Sample design is a foundational step in survey research, but when sampling frames are unavailable, sampling strategies must become more strategic. This is a classic research challenge when it comes to studying hidden or hard-to-reach populations and illicit behaviors. Researchers at NORC at the University of Chicago decided to employ respondent-driven sampling (RDS), a variation of snowball sampling which uses mathematical modeling to adjust for the typical biases associated with referral-based sampling approaches.
Incentives are often important drivers of survey response rates, but in the context of RDS, a strategic, well-tracked incentive and referral scheme can also be instrumental in helping to collect the data required for population size estimation. To this end, unique coupons, redeemable once after an interview, can serve this purpose. As coupons are distributed to a hidden population through friends and acquaintances, links in the resulting dataset between respondents are established. After successive waves of interviews and coupon distributions, a population estimate can be derived using a variation of RDS and mark-recapture called link-tracing.
Join us on July 7th to learn about the following:
- NORC’s study of Commercial Sexual Exploitation of Children (CSEC) in coastal Kenya
- An introduction to RDS, mark-recapture, and link-tracing methods and sampling strategies, and why these methods were selected for this study
- QR code and coupon generation
- How SurveyCTO server dataset technology was used to track, validate, and monitor coupons in real-time
- Challenges and lessons learned
Participants will gain valuable guidance on choosing non-probability sampling methods like RDS. This event will be especially helpful for anyone engaged in research on hard-to-reach populations.