多數偏差行為研究著重於偏差行為嚴重程度的探究，可能忽略青 少年群體異質性，而少數研究雖注意到偏差行為的群體異質性，但卻 忽略群體內青少年在偏差行為嚴重程度上的變異性。因素混合分析 （Factor Mixture Analysis, FMA）可在相同測量模式中，以潛在類別 分析（Laten Class Analysis, LCA）捕捉偏差行為的群體異質性，並 以因素分析（Factor Analysis, FA）來描述次群體內青少年在偏差行 為嚴重程度上的差異，但仍較少受到研究者的重視。有鑑於此，本研 究主要目的在於，評估LCA、FA 與FMA 應用於國中青少年偏差行為研 究的實益。研究結果發現：⑴ 2 類別1 因素FMA 模式，較適合描述國 中青少年偏差行為的群間與群內異質性。⑵在同時考量偏差行為的類 型與程度後，FMA 出現與LCA 不同的分類結果。最後於文末根據研究 結果，提出相應之建議。

Most research on deviance has emphasized investigating the severity of deviance. This may result in ignoring the heterogeneity in the population. Although, a few studies have addressed this heterogeneity, the variation in the severity of deviance within a class was still overlooked. Factor mixture analysis (FMA) incorporates latent class analysis (LCA) with factor analysis (FA) to capture the heterogeneity in a population and to depict the severity of deviance within a class, respectively, within a single model. However, it has received insufficient attention to date. The main purpose of this study was to evaluate the benefits of applying LCA, FA, and FMA to investigate adolescents’ deviance in junior high school. The results showed: (1) A twoclass, one-factor F MA model was suitable for depicting heterogeneities between and within classes for junior high school adolescents. (2) After taking type and degree of deviance into account, FMA had results differing from that of LCA. Corresponding suggestions were proposed based on the present results at the end of this article.