組織現象有鑲嵌的本質，以致研究資料難免有群集的特性，如果分析時忽略資料非獨立性的因素，對組織現象的探討往往失準。根據現有文獻，本研究先從知識論看層次議題，再從方法論探討資料非獨立的意義和其影響，繼之以家長式領導實徵分析爲案例，以同一筆資料做個人、群體和跨層次三種分析，比較並討論群集資料在不同分析方式下所反映的研究議題。最後，我們強調組織研究的理論、研究設計和分析都和層次有密切關係。

Most of the organizational phenomena are hierarchically imbedded, easily making research data clustered in nature. Data analysis without considering nonindependence often leads to misunderstanding of the phenomenon under examination. Based on extant literature, this study first addressed the level issues from an ontological perspective, followed by a discussion of methodological concerns of nested data. As an illustration, a set of data investigating the effects of paternalistic leadership and the moderation of group cohesiveness was analyzed with OLS and RCM techniques. The impacts of data non-independence were illuminated through comparing the results from single level (i.e., both individual and group) and cross level analyses. The relationships between constructs were largely inflated in single level analyses. Implications of analyzing nested data were addressed.