摘要:

Generally speaking, laws against discrimination prohibit treating similarly situated persons differently. Legal analysis or social policy supplies the criteria for ascertaining which persons are similarly situated. Statistical evidence has been used in cases, among others, alleging discrimination in criminal prosecutions, educational opportunities, jury selection, and employment practices. The United States Supreme Court has been ambivalent about the need for significance testing in such cases, and it has yet to consider carefully the use of formal inferential methods. Various techniques for conveying information to a court about the inferential value of sample statistics are illustrated in the context of two jury-selection discrimination cases. The role of statistical evidence in certain employment discrimination cases is also considered. It is suggested that the classical method of hypothesis testing used by the Supreme Court is not appropriate to testing whether a given defendant discriminated. Presentation ofpvalues, prediction or confidence intervals, and likelihood functions are shown to be preferable. Bayesian methods are also considered.

ISSN:0162-1459    

DOI:10.1080/01621459.1982.10477885

出版商:Taylor & Francis Group    

年代:1982

数据来源: Taylor

ISSN:0162-1459    

DOI:10.1080/01621459.1982.10477886

出版商:Taylor & Francis Group    

年代:1982

数据来源: Taylor

ISSN:0162-1459    

DOI:10.1080/01621459.1982.10477887

出版商:Taylor & Francis Group    

年代:1982

数据来源: Taylor

ISSN:0162-1459    

DOI:10.1080/01621459.1982.10477888

出版商:Taylor & Francis Group    

年代:1982

数据来源: Taylor

ISSN:0162-1459    

DOI:10.1080/01621459.1982.10477889

出版商:Taylor & Francis Group    

年代:1982

数据来源: Taylor

摘要:

Thestrengthof a significance test is defined as a weighted average of the power especially if the weights are a prior density. The concept is a compromise between Neyman-Pearsonian and Bayesian methods and is applied to tests for multinomial equiprobability and for no association in contingency tables. The differences of the power functions of two criteria,X2andG, are illustrated for the trinomial, whereGarose earlier from a partially Bayesian approach. Other methods for apprehending power functions of several variables are discussed. For 2 × 2 contingency tables with equal row or column totals the power and strength functions of several statistics are equal.

ISSN:0162-1459    

DOI:10.1080/01621459.1982.10477890

出版商:Taylor & Francis Group    

年代:1982

数据来源: Taylor

摘要:

Goodman recently presented a class of models for the analysis of association between two discrete, ordinal variables. The association was measured in terms of the odds ratios in 2 × 2 subtables formed from adjacent rows and adjacent columns of the cross-classification, and models were devised that allowed the odds ratios to depend on an overall effect, on row effects, on column effects, and on other effects. This article presents some generalizations of this approach appropriate for multiway cross-classifications, including (a) models for the analysis of conditional association, (b) models for the analysis of partial association, and (c) models for the analysis of symmetric association. Three cross-classifications are analyzed with these models and methods, and rather simple interpretations of the association in each are provided.

ISSN:0162-1459    

DOI:10.1080/01621459.1982.10477891

出版商:Taylor & Francis Group    

年代:1982

数据来源: Taylor

摘要:

Binary datad1,d2, …,dnare assumed to be generated by an underlying real-valued, strictly stationary process, {Xk}, and a response functionF.For a given monotone nondecreasing functionFfromRto [0, 1],Dktakes on 1 with probabilityF(xk) and 0 with probability 1 -F(xk), whereXk= xk.It is shown that all strictly stationary binary processes are characterized by such a procedure. Several approximations to then-dimensional joint probabilities ofDkare developed whenXkis a Gaussian first-order autoregressive process. Model-building procedures and methods by which to estimate parameters of a given model are discussed. The predictor ofdn+ 1that minimizes probability of error among all randomized rules is determined and for certain cases a bound for this probability is found.

ISSN:0162-1459    

DOI:10.1080/01621459.1982.10477892

出版商:Taylor & Francis Group    

年代:1982

数据来源: Taylor

摘要:

Jeffrey's rule for revising a probabilityPto a new probabilityP* based on new probabilitiesP* (Ei) on a partition {Ei}i= 1nis

ISSN:0162-1459    

DOI:10.1080/01621459.1982.10477893

出版商:Taylor & Francis Group    

年代:1982

数据来源: Taylor

摘要:

An example is given of a family of distributions on [— 1, 1] with a continuous one-dimensional parameterization that joins the triangular distribution (when Θ = 0) to the uniform (when Θ = 1), for which the maximum likelihood estimates exist and converge strongly to Θ = 1 as the sample size tends to infinity, whatever be the true value of the parameter. A modification that satisfies Cramér's conditions is also given.

ISSN:0162-1459    

DOI:10.1080/01621459.1982.10477894

出版商:Taylor & Francis Group    

年代:1982

数据来源: Taylor