再探臺灣的世代政治：交叉分類隨機效應模型的應用，1995–2010

Generational Politics in Taiwan Revisited: Application of a Cross-Classified Random Effects Model, 1995–2010

本文回顧年齡、時期與出生世代（Age-Period-Cohort，簡稱為APC）相關的臺灣世代政治文獻，發現數據資訊不足、測量不一致與多重共線性問題，交叉分類隨機效應模型（cross-classified random effects model，簡稱為CCREM）可改善之。本文以1995–2010 年台灣社會變遷調查數據分析國家認同與民主價值，發現時期效果最強：民眾臺灣認同加強、對統一支持下降、自由民主信念深化；年齡政治效果相對有限，世代政治效果顯著但複雜，與各政治世代的歷史經驗及集體記憶有關，並建議CCREM 模型值得推廣應用。

The generational politics studies in have Taiwan seriously suffered from“age-period-cohort (APC) problems,” that is, mixed information, unstandardizedmeasurement, and multi-collinearity problems, which lead to contradictoryresults. In this article, a Cross-Classified Random Effects Model (CCREM) isapplied to mitigate the APC problems. It investigates the APC differences of thepreferences on national identities and democratic values using a pooled datasetof national social surveys between 1995 and 2010. According to the CCREMresults, the “age effect” shows that the younger age-groups have stronger Taiwaneseidentity and more politically liberal attitudes, but the statistical effect isnot very significant. On the other hand, the “period effect” shows that after thedemocratic transition, the whole population has stronger Taiwanese identity andmore politically liberal attitudes, and the statistical effect is significant and robust.Finally, the “cohort effect” is more important and complicated than the “ageeffect”.