ABSTRACT
Information on connected egocentric networks or even a population-wide network structure is needed when aiming to impute a social network in agent-based simulations. The snowball study employed a survey method and instrument that allows the collection of information on both personal networks as well as a population-wide network structure. This program contributes to the scientific discussion in two dimensions: There are empirical results on network structures and there are results on survey methods. A combination of the personal network approach and snowball sampling was chosen as the best fitting survey method. Snowball sampling includes three major sources for bias: Selection, homophily, and degree bias. The survey team developed different measures to address bias: A multi-contact strategy, a web-page, a greeting card, and a gratification strategy were employed successfully. Finally, the survey included a chain method-related strategy to increase the response rate.
