Social Sampling

This report provides the reader with an overview of the concept of social sampling. Social sampling is a class of methods where properties of a social network are used to obtain more accurate samplers than using classical polling algorithms. The main interest lies in the correlation between the sampling size and the sampling error and the systematic bias due to the network structure. I define the concept of a sampler and present a Naive as well as an Ideal sampling algorithm. Further I show and prove some of their most important properties regarding sample size and bias. Eventually I present more realistic and efficient estimators and their real-world performance and demonstrate that social sampling is a powerful tool to obtain accurate estimates with very few samples.