本研究旨在介紹階層多變項線性模式,與提出一個新的階層線性模式研究方法的概念:階層線性模式的多樣本分析多群體階層線性模式,利用階層線性模式中的階層多變項線性模式延伸應用在這新的研究領域。以階層多變項線性模式來處理多個結果變項的多層次模式,除了可以比較迴歸係數的差異比較外,尚可用來比較多群體階層線性模式中的結構係數,以達到不同以往交互作用的調節效果的迴歸係數恆等性比較。透過24家公司664位員工資料,利用組織創新氣氛為組織層次解釋變項,員工組織承諾、性別為個體層次解釋變項對三個結果變項工作滿意、公司績效與工作表現的影響外,亦以性別區分為兩個群體,進行組織創新氣氛、組織承諾對工作表現影響的多群體恆等性比較。最後討論階層多變項線性模式與多樣本分析在應用上的相關優勢與限制。
The main purpose of this study is to provide a new research method and concept of the hierarchical linear modeling: multiple-samples analysis or MGHLM (multi-group hierarchical linear modeling). In this article, we introduce how to use the HMLM (hierarchical multivariate linear modeling) module of HLM software to analyze the multiple outcome variables in multilevel research and extend it to multiple-group HLM. Through the demonstration of innovation climate, organizational commitment, employees' gender and 3 outcome variables of 664 employees from 24 organizations, we can simultaneously estimate the within-group and between-group correlations among variables and compare their fixed effects. In addition, we separate the sample into male and female groups and use HMLM to do MGHLM and examine the invariance of hierarchical structure models. Finally, we conclude the advantages and limitations of HMLM and MGHLM.