中文摘要：

研究选取四个参照群体（国家、民族、性别、职业），探讨个体内部的社会身份重要性、以及个体对其不同社会身份的社会认同度对群体参照效应的影响。结果表明，在个体内部，社会身份重要性越重要，社会认同度越高，相应的群体参照效应越大。这说明在个体的自我内部，不同的社会身份是按照重要性来排列，越重要身份越靠近个体自我概念的核心，而且这种重要性影响着集体自我相关信息的认知加工。此外，社会认同作为群体参照效应产生的先决条件，在个体自我概念内部，社会认同的高低也影响着集体自我的记忆加工优势。

英文摘要：

A substantial body of researches on the group-reference effect （GRE） have shown that the reliable memory advantage for information encoded about one＇s own group over material encoded about other groups. Although GRE has been confirmed in numerous studies, many issues still remain to be resolved. Among these issues, the most apparent one is what factors may affect the strength of GRE. Although some external factors affect GRE have been discussed in the previous studies, however, some other factors especially internal factors still need to be further explored. Therefore, the aim of the current study was to examine the importance of the intra-individual social identities, as well as the social identification of different social identities which effect on GRE. To explore whether these factors affect individual self-memory advantage or not, in the current study, we employed adjectives as materials in the GRE paradigm, there are four reference groups （national, ethnic, gender, occupation）. 25 healthy students participated in this experiment （15 females and 10 males, with a mean age of 23.7 years）. This experiment consists of three parts： the encoding phase, distraction, recognition. During the encoding phase, participants were given a list of personality traits, each of personality traits about which they were required to make simple judgments. Under the ingroup-reference condition, for example ＇Does this word generally describe Chinese？＇ or ＇Does this word generally describe boy/girl？＇ and so on. The questions for the outgroup-reference condition are ＇Does this word generally describe American？＇ or ＇Does this word generally describe middle school students？＇ and so on. The lists of words had been matched for valence, word frequency. There were 5 blocks, the first block were exercise, the other four blocks contained 30 trials for each reference condition, so there were 120 trials in total. Following the encoding task, participants completed a 5-min distraction task and then mov