父母離婚對子女短期及長期的影響—以長期追蹤資料為例

家庭對於子女的認知發展以及人格特質有重要的影響力，然而近年來臺灣的單親家庭日漸增加，使得這些子女必須面對家庭結構的轉變，許多文獻更證實父母離婚的子女無論在學業、情緒發展、身心理健康、婚姻態度以及教育程度上皆有負面的影響。本文使用「台灣青少年計畫」 (Taiwan Youth Project，TYP) 國一樣本 (J1) 在第一波的父母婚姻狀態、國一樣本第三波以及成人時期2017年的學生問卷，目的在探討父母離婚對子女短期以及長期的影響，並根據Chen et al. (2019) 的研究結果進一步延伸探討影響子女長期發展的非經濟面機制。　　首先，本文使用一般迴歸分析法 (ordinary least squears, OLS) 分別檢視父母離婚在短期以及長期下，對子女的個人表現、家庭選擇、學校表現、職場表現、以及父母離婚對居住環境的影響。OLS結果發現，父母離婚的子女在青少年時期有較多的學習適應力問題、在校成績較差，而成人時期的最高教育年數以及最高教育程度也低於未經歷父母離婚的個體；父母離婚長期下也影響了父母與子女的關係，子女在國三時自評與父親的關係較差，且成人時期的子女相較於父母已婚者更不滿意與父母的關係。　　其次，為了減輕可能產生的選取問題，本文亦使用傾向分數配對法 (Propensity Score Matching, PSM) 重新檢視父母離婚和父母婚姻完整子女的差異，並藉此作為OLS估計結果的穩健性檢驗 (robustness check)。PSM的結果指出國一前父母離婚不會影響子女的在校成績，但仍然對子女的長期教育成就有負面的影響，其餘項目的估計結果與OLS一致，顯示OLS具有一定的穩健性。　　最後，兩種實證方法均顯示，放任型的管教方式以及較低程度的父母參與分別可能是導致子女與離婚的父母長期關係較差的因素。除了父母管教以及父母參與之外，居住環境的不穩定、居住品質較差亦可能是父母離婚子女的教育成就比父母婚姻完整的子女教育成就較低的潛在機制。

關鍵字

父母離婚；單親家庭；家庭結構；傾向分數配對法

並列摘要

Families have an important influence on children's cognitive development and personality traits. However, the number of single-parent families in Taiwan has increased in recent years. Many children are facing the change in the family structure. Literature have revealed that parental divorce has negative effects on children's academic performance, emotional development, physical and mental health, marital attitudes, and educational attainment. The thesis analyzes the short-term and long-term effects of parental divorce on children’s outcomes. The data used in this thesis are from the waves one and three, and the wave 2017 of the Taiwan Youth Project (TYP), where the samples mean age is 13 (7th grade), 15 (9th grade), and 31 years, respectively. Inspired by Chen et al. (2019), this thesis further examines the likely non-economic factors that may negatively affect the long-term development of children. 　　First, the thesis analyzes the short-term and long-term effects of parental divorce on children's personal behaviors, family choices, performance in school, and performance in workplace as well as its effects on living environment by using ordinary least squares models (OLS). The OLS results showed that compared with those who did not experience parental divorce, individuals who had experienced parental divorce before 7th grade had more problems with learning adaptability and poorer school performance when they were teenagers, and lower educational attainment when they were about 31 years. They also reported poor and unsatisfactory relationships with their divorced parents when they were teenagers and adult. 　　Second, in order to alleviate possible selection problems, the thesis also uses PSM models to re-analyze the differences in outcomes between individuals who experienced parental divorce and who did not to check the robustness of the OLS results. The PSM results indicated that parental divorce before age 13 did not affect children's school performance, but still had a negative impact on children's educational achievement. The rest of the estimation results were consistent with the OLS results suggesting some degree of robustness of the OLS results. 　　Last, both OLS and PSM results showed that permissive parenting styles and lower levels of parental involvement may be the likely factors that lead to poorer relationships between children and divorced parents in the long run. In addition to parenting style and parental involvement, unstable and poorer quality of living environments during adolescence may be also the potential mechanisms that prevent children of divorced parents from achieving higher education attainment.