A property of coefficients of variation for ordered bivariate log-normal variables

This paper considers the maximum and minimum of a pair of log-normal variables with equal mean. It shows that either order statistic has a smaller coefficient of variation than the two original log-normal variables provided the latter are of equal variance. When the variances are unequal, as the variance ratio increases, the minimum (maximum), has a smaller coefficient of variation if the correlation coefficient of the log-normal variables is small (small) and the variances are large (small).     

On estimating the conditional probability of discovering a new species

Based on a capture-recapture sample of size $i;n+k,n≥ l,k ≥ 0, from a population of an unknown number of distinct species (or classes), the problem of estimating the total probability of the species unobserved in the first n selections is considered. As the estimand depends on both the unknown parameters and the data, the standard theory of estimation is inadequate for this problem A suitable definition of sufficiency is introduced and used to prove a Rao-Blackwell type result and discuss uniformly minimum mean squared error unbiased estimation. An alternative proof for an inadmissibility result is presented. The new proof gives more insight and a method for deriving improved estimators. The theoretical developments may be useful in other problems concerning inferences about random parametric functions.     

Best linear predictors for factor scores

From the literature three types of predictors for factor scores are available. These are characterized by the constraints: linear, linear conditionally unbiased, and linear correlation preserving. Each of these constraints generates a class of predictors. Best predictors are defined in terms of Lowner's partial matrix order applied to matrices of mean square error of prediction. It is shown that within the first two classes a best predictor exists and that it does not exist in the third.     

Admissible minimum bias estimation of response surfaces

The minimum bias estimator was introduced as an alternative to the least squares estimator for approximating response functions by low-order polynomials. Here we show how to obtain an admissible estimator with smaller squared bias.     

Estimation of mean vector in presence of non-response

Simultaneous estimation of means of several variables is considered for finite population in presence of non-response. Two types of nonresponses (partial and complete) are considered using the technique of sampling and subsampling with equal probabilities without replacement. The optimum sample size and the optimum value of subsampling fraction to be repeated from the nonresponding units of the sample have been obtained for fixed survey budget.     

Smoothing techniques via the bezier curve

Although the Bezier curve is very popular in the area of computational graphics it has rarely been used by statisticians. In this paper we develop methods and techniques for use of the Bezier curve in estimation of density and regression function. Also, asymptotic mean integrated square error for both estimators are derived. Comparisons with kernel estimator are conducted using simulation.     

Superiority comparisons between mixed regression estimators

This paper gives necessary and sufficient conditions for a mixed regression estimator to be superior to another mixed estimator. The comparisons are based on the mean square error matrices of the estimators. Both estimators are allowed to be biased.     

Mean intergrated squared error properties and optimal kernels when estimating a diatribution function

Let X₁,X₂,… X_n be a sample of independent identically distributed (i.i.d)random variables having an unknown absolutely continuous distribution function f with density f the twofold aim of his paper consists in, firstly deriving asymptotic expressions of the mean intergrated squared error (MISE) of a kernel estimator of F when f is either assumed to be continuous everywhere or problem of finding optimal kernels in these two cases is studied in detail.     

Covariate classification by post–stratification

The plug–in Anderson's covariate classification statistic is constructed on the basis of an initially unclassified training sample by means of posty–stratification. The asymptotic efficiency relative to the discriminant based on an initially classified training sample is evaluated for the case where a covariate is present. Effect of post–stratification is examined.