## Abstract

Assume independent random samples are drawn from K populations whose distributions are location, scale, or location-scale families. Let T_{1} be an estimator which is admissible for the parameter corresponding to the first population. Next assume that the parameters are ordered. The question addressed is does T_{1} remain admissible? For various special cases of the model we exhibit estimators which are better than T_{1}. The models include estimating a location parameter with squared error loss, estimating a scale parameter with normalized squared error loss, finding a confidence interval for a location parameter, and finding a confidence interval for a normal variance. In these latter models universal domination as defined by Hwang [4] is shown.

Original language | English (US) |
---|---|

Pages (from-to) | 201-214 |

Number of pages | 14 |

Journal | Statistics and Risk Modeling |

Volume | 7 |

Issue number | 3 |

DOIs | |

State | Published - Mar 1989 |

Externally published | Yes |

## All Science Journal Classification (ASJC) codes

- Statistics and Probability
- Modeling and Simulation
- Statistics, Probability and Uncertainty

## Keywords

- Location parameters
- admissibility
- confidence intervals
- ordered parameters
- scale parameters
- universal domination