Sankhya: The Indian Journal of Statistics

1998, Volume 60, Series B, Pt. 1, 82-100

A CLASS OF BIVARIATE HEAVY-TAILED DISTRIBUTIONS

By

HUILING LE and ANTHONY O'HANGAN, University of Nottingham, U.K.
and
ALAN E. GELFAND, University of Connecticut, Storrs

SUMMARY. One-dimensional heavy-tailed distributions have been used for modelling problems in which there may be outlying observations or parameters (c.f. O'Hagan(1990)). In this paper we tackel the corresponding problems in two dimensions, by studying a class of functions analogous to those used in the one-dimensional case. Unlike those in the latter case the functions we require are only partially ordered and this leads to more intricate results.

AMS (1991) subject classification. Primary 62F15, secondary 62F35.

Key words and phrases. Bayesian inference, conflict, domination, heavy-tailed distribution, mode, outlier, robustness, t-distribution

Full paper (PDF)

This article in Mathematical Reviews