Coverage of generalized confidence intervals

Generalized confidence intervals provide confidence intervals for complicated parametric functions in many common practical problems. They do not have exact frequentist coverage in general, but often provide coverage close to the nominal value and have the correct asymptotic coverage. However, in many applications generalized confidence intervals do not have satisfactory finite sample performance. We derive expansions of coverage probabilities of one-sided generalized confidence intervals and use the expansions to explain the nonuniform performance of the generalized intervals. We then show how to use these expansions to obtain improved coverage by suitable calibration. The benefits of the proposed modification are illustrated via several examples.     

多个偏正态总体共同位置参数的Bootstrap置信区间

该文基于Bootstrap 方法研究多个偏正态总体共同位置参数的区间估计和假设检验问题.首先,分别给出未知参数的矩估计和极大似然估计.其次,将徐礼文[1]对多个正态总体共同均值的探讨推广到多个偏正态总体,进而构造共同位置参数的Bootstrap 置信区间和Bootstrap检验统计量.Monte Carlo模拟结果表明...     

On the simultaneous anova and manova tests

Summary In the present paper, the simultaneous confidence intervals associated with the Simultaneous Multivariate Analysis of Variance (SMANOVA) tests are derived when the hypotheses can be tested in a “quasi-independentrd manner. With respect to the Simultaneous Analysis of Variance (SANOVA) test, the quasi-independence is weakened and the distribution problems are investigated when the joint distribution of the sums of squares associated with various hypotheses is a “multivariate chi-square ” distribution. It is shown that the lengths of the confidence intervals associated with the SANOVA test (dependent case) are shorter than the lengths of the corresponding confidence intervals obtained by using Scheffé’s method when the hypotheses on several contrasts are tested simultaneously. It is also shown that the SMANOVA test yields narrower confidence intervals than the MANOVA test when several quasi-independent hypotheses are tested simultaneously. Part of the work was done under U.S. Air Force Contract No. AF 49 (638) 213 when the author was at the University of North Carolina. Part of the author’s Ph. D. thesis submitted to the University of Minnesota.     

Computational methods for comparing two binary diagnostic tests in the presence of partial verification of the disease

In the presence of partial verification of the disease, confidence interval estimation for the difference of sensitivities and specificities cannot be carried out by applying confidence intervals for the difference of binomial proportions, and the comparison of the accuracy of the two binary tests cannot be carried out by applying McNemar’s test. In this article we propose two methods for comparing the accuracy of two binary tests in the presence of partial verification of the disease. The first method is based on the application of the EM and SEM algorithms, and the second method consists of the calculation of confidence intervals for the difference in sensitivities and specificities applying confidence intervals for the difference in the two binomial proportions from the last table obtained applying the EM algorithm. We carried out simulation experiments in order to study and compare the coverage of several confidence intervals for the difference of the sensitivities and specificities.     

Panel模型中的精确检验和置信区间

利用广义p-值和广义置信区间的概念,研究了Panel模型中未知参数的检验和置信区间问题.对于回归系数,分别考虑了单个情形和多个线性无关情形下的检验和置信区间问题,得到了精确检验和置信区间.对于方差分量,研究了其任意线性组合的检验和置信区间问题,建立了精确检验和置信区间.基于广义p-值和广义置信区间,获取精确检验和置信区间的方法具有计算方便、易应用于小样本问题的特点.最后,分别从理论和数值上研究了这些精确检验和置信区间的统计性质.     

�������ŷֲ��Ķ��������̬����Ĺ�����ֵ����������Ĺ���

The log-normal distribution is a common choice for modeling positively skewed data arising from many practical applications.This article introduces a new method of constructing confidence interval for a common mean shared by several log-normal populations through confidence distributions, which combines all information from independent sources. We develop a non-trivial weighting approach by taking account of the sample variances of related quantities to enhance efficiency. Combined confidence distributions are used to construct confidence intervals for the common mean and a simplified version of one existing method is also proposed. We conduct simulation studies to evaluate the performance of the proposed methods in comparison with existing methods. Our simulation results show that the weighting approach yields shorter interval length than the non-weighting approach. The newly proposed confidence intervals perform very well in terms of empirical coverage probability and average interval length. Finally, applications of the proposed methodology is illustrated through three real data examples.     

