What It Needs for a Social Ego: A Survey-Based Comparison of Social Network Measures

Wednesday, July 16, 2014: 6:00 PM
Room: 416
Oral Presentation
Volker LANG , University of Tuebingen, Tuebingen, Germany
Steffen HILLMERT , University of Tuebingen, Tuebingen, Germany
In this paper we compare different methods of measuring social capital in conjunction with alters’ resources. Our analysis is based on a survey of two succeeding, complete cohorts of sociology students from one university. This survey includes detailed information on network ties between students. In addition, we use a resource generator and a position generator to assess ego’s social network beyond his/her fellow students. For all social network instruments, we differentiate between relationship strength, the basis for social capital, and the level of resources that are accessible via these social ties. We implement the same distinction with respect to measures of family background.

This unique design allows us to construct methodologically interesting counterfactuals. We can assess how the measured distribution of ego’s network resources differs if the accessibility of resources is not taken into account; how well we approximate ego’s resource distribution when based exclusively on indicators of relationship strength; and under which conditions – regarding ego’s network structure – we can measure social capital and related resources by using solely name generator based instruments. Furthermore, we compare these counterfactuals between different sampling designs. We analyze which path length is necessary to approximate the full cluster solution using a respondent-driven cluster sampling design.

Our study is constructed as a generic methodological study. Based on our results, we can specify which conditions regarding sampling designs and measurement instruments are necessary to adequately represent the social embeddedness of egos. These results are of general importance to researchers conducting larger-scale surveys, e.g. international survey projects which rely on cluster sampling designs and/or name generator based instruments when assessing the social networks of respondents.