本研究以Greenwald(1998，2000)所發展出來的「內隱聯結測驗」(implicit association test, IAT)來測量國人的自尊。「內隱聯結測驗」測量自尊的邏輯為;若個體的自與愈高，則其知識結構中，與自我相關的概念(例如:姓名、生日、出生地等)與正向概念的聯結也就愈強。因此，受測者將自我相關的概念詞彙與正向概念詞彙歸為一類的判斷作業之反應時間，將會顯著地快於將自我相關的概念詞彙與負向詞彙歸為一類的判斷作之反應時間，職是之故，研究者可以由後者與前者的平均反應時間之差來界定一個受測者的內隱自尊。由於內隱測量正使用隱藏間接的方式來測驗個體對自我的態度，故相對地較能避免研究參與者為了印象整飾而有意或無意的回答出社會期望的反應。本研究的預備性研究從國語辭典簡編本編輯資料字詞頻統計報告(1997)中挑選四百個具有明顯情緒性的正/負向事物(如:真愛或強暴)或正/負向特質(如:智慧或虛偽)，並透過101位受試者的喜好度評定選出極端正/負向的詞彙各26個，以做為正式研究中之內隱聯結作業使用。正式研究使用Greenwald(2000)的IAT自尊測量程序，由79位受測者同時接受IAT以及其他四個外顯自尊量表的測量。結果發現，國人與美國的受測者一樣，普遍具有正向的內隱自尊(自我概念與正向概念的聯結強過自我概念與負向概念的聯結)。此外，IAT內隱自尊與外顯自尊量表的得分間並無任何顯著關，這顯示兩者測量到不同的建構。有趣的是，自我欺驗及印象整飾量表與外顯自尊量表的得分有顯著的相關，但與內隱自尊量表的得分則無顯著相關，支持了內隱自尊測量能避免自我欺驗及印象整飾的假設。總之，本研究除了為國內IAT內隱測量建立適合國人的標準材料及程序外，也支持了其測量的效度，故IAT內隱測量應是未來深具潛力的新測量工具。

　　　　　 This article uses the Implicit Association Test (IAT, Greenwald, 1998, 2000) to measure the self-esteem of Chinese people. Because the IAT uses participants' reaction tie of judgment tasks to assess their self-esteem, the Social Desirable Responses and Impression management which often occurs in self report test will be eliminated as much as possible. In the pilot study, we collect 400 two-character, high frequency Chinese words which express strong emotional concepts (such as ‘ture love’真愛 and‘rape’強暴) or personal traits (such as’wisdom’智慧 and‘hypocrisy’虛偽) form the Chinese dictionary and corpus After rating of pleasantness by 101 college participants, 26 extremely positive and 26 extremely negative Chinese words were selected for the IAT. In the formal study, 79 Chinese undergraduate students at the National Taiwan Normal University accepted the self-esteem IAT, followed by four other kinds of self-report self-esteem inventories (the Rosenberg Self-esteem Scale, the Feeling of Inferiority Scale, the thermometer scale of self-esteem, the Semantic Differential Scale of Five Self-Esteem Items) and The Balanced Index of Desirable Responding (BIDR, Paulhus, 1988). The self-esteem IAT measurement results revealed self-positive tendencies similar to those found by Greenwald in western populations, and support the construct validity of IAT. In addition, there is no significant correlations between the self-esteem IAT and any explicit self-esteem measurements, these results show that IAT and self-report test measured totally distinct constructs, and demonstrate strong evidence of discriminative validity of IAT. Lastly, Although Self-deception and Impression Management scores of BIDR were not found to have any significant correlation with self-esteem IAT score, there were found to have significant correlation with each of the self-report self-esteem measurements. The above result supports the hypothesis that IAT can prevent Social Desirable Responses normally occurring in self repot measurement, and assess the true self-esteem of participants.