83178 from EVIDENCE to POLICY Learning what works, from the Human Development Network November, 2013 Cambodia: Challenges in Scaling Up Preschools School is an essential part of giving children the chance to The World Bank is committed to ending extreme poverty acquire the skills they need to reach their potential, raise and building shared prosperity and ensuring that children are healthy families and live a life free of poverty. But what’s healthy and ready to learn is a critical part of this. Recently, the right time to start school? A growing body of evidence researchers worked with the Government of Cambodia to shows that children who evaluate the impact of three pilot early childhood develop- first enter a classroom at ment programs that were being scaled up with assistance age five or six don’t al- from the World Bank. The evaluation found that expand- ways have the necessary ing programs can lead to new challenges that might not be basic learning skills, apparent when working on smaller pilots. In this case, un- putting them at a dis- foreseen implementation problems—some related to delays advantage that can lead in building schools or problems paying teachers—limited the to failure later on. In program’s reach and effectiveness. Parents who worked all day, North America and Eu- or worked far from the village center, also found the part-time rope, research has found that preschools can have a positive hours of the preschools difficult to manage, contributing to EDUCATION impact on getting children ready for the demands of pri- low use of the preschool services. The results of the evalua- mary school, especially among low-income children. But in tion highlight challenges governments can face as they move the developing world, preschools haven’t been as widely im- from pilot to scale up. Building on the lessons learned from plemented or evaluated. Pilots that have been implemented this evaluation, the government of Cambodia is planning in several countries in Latin America, Africa and Asia have a revised approach in its 2014-2018 strategic education resulted in improved skills for children, but it’s still unclear plan to strengthen the quality of preschools and ensure how to replicate these at scale. demand for these services. Context With assistance from the World Bank and the Education To scale up the program, the government decided that Fast Track Initiative Catalytic Fund (now the Global Part- communities with an existing primary school that needed nership for Education), the Government of Cambodia to be upgraded and/or expanded would receive a formal pre- sought to improve early childhood programs for the rural school as part of the renovations; and communities that didn’t poor by expanding and evaluating three early childhood qualify for renovation work, but had a high poverty rate and a development options: formal preschools, run by the Min- large number of under age five children, would get a commu- istry of Education, informal community-based preschools, nity-based preschool or a home-based program aimed at im- and home-based programs. The latter two were being pi- proving parenting practices. The formal preschools, staffed by loted by UNICEF and Save the Children Norway in a few government teachers, were for children aged three to five years provinces. old and operated four hours a day. The community-based pre- schools, covering the same age, operated for three hours a day in 480 new community preschools; and creating 450 new home- a space chosen by the community and were staffed by commu- based programs. The scale-up occurred throughout the country. nity members who received training and a small stipend. The home-based program targeted parents of children from birth Did You Know… Preschool offerings are expanding rapidly in Cambodia, but they are ac- to five and were run by trained women from the community. cessible to just a minority of the country’s children. Some 170,000 children The women were supposed to hold monthly group meetings to age three to five were enrolled in early childhood education in 2009–2010, discuss good parenting and developmentally appropriate activi- which was 20 percent of all kids in that age group. This is up from 2005–2006, when an estimated 120,000 children were enrolled, or about ties to do at home. 13 percent of the age group. The expansion was implemented between 2009 and 2011, There’s also a large discrepancy between urban and rural areas. While 15 with the goal of doubling access to early childhood development percent of Cambodia’s children live in urban areas, these children make up 25 services by: building 650 new formal preschool classrooms in percent of the total preschool enrollment. the same number of newly renovated primary schools; creating (Source: UNESCO) Evaluation The impact evaluation was designed to help the government ing them to be used as a control group. Baseline surveys in De- EDUCATION determine which preschool model worked best and researchers cember 2008 and endline surveys in June 2011 were conducted relied on randomization to identify impact. Randomization in 26 treatment villages and the 19 control villages for a total of was implemented separately for the formal and informal pre- 1,553 households. To evaluate the informal preschool models, school programs because the criteria used to select participat- 450 villages were randomly selected in 10 provinces and were ing communities differed. In practice, the experimental design equally divided between the control group and the two informal included five groups: three treatment groups (for each of the models. Researchers surveyed 32 randomly-selected villages in three interventions) and two control groups (one for the for- each group, for a total of 3,807 households. The baseline was mal sample and one for the informal sample). Randomization collected in May 2008, with an endline in January 2011. made it possible to compare outcomes in the different com- For the data, researchers conducted a household survey, a munities to determine the impact of each intervention. mother/caregiver survey and tests that measure a child’s cogni- To create a control group for the formal preschools, re- tive, socio-emotional, linguistic, gross motor and fine motor searchers relied on the program’s phase-in timeline. Not all skills. These were done at baseline and then repeated for the schools could be renovated at the same time, so 19 were follow-up two and a half years later. picked to be upgraded in the third year of the program, allow- Findings The scale-up failed to enroll the majority of with home-based program had a 12 percent participation children in either the formal preschools or the rate. informal programs. Reasons for the low enrollment varied. In interviews, parents cited the cost of buying supplies—such as pencils, In villages where a formal preschool was built, 40 percent pens, notebooks, and also clothes as a reason for not send- of children participated. Villages with a community-based ing their children to the formal preschools or the community- preschools had a 34 percent enrollment rate and villages based ones. Families that normally took their young children This note is based on the paper, Impact evaluation of three types of early childhood development interventions in Cambodia, by Adrien Bouguen; Deon Filmer; Karen Macours; and Sophie Naudeau. 