本研究針對Saaty 分析層級程序法（AHP）評比尺度縮減問題進行模擬實驗及實證研究，以探討分析層級程序法尺度縮減對於權重評估結果的可能影響，進而評析尺度縮減的可行性。在研究方法上，本研究僅考慮比九尺度更為精簡的七尺度、五尺度形式，而分別針對九尺度、七尺度、五尺產生模擬實驗之亂數資料及蒐集實證研究之意見資料；資料分析則著重在檢視「九一七」與「九一五」尺度縮減時，權重向量絕對差值、RMS值、MAD值等的變異情形，並進行Spearman等級相關考驗。研究結果發現：分析層級程序法尺度縮減對於權重評估結果具有小幅度之影響：不過，各項差異值大致上均在設定範圍內，變異不大，權重之等級次序亦多維持不變。尺度縮減幅度愈大，其權重評估結果與原九尺度之評估結果差異愈大；同樣的縮減幅度，對於因素數目較少的題組而言，其權重評估結果與原九尺度之評估結果差異較大。此外，在權重值相近的評比項目間，尺度縮減較易生權重排序變動的現象。基於研究發現，本研究認為：在使用分析層級程序法時，可將原九尺度之設計酌情縮減為七尺度或五尺度。若各題組之因素數目較多時，可較大幅地縮減成五尺度：反之，若各題組之因素數目較少時，則宜小幅縮減為七尺度或維持使用九尺度。

The design of a 9─point─scale presents a problem for Saaty’s Analytic Hierarchy Process （AHP）. This study tries to analyze the effect of scale─reduction on the use of AHP by a simulation experiment and empirical research. In this study, we only take account of 9─point─scales, 7─point─scales, and 5─point─scales；and give random data for a simulation experiment and collect responsive data for empirical research in 9─point─scales, 7─point─scales, and 5─point─scales individually. The comparisons focus on the variance of weight─values of “9─7” and “9─5”scale─reductions. We find scale─reduction of AHP will cause some variance in weight─values, but the variance is smaller than 0.1 mostly, the ranking of weight─values also remaining the same mostly. The more the scale is reduced, the more the variance；the greater the factors in weight─values assessment, the less the variance；the closer the weight─values, the greater the possibility of change of ranking of weight─values. Based upon the findings of this study, we can reduce the 9─point─scale to a 7─point─scale or a 5─point─scale when the weight─values assessment has more factors；otherwise we can only reduce it to a 7─point─scale, or still use a 9─point─scale for the assessment of weight─values in AHP.