Jens Malmros: Some advances in Respondent-driven sampling on directed social networks

Respondent-driven sampling (RDS) is one of the most commonly used methods when sampling from hidden or hard-to-reach populations. The RDS methodology combines an improved snowball sampling scheme with a mathematical model that is able to produce unbiased population estimates given that some assumptions about the actual recruitment process are fulfilled. One critical assumption, which is not likely to hold in most cases, is that the underlying social network of the population is undirected. The papers in this thesis provide extensions of RDS theory to populations with partially directed social networks.