在決策分析領域，分析層級程序法常被用於多評估準則中，以排序可選擇方案，做為決策決定的參考。當資料是以問卷調查方式取得時，如何將多份問卷資料彙總，是一值得探討的問題。所以，本文探討分析層級程序法中以問卷調查取得成對比較矩陣結果時，透過隨機變數來處理成對比較所隱含的不確定性，以統計分析方法來求取決策屬性相對權重估計式，並建立檢定統計量來檢測資料的一致性程度。最後，再經由分析層級架構，串聯各層級所求得的決策屬性相對權重，求出各可選擇方案的相對權重用於排序各可選擇方案，或決定各可使用資源的配置比例，以做為決策的參考。

　　　　　In decision analysis, the Analytic Hierarchy Process Z(AHP) is one of the popular methods that rank orders among a set of alternatives. When we collect the data by questionnaire, the key problem is how to summarize the data of all questionnaires. In this paper, ooit proposes that the variation of responses may be viewed as random variable. We discuss an estimator of relative weights and its consistency testing of the pairwise comparison matrix collected by questionnaire in AHP. Through properties of good estimator, we find the estimator and the hypothesis testing of consistency in a set of alternatives. By the way, through the connection of the hierarchy, we can find the priorities of alternatives. With these results, one could rank a set of alternatives in the order of priorities.