The Who and the Why? Selection Bias in an Unconditional Basic Income Inspired Social Assistance Experiment

It is well documented that it is important to take selection effects into account when analysing social experiments. A Randomised Controlled Trial (RCT) design usually prevents selection bias, but not when participation is voluntary. Despite the abundance of literature on the existence of selection bias, few studies provide in-depth insights on how these selection effects take place in practice, and what causes different groups to be over—or under-represented. Nijmegen is one of a number of Dutch municipalities that conduct an experiment with the social assistance system, loosely inspired by the Universal Basic Income (UBI). Participants receive their allowance with less conditions, and get the opportunity to earn additional income. In this chapter, selection effects are tested, using registry data of participants and all non-participants in the population. In addition, qualitative data are used to interpret the selection effects we find. Several characteristics turned out to increase or decrease participation: education, country of origin, being single, having an exemption of the obligation to work, and part-time work providing additional income. Further, we propose that stress is an important deterrent for people to participate in an experiment like this, even when the experiment is aimed at (among others) stress reduction, like the one in Nijmegen.