Using data from the National Longitudinal Study of Adolescent to Adult Health (Add Health), I investigate the influence of adolescent friendship network on the likelihood of college enrollment, and whether and how this influence is affected by stratification factors (e.g., gender, race/ethnicity, and socioeconomic status). However, there is a challenge in evaluating this influence process since adolescents usually non-randomly select their friends. A selection process needs to be taken into consideration simultaneously with the influence process of adolescents’ friendship network on their likelihood of college enrollment. Previous research on peer effects has methodological issues and limitations. Traditional methods (e.g., multivariate regression, multilevel modeling, or propensity score matching) using limited data (e.g., cross-sectional) and measures of friendship network (e.g., one best friend) could not solve the problem of integrating selection process and influence process in one model. In addition, the dyadic and triadic (or even higher level) dependency among friends in the network makes it more difficult to estimate selection and influence processes using traditional methods. To address these concerns, I employ longitudinal network analysis with stochastic actor-based models (SABMs) to account for the influence of friendship network on adolescent college enrollment when simultaneously considering the selection of friendship. The co-evolution model of network dynamics (selection) and behavioral dynamics (influence) also addresses the problem of endogeneity between network change and behavioral change. However, the co-evolution model requires network data and behavioral data measured in multiple time points, so in the first stage of this research, I generate the predicted probability of college enrollment at three time points of Add Health using traditional logistic regression. Then in the second stage of this research, I use the transformed likelihood of college enrollment, a statistical artifact, as the behavior variable in the co-evolution model to examine how the likelihood of college enrollment affect the friendship selection and in turn friend’s average likelihood of college enrollment in the network influences an adolescent’s own likelihood of college enrollment. In the first stage, I find that there are some levels of gender, race/ethnicity, and SES inequalities in the college enrollment, even after controlling for previous academic achievement, other individual characteristics, family backgrounds, and school level variables. In the second stage, the results of dynamic network analysis indicate significant selection (partial deselection) and influence effects of adolescent friendship networks on the likelihood of college enrollment. In the selection process, adolescents have high tendency to select friends who are similar to them in the likelihood of college enrollment, or terminate friendships with other students of dissimilar likelihood of college enrollment. In the influence process, the average alter effect is found consistently significant and positive across all models and schools, which indicates that there is strong social influence of friendship network on adolescents’ likelihood of college enrollment. The higher the average friends’ likelihood of college enrollment, the more likely the adolescent will increase own likelihood of college enrollment. I also discuss the significance of results and many important policy and practical implications.

對於零碎且結構複雜的資料來源時，社會網路分析能夠給予整體性的觀察，還能檢視個體之間的關係。目前社會網路分析研究中，因為將網路退化至簡單連結關係，所以會遺失許多珍貴的資訊。而網路規模和型態隨著研究議題的不同，也會跟著增大或趨於多變，但動態網路分析能夠提供我們檢視每個時期，網路的變化或社群的形成或消失，甚至能知道節點間的互動影響。本論文研究，以政府人事異動資料為主，並且加入了其他政府組織的相關資料，建構出政府組織的從屬網路，並在每個網路快照中，擷取出重要的官員異動；每一筆人事異動都是一個事件的發生，而特任或簡任官員在本研究中視為重要事件，從這些重要事件的發生，我們能夠對每個時間的官員，使用EventRank的演算法做排名計算。最後能從時間的變化中，觀察出每個時期的佔有重要影響力的官員。 / To fragmented and complex structure data, social network analysis (SNA) can give an overall observation, but also view the relationship between individuals. Recent research in SNA is the degradation of the network link to a simple relationship but it will lose a lot of valuable information. The size and type of network with different research topics will follow the increase or rapidly changing, dynamic network analysis can provide our view of changes in the network or community to form or disappear in every period, even know the impact of the interaction between nodes. This thesis is based on the government official changes and other related data to construct manager-subordinate network of the government organization and capture the important interactions between officials in every network snapshot. An official change is the occurrence of an event and special level official changes in this study as a critical event. From these critical events, we can use the Event Rank algorithm to rank the officials. Finally, we can observe which official has more influence from the time changes.