The ‘Tail Method’ for Constructing Frequentist Confidence Intervals that Utilise Uncertain Prior Information

La Trobe Statistics and Stochastic Colloquium talk

Seminar Abstract:

Bayesian credible intervals that utilise uncertain prior information can be constructed using an appropriate prior distribution.

What is not widely realised is that it is also possible to construct frequentist confidence intervals that utilise uncertain prior information.

The ‘tail method’ for constructing such confidence intervals has been around for quite some time. It is easy to understand and implement. Yet it was very limited in scope until a new development published in Biometrika in 2018. We describe both the original ‘tail method’ and this new development.

Host University: La Trobe University

Seminar Convener: Assoc. Prof. Paul Kabaila (La Trobe University)

How to participate in this seminar

1. Book your University’s Access Grid Room, or a university or APAC etc. Access Grid Room that you are able to use, and

2. Send an email to the seminar convener at La Trobe University (Paul Kabaila) to advise that you will be attending the seminar.