关键词: 团体依恋, 关系依恋, 组织认同, 项目因子分析, Fisher信息函数

Abstract: In accordance with Smith et al.’s (1999) paradigm of developing group attachment measurements, the current research developed the group attachment scale (GAS) based on the experience of close relationship inventory which was widely considered as standard measurement of dyadic attachment. To confirm such conjecture, GAS with several criterion scales were tested in a sample of 1571 employees. Data analysis incorporated item factor analysis (IFA) modeling instead of using traditional confirmatory factor analysis (CFA) based on linear regression. Within the framework of structural equation modeling (SEM), IFA could be formalized as a CFA model with latent response indicator which was supposed to be the underlying score externally manifested by categorical response, so it was also called as categorical confirmatory factor analysis (CCFA). The current research applied maximum likelihood estimator to fit a two-factor non-orthogonal CCFA model for GAS. In terms of univariate standardized Pearson residuals, the CCFA model fitted well to the empirical data. However, item and test information still had much to be desired. Validity analysis showed that group attachment moderately correlated with dyadic attachment, which probably indicated their common antecedent attachment theory. Besides, group attachment avoidance had stronger association than group attachment anxiety with group identification measures, but unexpectedly stronger with the sum-score of Rosenberg self-esteem scale. Possible reasons for such observations were further discussed.

Key words: group attachment, dyadic attachment, group identification, item factor analysis, Fisher information function