AFRICA REGION HUMAN DEVELOPMENT U1 W OR KI N G PAPER SE R I ES THE WORLD BANK Edrly Cildhood Development in AfriCd Cdn WVe Do More for Less? A Look dt the impdCt dnd ImpliCdtions of Preschools in Cdpe Verde dnd Guined 22614 Hf- , )l 7~~~~~~~~~~J u2 _ j2g00 N , Other Titles in This Series Dynamic Risk Management and the Poor-Developing a Social Protection Strategy for Africa Engaging with Adults-The Case for Increased Support to Adult Basic Education in Sub-Saharan Africa Inclure les adultes - Pour un appui a 1'education de base des adultes en Afrique subsaharienne Enhancing Human Development in the HIPC/PRSP Context-Progress in the Africa Region during 2000 Africa Region Human Development Working Paper Series Early Childhood Development in Africa: Can We Do More for Less? A Look at the Impact and Implications of Preschools in Cape Verde and Guinea Adriana Jaramillo Karen Tietjen Africa Region The World Bank ii AFRICA REGION HUMAN DEVELOPMENT WORKING PAPER SERIES © July 2001 Human Development Sector Africa Region The World Bank The views expressed herein are those of the authors and do not necessarily reflect the opinions or policies of the World Bank or any of its affiliated organizations. Cover design by Tomoko Hirata. EARLY CHILDHOOD DEVELOPMENT IN AFRICA: CAN WE DO MORE FOR LESS? Contents Foreword v Executive Summary 1 I Introduction 5 2 Early Childhood Care and Development Background 7 What is early childhood development and why is it important? 7 What are early childhood care and development programs? 7 Why invest in ECCD? 8 Does Africa need early childhood development programs? 11 3 Study Methodology 14 Approach and sample selection 14 Limitations of the study 15 4 Supply: Characteristics of the Preschool Sector in Guinea and Cape Verde 16 The preschool sector in Guinea 16 The preschool sector in Cape Verde 17 5 Demand: Who Goes to Preschool and Factors Influencing Attendance 20 The story 20 Some policy implications 23 6 Impact: The Effects of Preschool on Child Development 25 The story 25 Some policy implications 29 IV AFRICA REGION HUMAN DEVELOPMENT WORKING PAPER SERIES 7 Effectiveness: Programs or Program Characteristics That Appear To Be Most Effective in Terms of Child Development 31 The story 31 Some policy implications 36 8 Conclusions and Recommendations 39 Preschool education can be a double-edged sword that can both alleviate and exacerbate disparity 39 More and better can be done with less 39 Government can play a critical and essential role in supporting early childhood care and development programs 40 What are the next steps in supporting ECCD in Africa? 41 Annex 1: Study Method 43 Approach 43 Sample and samplingframe in Guinea and Cape Verde 44 Instruments 45 References 47 EARLY CHILDHOOD DEVELOPMENT IN AFRICA: CAN WE DO MORE FOR LESS? V Foreword O ne of the major challenges education policy- rationales-human rights, basic needs, economic and 0 _ makers in Africa face is how to reduce ineffi- social benefits-have been presented and docu- ciencies as well as inequity in the allocation mented, and in many developing countries, national and use of financial and human resources in educa- early childhood care and development programs tional systems. Often it is found that relatively high- have been incorporated into the social sector invest- cost interventions do not produce the expected ment plans. The Declaration of the World Education outcomes in terms of learning achievements, or even Forum in Dakar in 2000 stated as one of its goals to in keeping children in school. expand and improve comprehensive early childhood Children are Africa's most vulnerable population. care and development, especially for the most vul- They are most vulnerable to risk, whether economic, nerable and disadvantaged children. This is not yet environmental, geographic, political, or cultural. the case in Africa, where a very small percentage of Thirty-six percent of the annual number of under-five children have access to Early Childhood Care and deaths are African. The prevalence of malnutrition Development (ECCD) programs. While the benefits among African children is still very high, and Africa of ECCD are increasingly well known to policy- is the only region where malnutrition rates did not makers, the ability of both donors and governments decline in the 1980s and 1990s. Inoculation rates con- to act is constrained by insufficient public-sector tinue to lag behind the rest of the world, with about budgets, costs, management, and financing strategies 51% of children fully immunized. within the African context. As African governments struggle to provide edu- Interest is growing in ECCD in Africa, and the cational opportunities to their school-aged children, World Bank is currently supporting ECCD and the extreme poverty of many families is a major con- related school health and nutrition programs in more straint to enrolling their children in school or sup- than 14 countries. Despite the