本研究區分中介之精熟經驗與感知之精熟經驗。前者為玩家在角色扮演遊戲中所獲得之成功經驗，後者指玩家對此中介經驗之整體評估感受，且包含過去經驗。玩家在遊戲中不斷接受挑戰並破關的過程，即包含了成功的中介精熟經驗，此經驗與感知之精熟經驗應為正向相關，且為遊戲中必要之一環。角色依附是探討玩家與其角色關係之重要變項，當玩家越依附於其角色時，越能將此中介精熟經驗轉換成其感知之精熟經驗，越能對其效能感產生正面影響。本研究因此嘗試探討嚴肅遊戲中提供的中介精熟經驗是否會影響內部政治效能感，並探討在這影響機制中，感知精熟經驗，角色依附與內在政治效能感之路徑關係。 本研究採對照組前後測實驗設計。參與者為131位大學生，在本研究發展之政策嚴肅遊戲中，參與者被隨機分派到實驗組與控制組。兩組皆接受遊戲之中介精熟經驗處理，但為了檢驗遊戲中介之精熟經驗與玩家感知之精熟經驗之關係，實驗組提供不同場景以演練技巧之遊戲，控制組則未提供不同場景以演練技巧之遊戲。此實驗過程費時大約60分鐘。玩家在實驗前後於線上填寫問卷。本研究首先使用重複量數單因子變異數分析以檢驗不同性別與組別之實驗結果，接著，再以結構方程模型探討感知精熟經驗，角色依附與內在政治效能感之路徑關係。 研究結果發現：(一) 實驗組與控制組之內在政治效能感皆明顯提升，然而，實驗組之提升幅度並未顯著高於控制組。(二) 在路徑模式中，角色依附會透過感知之精熟經驗間接影響內在政治效能，而前測內在政治效能會透過感知精熟經驗間接影響後測內在政治效能，亦會對後測內在政治效能造成直接影響。 本研究有助於了解嚴肅遊戲之心理機制，其結果對於應用嚴肅遊戲設計與公民教育有重要啟示。 / This study distinguished “mediated enactive experience” and “perceived mastery experience”. The former referred to the experience when a player plays an avatar in a game and the latter referred to a player’s psychological evaluations of the experience including past experience. Mediated enactive experiences is positively related to perceived mastery experiences when players play and make progresses in a game. The experiences are an essential part of a game. Yet character attachment may influence the relationship between players and avatars. If a player attaches to his or her own avatar more, he or she may perceive the mediated enactive experiences more his or her mastery experience, therefore, character attachment should positively contribute to internal political efficacy through perceived mastery experience. A control group pretest-posttest experimental design was conducted in this study. Both groups received policy-related mediated enactive experiences from the serious game. In order to test if players who received more policy-related mediated enactive experiences will upgrade their internal political efficacy to a higher level, 113 college students were randomly assigned to one of the two groups: In the experimental group, players played an avatar with chances for practicing learned skills in three different settings; In the control group, players played an avatar without chances for practicing learned skills. Two repeated-Measure ANOVAs were conducted to analyze if there were gender and group differences on manipulation effects. Moreover, structural equation modelling (SEM) was employed to analyze the proposed path model of character attachment, perceived mastery experiences, prior and posttest internal political efficacy. The results showed that the internal political efficacy of both groups was significantly enhanced. However, the experimental group did not upgrade to a higher level than the control group. Furthermore, in the path model, character attachment influenced posttest internal political efficacy indirectly through perceived mastery experiences; moreover, prior internal political efficacy influenced posttest internal political efficacy through perceived mastery experiences indirectly and also influenced posttest internal political efficacy directly. To conclude, the findings of this study can help understand psychological mechanisms related to gaming as well as can be applied to civic education and serious game design in the future.