A NOTE ON OPTIMUM STRATIFICATION IN SAMPLING WITH VARYING PROBABILITIES1

Singh and Sukhatme [4] have considered the problem of optimum stratification on an auxiliary variable x when the units from the different strata are selected with probability proportional to the value of the auxiliary variable and the sample sizes for the different strata are determined by using Neyman allocation method. The present paper considers the same problem for the proportional and equal allocation methods. The rules for finding approximately optimum strata boundaries for these two allocation methods have been given. An investigation into the relative efficiency of these allocation methods with respect to the Neyman allocation has also been made. The performance of equal allocation is found to be better than that of proportional allocation and practically equivalent to the Neyman allocation.     

An exact small sample theory for post-stratification

A genuine small sample theory for post-stratification is developed in this paper. This includes the definition of a ratio estimator of the population mean ?, the derivation of its bias and its exact variance and a discussion of variance estimation. The estimator has both a within strata component of variance which is comparable with that obtained in proportional allocation stratified sampling and a between strata component of variance which will tend to zero as the overall sample size becomes large. Certain optimality properties of the estimator are obtained. The generalization of post-stratification from the simple random sampling to post-stratification used in conjunction with stratification and multi-stage designs is discussed.     

Optimum multivariate stratified double sampling design: Chebyshev's Goal Programming approach

In stratified sampling when strata weights are unknown a double sampling technique may be used to estimate them. A large simple random sample from the unstratified population is drawn and units falling in each stratum are recorded. A stratified random sample is then selected and simple random subsamples are obtained out of the previously selected units of the strata. This procedure is called double sampling for stratification. If the problem of non-response is there, then subsamples are divided into classes of respondents and non-respondents. A second subsample is then obtained out of the non-respondents and an attempt is made to obtain the information by increasing efforts, persuasion and call backs. In this paper, the problem of obtaining a compromise allocation in multivariate stratified random sampling is discussed when strata weights are unknown and non-response is present. The problem turns out to be a multiobjective non-linear integer programming problem. An approximation of the problem to an integer linear programming problem by linearizing the non-linear objective functions at their individual optima is worked out. Chebyshev's goal programming technique is then used to solve the approximated problem. A numerical example is also presented to exhibit the practical application of the developed procedure.     

THREE-WAY STRATIFICATION SAMPLING DESIGN WHEN ONE OF THE STRATIFYING VARIABLES IS TIME

Bryant, Hartley & Jessen (1960) presented a two‐way stratification sampling design when the sample size n is less than the number of strata. Their design was extended to a three‐way stratification case by Chaudhary & Kumar (1988) , but this design does not take into account serial correlation, which might be present as a result of the presence of a time variable. In this paper, a new sampling procedure is presented for three‐way stratification when one of the stratifying variables is time. The purpose of such a design is to take into account serial correlation. The variance of the unweighted estimator of the population mean with respect to a super population model is used as the basis for comparison. Simulation results show that the suggested design is more efficient than the Chaudhary & Kumar (1988) design.     

Optimum stratification for allocation proportional to strata totals for simple random sampling scheme

The paper considers the problem of finding optimum strata boundaries when sample sizes to different strata are allocated in proportion to the strata totals of the auxiliary variable. This variable is also treated as the stratification variable. Minimal equations, solutions to which provide the optimum boundaries, have been obtained. Because of the implicit nature of these equations their exact solutions cannot be obtained. Therefore, methods of obtaining their approximate solutions have been presented. A lim¬iting expression for the variance of the estimate of population mean, as the number of strata tend to become large, has also been obtained.     

Model selection with data-oriented penalty

We consider the problem of model (or variable) selection in the classical regression model using the GIC (general information criterion). In this method the maximum likelihood is used with a penalty function denoted by C_n, depending on the sample size n and chosen to ensure consistency in the selection of the true model. There are various choices of C_n suggested in the literature on model selection. In this paper we show that a particular choice of C_n based on observed data, which makes it random, preserves the consistency property and provides improved performance over a fixed choice of C_n.     

An improved Cp criterion for spline smoothing

Spline smoothing is a popular technique for curve fitting, in which selection of the smoothing parameter is crucial. Many methods such as Mallows’ C_p, generalized maximum likelihood (GML), and the extended exponential (EE) criterion have been proposed to select this parameter. Although C_p is shown to be asymptotically optimal, it is usually outperformed by other selection criteria for small to moderate sample sizes due to its high variability. On the other hand, GML and EE are more stable than C_p, but they do not possess the same asymptotic optimality as C_p. Instead of selecting this smoothing parameter directly using C_p, we propose to select among a small class of selection criteria based on Stein's unbiased risk estimate (SURE). Due to the selection effect, the spline estimate obtained from a criterion in this class is nonlinear. Thus, the effective degrees of freedom in SURE contains an adjustment term in addition to the trace of the smoothing matrix, which cannot be ignored in small to moderate sample sizes. The resulting criterion, which we call adaptive C_p, is shown to have an analytic expression, and hence can be efficiently computed. Moreover, adaptive C_p is not only demonstrated to be superior and more stable than commonly used selection criteria in a simulation study, but also shown to possess the same asymptotic optimality as C_p.     

