Personal Network Analysis

Personal Network Data Collection

In contrast to sociocentric network analysis, where the goal is to collect data from all actors within a bounded setting, much of the personal network data collection process is similar to any population-based survey. It begins with the identification of a population, the selection of a survey mode and a sampling frame, and power analyses to determine sample sizes and the process of sample selection. It differs from most surveys in the elicitation of a list of network alters from the respondent using name generators and by asking the respondent to report on characteristics of each alter (name interpreters) and about the ties between alters. Name generators vary widely in wording and design, which has consequences for estimates of network size and of several network compositional and structural characteristics. The variations in name generators partly reflect differences in research objectives, such as whether the focus is on social support or on a representation of the wider social context. Variations also concern the use of a single versus multiple name generator, the use of [Page 702]bounded recall (e.g., “with whom did you interact in the past week?”) versus prompts (e.g., “with whom do you discuss important matters?”), probing free versus fixed numbers of nominations, and question order. It is not unusual for personal network elicitation to be restricted to certain types of alters, such as those providing social support or with whom the respondent shares certain behaviors. Most personal network studies do not limit the domains within which these relations exist. It is often the variability in the number and content of these domains that provides the most predictive power.

Once the data have been collected, personal networks are typically analyzed at the ego-alter relationship level and/or the network level. Analyses at the relationship level tend to focus on the explanation of variations in relationship strength and quality and typically involve multilevel analysis, with relationships nested within personal networks. Some researchers are exclusively interested in the network (or ego) level. In that case, the alter characteristics are summarized into personal network composition variables and alter-alter ties into structural variables using social network metrics. The analyses then focus on the explanation of variations in network size, composition, and structure, and on using these variations to predict individual outcomes that are thought to be related to the social context.

...