THE CHOICE BETWEEN SETS OF REGRESSORS

This paper examines the choice of critical values for testing both non-sequential and nested sequential sets of constraints in the standard linear regression model. Modest increases in (e.g.) t-ratio critical values relative to their one-off values are often sufficient to maintain proper size. A Bayesian decision-theoretic approach, highlighted by the Schwarz (1978) criterion, provides a framework for deriving consistency and asymptotic local power properties of both forms of testing (data mining) algorithms.     

A BAYESIAN INTERPRETATION OF MULTIPLE POINT ESTIMATES

Consider a large number of econometric investigations using different estimation techniques and/or different subsets of all available data to estimate a fixed set of parameters. The resulting empirical distribution of point estimates can be shown - under suitable conditions - to coincide with a Bayesian posterior measure on the parameter space induced by a minimum information procedure. This Bayesian interpretation makes it easier to combine the results of various empirical exercises for statistical decision making. The collection of estimators may be generated by one investigator to ensure the satisfaction of our conditions, or they may be collected from published works, where behavioral assumptions need to be made regarding the dependence structure of econometric studies.     

Dynamic factor models

Summary Factor models can cope with many variables without running into scarce degrees of freedom problems often faced in a regression-based analysis. In this article we review recent work on dynamic factor models that have become popular in macroeconomic policy analysis and forecasting. By means of an empirical application we demonstrate that these models turn out to be usefu in investigating macroeconomic problems.     

B2C电子商务顾客满意度测评研究

文章在参考已有研究成果的基础上,构建了B2C电子商务顾客满意度测评模型及其评价指标体系,并运用模糊综合评判法对某电子商务公司进行实证测评,最后对测评结果进行分析,并提出相应的对策以提升顾客满意度.     

Models for the analysis of C-reactive protein in statin trials

Elevation in C-reactive protein (CRP) is an independent risk factor for cardiovascular disease progression and levels are reduced by treatment with statins. However, on-treatment CRP, given baseline CRP and treatment, is not normally distributed and outliers exist even when transformations are applied. Although classical non-parametric tests address some of these issues, they do not enable straightforward inclusion of covariate information. The aims of this study were to produce a model that improved efficiency and accuracy of analysis of CRP data. Estimation of treatment effects and identification of outliers were addressed using controlled trials of rosuvastatin. The robust statistical technique of MM-estimation was used to fit models to data in the presence of outliers and was compared with least-squares estimation. To develop the model, appropriate transformations of the response and baseline variables were selected. The model was used to investigate how on-treatment CRP related to baseline CRP and estimated treatment effects with rosuvastatin. On comparing least-squares and MM-estimation, MM-estimation was superior to least-squares estimation in that parameter estimates were more efficient and outliers were clearly identified. Relative reductions in CRP were higher at higher baseline CRP levels. There was also evidence of a dose-response relationship between CRP reductions from baseline and rosuvastatin. Several large outliers were identified, although there did not appear to be any relationships between the incidence of outliers and treatments. In conclusion, using robust estimation to model CRP data is superior to least-squares estimation and non-parametric tests in terms of efficiency, outlier identification and the ability to include covariate information.     

Inference of non-centrality parameter of a truncated non-central chi-squared distribution

Non-central chi-squared distribution plays a vital role in statistical testing procedures. Estimation of the non-centrality parameter provides valuable information for the power calculation of the associated test. We are interested in the statistical inference property of the non-centrality parameter estimate based on one observation (usually a summary statistic) from a truncated chi-squared distribution. This work is motivated by the application of the flexible two-stage design in case–control studies, where the sample size needed for the second stage of a two-stage study can be determined adaptively by the results of the first stage. We first study the moment estimate for the truncated distribution and prove its existence, uniqueness, and inadmissibility and convergence properties. We then define a new class of estimates that includes the moment estimate as a special case. Among this class of estimates, we recommend to use one member that outperforms the moment estimate in a wide range of scenarios. We also present two methods for constructing confidence intervals. Simulation studies are conducted to evaluate the performance of the proposed point and interval estimates.     

Unification of Dirichlet Methodology

In this article we provide a unified framework for solving Dirichlet related probability and waiting time problems. We consider a Pólya sampling scheme in which each time an object is selected, it is put back into the population along with c additional objects of the same type. By considering both fixed sample size and inverse sampling procedures, we unify the Dirichlet I, J, C, and D functions with their hypergeometric counterparts by extending these functions to Pólya sampling. We then use these functions to unify and extend the corresponding expected waiting time results.     

A specification strategy for order determination in arma models

In this paper a specification strategy is proposed for the determination of the orders in ARMA models. The strategy is based on two newly defined concepts: the q-conditioned partial auto-regressive function and the p-conditioned partial moving average function. These concepts are similar to the generalized partial autocorrelation function which has been recently suggested for order determination. The main difference is that they are defined and employed in connection with an asymptotically efficient estimation method instead of the rather inefficient generalized Yule-Walker method. The specification is performed by using sequential Wald type tests. In contrast to the traditional testing of hypotheses, these tests use critical values which increase with the sample size at an appropriate rate     

(M,S)-optimal row-column designs

Row-column designs may be considered to have two blocking schemes, namely the treatments by rows and treatments by columns component block designs. The (M,S)-optimality criterion is applied to row-column designs, and che connection between the (M,S)-optimal design and its component block designs is demonstrated.     

On some tests for independence in nonnormal situations: neyman'S C(A)test

A study is made of Neyman's C(a) test for testing independence in nonnormal situations. It is shown that it performs very well both in terms of the level of significance and the powereven for smallvalues of the samplesize. Also, in the case of the bivariate Polsson distribution, itis shown that Fisher's z and Student's t transforms of the sample correlation coefficient are good competitors for Neyman's procedure.