2013. Policy Research working paper, no. WPS 6540; Impact Evaluation series; no. IE 97. Washington, DC: World Bank. with them when they went to work outside the village said Treatment and control groups were also mixed up the early ending times for preschool was an obstacle because when implementing the community-based preschool they couldn’t come to get their children. Insufficient public program, which “contaminated” the samples. outreach was also part of the problem. Some parents thought their children were too young to qualify for preschool, while Thirteen percent of control students lived in a village with others didn’t even know a preschool was available. access to a formal preschool program. The same happened Teacher retention in the community-based preschools also in the informal sample: 11 percent of households in the emerged as a key obstacle to running the program. During control villages had access to a community-based program, the first year, many teachers in community preschools didn’t while 28 percent had access to a home-based program. receive promised stipends and many left their jobs. Villages had a hard time finding replacements because stipends were Still, some results that could be measured were so low—about $8 a month—and not paid regularly. Without noteworthy: For example, children in communities teachers, the community-based preschools couldn’t enroll stu- with a formal preschool registered for primary dents. Because the programs needed to find a structure, space school closer to the required age, as opposed to was often also an issue and some families believed their child the others who signed up a little earlier. could not enroll as a result. Home-based programs, which had the lowest enroll- On average, children in villages with formal preschools were ment, didn’t appear to have been implemented in most areas. 71 months old when they enrolled in primary school (offi- Among four villages visited in May 2012, only one seemed cially, children should be enrolled at 70 months of age,). But EDUCATION to have a home-based program. In addition, even when there children in villages without preschools enrolled earlier, at an was a program, parents confused it with other community average of 68 months. programs related to education and health. Researchers discovered that formal preschool led fami- lies to comply more closely with formal school registration The scale-up also faced other implementation age requirements. It seemed that children without access to problems, which affected the evaluation itself. preschool often were enrolled earlier in primary school. In villages where there wasn’t a preschool, more than half of Construction teams didn’t always follow plans when it came children registered before the official age of 70 months; in to the schedule for renovating primary schools. In some cases, villages where a preschool was available, only 40 percent of communities that weren’t supposed to receive schools until children were younger than the official age when they en- the last year got them earlier, while schools that were sup- rolled in primary school. It may be that in villages with pre- posed to be at least partly ready by the 2009-2010 school year schools, school administrators are stricter in enforcing the weren’t ready until the next school year. As a result, many stu- minimum age requirement. dents in the formal sample were evaluated after having been in the program for an average of just nine months instead of Researchers found that test scores of children with the anticipated two years. access to the formal preschools were for the most Overall, only 56 percent of villages that were slated to part unaffected. However, those between 66 and receive a community-based program actually had a work- 78 months at endline did worse than the control ing program. The home-based program, which consisted group when it came to cognitive skills important for of monthly meetings, was also not implemented accord- school readiness. ing to plan. While monitoring data based on answers provided by village chiefs showed that 70 percent of vil- Because much of the treatment group either did not par- lages in the home-based group implemented a program, ticipate in the preschool program or participated for just a the true number appeared to be much lower. In fact, in few short months, researchers expected to see little impact. several of these villages, not a single parent interviewed School readiness outcomes—cognitive, socio-emotional, had even heard of the program—despite the fact that the linguistic and physical—remained unchanged. Nevertheless, village chief (who was responsible for implementation) there was a negative impact on these school readiness skills for said it existed. children between 66 and 78 months at endline. This seemed to be because of delayed enrollment: Instead of attending primary The informal programs, meanwhile, had almost school like many in the control group, children in the treat- no impact on participants’ development or school ment enrolled in preschool instead. This may have put them at readiness. a disadvantage if preschool teachers were not as prepared and motivated as primary school teachers or if the preschool cur- Researchers tested students’ cognitive, socio-emotional, riculum was geared towards the youngest children in the class- linguistics, and physical development, but these indica- room. For this specific age group, being in the control group tors remained largely unchanged as a result of the inter- and attending primary school (instead of going to preschool) vention, likely due to implementation problems and low appears to have been better. participation. Conclusion Early learning opportunities for children in the developing munities to ensure parents enroll their children and take world are key to ensuring they have the tools they need to advantage of the opportunities available. Some possibilities learn and succeed. Several programs and pilots across the de- include communications campaigns that target parents, so veloping world have had encouraging results, but more evi- that they better understand the potential benefits of early dence is needed to determine not just what works—but how childhood development programs. Conditional cash trans- to make the programs work on a large scale. As this impact fers may also help families offset financial constraints. At the evaluation shows, what works on a small scale does not al- same time substantial effort needs to be made to ensure that ways work on a larger one. When scaling up such programs, the services delivered are at high quality and effective. policymakers should consider working closely with com- The Human Development Network, part of the World Bank Group, supports and disseminates research evaluating the impact of development projects to help alleviate poverty. The goal is to collect and build empirical evidence that can help governments and development organizations design and implement the most appropriate and effective policies for better educational, health and job opportunities for people in developing coun- tries. For more information about who we are and what we do, go to: http://www.worldbank.org/sief. The Evidence to Policy note series is produced by SIEF with generous support from the British government’s Department for International Development. THE WORLD BANK, HUMAN DEVELOPMENT NETWORK 1818 H STREET, NW WASHINGTON, DC 20433 Produced by Office of the Chief Economist, Human Development Network, Communications/Aliza Marcus and Daphna Berman