摘要:

Suppose we are givennindependent observations (X1,Y1), …, (Xn, Yn) from a conditionally specified distribution with densityf(x, y). The problem of estimating the unknown parameters off(x, y) is complicated by the presence of an intractable normalizing constant that, in this conditional approach, is chosen so that the density integrates to 1. An approach to estimation is used here that is known to result in asymptotically efficient estimators of the unknown parameters whenf(x, y) is from an exponential family. It is an application of a method that has appeared in the literature and is due to J. K. Lindsey. It very conveniently avoids the dependence on the normalizing constants in the joint distribution. The usual maximum-likelihood estimates of the parameters can be obtained using software readily available for Poisson regression. The usefulness of this estimation method is illustrated for a model resulting from specifying that the conditionals are two-parameter, shape and scale, gamma. It is assumed that only the scale parameter depends on the conditioning variable. This model subsumes the BEC class and several characteristics of the model extend those of the BEC class.

ISSN:0003-1305    

DOI:10.1080/00031305.1994.10476078

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor

摘要:

A correlation matrix between three variables has to satisfy certain conditions. Such a matrix essentially contains three numbers and thus can be represented by a point in three dimensions. The set of all possible correlation matrices yields a convex solid body with an uncommon shape. All its cross sections perpendicular to the axes are ellipses. At the same time, its surface contains the vertices and edges of a regular tetrahedron. Another unusual shape is obtained for banded correlation matrices between four variables.

ISSN:0003-1305    

DOI:10.1080/00031305.1994.10476079

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor

摘要:

Biostatisticians have become an integral part of medical research for the design and analysis of laboratory, epidemiology, and clinical studies. The demand for this expertise has increased over the past several decades, while the supply has remained relatively constant. The skills required to be a successful statistical collaborator go beyond the traditional statistics courses taught in graduate school. We describe the challenges facing those training the next generation of biostatisticians.

ISSN:0003-1305    

DOI:10.1080/00031305.1994.10476080

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor

摘要:

This article is a follow-up to a report of a 1984 conference on statistical education for engineers. We agree with most of that report but recognize that the course outlined there contains too much material. Accordingly, we modify those recommendations by reducing the number of topics considered. Suggestions for improving course presentations are also made.

ISSN:0003-1305    

DOI:10.1080/00031305.1994.10476081

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor

摘要:

The extension of classical linear models to generalized linear models has had an important unifying impact on the exposition, teaching, and practice of statistical modeling. This article gives a new and simple justification of the commonly used iteratively reweighted least squares procedure for obtaining maximum likelihood parameter estimates.

ISSN:0003-1305    

DOI:10.1080/00031305.1994.10476082

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor

摘要:

Simulation is simpler intellectually than the formulaic method because it does not require that one calculate the number of points in the entire sample space and the number of points in some subset. Instead, one directly samples the ratio. This article presents probabilistic problems that confound even skilled statisticians when attacking the problems deductively, yet are easy to handle correctly, and become clear intuitively, with physical simulation. This analogy demonstrates the usefulness of simulation in the form of resampling methods.

ISSN:0003-1305    

DOI:10.1080/00031305.1994.10476083

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor

摘要:

The probabilityP(λ, μ) that a team with mean score λ beats a team with mean score μ is calculated when the score of each team is Poisson distributed. It is found that ∂P(λ, μ)/∂λ is equal to the probability of a tie. When this equality holds for any distribution of the score of the team with mean μ, it is shown that the score of the team with mean λ must be Poisson distributed. The Poisson distribution is shown to fit certain baseball data, and it is also applied to some soccer data.

ISSN:0003-1305    

DOI:10.1080/00031305.1994.10476084

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor

摘要:

Longitudinal studies play a prominent role in biomedical and sociological research. Generalized estimating equations (GEE) provide a regression methodology to analyze the correlated data that often result from a longitudinal study. Many applied researchers are attracted to the informative and valid analyses GEE provides but cannot clear the hurdle of understanding the literature. This article places GEE in more familiar territory by building a link from the well-known least squares regression methodology.

ISSN:0003-1305    

DOI:10.1080/00031305.1994.10476085

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor

摘要:

Although sunflower plots are highly effective for displaying bivariate data with coincident observations, they possess certain disadvantages involving graphical perception of data. Moreover, sunflower plots for data that arrive “on line” can be updated only by completely redrawing the affected sunflowers. We propose a variation of the traditional sunflower plot that addresses these issues.

ISSN:0003-1305    

DOI:10.1080/00031305.1994.10476086

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor

ISSN:0003-1305    

DOI:10.1080/00031305.1994.10476087

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor