心隨境轉？情境式問卷題目測量軍事人員貪腐容忍度之研究

Mind to Heart? Hypothetical Situation Questions and Measurement of the Tolerance of Corruption among Military Personnel in Taiwan

本文認為要有效測量「貪腐容忍」，理應考量社會脈絡因素及受訪者的背景因素，透過特定情境的錨定效果，使其有一致的回應基礎。本研究使用一項有關「國軍對國防廉潔程度認知」的調查資料，比較受訪者在兩種不同問法下，如何判斷或評價特定概念所具有之意義。我們以「檢舉貪腐意願」為預測效標，評估這兩種問法的效標關聯效度，試圖瞭解受訪者回答問題時，是否因提供共同的判斷基準，而改善測量效度。
研究結果顯示，在3種有關國軍貪腐的情境中，受訪者對「公款公用」和「送禮風氣」的認知和回應模式較一致，其比例較為接近。在貪腐整體容忍度的測量中，受訪者在「抽象式問法」的回應模式高度集中在0（即零容忍）的類別，其他類別的回應比例偏低；「情境式問法」的變異性相對較高，受訪者較能針對不同情境作出客觀判斷，且兩種問法的相關係數並不高。在個人背景因素方面，不同性別受訪者在不同問法或不同情境的回應模式較接近；年齡層愈高的受訪者，在兩種問法的平均數差距愈小，回應模式愈趨一致；不同學歷之受訪者的回應模式一致性，不如不同年齡層的受訪者來得明顯；而不同階級之受訪者的回應模式較不一致。若從效標關聯效度的觀點，貪腐容忍度的「情境式問法」對檢舉貪腐意願的解釋力略高於「抽象式問法」。在回應一致性方面，年齡層、教育程度、階級及服務年資愈高的受訪者，其「回應一致」的比例愈高。

This study believes that in order to effectively measure“corruption tolerance”, social context factors and background factors of the interviewees should be taken into consideration. We use data from a survey on Taiwan’s“military personnel’s perception of national defense integrity”to compare how the interviewees judge or evaluate the properties of certain concepts under two different forms of questioning, namely,“abstract descriptive”and“simulated situational”. By using“willingness to report corruption”as the prediction criterion to evaluate the criterion-related validity of these two different forms of questions, this study tries to determine whether the interviewees were answering the questions because they had common benchmarks for judgment so as to improve the validity of measurements.
The research results show that among the three scenarios related to military corruption, the respondents’perceptions and response patterns for the two scenarios of“public use of public funds”and“gift-giving culture”are relatively consistent, and the percentages of each option are relatively close. In the measurement of the overall tolerance of corruption, the response pattern of the respondents in the“abstract question”, which is highly concentrated in the category of 0 (i.e. zero tolerance), the proportion of responses in the other categories is relatively low. The responses to the“situational question”, by contrast, are characterized by high variability, and the respondents are better able to make objective judgments on different situations. The correlation coefficient of the scores obtained by the two questioning methods is not high. In terms of the relationship between the demographic factors and different questioning methods, the response patterns of respondents of different genders in different questioning methods or different situations are relatively close; The higher the age group of the respondents, the smaller the difference between the averages of the two questioning methods, that is, the more consistent the response patterns are. The consistency of the response patterns of respondents with different educational levels is not as obvious as that of respondents of different age groups; The response patterns of respondents from different military ranks were less consistent. As far as criterion-related validity is concerned, the“situational question”of corruption tolerance has slightly higher explanatory power than the“abstract question”. In terms of response consistency, the older the age group, the higher the education level, the higher the military rank and the more years of service, the higher the proportion of“consistent responses”.