Early Years Preschool Program Impact Evaluation 2017, Baseline Survey

This study is a randomized control trial (RCT) of the EYPP to determine its impacts on children's learning and development. An RCT is the most rigorous type of study design. In 2016, we randomly assigned 100 schools in the Meherpur district of Bangladesh to either receive the EYPP (n = 50) or to a no-program control group (n = 50). The children participating in the study from these communities are expected to enroll in government preprimary in 2019 and enter grade 1 in 2020. In the 50 treatment school catchment areas, children selected for the study will be invited to participate in the EYPP program at their local school in 2018, and then will go on to government preprimary as usual in 2019. In the 50 control school catchment areas, children selected for the study will be eligible to enroll in the government preprimary as usual in 2019, but will not have EYPP available to them the year before. This allows us to estimate the net effects of adding the second year of pre-primary education (EYPP) compared to having only one year of pre-primary education (business as usual).

We collected baseline data in all 100 communities in December 2017 and January 2018, with a sample of 1,856 children. We will conduct a mid-term outcome assessment in approximately November 2018 (just before the study children are expected to start the one-year government pre-primary program), and will conduct an end-line assessment just prior to on-time enrollment in grade one (approximately November 2019). The EYPP program will serve the children assigned to it between February and December 2018.

Sampling of Children

During an October 2017 visit to Meherpur, we learned that EYPP schools typically accepted approximately 18-20 children, and no more than 25 children. The EYPP staff expressed a preference for enrolling children within proximity to the school, and giving priority to children who live closer to the school or center. This preference is guided by the experience that children who live further away are less likely to regularly attend and their parents are less likely to be involved in the program. All schools visited stated that they did not expect any children to participate who lived further than a 15-minute walk from the EYPP class.

Data International conducted a census of every household within a 15-minute walk of the primary school. The resulting census included a total of 36,806 households across the 100 study communities. For each household, if there were any children ages 3-6 years old, enumerators recorded the child's name and date of birth, father's name, whether the child was currently in an education program (and if yes, what type), and what the family's plan was for the child in 2018 (stay home, or participate in the educational program). Enumerators also recorded the exact household location using GPS coordinates, and asked how many minutes it will take the child to walk from the home to the primary school.

The target sample for our study included all children in the census areas born from January 1, 2013 - December 31, 2013 (because on-time enrollment in government pre-primary school for these children would be in January 2019). In most cases (exact figure unknown but in a substantial majority), children's dates of birth were verified with the Extended Program of Immunization (EPI) card or a birth certificate. If these documents were unavailable (even after parents were encouraged to search), enumerators recorded what the parent reported as the child's date of birth. We identified a total of 1,986 children born in 2013. We did not exclude any age-eligible children based on any other criteria (for example, children with disabilities were included in our sample pool). See Appendix B for recruited sample size and percentage of target for each school/community, by upazila. See Appendix C for a copy of the informed consent for family recruitment into the study.

AIR agreed with the World Bank that we would sample an average of 20 children in each of the 100 study communities. Many communities had fewer than 20 eligible children. Because EYPP centers will typically enroll up to 25 children, for both treatment and control communities with 25 or fewer children, we included all eligible children in the study (with parental consent). In the 20 communities (14 treatment and 6 control) with over 25 children in the target age range, we drew a random subsample of 25 for inclusion in this sample.
Exhibit 2 shows the sample recruited for this study. Recruitment rates were very high among children sampled for this study. All communities and EYPP schools included in the sample participated in baseline data collection as planned.

Of the 1,856 children recruited for this study, 908 were girls and 948 were boys.