Conditional statistical inference and quantification of relevance

We argue that it can be fruitful to take a predictive view on notions such as the precision of a point estimator and the confidence of an interval estimator in frequentist inference. This predictive approach has implications for conditional inference, because it immediately allows a quantification of the concept of relevance for conditional inference. Conditioning on an ancillary statistic makes inference more relevant in this sense, provided that the ancillary is a precision index. Not all ancillary statistics satisfy this demand. We discuss the problem of choice between alternative ancillary statistics. The approach also has implications for the best choice of variance estimator, taking account of correlations with the squared error of estimation itself. The theory is illustrated by numerous examples, many of which are classical. Copyright 2003 Royal Statistical Society.

[1]  George Casella,et al.  Frequentist post-data inference , 1995 .

[2]  G. Upton.  Fisher's Exact Test , 1992 .

[3]  David R. Cox,et al.  Prediction and asymptotics , 1996 .

[4]  R. Fisher.  Statistical methods for research workers , 1927, Protoplasma.

[5]  William G. Cochran,et al.  Experimental Designs, 2nd Edition , 1950 .

[6]  I. Helland.  Simple Counterexamples against the Conditionality Principle , 1995 .

[7]  T. Rao,et al.  Tensor Methods in Statistics , 1989 .

[8]  T. Severini.  Local ancillarity in the presence of a nuisance parameter , 1993 .

[9]  A. Ledford,et al.  Finite Sample Inference for Extreme Value Distributions , 1997 .

[10]  Peter McCullagh,et al.  Conditional inference and Cauchy models , 1992 .

[11]  Nancy Reid,et al.  The Roles of Conditioning in Inference , 1995 .

[12]  Recovery of Ancillary Information , 1988 .

[13]  G. Hillier,et al.  Ill-posed Problems and Instruments' Weakness , 2004 .

[14]  G. K. Robinson.  Conditional Properties of Statistical Procedures , 1979 .

[15]  John D. Kalbfleisch,et al.  Sufficiency and conditionality , 1975 .

[16]  R. Sundberg.  Precision estimation in sample survey inference: A criterion for choice between variance estimators , 1994 .

[17]  On Cox's Criterion for Discriminating between Alternative Ancillary Statistics , 1983 .

[18]  N. Reid,et al.  Information, ancillarity, and sufficiency in the presence of nuisance parameters† , 1994 .

[19]  David R. Cox,et al.  The Choice between Alternative Ancillary Statistics , 1971 .

[20]  Ronald Aylmer Sir Fisher,et al.  Statistical Methods, Experimental Design, and Scientific Inference , 1990 .

[21]  Else Sandved.  Ancillary Statistics and Estimation of the Loss in Estimation Problems , 1968 .

[22]  G. Upton.  A Comparison of Alternative Tests for the 2 Times 2 Comparative Trial , 1982 .

[23]  R. Buehler.  Some Ancillary Statistics and Their Properties , 1982 .

[24]  D. Hinkley.  Likelihood as approximate pivotal distribution , 1980 .

[25]  B. Lindsay,et al.  On second-order optimality of the observed Fisher information , 1997 .

[26]  R Sundberg,et al.  Comparison of Confidence Procedures for Type I Censored Exponential Lifetimes , 2001, Lifetime data analysis.

[27]  B. Efron,et al.  Assessing the accuracy of the maximum likelihood estimator: Observed versus expected Fisher information , 1978 .

[28]  A. Dawid.  Fisherian Inference in Likelihood and Prequential Frames of Reference , 1991 .

[29]  R. Fisher,et al.  The Logic of Inductive Inference , 1935 .

[30]  F. Yates,et al.  Tests of Significance for 2 × 2 Contingency Tables , 1984 .

[31]  D. Cox,et al.  Inference and Asymptotics , 1994 .

[32]  J. F. C. Kingman,et al.  Information and Exponential Families in Statistical Theory , 1980 .

[33]  R. Sundberg.  Ancillarity and conditional inference for ML and interval estimates in some classical genetic linkage trials , 2001 .

[34]  Trevor J. Sweeting,et al.  Discussion of A. P. Dawid: Fisherian inference in likelihood and prequential frames of reference , 1991 .