An optimization approach applied to the optimal stratification problem

A new optimization algorithm is presented to solve the stratification problem. Assuming the number L of strata and the total sample size n are fixed, we obtain strata boundaries by using an objective function associated with the variance. In this problem, strata boundaries must be determined so that the elements in each stratum are more homogeneous among themselves. To produce more homogeneous strata, this paper proposes a new algorithm that uses the Greedy Randomized Adaptive Search Procedure (GRASP) methodology. Computational results are presented for a set of problems, with the application of the new algorithm and some algorithms from literature.     

Allocation Constraints in Stratification

When a finite population is to be stratified, one of constraints in stratification is that sample sizes from strata may not be greater than the corresponding stratum sizes and may not be smaller than two. There are several ways of treating this allocation constraint, each providing an alternative approach to stratification. In this article, it is shown that a choice of the approach has a bearing on stratification efficiency. Unfortunately, no particular approach out of the four compared is shown to be the most efficient for each population studied. In addition, the approaches are applied to stratify a real population.     

Efficient estimators for adaptive stratified sequential sampling

In stratified sampling, methods for the allocation of effort among strata usually rely on some measure of within-stratum variance. If we do not have enough information about these variances, adaptive allocation can be used. In adaptive allocation designs, surveys are conducted in two phases. Information from the first phase is used to allocate the remaining units among the strata in the second phase. Brown et al. [Adaptive two-stage sequential sampling, Popul. Ecol. 50 (2008), pp. 239–245] introduced an adaptive allocation sampling design – where the final sample size was random – and an unbiased estimator. Here, we derive an unbiased variance estimator for the design, and consider a related design where the final sample size is fixed. Having a fixed final sample size can make survey-planning easier. We introduce a biased Horvitz–Thompson type estimator and a biased sample mean type estimator for the sampling designs. We conduct two simulation studies on honey producers in Kurdistan and synthetic zirconium distribution in a region on the moon. Results show that the introduced estimators are more efficient than the available estimators for both variable and fixed sample size designs, and the conventional unbiased estimator of stratified simple random sampling design. In order to evaluate efficiencies of the introduced designs and their estimator furthermore, we first review some well-known adaptive allocation designs and compare their estimator with the introduced estimators. Simulation results show that the introduced estimators are more efficient than available estimators of these well-known adaptive allocation designs.     

The gini-simpson index of diversity: estimation in the stratified sampling

In previous papers the problem of estimating the Gini-Simpson index of diversity for large populations has been considered by using random samplings with and without replacement, Nevertheless, the populations to which this estimation is usually applied (e.g., anthropoiogicai, ecological, linguistic and sociological populations) often arise naturally stratified.

In this paper we first construct unbiased estimators of the Gini-Simpson index from a sample drawn according to a stratified sampling with proportional allocation and independently in different strata. Then, we determine the standard error of such estimators. The advantages of the stratification in estimating diversity are later confirmed by means of a practical example. We finally suggest complementary studies that could be additionally developed.     

Compromise allocation in multivariate stratified surveys with stochastic quadratic cost function

In many real life situations the linear cost function does not approximate the actual cost incurred adequately. The cost of traveling between the units selected in the sample within a stratum is significant, instead of linear cost function. In this paper, we have considered the problem of finding a compromise allocation for a multivariate stratified sample survey with a significant travel cost within strata is formulated as a problem of non-linear stochastic programming with multiple objective functions. The compromise solutions are obtained through Chebyshev approximation technique, D ₁- distance and goal programming. A numerical example is presented to illustrate the computational details of the proposed methods.     

Linear model selection by cross-validation

We consider the problem of model (or variable) selection in the classical regression model based on cross-validation with an added penalty term for penalizing overfitting. Under some weak conditions, the new criterion is shown to be strongly consistent in the sense that with probability one, for all large n, the criterion chooses the smallest true model. The penalty function denoted by C_n depends on the sample size n and is chosen to ensure the consistency in the selection of true model. There are various choices of C_n suggested in the literature on model selection. In this paper we show that a particular choice of C_n based on observed data, which makes it random, preserves the consistency property and provides improved performance over a fixed choice of C_n.     

