×

A Bayesian treatment of multivariate normal data with observations missing at random. (English) Zbl 0664.62026

Statistical decision theory and related topics IV, Pap. 4th Purdue Symp., West Lafayette/Indiana 1986, Vol. 1, 225-233 (1988).
[For the entire collection see Zbl 0638.00031.]
The problem of missing observations in multivariate normal data is important both because of the central role that the normal distribution plays in statistics and because it provides a base from which to test a variety of heuristics for handling missing data. In this paper a Bayesian approach is employed in order to provide a conjugate analysis for the missing data problem. Attention is restricted to the special case in which the data are missing at random, not because this is necessarily the most important case empirically, but because it is a relatively simple, tractable case. It is assumed that the parameters of the process causing the data to be missing are distinct from those of the normal distribution which generates the data.

MSC:

62F15 Bayesian inference
62H10 Multivariate distribution of statistics
62H05 Characterization and structure theory for multivariate probability distributions; copulas
62H99 Multivariate analysis

Citations:

Zbl 0638.00031