Recruiting experimental subjects using WhatsApp

The aim of many experiments is to estimate the effect of different interventions on subjects' decision making. However, obtaining large samples and internal validity is challenging. This paper presents an alternative device at almost no cost that can easily provide a very large number of participants (700 in 1 hour). We asked 14 students to invite their WhatsApp contacts to participate in an online experiment. The students created a total of 80 diffusion groups with 25 contacts each. Using the diffusion groups as clusters, we ran a cluster randomization procedure in order to assign subjects to a framing experiment (treatment + control). We obtained the same level of attrition, duplicates and uninvited subjects across the treatment and control groups. Moreover, the experiment yielded consistent results in line with the framing literature.