再论统计平均数大小关系的证明及其推广

一、引言统计平均数是统计分析和一般经济分析中广泛运用的指标形式,在统计学中有重要的地位。统计平均数按代表意义和计算方式的不同,可以分为数值平均数和位置平均数,其中常用的数值平均数有算术平均数(-X)、几何平均数(G)和调和平均数(H),三者源于同一个家族———幂平均数(     

Constructing copula functions with weighted geometric means

The weighted arithmetic mean of two copulas is a copula. In some cases, geometric and harmonic means also provide copulas. There are copulas specially appropriate to be combined by using weighted geometric means. With this method of construction we combine Farlie–Gumbel–Morgentern and Ali–Mikhail–Haq copulas to obtain families of copulas which can be expressed in terms of double power series. The Gumbel–Barnett copula is also considered and a new copula is proposed, which arises as the first order approximation of the weighted geometric mean of two copulas. Invariance of two multivariate distributions (Cuadras–Augé and Johnson–Kotz) by weighted geometric and arithmetic means is also studied.     

Geometry of the log-likelihood ratio statistic in misspecified models

We show that the asymptotic mean of the log-likelihood ratio in a misspecified model is a differential geometric quantity that is related to the exponential curvature of Efron (1978), Amari (1982), and the preferred point geometry of [Critchley et al., 1993] and [Critchley et al., 1994]. The mean is invariant with respect to reparameterization, which leads to the differential geometrical approach where coordinate-system invariant quantities like statistical curvatures play an important role. When models are misspecified, the likelihood ratios do not have the chi-squared asymptotic limit, and the asymptotic mean of the likelihood ratio depends on two geometric factors, the departure of models from exponential families (i.e. the exponential curvature) and the departure of embedding spaces from being totally flat in the sense of Critchley et al. (1994). As a special case, the mean becomes the mean of the usual chi-squared limit (i.e. the half of the degrees of freedom) when these two curvatures vanish. The effect of curvatures is shown in the non-nested hypothesis testing approach of Vuong (1989), and we correct the numerator of the test statistic with an estimated asymptotic mean of the log-likelihood ratio to improve the asymptotic approximation to the sampling distribution of the test statistic.     

Generalized definition of the geometric mean of a non negative random variable

The first probabilistic definition of the geometric mean of a non negative random variable under certain assumptions was given in Feng et al. (2013 Feng, C., Wang, H., Tu, X. (2013). Geometric mean of nonnegative random variable. Commun. Stat.—Theory Methods 42:2714–2717.[Taylor &; Francis Online], [Web of Science ®] , [Google Scholar]). In this paper, we generalize the definition to a larger class of random variables. We also show the basic properties of the geometric mean and point out its discontinuity and instability. Some convergence properties are studied as well, for which we emphasize its link to the positive moments of the random variable. A discussion of potential applications of the new definition in biomedical research and open questions to complete the theory of geometric mean is highlighted.     

Robust estimators based on generalization of trimmed mean

In this article, we propose new estimators of location. These estimators select a robust set around the geometric median, enlarge it, and compute the (iterative) weighted mean from it. By doing so, we obtain a robust estimator in the sense of the breakdown point, which uses more observations than standard estimators. We apply our approach on the concepts of boxplot and bagplot. We work in a general normed vector space and allow multi-valued estimators.     

Bayesian prediction of order statistics with fixed and random sample sizes based on k-record values from Pareto distribution

In this paper, the two-parameter Pareto distribution is considered and the problem of prediction of order statistics from a future sample and that of its geometric mean are discussed. The Bayesian approach is applied to construct predictors based on observed k-record values for the cases when the future sample size is fixed and when it is random. Several Bayesian prediction intervals are derived. Finally, the results of a simulation study and a numerical example are presented for illustrating all the inferential procedures developed here.     

On the characterization of maximum likelihood estimators for location-scale families

Given maximum likelihood equations for location and scale parameters, one determines conditions under which there exists a uniquely defined parametric statistical model, whose location and scale maximum likelihood estimators are the given ones. The constructive approach is exemplified at several kinds of mean estimators including the mean, mean square, mean mean and stretched power mean. The possible extension of the method to more general situations is discussed and illustrated at the sample median maximum likelihood estimator.     

Scale estimators for symmetric stable distributions

This paper reports on a simulation comparison of scale estimators for symmetric stable distributions in terms of their ability to identify the population with the greater scale. The modified geometric mean is found to be superior to the sample standard deviation and the Fama-Roll estimator for the larger values of the characteristic exponent, while the Fama-Roll estimator is judged superior for the smaller values. Further, this study sheds some light on the question of the appropriate sample size for discriminating risk measurement in investment analysis when the samples are taken from symmetric stable distributions.     

Compound Regression and Constrained Regression: Nonparametric Regression Frameworks for EIV Models

Abstract

Errors-in-variable (EIV) regression is often used to gauge linear relationship between two variables both suffering from measurement and other errors, such as, the comparison of two measurement platforms (e.g., RNA sequencing vs. microarray). Scientists are often at a loss as to which EIV regression model to use for there are infinite many choices. We provide sound guidelines toward viable solutions to this dilemma by introducing two general nonparametric EIV regression frameworks: the compound regression and the constrained regression. It is shown that these approaches are equivalent to each other and, to the general parametric structural modeling approach. The advantages of these methods lie in their intuitive geometric representations, their distribution free nature, and their ability to offer candidate solutions with various optimal properties when the ratio of the error variances is unknown. Each includes the classic nonparametric regression methods of ordinary least squares, geometric mean regression (GMR), and orthogonal regression as special cases. Under these general frameworks, one can readily uncover some surprising optimal properties of the GMR, and truly comprehend the benefit of data normalization. Supplementary materials for this article are available online.     

