Mazzeo Suzanne E, Aggen Steven H, Anderson Charles, Tozzi Federica, Bulik Cynthia M
Department of Psychology, Virginia Commonwealth University, Richmond 23284-2018, USA.
Int J Eat Disord. 2003 Sep;34(2):255-64. doi: 10.1002/eat.10180.
To investigate the internal structure of the Eating Inventory using confirmatory factor analytic (CFA) methods.
Female participants in a population-based twin study (N = 1,510) completed a reduced version of the Eating Inventory as part of a larger study of eating behaviors. Two CFA estimation methods were implemented using LISREL and Mplus to evaluate three different factor structures reported in the literature.
Overall results for Stunkard and Messick's (1985) three-factor model indicated a poor fit to these data. The structure reported by Ganley resulted in an improper solution with factor correlations over 1 and several loadings estimated near 0 or with negative signs. The model proposed by Hyland et al. also fit poorly for all three models, the pattern of loadings on the Restraint factor proved most difficult to interpret.
This study represents the first application of these confirmatory methods to the three subscales of the Eating Inventory. The instrument, as administered in this study, did not conform well to the structures proposed in previous studies. Because the Eating Inventory is recommended as a treatment planning tool, poor replication of the constructs purported to be assessed by this instrument underscores the importance of further internal validation research.
采用验证性因素分析（CFA）方法研究饮食量表的内部结构。
在一项基于人群的双生子研究中，女性参与者（N = 1510）完成了饮食量表的简化版，作为一项关于饮食行为的大型研究的一部分。使用LISREL和Mplus实施了两种CFA估计方法，以评估文献中报道的三种不同因素结构。
对于Stunkard和Messick（1985年）的三因素模型，总体结果表明该模型与这些数据拟合不佳。Ganley报告的结构导致了一个不合适的解决方案，因素相关性超过1，并且几个负荷估计接近0或带有负号。Hyland等人提出的模型对所有三个模型的拟合也很差，抑制因素上的负荷模式最难解释。
本研究是这些验证性方法首次应用于饮食量表的三个分量表。在本研究中使用的该量表与先前研究中提出的结构不太相符。由于饮食量表被推荐作为一种治疗计划工具，该量表声称要评估的结构缺乏良好的重复性，这突出了进一步进行内部效度研究的重要性。