recognized benefits porting them once enrolled. With an average primary of early interventions, many questions remain. gross enrollment ratio of 77%, and secondary enroll- Governments and donors are struggling with how ment ratios as low as 29% for boys and 23% for girls, to best serve the developmental needs of these education policy-makers in Africa constantly face the youngest citizens, and important questions are yet challenge of increasing access to schools, keeping to be answered. For example, how does a policy- children in school, and improving their achievement maker ensure that the poorest, most vulnerable levels, all within realistic investment frameworks. children benefit from ECCD? What is the respon- The potential positive rates of return and cost sav- sibility and role of government in providing ECCD, ings in terms of better school performance and health and what are the optimal models in terms of pro- associated with interventions in the early years have viding the best outcomes at affordable and sustain- received attention in the past decades. A variety of able costs? Vi AFRICA REGION HUMAN DEVELOPMENT WORKING PAPER SERIES This study attempts to address these and related fited from contributions by Marito Garcia, Marlaine issues in order to extract lessons as well as implications Lockheed, Alain Mingat, Alan Ruby, Eduardo Velez, for policy-making and to provide some guidance to Adriaan Verspoor, and Mary Eming Young (all of education-sector decision makers. The study examines World Bank), and David Weikart (former President of tvo countries in Sub-Saharan Africa: Cape Verde, High Scope). which is relatively wealthy, and Guinea, which is poor. We thank the Government of Norway for provid- Among the lessons learned in this study are that there ing the financial support to conduct this study and is demand for ECCD among households of all socioe- produce the report through the Norwegian Education conomic levels; that the provision of ECCD programs Trust Fund. can reduce disparities in school readiness among chil- We hope that this report is useful to policy-makers dren from different socioeconomic backgrounds; and in Africa. We also hope that it furthers the dialogue that high-cost programs do not necessarily produce that task managers in the World Bank and other the best results. The study also points out some of the donor organizations are engaged in to find how best challenges ahead, including room for improvement in to meet the educational and health needs of Africa's terms of efficiency, equity, and quality. children and to contribute to their intellectual, physi- This report was prepared by Adriana Jaramillo, cal, and emotional development. Education Specialist at the World Bank, and Karen Tietjen, consultant, and is based on research con- ducted by Bruno Suchaut from the University of Birger Fredriksen Bourgogne in collaboration with the Governments of Sector Director, Human Development Guinea and Cape Verde. The final document bene- Africa Region, The World Bank This report was prepared by Adriana jarlaillo and Karen Titjen, with speial contrib Marlaine Lockheed, Alain , A Ruby Ediuardo Velez, Adriaan Verspooan ai Wiat It is based on research condctedb o Shaut, in cooeratiun with Salim Dialin Ginea, a Malou Paul in Cape Ver. T h tt n E YogMtGt Prouty for their comments a uios. Thi resea wasnted in c tth the G emments of Guinea and Cape erde.ielwoulsalso likel thanks to the l teams, preschools, prns, and,omuitiesd who patiipte in the study. Early Childhood Development in Africa: Can We Do More for Less? A Look at the Impact and Implications of Preschools in Cape Verde and Guinea Africa Region Human Development Working Paper Series Executive Summary M any of the problems constraining the devel- acteristics that are most effective in promoting child hi opment of Africa's human resource base are development, and discusses the policy implications rooted in the early childhood (0-6 years old) of these programs. of its impoverished and disadvantaged populations. The early childhood years offer an unparalleled Supply: Characteristics of the preschool sector in window of opportunity to exercise a positive and Guinea and Cape Verde lasting influence on the health, intelligence, and future development of the individual child. Early In Guinea, about 2% of the 1.4 million children childhood care and development (ECCD) is the pro- under the age of six have access to the 500 preschools vision of services that respond to young children's in the country. Only three are public; the rest are oper- basic needs for cognitive stimulation, early learning, ated through either private or NGO initiatives and nutrition, and basic health care. Yet, today most chil- most are found in urban areas. Of the nearly 1100 dren in Africa-almost 98%-grow up without bene- "preschool teachers," only 8% have a higher educa- fit of ECCD programs or interventions. Moreover, the tion and 13% have been trained in