Calling All Parents: Leveraging Behavioral Insights to Boost Early Childhood Outcomes in the Developing World

Abstract

Background Parental stress and poverty have been associated with reduced cognitive and socioemotional development in children, as well as poorer physical and mental health. While there is evidence that home visits contribute to improve the childrearing environment, these programs are expensive and difficult to scale. The large prevalence of risk factors and limited resources among families with small children in developing countries calls for the development of effective scalable parenting support programs. Aim To develop and experimentally evaluate an over-the-phone support and messaging program for highly vulnerable families with children aged 0 to 3. The intervention aims at helping families access government benefits, promoting positive parenting practices at home, fostering language development in children, and providing personalized assistance to families. The intervention The program combines weekly phone calls to parents by a program facilitator, with text and audio messages, a chatbot, and an artificial intelligence tool that provides parents with automatic feedback on their language interactions with their children. Targeted at vulnerable families with children aged 0 to 3 who qualify for support from the Ministry of Social Development in Uruguay, the intervention spans 8 months. In the first 4 months it connects families to social benefits to which they are eligible, and addresses general parenting beliefs, attitudes, and practices. In the last 4 months, it focuses specifically on promoting language stimulation. The program design is rooted on behavioral economics tools. Methods We evaluate the program through a randomized controlled trial involving 1,360 families. The treatment group received calls and text and audio messages for eight months, with free access to the chatbot and feedback tool. The control group did not receive any teleassistance or messaging but had access to a restricted version of the chatbot containing information on local resources. Our evaluation includes two follow-up telephone surveys, at four months and eight months after program initiation, and language metrics derived from WhatsApp audios of parent-child conversations submitted after the intervention. Findings The program increases families’ access to government benefits and labor programs by 0.3 standard deviations (SD). It also improves parental knowledge of language development and parental involvement in cognitive stimulation (by 0.19 SD and 0.15 SD, respectively). Furthermore, treated mothers increase their average number of words per minute in audios with their children (by 0.37 SD), and augment their voice pitch range (by 0.36 SD), a feature of linked to communication quality. Finally, mothers in the treatment group show higher well-being (0.16 SD) and lower levels of stress (0.2 SD). Our analysis of mechanisms suggests that access to cash transfers is the main reason behind stress reduction and wellbeing improvement. The intervention also appears to increase parental involvement more effectively among families facing more economic difficulties at baseline. More educated families get more involved in the program and experience larger effects in terms of wellbeing and child stimulation. Conclusion We show that an over-the-phone intervention, combining messages, calls and AI, can have substantive positive effects on family’s wellbeing and the childrearing environment, with promising scalable features.