Respondent Driven Sampling (RDS)

Respondent Driven Sampling (RDS) is a network-based sampling method that is similar to the snowball method. Both methods are particularly suitable for research areas in which access to potential study participants from certain groups is difficult. In contrast to the snowball method, in which participants recommend other people from their social network as potential participants, the RDS extends this process to include mathematical models for weighting the sample. This makes the recruitment process more controlled and less random. This leads to a more representative sample, which means that the selected participants better represent the entirety of the target group. This makes the RDS a valuable tool in empirical research, especially in situations where conventional sampling methods reach their limits

Respondent-driven sampling (RDS)

Respondent-driven sampling (RDS) is based on the participation of individual people in a survey, who are called "seeds" in English, as they figuratively represent the seed that is first sown in order to then attract further participants for the survey. These people share the survey in combination with links or vouchers in their networks. The newly acquired study participants in turn forward the survey in their networks, creating chains of connections with multiple links. Such link chains contribute to the diversity of the sample and enable its growth, both in terms of size and quality.

In order to obtain as diverse a sample as possible, the first step is to ask as many different people ("seeds") as possible to take part in the survey. To increase the willingness to participate, an incentive (usually in the form of a sum of money) is paid both for participation in the survey and for the successful recruitment of further participants.

Further questions about the method:

The integration of the RDS sample into the NaDiRa.panel is a valuable addition to the other quantitative surveys: If carried out successfully, this can provide estimates for the entire Black, African and Afro-diasporic population in Germany, for example. The RDS method also enables probability-based estimates to be made in order to carry out in-depth analyses of differences between the groups mentioned above.

RDS has been shown to work (relatively) well when (1) the process starts with a small number of highly motivated individuals ("seeds") (e.g. intrinsically motivated community members who strongly believe in the cause); (2) they receive adequate training (e.g. in a workshop); (3) the study population is relatively homogeneous, highly networked and digital. However, RDS can fail if (1) the individuals ("seeds") have no direct contact with the researchers or instructors; (2) they are selected from a group of survey participants (instead of volunteers); (3) the study has no (immediate) impact on the respondents. It is currently unclear whether we can overcome these challenges, although the Afrozensus has shown that network-based (i.e. snowball) sampling methods are possible within the Black, African and Afro-diasporic community in Germany.

In addition, due to the need for linkage tracking for statistical inference, the survey is confidential rather than truly anonymous. However, personal data is only available to respondents, not to recruits; until a person completes and submits the survey, they remain anonymous. We believe that this fact, combined with the support and involvement of opinion leaders within the target group, will reduce respondents' concerns and sensitivities and encourage participation.

Contact persons

Zaza Zindel Researcher

Zaza Zindel

 Researcher

Selected literature

  • Leonard, M.M. Evaluating sampling methods for ethnic minorities. (2023). Statistics Canada Symposium Proceedings. Forthcoming.