個人網絡成員彼此之間的聯繫強度如何，又受到什麼結構條件影響，是探究網路結 構的一項重要取向。本文依據「臺灣地區社會變遷基本調查」1997年第三期第三次資料，以 個人網絡中他人（alter）成員彼此之的聯繫強度作為分析對象，探討他人間如何隨著個別和 自我（ego）之間的聯繫，以及彼此在人口特徵和社經地位上的同質性，而增強首減弱互相熟 識的強度。分析單位為他人間的聯繫，樣本聯繫數合計24,415筆。研究發現呼應社會網絡理 論的強度命題：兩位他人之間的聯繫，隨著自我和這兩位他人之間的聯繫強度而增強；們我 與他人聯繫效應之強度，又因為他人在個人網絡結構中所處的位置而有所不同──這種聯繫 效應，在自我和關系較弱（後一位）他人之間的聯繫上顯得特別強烈而一致。研究另一垂點 在區辨他人之間的同質性對彼此的聯繫有何效應。他人同質性隨著情感性網絡首工具性網絡 的分野，以及自我的親屬圈、非親屬圈、混合型（親屬和非親屬交錯圈）的區別，對他人間 的聯繫特質來解釋相對應的聯繫特質，並彰顯個人網絡中那些佔據邊陲置的他人聯繫，顯然 扮演著更為關鍵的角色。

　　　　　Using a network structure approach, the author examines how two alters in personal network vary in their ties with each other to create a triad consisting of two alters and an ego. The study focuses on tow circumstances that determine the strength of alter-to-alter ties: how closely each alter is tied to the ego, and how similar the alters are to each other in terms of demographic and socioeconomic characteristics. Data were taken from the 1997 "Taiwan Social Change Surveys". Respondent listed up to five alters each in one expressive and one instrumental network. Using alter-to-alter ties as the unit of analysis, the author extracts 24,415 ties from 2,835 respondents. The dependent variable (tie strength between two alters in each triad) is measured in terms of "very familiar", "somewhat familiar", and "total stranger". The first independent variable, "strength of ego-to-alter tie", is measured in terms of a composite score from a factor analysis of length of acquaintance, frequency of interaction, closeness, and significance. The second group of independent variables, "alter-to-alter homophily", includes seven demographic and socioeconomic indices. The findings correspond well with a proposition common to network analysis: the strength of ties between two alters depends largely on the strength of ties between the ego and each alter. However, further analysis indicates that the tie between ego and a "latter alter" (whose tie to ego is inherently weaker) plays a more substantial role than the tie between ego and the "former alter" (usually a closer tie). This qualification highlights the critical role of network location- that is, peripheral structure is more crucial in shaping other features within the network. Alter-to-alter ties also tend to increase when the alters are more homophilous. Nonetheless, demographic homophilous effects are more limited among kin alters, especially in expressive networks. In contrast, socioeconomic homophilous effects are strong and consistent across both types of networks, as long as at least one non-kin alter appears as a network member.