Statistical Reversion toward the Mean: More Universal than Regression toward the Mean

Schmittlein discussed the lack of universality of regression toward the mean. The present note emphasizes the universality of a similar effect, dubbed “reversion” toward the mean, defined as the shift in conditional expectation of the upper or lower portion of a distribution. Reversion toward the mean is a useful concept for statistical reasoning in applications and is more self-evidently plausible than regression toward the mean.     

On estimating the mean of the selected population with unknown variance

This paper compares four estimators of the mean of the selected population from two normal populations with unknown means and common but unknown variance. The selection procedure is that the population yielding the largest sample mean is selected. The four estimators considered are invariant under both location and scale transformations. The bias and mean square errors of the four estimators are computed and compared. The conclusions are close to those reported by Dahiya ‘1974’, even for small sample sizes     

A new nonparametric estimator for the mean of the selected population

A modified bootstrap estimator of the mean of the population selected from two populations is proposed which is a convex combination of the two sample means, where the weights are random quantities. The estimator is shown to be strongly consistent. The small sample behavior of the estimator is investigated and compared with some competitors by means of Monte Carlo studies. It is found that the newly proposed estimator has smaller mean squared error for a wide range of parameter values.     

Locally optimal test for simultaneously testing the mean and variance of a normal distribution

A generalization of the locally most powerful unbiased (LMPU) test for the single parameter case to the k-parameter case was proposed by SenGupta and Vermeire (1986). In particular we defined a locally most mean power unbiased (LMMPU) test based on the mean curvature of the power hypersurface. Compared to the type C tests of Neyman and Pearson and the type D tests (Isaacson, 1951), LMMPU tests possess better theoretical properties and enjoy ease of construction of critical regions. In this paper we present an interesting example of a two-parameter univariate normal population for which Isaacson (1951, p. 233) was unsuccessful in finding a type D test. For the case of one observation, we prove that no Type D region exists but the LMMPU test is obtained - it is an example of a test with singular Hessian matrix for its power but is nevertheless a strictly locally unbiased (LU) test.     

A modified bootstrap estimator for the mean of an asymmetric distribution

A modified bootstrap estimator of the population mean is proposed which is a convex combination of the sample mean and sample median, where the weights are random quantities. The estimator is shown to be strongly consistent and asymptotically normally distributed. The small- and moderate-sample-size behavior of the estimator is investigated and compared with that of the sample mean by means of Monte Carlo studies. It is found that the newly proposed estimator has much smaller mean squared errors and also yields significantly shorter confidence intervals for the population mean.     

Estimation and testing under exchangeable exponential model with a single outlier

Several estimators of mean of an exponential distribution, when an unidentified single outlier in a sample of size n is present* are discussed. It is assumed that n?1 of these observations have a mean σ, While one could have a mean σ/α. The estimation of σ has been considered in detail with some reference to the estimation of α. Finally, tests of hypotheses about σare briefly mentioned.     

Mean square error efficient estimation of an exponential mean under an exchangeable single outlier model

This paper is concerned with estimating θ, the mean of an exponential distribution under a single outlier exchangeable model. It is a.ssumed that the single outlying observation is also exponential with mean θ/α, where 0 < α < 1. The estirnators proposed are weighted averages of the order statistics. The formulas for the weights minimizing the mean square error are presented. These weights are calculated for certain combinations of the sample size n and of α. It is found that the optimal weights very nearly have a certain form. The mean square errors of a simplified estitnator are compared lo those of Joshi (1972, 1988) and of Clhikkagoudar and Kunchur (1980). A nlodification of Joshi's iterative procedure is suggested.     

Robustness of tests for directional mean

In this paper we study the robustness of the likelihood ratio, circular mean and circular trimmed mean test functionals in the context of tests of hypotheses regarding the mean direction of circular normal and wrapped normal distributions. We compute the level and power breakdown properties of the three test functionals and compare them. We find that the circular trimmed mean test functional has the best robustness properties for both the above-mentioned distributions. The level and power properties of the test statistics corresponding to these functionals are also studied. Two examples with real data are given for illustration. We also consider the problem of testing the mean direction of the von-Mises–Fisher distribution on the unit sphere and explore the robustness properties of the spherical mean direction and likelihood ratio test functionals.     

On estimation of population mean based on two measurements

There are many estimators for population mean: arithmetic mean, median, root mean square, geometric mean, harmonic mean, and so forth. Which one is the best? Here, we answer this question given two measurements from a uniform distribution.     

Ranked-Set-Sample-Based Tests for Normal and Exponential Means

We consider the problems of testing a normal mean and an exponential mean under the median ranked set sampling (MRSS) scheme. The tests based on MRSS outperform those based on the usual ranked-set samples. In addition, through heuristic arguments we postulate working formulas for computing cut-off points for RSS-based and MRSS-based tests, respectively. Our simulation studies indicate these formulas perform surprisingly well, even for small samples, and solve the task of computation for large samples.     

Imputation by power transformation

In this paper, a new power transformation estimator of population mean in the presence of non-response has been suggested. The estimator of mean obtained from proposed technique remains better than the estimators obtained from ratio or mean methods of imputation. The mean squared error of the resultant estimator is less than that of the estimator obtained on the basis of ratio method of imputation for the optinum choice of parameters. An estimator for estimating a parameter involved in the process of new method of imputation has been discussed. The MSE expressions for the proposed estimators have been derived analytically and compared empirically. Product method of imputation for negatively correlated variables has also been introduced. The work has been extended to the case of multi-auxiliary information to be used for imputation.