An ancillarity paradox which appears in multiple linear regression
Consider a multiple regression in which \(Y_ i\), \(i=1,...,n\), are independent normal variables with variance \(\sigma^ 2\) and E \(Y_ i=\alpha +V'_ i\beta\) where \(V_ i\in {\mathbb{R}}^ r\) and \(\beta \in {\mathbb{R}}^ r\). Let \({\hat \alpha}\) be the usual least squares estimator of \(\alpha\). If \(V_ i\) is fixed than \({\hat \alpha}\) is admissible under squared error loss. However, if \(V_ i\) are random observations, independently distributed according to a multivariate normal distribution with mean \(\underset{\tilde{}} 0\) and known nonsingular covariance matrix \(\theta\), then (it is shown that) \({\hat \alpha}\) is inadmissible, when \(r\geq 2\). Several estimators dominating \({\hat \sigma}\) when \(r\geq 3\) are given. Analogous results are given for the case when \(\theta\) and \(\sigma^ 2\) are unknown. It is interesting to note that \(\{V_ i\}\) are ancillary statistics in the above setting. Hence the admissibility of \({\hat \alpha}\) depends on the distribution of the ancillary statistics. This fact constradicts the notion, widely held, that statistical inferences should be carried out, conditional on the value of any ancillary statistic. That is, the distribution of the ancillary statistic should be irrelevant to the statistical inference.
- On-line predictive linear regression
- Estimated confidence under ancillary statistic everywhere-valid constraint
- Admissibility of the constant-coverage probability estimator for estimating the coverage function of certain confidence interval
- The emperor's new tests. (With comments and a rejoinder).
- Estimators of slopes in linear errors-in-variables regression models when the predictors have known reliability matrix
- Algebraic algorithms for sampling from conditional distributions
- On estimating domain totals over a subpopulation
- Brown's paradox in the estimated confidence approach
- Dependence of variance on covariate design in nonparametric link regression
- Semi-supervised empirical risk minimization: using unlabeled data to improve prediction
- Larry Brown's contributions to parametric inference, decision theory and foundations: a survey
- Larry Brown's work on admissibility
- Fisher and regression
- Ancillaries and conditional inference (with comments and rejoinder)
- An ancillary paradox in testing
- Directional tests and statistical frames
- Optimal equivariant prediction for high-dimensional linear models with arbitrary predictor covariance
- Critical issues in different inferential paradigms
- Applied regression analysis bibliography update 1990-91
- Construdtion of shrinkage estimators for the regression coefficient matrix in the gmanova model
- Small area estimation with correctly specified linking models
- scientific article; zbMATH DE number 922437 (Why is no real title available?)
- High-dimensional inference in misspecified linear models
- Shrinkage estimators for prediction out-of-sample: conditional performance
- Covariate-assisted ranking and screening for large-scale two-sample inference
- The unconditional distributions of the OLS, TSLS and LIML estimators in a simple structural equations model
- Admissibility of confidence estimators in the regression model
- Ridge regression and asymptotic minimax estimation over spheres of growing dimension
- Semi-Supervised Linear Regression
- Maximal co-ancillarity and maximal co-sufficiency
- The estimation of stratum means vector with random sample sizes
- D. R. Cox: aspects of scientific inference
- Gamma-minimax prediction for the multinomial distribution
- An efficient method of estimating the true value of a population characteristic from its discrepant estimates
This page was built for publication: An ancillarity paradox which appears in multiple linear regression
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2641018)
