【作者】 罗佳；

【导师】 张进辅；

【作者基本信息】 西南大学 ， 基础心理学， 2008， 硕士

【摘要】 随着现代认知心理学的发展,尝试从认知的立场来理解自尊为其测量和评价提供了一个新的理论基础和思路,使自尊研究从意识领域伸展到了无意识领域。这种个体对自身在意识之外的评价被称为内隐自尊。Greenwald和Banaji(1995)将内隐自尊定义为:人们在对与自我相关或无关的客体进行评价时的一种态度表现,而这种态度是无法通过内省的方式被意识到的或被正确识别的。内隐自尊一经提出就受到了研究者的广泛重视并得到了大量实证研究的支持。由于内隐自尊是无意识的,无法通过内省的方式被意识到,传统的自陈量表是无法直接对内隐自尊进行测量的。因此,内隐自尊的测量方法——内隐测验也成为了其实证研究的主要趋势之一。目前,已经有很多内隐测验作为测量工具对内隐自尊进行测量,其中被广泛采用的一种方法是Greenwald等(1998)在反应时范式的基础之上,把实验设计的思想运用于测量之中,以测量个体本身已存在的联想倾向达到测量目的的内隐联想测验(Implicit Association Test,IAT)。IAT测量内隐认知的信效度得到了研究者不同程度的支持与质疑,但专门针对其测量内隐自尊的信效度研究却很少。因此本研究从测量效度的角度出发,对IAT测量内隐自尊的适用性进行了探讨,具体关注的是IAT测量内隐自尊的内部效度,并在此基础上对IAT测量内隐自尊的实际应用提出建设性意见。对IAT内部效度研究的一个主要方面是对其程序当中出现的刺激词进行的研究。IAT的刺激词包括概念词和属性词两类。本研究正是通过不同的程序变式对这两类刺激词对内隐自尊测量结果的影响进行探讨。实验一主要是通过IAT程序属性词的不同设置来探讨属性词对测量结果的影响,结果显示(1)属性词设置对测量结果的影响是复杂的,与“非特定他人”存在交互作用。当概念词中存在“非特定他人”的时候,属性词设置对测量结果的影响不显著,而当概念词中不存在“非特定他人”的时候,属性词设置对测量结果的影响显著;(2)“非词”的效价可能是不稳定的,受到“非特定他人”存在与否的影响,在存在“非特定他人”的情况下,“非词”的效价是消极的;(3)在IAT测量结果中“非特定他人”或是其与属性词设置的交互作用带来了大部分的误差;(4)内、外自尊的低相关主要是由低外显自尊个体内、外自尊的分离所导致的。这些结论都还需要进一步实证研究的支持。因此,实验二主要是探讨概念词设置对内隐自尊测量结果的影响并进一步验证实验一得到的研究结果。通过对概念词的不同设置,实验发现(1)“非特定他人”对测量结果产生影响的方式是复杂的,与属性词的设置有密切联系;(2)进一步证实了“非词”的效价受到“非特定他人”存在与否的影响,当概念词中不存在“非特定他人”时,“非词”的效价是中性的,并在此基础上证明了“非特定他人”的效价是消极的;(3)SC-IAT与IAT测量的不是同一内隐自尊成分,SC-IAT的测量结果可能更符合对“内隐自尊”的定义;(4)同时,进一步证实了当概念词中不存在“非特定他人”的时候,属性词设置对测量结果的影响显著;(5)进一步证实了低外显自尊组在更大的程度上影响着内、外自尊的关系。因此,结合实验一与实验二的结果,就刺激词对内隐自尊测量结果的影响而言,本研究的主要结论是(1)“非特定他人”的效价是消极的,因此,当概念词中含有“非特定他人”的时候,测量的结果大于真实值;(2)当概念词中含有“非特定他人”的时候,属性词的设置即使采用非两极设置(积极-非词),其性质也将被两极化(“非词”的效价变为消极的从而与“积极属性词”成对立两极),因此对结果的影响,概念词设置的作用大于属性词设置的作用;(3)概念词设置与属性词设置这两个因素的交互作用所带来的测量误差显著的使测量值大于真实值。因此,在使用IAT测量内隐自尊时,研究者对刺激词的作用应更加谨慎,而要避免这一误差的影响,本研究显示SC-IAT2能够较好地满足这一目的,即采用概念词中不包括“非特定他人”,属性词设置为非两极的测验程序测得的结果可能更接近个体的真实内隐自尊水平。更多还原

