Performance of Model Averaged Tail Area Confidence Interval

Abstract:

Every model has an uncertainty in the variables that it should include. Model averaging is considered as a promising method that could be used to perform inference in the presence of model uncertainty. The performance of this method heavily depends on the data-based model weights used. Traditionally, this weight is chosen to be proportional to the exponential of minus the Generalized Information Criterion divided by two. We observe that the model-based confidence interval performs better, in terms of coverage and expected length, in the case of two nested linear regression models when this division by two is replaced by multiplied by a positive tuning constant. In the second part of the talk, we extend the analysis of the performance of Model Averaged Tail Area confidence interval by Kabaila, Welsh and Abeysekara, Scandinavian Journal of Statistics, 2016, to the case of three or more nested linear regression models. We also assess the influence of the weight function on the performance of this confidence interval for three nested linear regression models applied to the ‘Cholesterol’ data set.

Join from a PC, Mac, iOS or Android: https://latrobe.zoom.us/j/98357628534

Or iPhone one-tap (Australia Toll):  +61280152088,98357628534#

Or Telephone:

Dial: +61 2 8015 2088

Meeting ID: 983 5762 8534

International numbers available: https://latrobe.zoom.us/u/acAxlLgAVI

Or a H.323/SIP room system:

Dial: 98357628534@zoom.aarnet.edu.au

or SIP:98357628534@zmau.us

or 103.122.166.55

Meeting ID: 98357628534

Or Skype for Business (Lync):

SIP:98357628534@lync.zoom.us