Confident estimation for density of a biological population based on line transect sampling

Line transect sampling is a very useful method in survey of wildlife population. Confident interval estimation for density D of a biological population is proposed based on a sequential design. The survey area is occupied by the population whose size is unknown. A stopping rule is proposed by a kernel-based estimator of density function of the perpendicular data at a distance. With this stopping rule, we construct several confidence intervals for D by difference procedures. Some bias reduction techniques are used to modify the confidence intervals. These intervals provide the desired coverage probability as the bandwidth in the stopping rule approaches zero. A simulation study is also given to illustrate the performance of this proposed sequential kernel procedure.     

条件密度置信区间的覆盖精度

本文用经验似然方法讨论了条件密度的置信区间的构造. 通过对覆盖概率的Edgeworth展开得到了经验似然置信区间的覆盖精度, 同时证明了条件密度的经验似然置信区间的Bartlett可修正性     

Inference for constant-stress accelerated degradation test based on Gamma process

This paper proposes a Gamma constant-stress accelerated degradation model based on the principle of the degradation mechanism invariance. The maximum likelihood estimators of the parameters of the proposed model are derived. Based on Cornish–Fisher expansion, the approximate confidence interval for the shape parameter of the Gamma degradation process is developed. Since it is difficult to obtain the exact confidence intervals for other model parameters and some quantities such as the mean degradation in unit time, the quantile and the reliability function of the lifetime at the normal stress level, the generalized confidence intervals for these quantities are proposed. The percentiles of the proposed generalized pivotal quantities can be obtained by the simulation. The performances of the proposed confidence intervals are evaluated by the Monte Carlo simulation method. In the simulation study, the proposed confidence intervals are compared with the Wald and the bootstrap-p confidence intervals. The simulation results show that the proposed confidence intervals outperform the Wald and the bootstrap-p confidence intervals in terms of the coverage percentage. Finally, a real example is used to illustrate the proposed procedures.     

Two-sample inference based on one-sample Wilcoxon signed rank statistics

A two-sample test is studied which rejects the null hypothesis of equal population medians when two Wilcoxon distribution free confidence intervals are disjoint. A confidence interval for the difference in population medians is constructed by subtracting the endpoints of two one-sample confidence intervals. Two different ways to select the one-sample intervals are presented. A solution that specifies equal confidence coefficients for the one-sample intervals is recommended. All solutions are shown to have the same asymptotic (Pitman) efficiency as the Mann-Whitney two-sample test.     

Exact confidence coefficients of simultaneous confidence intervals for multinomial proportions

Simultaneous confidence intervals for multinomial proportions are useful in many areas of science. Since 1964, approximate simultaneous 1-α confidence intervals have been proposed for multinomial proportions. Although at each point in the parameter space, these confidence sets have asymptotic 1-α coverage probability, the exact confidence coefficients of these simultaneous confidence intervals for a fixed sample size are unknown before.In this paper, we propose a procedure for calculating exact confidence coefficients for simultaneous confidence intervals of multinomial proportions for any fixed sample size. With this methodology, exact confidence coefficients can be clearly derived, and the point at which the infimum of the coverage probability occurs can be clearly identified.     

Coverage Accuracy of Confidence Intervals in Nonparametric Regression

Point-wise confidence intervals for a nonparametric regression function with random design points are considered. The confidence intervals are those based on the traditional normal approximation and the empirical likelihood. Their coverage accuracy is assessed by developing the Edgeworth expansions for the coverage probabilities. It is shown that the empirical likelihood confidence intervals are Bartlett correctable.     

On simultaneous confidence intervals for all contrasts in the means of the intraclass correlation model with missing data

In this paper, we consider simultaneous confidence intervals for all contrasts in the means when the observations are missing at random in the intraclass correlation model. An exact test statistic for the equality of the means and Scheffé, Bonferroni and Tukey types of simultaneous confidence intervals are given by an extension of Bhargava and Srivastava [On Tukey's confidence intervals for the contrasts in the means of the intraclass correlation model, J. Royal Statist. Soc. B35 (1973) 147-152] when the missing observations are of the monotone type. Finally, numerical results of simultaneous confidence intervals are presented.     

