本研究採用一項有關青少年成長與發展調適問題的長期貫時性追蹤資料(1996-1999)，試圖突破過去討論青少年憂鬱症狀發展時，所用之横斷式資料的囿限，嘗試應用潛在成長曲線模型(Latent growth curve model, LGC Model)的分析方法，加入歷史時間的縱深，捕捉青少年憂鬱症狀的「起始狀態」、與「個別的成長軌跡發展」。以不扭曲地將所有受試青少年在三年間的內化症狀變化情形，忠實地描述出來。而後，加入「改變」因素的討論，企圖尋找能影響青少年憂鬱症狀發展軌跡的關鍵機制。 此研究主要目的即在「具象化」復原力的理論觀點，企圖加入動態的時間面向，確認負向生活事件與青少年憂鬱症狀發展軌跡之間的因果關聯，並探討來自個人、與環境脈絡中的關係運作，對青少年憂鬱症狀平均數、變化方向與速率的跨時間影響。研究結果明確回答：為什麼有些青少年在受到憂鬱症狀的負向影響之後，尚能有回復機會並「表現地比預期好」的疑問。至於針對一群憂鬱症狀發展呈現改善、或惡化的少數青少年樣本，在性別、自尊、負向生活事件、家庭親子互動、學校好朋友關係等特性上的差異，本研究亦逐一說明。 在理論層次上，本項研究結合適切的研究方法，從「靜態」到「動態」地觀察青少年的身心發展、自「個人」到「家庭系統內外」討論內外在資源對青少年復原的短暫以及長久影響效果，並以一般青少年為研究對象的作法，擴增了復原力理論的推論範疇與解釋深廣。研究顯示，青少年的「改善」或「惡化」憂鬱症狀發展軌跡，確實在環境脈絡的節制之下，存在著個別差異。此外，青少年起始的憂鬱狀態並不影響憂鬱症狀軌跡發展的變化率。家庭經濟不利這項負向生活事件，對於青少年憂鬱症狀的預測，只呈現短暫的初始影響。自尊和好朋友關係皆是青少年可以主動建構與可為之舉，為兩個最重要能影響青少年憂鬱症狀變化的關鍵因素。至於學校脈絡，則可視為在家庭脈絡之外，能提供青少年憂鬱症狀改變效果的新路徑，以及讓青少年可以順利「轉大人」之雙重機會的結構因素。 / Using data derived from a panel study (1996-1999) of long-term Taiwanese adolescent development and adaptation, this study intended to break through the limitations of cross-sectional studies, which plagued past studies of adolescents’ developing depressive symptoms. By employing the Latent Growth Curve Model (LGC Model), this study mainly attempted to feature the individual initial status and the trajectory of every adolescent’s developmental depressive symptoms, which concerned about the important functions of the dynamic historical time and space on youth internalizing symptoms, for the research purpose to reflect the real resilient outcome each adolescent displayed. Besides, in order to understand the key factors that were taken as positive and effective mechanisms to influence the initial status and rates of changes on youth trajectories of depressive symptoms, several latent constructs such as self-esteem and protective factors developed from family and school contexts were taken into accounts. Further, specified characteristics were noted to highlight the basic differences gradually showed between resilient improved adolescents and worsen ones. A positive-psychological stance was taken as the leading research perspective in this study. The results shows that familial economic hardship only affects the initial status of adolescent depressive symptoms, implying that this negative event just had a short-term effect on youth’s psychological well beings. Those who were initially vulnerable to familial negative event had opportunity to become resilient over time. As to the protective factors, self-esteem and cohesive good-friendship were two crucial facets adolescents could actively construct and make efforts for further resilient performances to be better than expected. The analyzing results indicated, interestingly, that parent-child relationship early obtained in family context and adolescent’s satisfaction with parenting merely counted for the initial impact on adolescent trajectories of depressive symptoms. Concerns and cohesive relationships acquired in school contexts, especially in classes, provided dual chances for adolescents to become resilient in a long run.

This thesis focuses on statistical and machine learning methods designed for sequential and repeated measurement data. We start off by considering the classic general linear model (MANOVA) followed by its generalization, the growth curve model (GMANOVA), designed for analysis of repeated measurement data. By considering a binary classification problem of normal data together with the corresponding maximum likelihood estimators for the growth curve model, we demonstrate how a classification rule based on linear discriminant analysis can be derived which can be used for repeated measurement data in a meaningful way. We proceed to the topics of neural networks which serve as our second method of classification. The reader is introduced to classic neural networks and relevant subtopics are discussed. We present a generalization of the classic neural network model to the recurrent neural network model and the LSTM model which are designed for sequential data. Lastly, we present three types of data sets with an total of eight cases where the discussed classification methods are tested. / Den här uppsatsen introducerar klassificeringsmetoder skapade för data av typen upprepade mätningar och sekventiell data. Den klassiska MANOVA modellen introduceras först som en grund för den mer allmäna tillväxtkurvemodellen(GMANOVA), som i sin tur används för att modellera upprepade mätningar på ett meningsfullt sätt. Under antagandet av normalfördelad data så härleds en binär klassificeringsmetod baserad på linjär diskriminantanalys, som tillsammans med maximum likelihood-skattningar från tillväxtkurvemodellen ger en binär klassificeringsregel för data av typen upprepade mätningarn. Vi fortsätter med att introducera läsaren för klassiska neurala nätverk och relevanta ämnen diskuteras. Vi generaliserar teorin kring neurala nätverk till typen "recurrent" neurala nätverk och LSTM som är designade för sekventiell data. Avslutningsvis så testas klassificeringsmetoderna på tre typer av data i totalt åtta olika fall.

This research highlights the importance of considering the degree students’ integration into school classes when estimating school effects. Combining and using two different datasets collected before and after education reform in Taiwan, the study compares school effects under two different education systems in order to answer the question about the efficiency of education reform. I estimate multilevel growth models to assess how school environments affect changes in students’ initial and change rate of their academic performance across junior high school years. Besides, two-part random-effects models are also introduced into the analyses to testify how school environment influence adolescent performance in their high school enrollments. My results support and extend Blau’s structural theory, revealed that school contexts and school networks directly and indirectly influences students’ performance in their school classes and in their high school enrollments, suggesting students’ outcome are conditioning by the local structure, the school environments. However, through making more friends inside and outside school classes, students still have their own power to modify the environmental impacts on themselves. With respect to the comparisons of school effects on individuals’ performance under two different education systems in Taiwan, the decreasing peer influences and the decreasing significance of school networks indicate that the school effects gradually decline after the administration of education reform. One should note that simply a little change on education system might alter students, parents, and teachers’ behaviors. The decreasing peer effects and the decreasing school effects on students’ academic performance suggesting that students might change their behaviors on interacting with their friends and change their behaviors at schools in order to jostle higher education after education reform. The increasing cram schooling and the increasing significance of family SES support the inference that students modify their behaviors to come up against the education reform in Taiwan. These findings suggest the need for more panel datasets collected from the newly cohorts after education reform was administrated for a period and the need for more studies of education reform and school effects, to have more understanding about the mechanisms of school efficiency.