#SafeSigns

Reach Digital Health (previously known as Praekelt.org) partnerered with The Behavioural Insights Team (BIT) to explore digital solutions to support young women in South Africa to navigate unhealthy relationships and to expand knowledge on tackling intimate partner violence (IPV). Drawing on a review of relevant IPV and behavioural science literature, we designed #SafeSigns launching it under the name ChattyCuz, an interactive, WhatsApp-based chatbot to deliver content to young women (18 - 24 year olds) to improve attitudes related to gender power dynamics. We conducted a randomized control trial (RCT) with experimental design to identify the most effective chatbot experience to improve attitudes, build skills and ultimately lower the prevalence of IPV. 

We tested the impact of delivering content using two interaction designs, namely gamified and narrative chatbots, and a chatbot that only provides information (Business-as-usual control), compared to receiving no intervention (Pure control). The intervention was designed through a participatory approach, including the perspectives from end-users in the initial design stage. 

While promising evidence exists on the use of traditional media or videos to promote help-seeking or help-giving behaviors or gamified platforms to improve health outcomes, this study is among the first to assess IPV exposure after randomizing participants to digital, interactive platforms. Our findings will inform programmatic decisions for digital interventions in this space moving forward.

Impact or change made

During the RCT, 19,643 young women were recruited to ChattyCuz over one month. We found that the three chatbots effectively improved users’ attitudes and beliefs in power equity compared to the no-chatbot control. The gamified chatbot was the most effective in eliciting the desired outcomes overall, with moderate and significant effects on young women’s gender beliefs and skills related to identifying unhealthy behaviors.

Both interaction designs (gamified and narrative) also had twice as high user retention rates (56% and 60%, respectively), compared to the information-only chatbot (24%). This in itself is an encouraging finding regarding our intervention design, and could offer lessons for future digital interventions for young women in low- and middle-income settings.

These findings have implications for digital health, adolescent health, and IPV prevention fields. As South Africa reports among the highest rates of IPV exposure globally (five times the global average), designing interventions for this setting is an urgent and pressing challenge.