On some confidence intervals for estimating the mean of a skewed population

A number of methods are available in the literature to measure confidence intervals. Here, confidence intervals for estimating the population mean of a skewed distribution are considered. This note proposes two alternative confidence intervals, namely, Median t and Mad t, which are simple adjustments to the Student's t confidence interval. In order to compare the performance of these intervals, the following criteria are considered: (i) coverage probability; (ii) average width; and (iii) ratio of coverage to width. A simulation study has been undertaken to compare the performance of the intervals. The simulation study shows that for small sample size and moderate to highly skewed distributions, the proposed Median t performs the best in the sense of higher coverage, and the Mad t performs best in the sense of smaller confidence width. The proposed methods are very easy to calculate and are not overly computer-intensive, like Bootstrap confidence intervals. Some real-life examples have been considered that support the findings of the paper to some extent.     

Multistep stochastic mirror descent for risk-averse convex stochastic programs based on extended polyhedral risk measures

We consider risk-averse convex stochastic programs expressed in terms of extended polyhedral risk measures. We derive computable confidence intervals on the optimal value of such stochastic programs using the Robust Stochastic Approximation and the Stochastic Mirror Descent (SMD) algorithms. When the objective functions are uniformly convex, we also propose a multistep extension of the Stochastic Mirror Descent algorithm and obtain confidence intervals on both the optimal values and optimal solutions. Numerical simulations show that our confidence intervals are much less conservative and are quicker to compute than previously obtained confidence intervals for SMD and that the multistep Stochastic Mirror Descent algorithm can obtain a good approximate solution much quicker than its nonmultistep counterpart.     

A comparison of testing and confidence interval methods for the median

There are only relatively few confidence coefficients and levels available for distribution-free confidence intervals and test for the median based on the sign statistic. This problem can be overcome by interpolating confidence intervals or by studentizing the median by an estimate of its standard error. Such methods are discussed and then compared via simulation.     

Improved confidence intervals for the scale parameter of Burr XII model based on record values

In this paper some different sorts of confidence intervals are considered for the scale parameter of the Burr type XII distribution based on the upper record values. In this regard, the coverage probability is adopted as a measure of improvement when the endpoints are the same for all types of confidence intervals. Proposed confidence intervals are based on the preliminary test estimator, Thompson shrinkage estimator and Bayes estimator with conjugate prior information. It is nicely demonstrated that the confidence intervals based on the above methodologies are superior to the equal tail confidence interval on specific intervals. Subsequently, to construct a uniformly dominant confidence interval, the result of Kubokawa (Ann Stat 22(1):290–299, 1994) is extended for dependent observations by making use of the information that exists in a covariate record value.     

Fixed-width multiple-comparison procedures using common random numbers for steady-state simulations

Suppose that there are k ? 2 different systems (i.e., stochastic processes), where each system has an unknown steady-state mean performance. We consider the problem of running a two-stage simulation using common random numbers to construct fixed-width confidence intervals for two multiple-comparison problems. Under the assumptions that the stochastic processes representing the simulation output of the different systems satisfy a functional central limit theorem and that the asymptotic covariance matrix satisfies a condition known as sphericity, we prove that our confidence intervals are asymptotically valid (as the desired half-width of the confidence intervals tend to zero). We develop both absolute- and relative-width confidence intervals. Empirical results are presented indicating the procedures’ robustness to violations of the sphericity assumption.     

Confidence intervals for ratio of two Poisson rates using the method of variance estimates recovery

Inference based on ratio of two independent Poisson rates is common in epidemiological studies. We study the performance of a variety of unconditional method of variance estimates recovery (MOVER) methods of combining separate confidence intervals for two single Poisson rates to form a confidence interval for their ratio. We consider confidence intervals derived from (1) the Fieller’s theorem, (2) the logarithmic transformation with the delta method and (3) the substitution method. We evaluate the performance of 13 such types of confidence intervals by comparing their empirical coverage probabilities, empirical confidence widths, ratios of mesial non-coverage probability and total non-coverage probabilities. Our simulation results suggest that the MOVER Rao score confidence intervals based on the Fieller’s theorem and the substitution method are preferable. We provide two applications to construct confidence intervals for the ratio of two Poisson rates in a breast cancer study and in a study that examines coronary heart diseases incidences among post menopausal women treated with or without hormones.     

Modeling a Simultaneous Confidence Band of the Mean Value of Multiple Responses with a Rectangular Domain for Predictors

The problem is considered of modeling simultaneous confidence intervals for the mean values of multiple responses in a linear multivariate normal regression model with predictor variables defined in intervals. To solve it, a numerical way of calculating the critical value that determines the simultaneous confidence interval of a given level is used. Simultaneous confidence intervals are numerically modelled and analyzed by comparison for regression, the mean value of multiple responses, and individual observation.