Hau K T
Faculty of Education, The Chinese University of Hong Kong, Shatin, New Territories.
J Pers Assess. 1995 Aug;65(1):117-32. doi: 10.1207/s15327752jpa6501_9.
Confirmatory factor analyses with congeneric models were used to examine whether seven common locus of control instruments were measuring the same single construct. The subjects were 191 adolescents, and the analyses were based on the intercorrelations among the scales as reported by Furnham (1987). In addition to the one-factor congeneric model, other first- and second-order factor models were posited to reflect the effects due to target subject (adult vs. child), dimensionality (uni- vs. multidimension), and scaling method (yes/no, forced-choice, 7-point). A recently developed strategy in multitrait-multimethod analyses was also adopted in the analyses. In general, results of this study did not substantiate the claim that all scales were measuring the same construct. Even among the more strongly correlated unidimensional measures for children, the support for a one-factor model was still rather weak. Furthermore, results did not justify the categorization of the measures by their target subject, dimensionality, and scaling method.
采用同属模型进行验证性因素分析，以检验七种常见的控制点量表是否测量的是同一个单一结构。研究对象为191名青少年，分析基于弗恩汉姆（1987年）报告的各量表之间的相互关系。除单因素同属模型外，还提出了其他一阶和二阶因素模型，以反映目标对象（成人与儿童）、维度（单维与多维）和量表编制方法（是/否、迫选、7点量表）的影响。分析中还采用了多特质多方法分析中最近开发的一种策略。总体而言，本研究结果并未证实所有量表测量的是同一结构这一说法。即使在儿童的相关性更强的单维测量中，对单因素模型的支持仍然相当薄弱。此外，研究结果也无法证明根据测量的目标对象、维度和量表编制方法对这些测量进行分类的合理性。
