependent variables can be evaluated in closed form; for the independent variables, a single numerical integration is required. Higher moments are also available. We use mathematical programming to find parameters for the processes, so they will give bounds on the expected maximum, rather than approximations of unknown accuracy. Our original application is to the maximum number of people on-line simultaneously during the day in an infinite-server queue with a time-varying arrival rate. The upper and lower bounds are tighter than previous bounds, and in many of our examples are within 5% or 10% of each other. We also demonstrate the bounds’ performance on some PERT models, AR/MA time series, Brownian motion, and product-form correlation matrices.     

自相关对常规控制图影响的模拟研究与案例分析

常规统计控制图的基本假设前提是观测值独立同分布,而在实际生产过程中,质量指标值常表现出自相关现象,违背独立性假定。本文运用平均链长(ARL)研究自相关过程为AR(1)时对常规控制图的影响,并比较了常规控制图和残差控制图对序列相关过程的控制效果。模拟结果和实例分析表明:当过程序列相关时,使用常规作图法估计出的标准差是有偏的,致使控制限设置错误和常规控制图检测能力降低。因此,在一些统计过程控制中,须考虑自相关现象并采用适当的控制图方法。     

Nonparametric estimation of the spectral density of amplitude-modulated time series with missing observations

Consider a real-valued and second-order stationary time series with mean zero. The aim is to estimate its spectral density. A minimax solution of this problem is known when either the time series is observed directly, or some observations are missed according to an independent Bernoulli process, or for some special cases when the time series is multiplied by an amplitude-modulating time series with known distribution. It is shown that if a time series of interest, a Bernoulli time series defining missing mechanism, and an amplitude-modulating time series are mutually independent, then the shape of spectral density of an underlying time series of interest can be estimated with the minimax rate known for the case of direct observations. Furthermore, in some special cases the spectral density can be estimated with the minimax rate known for directly observed time series of interest.     

Asymptotically optimal estimation in a linear regression problem with random errors in coefficients

We consider the problem of estimating an unknown one-dimensional parameter in the linear regression problem in the case when the independent variables (called coefficients in the article) are measured with errors, and the variances of the principal observations can depend on the main parameter. We study the behavior of two-step estimators, previously introduced by the authors, which are asymptotically optimal in the case when the independent variables are measured without errors. Under sufficiently general assumptions we find necessary and sufficient conditions for the asymptotic normality and asymptotic optimality of these estimators in the new setup.     

Asymptotically optimal estimation in a linear-fractional regression problem with random errors in coefficients

Considering the linear-fractional regression problem with errors in independent variables, we construct and study asymptotically optimal estimators for unknown parameters in the case of violation of the classical regression assumptions (the variances of the observations are different and depend on the unknown parameters).     

Estimating volatility on overlapping returns when returns are autocorrelated

Overlapping financial returns are sometimes used to increase the efficiency and power of statistical tests and for Value-at-Risk analysis. This is particularly useful when there are not many observations, such as daily returns for emerging markets. Sometimes, returns show autocorrelation. In this paper, unbiased variance estimators are derived for overlapping returns when the returns are generated by AR(1) or MA(1) processes. A limited Monte Carlo experiment reveals that alternative estimators can suffer from substantial bias. The relevance of using proper estimators is emphasized by considering daily returns for six emerging markets.     

Periodogram analysis with missing observations

Estimation of the spectral measure, covariance and spectral density functions of a strictly stationaryr- vector valued time series is considered, under the assumption that some of the observations are missed. The modified periodograms are calculated using data window. The asymptotic normality is studied.     

The Order of the Error Term for Moments of the Log Likelihood Ratio Unit Root Test in an Autoregressive Process

This paper investigates the asymptotics of the log likelihood ratio test for a unit root in an autoregressive (AR) process of general order. The main result is that the expectation and variance (in fact, all moments) of the test statistic may, to the order of T-1, where T is the number of observations, be approximated by the expectation and variance of the corresponding test in an AR(1) process. This result has obvious implications for the asymptotics of unit root tests for panels. An explicit formula for the approximation error of a test in an AR(2) process is also given.     

On some representations of sample variance

The usual formula for variance depending on rounding off the sample mean lacks precision, especially when computer programs are used for the calculation. The well-known simplification of the total sums of squares does not always give benefit. Since the variance of two observations is easily calculated without the use of a sample mean, and the variance of a sample of n observations is the average of the variances of observations based on n(n-1)/2 distinct subsets of units of size 2 from the sample, it is argued that this sense of pairing may result in precision. Some other forms of variance are presented which provide some insight into it. The contribution of a new observation of variance is highlighted, which is important in sequential sampling. Notions are illustrated with examples.