ECCD. The most quality of many of the ECCD programs is low. Gov- common preschool type is the 'cole maternelle" emments as well as donors are struggling with how which follows the traditional French preschool model to best serve the developmental needs of their and targets children aged three to six; followed by the youngest citizens. "jardin d'enfants", which enrolls children from the age This study examines the preschool programs-the of two. The 60 Community Education Centers (CEC) predominant model found in the two countries-of are a recent initiative found in rural areas, operated early childhood care and development-in a compar- by local community organizations and financed by atively wealthy country (by African Standards)- UNICEF and Action-Aid. In all cases, parents pay Cape Verde-and a very poor country-Guinea. The fees, which range from $US3 to $US15 a month. purpose of this study is to build a knowledge base In Cape Verde, about 40% (24,000) of the children about the status of ECCD in West Africa, to identify from two to six years of age are enrolled in preschool, the building blocks of affordable and effective pre- although five-to-six year olds constitute over 46% of school or ECCD programs within an African context, the preschool students. A total of 316 preschools are and to better understand the role that governments found in all nine administrative zones (conselhos), and donors can play to support ECCD initiatives. The with the majority concentrated on the island of Santi- study develops a profile of the preschool/ECCD ago. Most of the 600 preschool personnel lack special- sector, explores the factors determining preschool ized training. Fifty-three percent of the preschools enrollment, analyzes the influence of preschool on receive public funds, and are run by municipal gov- child development, identifies the programs and char- ernments or the National Social Protection Institution 2 AFRICA REGION HUMAN DEVELOPMENT WORKING PAPER SERIES (ICS). Twenty percent are run by the Red Cross, 10% dents in each SES grouping in each country attain by religious institutions, 8% by private sector entities, higher raw test scores than the control group children 7% by NGOs, and 2% by local communities. who have not attended preschool. Although family background exercises the most influence on scores, Demand: Who goes to ECCD/preschool and preschool increases the explanatorv power to approx- factors influencing attendance imately 16 to 20%. Preschool can have an equalizing effect on cognitive The provision of preschool education can exacerbate dis- development. Failure to enroll lower SES group chil- parities between rich and poor children, as the wealthy are dren in preschool may result in even wider dispari- its biggest users and beneficiaries. Because children from ties in cognitive development and school readiness wealthier households are more likely to enroll in pre- skills. Vulnerable and at-risk children-whether the school, they will reap the developmental benefits it urban or rural poor, disadvantaged minorities, or provides. Meanwhile, the poorer children struggle girls-are likely both to need and to profit from the with the inherent disadvantages associated with "value-added" of preschool, in order to overcome poverty and without the added and remedial benefits their deprived environments and enter primary of preschool. school on a more equal footing with their more The double-edge sword: The gap may widen due to the advantaged peers. type antd quality of preschools affordable to different SES- Government should conisider encouraging the develop- level households. Wealthier children attend more ment of multiple-year ECCD programs. Public policy expensive preschools with "academic" orientation. could facilitate multiple-year attendance-associated Even if children from poorer households have the with higher cognitive development scores-by target- opportunity to attend preschool, they may not bene- ing subsidies at and encouraging the development of fit as much as their wealthier peers because they may preschools that serve poor children, who are less not receive the same preparation for primary school, likely to attend multiple years, and by offering pro- putting them at a disadvantage. However, this gap can grams that are gauged to serve the development be narrowed as was the case in Guinea, where children needs of different age groups. from lewer SES level derived greater benefits from pre- Preschools-particuilarly public on7cs servinig all SES school/CEC programs than their peers from the highest levels of children-should tailor thiri7' programs to suit the SES. The fact that there is no gender bias evident in needs of their studenzts or of the most vulnerable popula- enrollments also suggests that efforts to ensure tion. Guinean preschools have the greatest success in affordable, quality preschool for poor children will developing pre-reading or visual discrimination benefit girls