Behavioral Transmission

Behavior change programs, including much of early education curricula, assume the positive transmission of behavior from one context to another. We randomize a hand hygiene edutain-ment program in schools in Bangladesh to trace school-to-home transmission of handwashing behavior and randomize the proportion of students who receive handwashing resources at home to track home-to-school transmission. We find that children induced to wash more at home exhibit less washing at school. Likewise, children induced to wash more at school wash less at home. This negative transmission spills over to other household members and non-school days, such that the cumulative impact of school edutainment on total washing is negative and potentially substantial: treated children exhibit higher rates of respiratory infection and loose stool, though these estimates are imprecise. Our results are consistent with the mechanisms of crowd-out, cue-based habit formation, and ‘reverse’ vertical transmission of behavior. They highlight an unintended consequence of behavior change interventions, like those often imple- mented in education, that presume complementarities in behavior across contexts but evaluate effects only at the site of intervening. contexts.