Rosenberg自尊量表中的方法因素及其相关物

Method Effects in Rosenberg Self Esteem Scale and Their Correlates

摘要: Rosenberg自尊量表的因素结构仍然是学术界争论的焦点问题。本研究拟解决如下问题：量表测量的是单维还是双维结构，中国被试中是否存在正向和反向方法因素，这两种方法因素在多大程度上影响了测验的结构效度和效标关联效度，不同测量工具中的方法因素是否同质，是否可解释为一种稳定的反应风格或人格变量，中国人的自尊量表得分有哪些不同于西方人的独有特征等。对525名大学生被试测试了自尊、大五人格、一般健康等人格量表，结果发现自尊量表中的正向和反向题有相似的心理测量学性能，测量相同特质。正向和反向方法因素在中国被试中都存在，但没有破坏自尊的单维结构。反向方法因素对测验结果的影响更强。不同测量工具中的方法因素具有同质性，属于系统误差，具有高抑郁、高神经质等特征的人容易受方法因素的影响。中国被试在自尊上表现出了谦虚、含蓄，能辩证地认识积极和消极自我等东方文化的特质。

Abstract: There is an ongoing debate on the use of negatively worded items in Rosenberg Self Esteem Scale (RSES) and other personality measures. This study was designed to explore the following questions: Do the five positively and five negatively worded RSES items measure one factor or two oblique factors? Do negative wording effect and positive wording effect exist in Chinese culture? If the answer is yes, to what extend do they affect structural and criterion related validity of RSES? Do negative and positive wording effect abstracted from different personality measures exhibit congruent validity? Can they be considered response style or meaningful personality variables? What features can be observed in Chinese participants that differentiate them from Western ones? 525 undergraduates were recruited and were administrated RSES, BFI, STAI, CES-D, GHQ-12, IRI, WLEIS, GWB, and MCSD. The results showed: the positively and negatively worded items in RSES perform similar psychometric properties; Method effect can be observed both in positively and negatively worded items, which are detrimental to the construct validity of RSES; When one trait factor and two method factors are defined in CTCM, close model data fit result in; In RSES the size of positive method effect is much smaller than negative method effect; Method effects in RSES and other personality measures are similar in nature; Modest bias is confirmed in the present study, Chinese people show stronger tendency to admit negative aspects of their self-concepts. As a con-clusion, balanced use of positively and negatively worded items is encouraged. Advanced statistic method should be employed to control method factors and estimate the true relationships between constructs measured by different personality measures. Chinese people are dialectical and are more modest when assessing their self-esteem.