as well as boys. skills and oral language skills, while Cape Verde Lowering the price barriers to entry could benefit chil- preschools enjoy the most success in developing basic cdren and improve household welfare. Since children from concepts and language skills. Multiple years in pre- poorer households do attend ECCD/preschool in sig- school will increase a child's total score. Preschool nificant numbers, there is little evidence to suggest programs may need to change emphasis among basic that constraints other than economic would prevent concepts, language and reading, depending on the lower SES households from enrolling their children in SES group being served, given their different per- preschool, if the price were affordable. Household formances. welfare could increase as a greater percentage of pre- school students have mothers who work outside of Effectiveness: Programs or program their homes or engage in professional activity. characteristics that appear to be most effective in terms of child development Impact: The effects of ECCD/preschool on child development Not all preschool programs are equal. Various pre- school programs, organizational structures, and The gains in cognitive development associated attributes will affect child development differently, with preschool suggest that government should with the preschools themselves accounting for 55% of encourage the provision of preschool. Preschool stu- variance in Guinea and 42% in Cape Verde. There EARLY CHILDHOOD DEVELOPMENT IN AFRICA: CAN WE DO MORE FOR LESS? 3 appears to be no clear pattern in terms of the factors nitive development, compensating for disadvantaged and/or models that make a clear difference in cogni- backgrounds. Even if preschool cannot entirely close tive development. Frequently, the school characteris- the cognitive development gap between rich and poor tics will have opposite effects in Guinea and Cape children, a lack of preschool can certainly widen the Verde, indicating that preschool effectiveness is gap in a relatively well-off country. Public policy highly contextualized. should be formulated to emphasize the support of The teacher's cultural background and proximity to the preschool or ECCD programs for poor and disadvan- school are more likely to have an impact than formal qual- taged children. In addition to reducing disparities in ifications. This suggests that focused, short-tern and school readiness among different SES levels, the pro- low-cost training may be more effective. vision of subsidized ECCD programs for lower Language of instruction in preschool must be consid- income families could also free mothers to find ered carefully, and policy will differ according to country. employment, thus improving household welfare. In both Guinea and Cape Verde, the language of More and better can be done with less. High-cost pro- instruction had a significant impact on cognitive grams may not necessarily produce the best results. development scores. But the same variable-bilin- Low-cost preschools need not sacrifice quality, as gual instruction-had different effects. The develop- many such preschools are effective, producing high ment of the child in his/her first language, the cognitive development scores at low unit costs. Tradi- attitudes and support of the parents for the language tional approaches-government-supported, highly chosen in school, and the perception of the wider paid teachers, formalized training, etc.-do not in community of the status of both the local and second themselves guarantee high cognitive development language are to be taken into account in choosing the scores. Short, focused, reliable and contextualized language of instruction. training can be more efficient and less costly than Preschools do not need full government subsidies but most available formal programs. Community-based will probably need some support if equity concerns are con- programs, such as the CECs in Guinea, appear prom- sidered. Communities-even poor ones-are willing ising, due in part to parental involvement and financ- to support preschools. Parents are willing to pay fees ing, and in part to the proximity of the locally-resident to enroll their children, as evidenced in both Guinea teacher, who is able to spend more time at the school. and Cape Verde. The wide variation in effectiveness of preschools, at Low-cost preschools serving poor communities can be many different price points, indicates that there are effective. Both high- and low- cost schools produced many factors that can influence a preschool's success. admirable (and lamentable) results. In Guinea, the There should be closer examination of what make lowest cost community education center (CEC) indigenous models of ECCD successful, particularly model produced highest cognitive scores. Moreover, in the case of the CEC programs in Guinea, as well as over 70% (on average 90%) of the children in these in other francophone African countries