Discussion

This report is the product of the Task Force on State and Local Government Statistics formed by the ASA Committee on Government Statistics. The Task Force began its work in March 1985 because of a growing concern among ASA members that state and local governments were undertaking many statistical programs and functions of which the ASA had little knowledge and for which it had little role in supporting, and because of the potential and real impacts of federal policy changes and budget reductions on state statistical programs. The report examines the need for state and local government statistical policy, reviews current practice, summarizes elements of a statistical policy, proposes alternative models for organizing the statistical policy function, and makes recommendations for further action.     

The use of Andrews curves for detecting the out-of-control variables when a multivariate control chart signals

The aim of this paper is to present a new method for solving the problem of detecting the out-of-control variables when a multivariate control chart signals. The main idea is based on Andrews curves. The proposed method is investigated thoroughly and is proved to have interesting results in comparison to a competing method.     

The Significance of a Square-Root Rule

In this article, I comment on a rough square-root rule for statistical significance which turns out to be equivalent to testing whether an observation lies within one standard deviation of that expected.     

统计分析在校服生产中的应用

 对小学生人体测量数据进行因子分析，得出在校服下装制作时，臀长和股上长是非常重要的采寸部位。通过对小学生年龄段的聚类分析，提出了一种新的校服生产分类方法，对小学生校服的组织生产有着重要的指导意义。     

On composite marginal likelihoods

Composite marginal likelihoods are pseudolikelihoods constructed by compounding marginal densities. In several applications, they are convenient surrogates for the ordinary likelihood when it is too cumbersome or impractical to compute. This paper presents an overview of the topic with emphasis on applications.     

Teaching Survey Sampling

In recent years the focus of research in survey sampling has changed to include a number of nontraditional topics such as nonsampling errors. In addition, the availability of data from large-scale sample surveys, along with computers and software to analyze the data, have changed the tools needed by survey sampling statisticians. It has also resulted in a diverse group of secondary data users who wish to learn how to analyze data from a complex survey. Thus it is time to reassess what we should be teaching students about survey sampling. This article brings together a panel of experts on survey sampling and teaching to discuss their views on what should be taught in survey sampling classes and how it should be taught.     

An unsupervised, ensemble clustering algorithm: A new approach for classification of X-ray sources

A large volume of CCD X-ray spectra is being generated by the Chandra X-ray Observatory (Chandra) and XMM-Newton. Automated spectral analysis and classification methods can aid in sorting, characterizing, and classifying this large volume of CCD X-ray spectra in a non-parametric fashion, complementary to current parametric model fits. We have developed an algorithm that uses multivariate statistical techniques, including an ensemble clustering method, applied for the first time for X-ray spectral classification. The algorithm uses spectral data to group similar discrete sources of X-ray emission by placing the X-ray sources in a three-dimensional spectral sequence and then grouping the ordered sources into clusters based on their spectra. This new method can handle large quantities of data and operate independently of the requirement of spectral source models and a priori knowledge concerning the nature of the sources (i.e., young stars, interacting binaries, active galactic nuclei). We apply the method to Chandra imaging spectroscopy of the young stellar clusters in the Orion Nebula Cluster and the NGC 1333 star formation region.     

Cumulative sum schemes for surgical performance monitoring

Summary.  The standard cumulative sum (CUSUM), risk-adjusted CUSUM and Shiryayev–Roberts schemes for monitoring surgical performance are compared. We find that both CUSUM schemes are comparable in run length performance except when there is a high heterogeneity of surgical risks, in which case the risk-adjusted CUSUM scheme is more sensitive in detecting a shift in surgical performance. The Shiryayev–Roberts scheme is found to be less sensitive compared with the CUSUM schemes in detecting a deterioration in surgical performance. Using the Markov chain method, the exact average run length of a standard CUSUM scheme can be computed whereas the average run length of a risk-adjusted CUSUM scheme is approximated. For a risk-adjusted CUSUM scheme, the accuracy of the average run length depends on the fineness of the discretization of CUSUM values, which relies on the chart limit, shift to be detected optimally and in-control surgical risk distribution. A sensitivity analysis shows that the risk-adjusted CUSUM and Shiryayev–Roberts schemes still perform moderately well in detecting a deterioration and an improvement in surgical performances respectively even though there is a misspecification of the in-control surgical risk distribution. In general, the run length performance of the Shiryayev–Roberts scheme is comparatively less sensitive to a misspecification of the in-control surgical risk distribution.     

Using historical data for Bayesian sample size determination

Summary.  We consider the sample size determination (SSD) problem, which is a basic yet extremely important aspect of experimental design. Specifically, we deal with the Bayesian approach to SSD, which gives researchers the possibility of taking into account pre-experimental information and uncertainty on unknown parameters. At the design stage, this fact offers the advantage of removing or mitigating typical drawbacks of classical methods, which might lead to serious miscalculation of the sample size. In this context, the leading idea is to choose the minimal sample size that guarantees a probabilistic control on the performance of quantities that are derived from the posterior distribution and used for inference on parameters of interest. We are concerned with the use of historical data—i.e. observations from previous similar studies—for SSD. We illustrate how the class of power priors can be fruitfully employed to deal with lack of homogeneity between historical data and observations of the upcoming experiment. This problem, in fact, determines the necessity of discounting prior information and of evaluating the effect of heterogeneity on the optimal sample size. Some of the most popular Bayesian SSD methods are reviewed and their use, in concert with power priors, is illustrated in several medical experimental contexts.     

国家统计机构比较研究

 内容提要：本研究报告在广泛收集和研究国外相关资料的基础上，介绍了国家统计机构的一般组织结构、官方统计基本原则在各国的实施和经验。在此基础上，剖析了官方统计10项基本原则在我国统计工作中的实施情况，通过对照、检查，指出我国统计工作在很多方面已经达到国际通行的官方统计基本原则要求，在一些方面甚至处于国际领先水平。但是，在某些方面与国际标准相比，仍有一定的差距，需要进一步改进、提高。为此，提出了完善我国统计工作的初步建议。