【Abstract】 With the development of Cognitive Psychology , the researches of self-esteem explored into the domain of nonconsciusness, which was called implicit self-esteem. Greenwald and Banaji (1995) defined it as automatic, overlearned, and nonconscious evaluation of the self that guides spontaneous reactions to self-relevant stimuli.Implicit self-esteem was highlighted in psychological research, and because it can’t be tested directly by traditional scale, the measurement of implicit self-esteem, implicit tests, became one of the main research trend.Nowadays, there were many indirect tests measuring implicit self-esteem, one of which was proposed by Greenwald et.al (1998), which was the Implicit Association Test (IAT). But since the IAT born, there had been many researches focusing on the reliability and validity of IAT, some of which questioned the IAT’s reliability and validity, some of which demonstrated that the reliability and validity were acceptable. Eventhough, the use of IAT has become relatively common in measuring unconscious attitude. But in the self-esteem domain, there was few research focusing on the validity of IAT. So this research intended to find if the IAT was suitable to measure implicit self-esteem from the perspective of validity, and it specially focused on the internal validity, based on which, it would propose useful suggestions in the practice of IAT measuring the implicit self-esteem.One of the main research aspects focuing on the internal validity of IAT was the research about stimuli in the procedure, including concept words and attrubitive words. This research intended to find out how this two kinds of words affect the results of IAT. Study 1 focused on how different arrangement of attributive words affect the results, and it showed (1) the affection of the arrangement of attributive words was complicated, which was interacted with "unspecificed other". When there existed "unspecificed other" in concept words, the arrangement of attributive words didn’t affect the result significantly, while it indeed affected significantly when there wasn’t "unspecificed other"; (2) the valence of "nonwords" in attributive words was not stable, which was affected by "unspecificed other", when there existed "unspecificed other" in concept words, it showed negative; (3) "unspecificed other" or the interaction with the attributive words brought main error in the result of IAT; (4) No clear relation was found between implicit-explicit self-esteem, which was reasonably believed to be caused by the discrepancy of implicit-explicit self-esteem of subjects who had low explicit self-esteem. All those findings needed further confirmation, so study 2 mainly focused on the affection of the different arrangement of concept words and repeated the results of study 1. The results showed (1) the affection of the arrangement of concept words was complicated, which was interacted with attributive words; (2) further confirmed that the valence of "nonwords" was affected by "unspecificed other", when there was no "unspecificed other" , it appeared neutral, based on that, it was believed the valence of "unspecificed other" was negative; (3) SC-IAT and IAT may not measure the same thing about implicit self-esteem, but the result of SC-IAT seemed more suitable to the definition of "implicit self-esteem"; (4) Meanwhile, further confirmation about how the different arrangement of attributive words affect the result; (5) further confirmation about subjects who had low explicit self-esteem had much more affection about the relationship between implicit self-esteem and explicit self-esteem.Therefore, focusing on the affection of stimuli combining the results of study 1 and study 2, it reasonably suggested that (1) the valence of "unspecificed other" was negative, which resulted in an IAT score was much higher than real level; (2) when there existed "unspecifice other" in concept words, the different arrangement of attributve words made no difference, so it was safe to say the influence of the arrangement of concept words was stronger than which of the arrangement of attributive words; (3) the interaciton between concept words and attributive words was responsible for the error of IAT, it made the test scores higher than they should be. So it’s necessary to be more cautious about the role of stimuli in IAT when measuring implicit self-esteem. And in order to avoid such error, this study suggested SC-IAT2 can be available, which charactered by no "unspecificed other " in concept words and nonbipolar arrangement of attributive words. The results of SC-IAT2 possibly approached to the real level of implicit self-esteem.更多还原