SPLASH School Outcome Study
The SPLASH School Outcome Study investigated the effect of a full WASH in Schools program on absenteeism and teacher-pupil contact time in the context of the USAID-funded Schools Promoting Learning Achievement through Sanitation and Hygiene (SPLASH) project in Zambia. The study was conducted in three intervention and three control districts. SPLASH implemented a comprehensive WASH in Schools program in the intervention districts (Mambwe, Chipata, and Lundazi), located in the Eastern Province. The control districts (Luangwa, Chongwe, and Rufunsa) were adjacent to the intervention districts but located in Lusaka Province. The intervention/control districts in the study were matched for cultural, economic, and ecological characteristics. No specific WASH in Schools program was implemented in the control districts while the study was underway.
The study was based on a quasi-experimental longitudinal design with data points collected during three school terms over a period of 10 months in 124 schools equally distributed into intervention and control study groups. A sampling framework was constructed in each province. Inclusion in the framework relied on the presence or absence of 11 criteria associated with a full school WASH package incorporating both infrastructure and hygiene education elements. Schools were randomly selected from each one of the regional sampling frameworks. Intervention schools met all 11 criteria, and control schools met only some or none at all. In selected schools, field enumerators visited odd grades (1, 3, 5, 7 and 9) and conducted roll call to determine pupil absenteeism and establish whether the teacher was present. In addition to roll call, enumerators also collected data on pupil absences during the two weeks prior to the school visit, identifying pupils who had been absent at least one day in those two previous weeks. They interviewed the parents of children so identified to understand the reason(s) for the child’s absence. To collect data, the researchers developed three instruments using ODK software and uploaded them onto six Samsung tablets running the Android operating system. Data were analyzed by month, school term, and season. The monthly analysis was descriptive. However, for school and season comparison researchers used repeated measures of analysis of variance. A gender analysis was conducted to determine differences between boys and girls. A confounder analysis was conducted using linear regression.
The study clearly establishes that improved WASH conditions and education in a school had a positive effect on pupil and teacher absenteeism and teacher-pupil contact time. Effects in the expected direction are clear during the measures taken in the 2015 school year but not in the last few months of the 2014 school year. The difference between intervention and control schools is evident when looking at the measure we termed de jure (past two weeks) absenteeism. In this case, the absenteeism differences detected may be at least threefold higher (or 300 percent) in control schools when the time dimension is seasons, but fivefold (or 500 percent) when the time dimension is school term. The differences when we look at de facto (day of data collection) absenteeism hover around 34 or 35 percent when we look at the data by seasons and may be as high as 100 percent when we look at the data by school terms. These differences occur in two of three seasons or terms examined. Further, pupil absenteeism findings remain even in the presence of three other potential confounders: the presence of a school feeding program, the presence of school improvement projects other than WASH, and the type of school. The WASH program seems to affect boys and girls equally. WASH in Schools programs positively impact school attendance for teachers and pupils but appear to have less of an impact on contact time.