Designing stratified sampling in economic and business surveys

In most economic and business surveys, the target variables (e.g. turnover of enterprises, income of households, etc.) commonly resemble skewed distributions with many small and few large units. In such surveys, if a stratified sampling technique is used as a method of sampling and estimation, the convenient way of stratification such as the use of demographical variables (e.g. gender, socioeconomic class, geographical region, religion, ethnicity, etc.) or other natural criteria, which is widely practiced in economic surveys, may fail to form homogeneous strata and is not much useful in order to increase the precision of the estimates of variables of interest. In this paper, a stratified sampling design for economic surveys based on auxiliary information has been developed, which can be used for constructing optimum stratification and determining optimum sample allocation to maximize the precision in estimate.     

Optimum allocation in multivariate stratified sampling design in the presence of nonresponse with Gamma cost function

In multivariate stratified sample survey with L strata, let p-characteristics are defined on each unit of the population. To estimate the unknown p-population means of each characteristic, a random sample is taken out from the population. In multivariate stratified sample survey, the optimum allocation of any characteristic may not be optimum for others. Thus the problem arises to find out an allocation which may be optimum for all characteristics in some sense. Therefore a compromise criterion is needed to workout such allocation. In this paper, the procedure of estimation of p-population means is discussed in the presence of nonresponse when the use of linear cost function is not advisable. A solution procedure is suggested by using lexicographic goal programming problem. The numerical illustrations are given for its practical utility.     

Optimal stratification in stratified designs using weibull-distributed auxiliary information

Sampling has evolved into a universally accepted approach for gathering information and data mining as it is widely accepted that a reasonably modest-sized sample can sufficiently characterize a much larger population. In stratified sampling designs, the whole population is divided into homogeneous strata in order to achieve higher precision in the estimation. This paper proposes an efficient method of constructing optimum stratum boundaries (OSB) and determining optimum sample size (OSS) for the survey variable. The survey variable may not be available in practice since the variable of interest is unavailable prior to conducting the survey. Thus, the method is based on the auxiliary variable which is usually readily available from past surveys. To illustrate the application as an example using a real data, the auxiliary variable considered for this problem follows Weibull distribution. The stratification problem is formulated as a Mathematical Programming Problem (MPP) that seeks minimization of the variance of the estimated population parameter under Neyman allocation. The solution procedure employs the dynamic programming technique, which results in substantial gains in the precision of the estimates of the population characteristics.     

Bayesian analysis of binary survey data

First a comprehensive treatment of the hierarchical-conjugate Bayesian predictive approach to binary survey data is presented, encompassing simple random, stratified, cluster, and two-stage sampling, as well as two-stage sampling within strata. For the case of two-stage sampling within strata when there is more than one variable of stratification, analysis using an unsaturated logit linear model on the prior means is proposed. This allows there to be cells containing no sampled clusters. Formulas for posterior predictive means, variances, and covariances of numbers of successes in unsampled portions of clusters are presented in terms of posterior expectations of certain functions of hyperparameters; these may be evaluated by existing methods. The technique is illustrated using a small subset of Canada Youth & AIDS Study data. A sample of students within each of various selected school boards was chosen and interviewed via questionnaire. The boards were stratified/poststratified in two dimensions, but some of the resulting cells contained no data. The additive logit linear model on the prior means produced estimates and posterior variances for boards in all cells. Data showed the additive model to be plausible.     

Optimum stratification and allocation in inventory sampling: An efficient two stage grid search procedure

Standard Methods of optimal stratification are solving the optimization problem as a function of strata boundaries and sample allocation only. In this paper we show that by means of a flexible two stage grid search procedure strata boundaries, sample allocation and furthermore number of strata can be taken into account in an effective way when optimizing stratification and allocation. By means of a Monte Carlo simulation we show that the proposed procedure is efficient compared to the well known standard procedures.     

Determination of a new mixed variable lot-size multiple dependent state sampling plan based on the process capability index

This article proposes a new mixed variable lot-size multiple dependent state sampling plan in which the attribute sampling plan can be used in the first stage and the variables multiple dependent state sampling plan based on the process capability index will be used in the second stage for the inspection of measurable quality characteristics. The proposed mixed plan is developed for both symmetric and asymmetric fraction non conforming. The optimal plan parameters can be determined by considering the satisfaction levels of the producer and the consumer simultaneously at an acceptable quality level and a limiting quality level, respectively. The performance of the proposed plan over the mixed single sampling plan based on C_pk and the mixed variable lot size plan based on C_pk with respect to the average sample number is also investigated. Tables are constructed for easy selection of plan parameters for both symmetric and asymmetric fraction non conforming and real world examples are also given for the illustration and practical implementation of the proposed mixed variable lot-size plan.