Handwashing behaviour and habit formation in the household: evidence of spillovers from a pilot randomised evaluation in South Africa

Handwashing with soap at critical times is a simple and effective way to prevent the spread of communicable diseases, such as diarrhoea and acute respiratory infection, which are major causes of morbidity and mortality in the developing world. However, rates of handwashing remain low, and interventions which attempt to improve handwashing behaviours have largely been unsuccessful in practice. In 2014, we conducted a pilot randomised evaluation in poor urban community in Cape Town to measure the impacts of HOPE SOAP© – a translucent bar of soap with a toy inside of it, designed to make handwashing fun for children. In the pilot, 229 households were randomly assigned to receive four deliveries of HOPE SOAP© over a period of eight weeks. Analysis found that that HOPE SOAP© had positive impacts on children's handwashing behaviours and health outcomes. In this study, we expand upon the previous analysis, and examine the spillover effects – or indirect effects – that the intervention had on other members of children's households. Specifically, we employ OLS regression analysis to investigate the impacts of HOPE SOAP© on the handwashing behaviours of children's primary caregivers (based observation, where a snack was provided to caregivers and data were recorded on handwashing prior to preparing the snack), and on the health outcomes of all nontreated household members. Evidence shows that a child's assignment to HOPE SOAP© had a positive impact on the handwashing behaviour of their caregiver: the intervention increased the probability that a caregiver washed their hands before preparing a snack (at a second instance of observation) by 13 percentage points on average (p-value 0.17). Investigation of the causal mechanisms for this effect suggests that HOPE SOAP© affected caregiver behaviour both by disrupting existing poor-hygiene habits, and strengthening handwashing norms within households. Specifically, the intervention effect was larger for caregivers whose households had a lower cleanliness score at baseline (b: 0.19, p-value 0.12); suggesting that the intervention may have been effective at disrupting bad hygiene habits. Despite its positive effects on behaviour, HOPE SOAP© had no discernible short-term impacts on the health of individual household members. Nonetheless, the positive impact of HOPE SOAP© on caregiver handwashing behaviour suggests that the intervention's effects extend beyond improvements in outcomes for children; and indicate that HOPE SOAP© may be successful in inducing habitual handwashing behaviours which can persist in the long-run. Thus, a future large-scale randomised controlled trial is warranted to test this promising intervention.