Skip to main content
Open Access Publications from the University of California

UC Berkeley

UC Berkeley Electronic Theses and Dissertations bannerUC Berkeley

A Tie-Centered Approach for Ego-centric social network studies

No data is associated with this publication.

Along with the development of multi-level statistical models, the dyadic social relationship has been getting attention again in the current quantitative ego-centric social network studies. While many empirical studies have demonstrated distinctive pathways of tie-level dynamics, the dominant way of operationalizing observed social ties often ignores the heterogeneity of dyadic relationships within a given relationship category. Instead, many of social relationship categories such as friends, confidants or social support relationships has been often treated as if each category represents a homogeneous social relationship. Yet, each term of social relationships in effect covers a various range of social relationships. The main goal of this dissertation is to reveal heterogeneous types of social relationships within a given relationship category and incorporate them into empirical examination on the association among social relationships, life course, and mental health status.

In this dissertation, I propose a tie-centered approach as an alternative way of studying ego-centric social networks. The social tie is a complex entity. Within a single dyadic relationship, many relational attributes, contexts, and histories are intertwined. Accordingly, the term “friend,”, “confidant”, or “social supporter” can be used in many different senses. In order to understand the multiple forms of a given relationship, the tie-centered approach suggests to inductively create typologies based on the multiple dimensions of any given type of social tie—that is, multiple dimensions of a tie between friends, of a tie between confidants, and so forth. For example, the strength of the relationship (e.g., closeness) can be combined with other variables (e.g., proximity, frequency of contact, length of the relationship, etc.) to form a multidimensional typology. Methodologically, I utilize clustering methods—specifically, a multi-level latent-class model—to investigate how these different attributes of social ties are configured within the hierarchical structure of egos and alters in survey data on personal networks.

Using data from the UC Berkeley Social Network Study (UCNets), a rich source of data on the personal networks of a representative sample of San Francisco Bay Area residents, I apply the tie-centered approach to studying three widely studied social relationships: “Confidant”, “Friend” and “Support relationship.” And I found the heterogeneous subtypes of social relationship within each relationship category: four different types of friends (“the active friend”, “the long-distance friend”, “the longtime-but-not-close friend”, and “the new friend”), four different types of confidants (“The strong-tie type”, “the companion type”, “the remote type”, and “the acquaintance type confidant”), and six different types of social support exchanges patterns (“the multiple engagements”, “exchanging help and socializing”, “counseling”, “socializing,” “receiving help,” and “providing help”). Further analyses show that over the life course, people have different types of friends, and change support exchanging patterns with their network members. And the different types of confidants have different effects in reducing depression.

The overall implication of my dissertation is that the social relationship cannot be simply interpreted by the general expectation of what a given relationship category would be. Rather, even though some social relationships are described as the same category of social relationship, each dyadic social relationship within the same relationship category has different quality and features from each other. And the varying quality or subtypes of relationships may matter more than the size or structure of their social networks for understanding how social context affects social networks and for estimating the effects of social ties on individuals’ outcomes.

Main Content

This item is under embargo until October 12, 2023.