The construction of appropriate estimators is complicated when data are sampled according to complex rules, or when multiple measurements with different inherent biases are available. This research will develop estimation procedures for certain of these cases. Then their statistical properties will be investigated when sampling is multistage, or when estimates are obtained by combining biased and unbiased estimators. Constructions of admissible estimators for spherically symmetric families will also be sought using extensions of James-Stein procedures.
