ISSN:0003-1305    

DOI:10.1080/00031305.1996.10473553

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor

摘要:

Neural networks have received a great deal of attention over the last few years. They are being used in the areas of prediction and classification, areas where regression models and other related statistical techniques have traditionally been used. In this paper we discuss neural networks and compare them to regression models. We start by exploring the history of neural networks. This includes a review of relevant literature on the topic of neural networks. Neural network nomenclature is then introduced, and the backpropagation algorithm, the most widely used learning algorithm, is derived and explained in detail. A comparison between regression analysis and neural networks in terms of notation and implementation is conducted to aid the reader in understanding neural networks. We compare the performance of regression analysis with that of neural networks on two simulated examples and one example on a large dataset. We show that neural networks act as a type of nonparametric regression model, enabling us to model complex functional forms. We discuss when it is advantageous to use this type of model in place of a parametric regression model, as well as some of the difficulties in implementation.

ISSN:0003-1305    

DOI:10.1080/00031305.1996.10473554

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor

摘要:

In meta-analysis, independently executed historical studies are viewed as imperfect replications of one overall but unplanned experiment. Using empirical parameter estimates of those studies as a unit of analysis, a linear model is specified to uncover systematic variation as a function of study design, data characteristics, model specification, etc. Because of the fallible nature of the historical studies, however, the vector of parameter estimates reflects cross-sectional estimation biases as well as inefficiencies. The outcome of many meta-analyses, therefore, is largely a description of those systematic biases and inefficiencies, a result with little intrinsic scientific value given the fundamental meta-analysis objective of improving understanding of the studied phenomenon. Unfortunately, estimating more scientifically interesting effects on the basis of systematic variation in historical parameter estimates is conditional on the fallible character of the historical studies. This paper provides clarity on what is being estimated in meta-analyses and on how scientifically interesting effects can be estimated unbiasedly using response surface extrapolation in a least squares framework.

ISSN:0003-1305    

DOI:10.1080/00031305.1996.10473555

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor

摘要:

This article exemplifies how to recast well-known statistical tests in terms of more general statistical models. The example used is the Bowker test, a popular test for assessing axial symmetry in square cross-tabulations. Researchers use this test to assess patterns of change in categorical variables across two occasions. This article reviews how the Bowker test can be equivalently expressed in terms of log-linear models with side constraints that specify (1) those pairs of cells that are supposed to be symmetrical with regard to the main diagonal and (2) that the main diagonal cells remain untouched when estimating expected cell frequencies. This article proposes expressing the model of axial symmetry through vectors in the design matrix. In addition, this article recasts quasi-symmetry models in terms of nonstandard log-linear models. One benefit from this new formulation is that estimation of parameters and expected cell frequencies can be performed without unfolding the cross-tabulation. The article also introduces multiple group models. In addition, it is shown how shift patterns can be made part of models. Benefits from recasting well-known tests in terms of more general statistical models include that parameters can be interpreted and that tests can be parts of more elaborate research designs.

ISSN:0003-1305    

DOI:10.1080/00031305.1996.10473556

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor

摘要:

Correlated binary data are frequently analyzed in studies of repeated measurements, reliability analysis, and others. In such studies correlations among binary variables are usually nonnegative. This article provides a simple algorithm for generating an arbitrary dimensional random vector of non-negatively correlated binary variables. In some frequently encountered situations the algorithm reduces to explicit expressions. The correlated binary variables are generated from correlated Poisson variables. The key idea lies in the property that any Poisson random variable can be expressed as a convolution of other independent Poisson random variables. The binary variables have desired correlations by sharing common independent Poisson variables.

ISSN:0003-1305    

DOI:10.1080/00031305.1996.10473557

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor

摘要:

Recent research has focused on studying the patterns in the digits of closely followed stock market indexes. In this paper we find that the series of 1-day returns on the Dow-Jones Industrial Average Index (DJIA) and the Standard and Poor's Index (S&P) reasonably agrees with Benford's law and therefore belongs to the family of anomalous or outlaw numbers.

ISSN:0003-1305    

DOI:10.1080/00031305.1996.10473558

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor

摘要:

In this article the problem of comparing two independent binomial populations is considered. It is shown that the test based on the confidence intervalpvalue of Berger and Boos often is uniformly more powerful than the standard unconditional test. This test also requires less computational time.

ISSN:0003-1305    

DOI:10.1080/00031305.1996.10473559

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor

摘要:

This article gives a simple and exact method of constructing one-sided tolerance intervals for the case of one-way analysis of variance (ANOVA) with random effects when the variance ratio is known. Because the method developed does not involve any approximations and it retains the maximum possible degrees of freedom, it is superior to any existing methods. Further, the method is applicable for both unbalanced and balanced designs. When the variance ratio is unknown, we suggest an approximate unbiased estimator to be used. Simulations are conducted to indicate the loss in precision with respect to the coverage and the confidence. An example is given to illustrate the methodology.

ISSN:0003-1305    

DOI:10.1080/00031305.1996.10473560

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor

摘要:

A class of statistics is introduced for testing stochastic ordering between two independent distributions. This class includes as a special case the celebrated Mann—Whitney—Wilcoxon statistic. The new class is shown to be asymptotically normal both under the null and nonnull hypotheses. It is distribution-free. Using Pitman's asymptotic efficacy it is shown that for some alternatives the Mann—Whitney—Wilcoxon statistic is the member with the highest efficacy, although for others it is not, and the member with the highest efficacy is identified.

ISSN:0003-1305    

DOI:10.1080/00031305.1996.10473561

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor

摘要:

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.

ISSN:0003-1305    

DOI:10.1080/00031305.1996.10473562

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor