Impact evaluations focused on school absenteeism commonly use school records of untested quality or expensive spot-check data. We use a large dataset on more than 30,000 unannounced random spot-checks for 6,000 female and male students across 30 schools in Western Kenya, with equivalent school record entries. First, we confirm that while absenteeism is common overall, pubescent girls miss more schooldays than boys because of high incidence of school transfers. Second, we test the quality of school records using Cohen’s kappa-coefficient, revealing non-random inconsistencies across the two data sets. Lastly, we conducted a three-arm cluster randomized control trial that provided sanitary products to schoolgirls to reduce absenteeism. Using the school record data, we would draw erroneous conclusions regarding the program treatment effects. Using the spot-check data, we confirm that providing sanitary pads reduces absenteeism by 5.4 percentage points.
Columbia SIPA. Center for Development Economics and Policy
Record created by