期刊名称：International Journal of Statistics and Probability
印刷版ISSN：1927-7032
电子版ISSN：1927-7040
出版年度：2019
卷号：8
期号：5
页码：1-12
DOI：10.5539/ijsp.v8n5p1
出版社：Canadian Center of Science and Education
摘要：This paper reports three simulation studies conducted to identify the contextual conditions leading to the observation of a difficulty factor in confirmatory factor analysis. The data of each study were generated to show one underlying source of responding only whereas the difficulties of the simulated items constituting the contextual condition were varied. The first study showed that a broad range of difficulties of items was insufficient for driving a difficulty factor. The second study revealed that very large and small difficulties of the same size could lead to a difficulty factor if the confirmatory factor model included two correlated factors. In the third study a subgroup of simulated items showed very large difficulties of the same size while the difficulties of the other simulated item were varied. In this study almost all combinations of difficulties led to the observation of a difficulty factor that was correlated or uncorrelated with the genuine factor.
