24  Collecting Ego-Network Data

Social scientists employ a variety of approaches to collect and analyze ego network data, which centers on an individual (the “ego”) and their direct connections (the “alters”), as well as the relationships among those alters. This specialized form of network analysis seeks to understand personal social structures by gathering specific information from the ego’s perspective.

24.1 Methods of Data Collection

While network data can be derived from surveys, direct observation, or large-scale digital traces (big data), surveys are considered the optimal method for collecting ego network data. This involves directly asking individuals about their relationships.

The information collected typically includes:

  • Details about the ego themselves.
  • The nature of relations (role, exchange, interaction) between the ego and their alters.
  • The relationships among the alters, as perceived by the ego.
  • Specific properties or characteristics of each relationship.

24.1.1 Measuring Ego Networks: A Four-Step Process

When conducting a survey to measure ego networks, social scientists typically follow a structured four-step process:

  1. Define the Relation of Interest: The first crucial step is to specify the type of relationship the researcher wants to study. This definition directly informs the name generator, which is the question or prompt used to elicit the names of alters from the ego.
  2. Specify a Maximum Number of Alters: Researchers typically set an upper limit on the number of alters an ego can name, usually ranging from 5 to 20. For instance, the NetHealth study allowed egos to name up to 20 alters.
  3. Identify Alter and Tie-Level Characteristics: This involves selecting specific characteristics about the alters and the relationships (ties) themselves that are relevant to the research question. This process utilizes the name interpreter, which consists of follow-up questions designed to gather detailed information about the named alters and their connections.
  4. Collect Information on Alter-to-Alter Relations (Optional): This final step involves gathering data on the direct relationships or connections that exist between the alters themselves, independent of their direct ties to the ego. This information is optional but vital for understanding the internal structure and clustering within the ego network.

24.2 Name Generators: Four Main Approaches

The choice of name generator is critical because different approaches can produce ego networks with distinct structural characteristics. The four primary approaches are:

  • Role-Relation Approach: This method asks egos to list individuals based on their social roles or pre-defined relationship categories. Examples include “List the names of up to [X] friends, co-workers, co-authors, pub-mates, etc.”.

    • Problem: This approach can suffer from low reliability, cultural specificity issues (e.g., how “friend” is interpreted can vary), and heterogeneity in how labels are understood across different respondents.

  • Interaction Approach: This approach focuses on the frequency or duration of contact between the ego and others. An example is, “List the names of up to [X] of the people you most frequently interact with”. The NetHealth study utilized an interaction-based name generator, allowing up to 20 alters.

    • Problem: While useful, frequency of interaction does not always equate to importance or closeness, suggesting it may need to be combined with other approaches for a more comprehensive understanding.

  • Affective Approach: This method elicits names based on subjective sentiments or cognitions. An example question is, “List the names of up to [X] of the people you feel closest to, trust most, etc.”.

    • Problem: Similar to the role-relation approach, it can have reliability issues. However, it is effective for identifying strong or subjectively significant ties, although it tends to elicit smaller networks.

  • Exchange Approach: This approach centers on the exchange of resources, information, favors, or support. An example query is, “List the names of up to [X] of the people you would go to for advice, financial support, etc.”. This approach is a natural fit for studies on “social capital”.

    • Advantages: It generally exhibits fewer reliability problems and can effectively capture both weak ties and a larger number of alters.

24.2.1 Famous Examples of Name Generators

  • Wellman (1979) East York Study: “Name the six persons outside your home that you feel closest to”.
  • Fischer (1982) Northern California Communities Study: “Name all persons who would provide any of eight types of aid”.
  • Marsden (1987) General Social Survey (GSS): “Named all persons with whom you have discussed matters important to you,” used to elicit core discussion networks.

24.3 Name Interpreters: Types of Measured Traits

After alters are named, name interpreters gather detailed information about them and the relationships. This includes:

  • Socio-demographic characteristics of alters: Such as sex/gender, age, race, years of education, and religion.
  • Relational traits: Like relationship to an organization, specific relationship types (e.g., family, friend, co-worker), shared dormitory/roommate status, and frequency of contact.
  • Sentiments and Affect: Measures of closeness (e.g., “especially close,” “merely close,” “less than close,” “distant”), trust, or negative sentiments. Subjective closeness is considered a reliable indicator of tie strength.
  • Exchanges and Support: Types of support provided (e.g., financial help, comfort, advice, social support), and topics of discussion.
  • Relationship Age or Duration: How long the ego has known the alter.
  • Subjective Similarity/Cultural Matching: Whether the alter shares tastes, attitudes, or engages in similar activities with the ego. This relates to the concept of cultural matching in networks.
  • Other Traits: Such as health, happiness, activity levels, and sleep patterns (as seen in the NetHealth study).

24.4 Alternative Approaches and Considerations

Beyond the standard name generator/interpreter surveys, other methods for collecting ego network-related data in social science include:

  • Position Generators: These surveys focus on an ego’s access to various social positions (e.g., occupations) within the network. They are often used in social capital studies and are advantageous for their high face validity, quickness, ease of standardization, and ability to capture significant information with minimal respondent strain. They can be followed by “position interpreters” to collect data on ties linking the ego to the position.
  • Resource Generators: Similar to position generators, these questions ask about access to specific resources (e.g., someone who can repair a car, plays an instrument) and through which types of alters these resources are accessed. They combine the practical advantages of position generators with the broader coverage of name generators, allowing for the capture of multiple dimensions of social support and the empirical linking of resources to role-relations.
  • Visual Network Scales: These involve presenting egos with idealized network diagrams and asking them to rate how closely their own network resembles these scenarios. This method can be used to understand how individuals perceive their position in a social network or social network change over time (e.g., moving towards the periphery or center of a network).
  • Cognitive Social Structures (CSS): While not strictly a “generator” for ego networks, CSS is a survey-based approach that measures social ties by asking individuals about their perceptions of relationships between other people in the network, going beyond just their direct ties. This can yield “slices” (network as perceived by one individual), locally aggregated structures (where both parties agree on a tie), or consensus structures (where a proportion of others agree on a tie), adding a cognitive-psychological dimension to network analysis