Abstract

Over recent decades, declining survey response rates—particularly among specific groups in society—have raised significant concerns. In nationally representative surveys that aim to make inferences about the general population, systematically undercovering or undersampling of particular groups, such as young adults, may lead to biased estimates. The difficulty of reaching and motivating young adults to participate in surveys calls for other data collection methods, including forms of non-probability sampling. According to the literature, one of the most promising non-probability sampling methods is Respondent-Driven Sampling (RDS). RDS is a chain-referral recruitment method that has been widely used for marginalized and hidden populations; more recently, it has also been applied to broader hard-to-reach groups. In theory, RDS offers advantages over other non-probability sampling methods because it employs estimators that attempt to correct for certain biases. This makes it a potentially interesting method for national statistical organisations, such as Statistics Netherlands.
In this presentation, we discuss an ongoing experiment on the use of RDS among young adults in the Netherlands to recruit participants for an online survey. Our aim is to explore the opportunities and challenges of applying RDS in the context of national statistics through a series of pilot studies. We describe the overall experimental design and focus in detail on the first pilot. Taking a practical perspective, we focus on the data collection process, addressing issues such as target group characteristics, recruitment strategies and social networks of young adults, the role of incentives, alternative modes of survey distribution, risks of survey fraud, and technical challenges. We outline the different design options considered for the first pilot and share insights from two focus groups with young adults regarding their perceptions of and ideas about this method.