Anzeige
Erschienen in:
2019 | OriginalPaper | Buchkapitel
Exact Model Averaged Tail Area Confidence Intervals
verfasst von : Paul Kabaila, Rheanna Mainzer
Erschienen in: Statistics and Data Science
Verlag: Springer Singapore
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
Abstract
Properties of the model averaged tail area (MATA) confidence interval proposed by Turek and Fletcher (CSDA 2012) depend critically on the data-based weights assigned to each tail area equation. By restricting attention to weights based on exponentiating minus AIC/2 and other similar weights it is not possible to find a MATA confidence interval with the desired minimum coverage probability. In the simple scenario that there are two nested normal linear regression models over which we average, a weight function is proposed that results in a MATA interval with correct minimum coverage for many combinations of the known quantity that the coverage depends on. This weight function is shown to outperform current popular choices of weight functions for the MATA interval.