尋找機制與過程：長期追蹤研究的功用

Looking for the Mechanism and the Process: The Function of Panel Study

吳齊殷(Chyi-In Wu) ； 張明宜(Ming-Yi Chang) ； 陳怡蒨(I-Chien Chen)

《αβγ量化研究學刊》 2卷1期 (2008/12) Pp. 1-26

摘要

調查研究經常被用來掌握社會動向及現況的發展，進而據以分析探查社會現象與社會問題。然而，近幾年長期性資料的蒐集與分析爲調查研究於資料蒐集、研究設計以及資料分析上帶來很大的突破。長期性貫時追蹤資料(panel data)的建立與蒐集提供研究者新的基本工具，重新思索、重新檢證社會現象問的因果關係。結合時問向度的經驗資料與分析技術將有助於我們釐清社會現象的基本時問次序，重建事件問的先後順序與因果關係的連結，此特點不僅可以彌補傳統量化研究的缺點，更可以把人重新放回所屬的歷史脈絡中又能不失個人有所變化成長的可能性。本文列舉三種實證研究的分析實例，潛在成長曲線模型、潛在類別模型以及階層線性模型。透過實例來說明沿承相同研究議題（青少年憂鬱症狀的發展變化）而來開展三種研究分析的策略以及模型應用的特點。結合長期資料的特性與長期資料的分析技術，研究者能尋找各種社會運作的機制與過程。

關鍵字

長期性貫時追蹤資料 ； 潛在成長曲線模型 ； 潛在類別模型 ； 階層線性模型

並列摘要

For social researchers, survey research was used to go over the social trends and social development, with this; they could capture social phenomenon and social problems. Nowadays, the collection and analysis of panel data make a breakthrough on survey research.Panel data becomes a fundamental resource to rethink and verify the causal relation of social phenomenon. By rebuilding the time-order of social events, the panel data can help us to find out the causal relation, which is useful for the researchers to realize the mechanism and process of social phenomenon. Together, the characteristics of panel data in time dimension with the statistical technology of data analysis, one could catch up the disadvantages of quantification research, but also put the person back his historical context without losing the considering about the possibility of individual change.In this article, we intend to takes three empirical researches as example, to demonstrate how Latent Growth Curve Model, Latent Class Model and Hierarchical Linear Model can be used to analyze panel data. By answering the extended research questions, we apply the three methods on data analysis and model application. We concluded that the researchers have to make good use of panel data and have a good strategy for data and the analytic methods, in that case, the researchers will have the best chance to access the mechanisms and a series of processes behind the social phenomenon.

並列關鍵字

Panel Data ； Latent Growth Curve Model ； Latent Class Model ； Hierarchical Linear Model