在進行決策分析時，分析層級程序法是被普遍使用的一種方法，但在應用時，常因人們思考上的限制或資訊取得的困難，使得各層級所列出的決策屬性，在涵意上往往不具互斥的特性，因此，在求算屬性權重估計值時，應將上述因素納入考慮。本文在層級中決策屬性具有相關性的情況下，以常態隨機誤差項來探討成對比較時因屬性相關干擾所造成的誤差，求出具有BLUE性質的屬性權重估計式，並利用分析層級程序中的層級架構，連接各層級的干擾誤差，得出可選擇方案權重的估計值，以及其對應之估計誤差變異數，再配合統計決策法則，選出最適的方案

In decision analysis, the analytic hierarchy process is one of the popular methods that rank orders among a set of alternatives. But men did it under the constraints of incomplete information and the boundedness to think of, therefore, one can hardly exhibit the decision attributes inclusively and exclusively in all levels of the hierarchy. In this paper, we discuss an estimator, with the properties of BLUE, of relative weights under the condition of decision attributes being interdependent. Through the connection of levels in the hierarchy, the estimators and the variances of the estimators of relative weights of a set of alternative are found. With these results, one could choose out the best alternative.