社群媒體與大學生的網絡界限：伴遊圈、談話圈、接觸圈

Social Media and Network Boundaries among College Students: Reconstructing Companions, Conversations, and Contact Circles

網絡大小及界限是探索社會互動的核心議題，但實證研究卻難以探索網絡界限如何區隔、有何明確標準得以劃分不同層次的社會圈。這項社會網絡文獻上的難題，隨著社群媒體的發展迎來另類挑戰和機會。社群媒體一方面透過即時頻繁的互動，讓網絡界限益加多變莫測，另一方面提供完整的接觸紀錄，有利於辨別網絡界限如何區隔。本研究試圖掌握社群媒體的研究契機，透過多元的實證資料及多層次分析方法，挑戰個人網絡研究的核心難題。研究議題依據網絡界限的論述基礎，利用社群媒體紀錄的優勢，嘗試區辨由核心網絡往外擴張到弱聯繫的不同社會圈，並以大學生在臉書上的「伴遊圈」、「談話圈」、「接觸圈」為例，分析不同社會圈如何有所區隔。實證資料取自全國各大學及學院，共58個科系、1,531位大學部應屆畢業生提供的網路調查及臉書活動記錄。多層次分析結果顯示臉書上的聯繫同質性和相似性均有助於區辨伴遊圈、談話圈、接觸圈等不同的個人網絡範圍，得以延伸社會網絡研究傳承中的重要概念，區辨清晰的網絡界限。研究成果展現如何藉由爬梳完整的數位足跡資料，重建個人在社群媒體留下的互動紀錄，再結合具代表性的調查樣本資料，揭示在數位時代中如何區隔個人接觸網絡的層層界限。

As a fundamental issue of social network studies, the boundaries of personal networks are always hard to specify and differentiate. Not only is it challenging to conceptualize how far an individual's social circles may reach, but it also poses a daunting task for researchers to design effective approaches that help construct comprehensive network structures with clear boundaries. Conventional network generators based on surveys and other methods help reveal network boundaries by relying heavily on participants' cognitive abilities to recall or estimate how they interact and connect with others, which can result in various biases or errors. With the surge of social media, researchers have unprecedented opportunities to avoid such recall biases and errors by using automatically recorded data of social networking, even though the new communication means often accommodate fleeting contacts that make network boundaries more contingent. To explore how network boundaries emerge on social media, this study integrates data from both sampling surveys and social media. Based on multi-stage clustered sampling, we interviewed 1,531 senior students from 58 college departments in Taiwan who completed an online survey and authorized the use of their contact records on Facebook. The interaction records at both contact and tie levels on Facebook helped us identify who may comprise each of three network subnets located from core to periphery: companions, conversations, and contact circles. Paired with sociodemographic variables taken from the online survey, these contact records further facilitated multilevel analyses on how network structures in a particular academic year affect network size and boundaries in the following year. Overall, while personal networks on Facebook tend to expand when a user updated his or her activity posts more frequently and regularly in the previous year, the size of core networks also increases if more network members in the previous year were similar to the user in terms of sociodemographic background and online behaviors. In addition to revealing how network structures help expand subsequent personal network size, this study shows how an integrated approach of combining survey data from a nationally representative sample with Facebook contact records helps delineate the boundaries of subsets in personal networks.