本研究為次級資料分析，運用「台灣長期教育追蹤資料庫」中接受四波次追蹤調查的樣本資料。本研究以家庭中父母是否曾發生離婚或分居事件，及其發生之時間點，作為區分不同組別的方式。將家庭結構組別區分為雙親家庭、高中時期成為單親家庭、國中階段成為單親家庭，以及國中之前成為單親家庭。在本研究中運用家庭系統、家庭內教育資源及家庭歷程等理論觀點，並在研究設計上，使用長期追蹤資料，並以固定效果模型及潛在成長模型進行研究分析。探究家庭結構的改變，對不同階段青少年子女之學習成就的影響。而所獲得相關分析結果亦驗證本研究之研究假設。 茲將本研究之重要發現摘要如下： 一、家庭結構的改變對子女學習具有負面影響。 二、當家庭結構發生改變之時點，如果是在子女幼年時期，會對子女學業能力後續發展會產生長期持續累積的負面影響，並與雙親家庭子女學習成就差距持續擴大。 三、子女在國中階段成為單親家庭對學習成就影響最大。 四、高中階段單親家庭組則存在當下近期效果之影響。 五、家庭結構亦會透過家庭資源、父母教育期望與家庭教育歷程，發揮間接影響青少年子女學習成就效果。 六、運用固定效果模型與傳統迴歸模型進行比較，固定效果模型所獲得之估計值較小，縮減了傳統迴歸模型之估計誤差。 綜合言之，本研究將時間概念納入家庭結構之動態分析，克服了過往研究所存在的內生性偏誤，並強化因果之說明，並補充過往理論觀點不足之處。 / The method to carry out this study was using secondary data, which included four waves with the same participants. The purpose of this study is to explain the effect of family structure the change were related to adolescent learning achievement. The secondary data is adapted from Taiwan Education Panel Survey (TEPS). This study by family in parents whether once to have the divorce event and the event timing takes the discrimination different groups. Family structure groups divide into “the intact family”, “the senior high school stage becomes the single-parent family”, “the junior high school stage becomes the single-parent family” and “early life becomes the single-parent family”. The research viewpoint is from family system theory, family educational resources theory and family process theory proposed. Inquires into the family structure the change, how to influence children learning achievement in different life cycle? The research was longitudinal study design. The study used the fixed effect model and latent growth model to carry on the research analysis. Accordingly, the results support our proposed model and its connection with academic achievement of students. Author summarizes the findings as follows: 1. Family structure's change for the children has the negative influence in learning achievement. 2. When the family structure change time point was in children childhood time. IT has the negative accumulates influence. And the learning achievement disparity with the intact family children to expand continually. 3. The children become the single-parent family in junior high school stage to learning achievement influence to be biggest. 4. The high school stage single-parent family group will have immediately in the near event influence of the effect. 5. The family structure will also penetrate the family resources, the parental educational expectation and the family process , make indirect influence on adolescent’s the learning achievement. 6. The fixed effects model and the tradition regression model carries on the comparison. The fixed effect model obtains the estimated value to be small. The fixed effect model reduced of error of estimation the tradition regression model. To sum up, this research integrates the time concept with the family structure in the dynamic analysis. Explanation of the strengthened causes and effects, the outcome supplements the passing theory viewpoint deficiency.