Skip to content

Evaluation of a Parental Engagement Program

Endline report on impacts of a parental engagement SMS intervention

Jonathan Torgovnik/Getty Images/Images of Empowerment

Endline report: Evaluation of a parental engagement SMS intervention in Lilongwe Rural - 370 KB

Download PDF


We present results of a randomised control trial aimed at estimating causal impacts of SMS nudges aimed to improve parental engagement with their children. The study took place in a single Traditional Authority (TA) in Lilongwe Rural, Malawi among 2,095 active GiveDirectly recipients. Half of GiveDirectly cash transfer recipients were randomised to receive messages on their phones with information on the importance of education and tips on how to engage with their children. The control group only received GiveDirectly transfers. We find no impacts across all pre-specified indicators. Contrary to experimental findings, we find overwhelming interest in the messages among recipients of the intervention and large self-reported impacts. We find substantial spillovers (much larger than anticipated) and about half of the treatment group reports sharing the messages with other people in their community. Roughly about the same percentage of the control group reports either receiving messages or discussing messages with other people. This limits our ability to conclusively state whether the intervention is impactful on the main indicators of interest– parental engagement and child school attendance. 

Given the inconclusive results, we cannot recommend scaling up the intervention with certainty. However, a few considerations such as relative cost of the intervention and heightened enthusiasm among recipients, could help GiveDirectly and interested funders make decisions. If the intervention is scaled, we advise further tailoring messages to the local context, conducting periodic descriptive studies to measure interest in the intervention by recipients, and systematically tracking important indicators (such as attendance rates) against pre-selected targets.