Abstract

The classical method of analysis of a series of experiments is somewhat
involved in being conditional on various, occasionally unrealistic, assumptions
such as homogeneity of variances of experimental error, lack of
interactions of treatments and places,etc. In this work, we adopt a Bayesian
view to account for such heterogeneities. Our appoach is illustrated by a real
series of experiments regarding the effect of fertilizers on crops of sugarbeet
in various places of Iran. Conjugate priors are considered for parameters of
the model and posteriors are obtained. Using iniormation from different
places, the unknown priors are estimated and the empirical Bayes procedure
is employed