where these schools came from poorer (SES 2 and SES 1) house- types of initiatives are underway. holds. In Cape Verde, it was found that many schools Government can play a critical and essential role in are efficient, producing high cognitive development supporting early childhood care and development pro- scores at low unit cost. grams. Both the services the government provides There is no single effective model of preschooling or and the policies it promulgates will depend on con- magic bullet characteristic that will lead to child develop- text and should work within a realistic investment ment. Further study/analysis must be conducted of framework. In Guinea, where preschool enrollment is what constitutes key features of the effective pro- low and there are scant public resources for provid- grams. ing ECCD programs, the government can establish a supportive policy framework that creates an environ- Conclusions and recommendations ment in which private sector and community ECCD initiatives can flourish. This could take many differ- Preschool education can be a double-edged sword that ent forms, such as supporting IEC and parental edu- can both alleviate and exacerbate disparity. Providing pre- cation campaigns about the importance of early school services can have an equalizing effect on cog- childhood education and care, piloting low-cost/ 4 AFRICA REGION HUMAN DEVELOPMENT WORKING PAPER SERIES community-supported ECCD models based on What are the next steps in supporting ECCD in promising indigenous programs, creating ECCD Africa? Several issues requiring further investigation start-up credit funds, or developing non-onerous are highlighted by this research, ranging from the guidelines and guidance for community programs. longer-term impact of preschool on primary student Improving efficiency and equity. In Cape Verde where performance and the elusive factors that make a pre- the public sector is already providing a significant school effective, to how best to expand access to proportion of the population with preschool services, preschools programs without raising public expendi- the government role should focus on the more effi- ture burdens or pricing poor parents out of the pre- cient use and distribution of public sector resources. school market. The government of Guinea-with Efforts should be made to further study the World Bank assistance-will initiate ECCD pilot pro- preschools that have enjoyed the greatest success at grams in the two poorest regions of the country to the least cost and to replicate them. Focused, ECCD- provide a more in-depth and qualitative look at the related training should be provided to public sector features of successful programs, as well as their costs, teachers. The redistribution of resources to benefit the and to replicate them within the framework sug- poor could be effected by levying user charges on gested by this study, i.e., community-based and wealthier households, limiting public preschool parent-supported centers. In Cape Verde, several of access to poor families, or providing them with sub- the research findings-in particular the equitable sidies (scholarships, uniforms, nutrition or health allocation of preschool resources and the effective- services) to encourage enrollment. ness of government preschools-are subjects of the Improving quality. Government can act to improve policy dialogue currently taking place between the the quality of ECCD services available either through Government and the World Bank. public- or private-sector providers. Government can help equip local organizations and parents with the knowledge of what to expect from ECCD providers lc implictions; in order to select or monitor their children's pre- What canand donors'do? school. Government and donors should think outside the box of formal, traditional preschool programs. E0r ounI:andpascto tiafves They should encourage and support communities to * i Iowi ct,ee arelisticod8 of ECCD engage in the creation of ECCD services, and they * LaunhlEGadpatlEn campaigns should target whatever assistance is feasible to those * Xork within a realistic invetIent frark communities that have acted to provide preschool or ECCD services. 1 Introduction M any of the problems constraining the devel- yet the case in much of Africa. While the benefits of im opment of Africa's human resource base are ECCD are increasingly well-known to policy-makers, rooted in the early childhood (0-6 years old) both government and donor ability to act is con- of its impoverished and disadvantaged populations. founded by insufficient public sector budgets and Inadequate nutrition and health care, unstimulating concern about diverting resources from existing edu- child-care environments, and insufficient parenting cation and health investment priorities, as well as by of young children can result-in the short-term-in lack of knowledge about the most effective stunted physical, cognitive, and psycho-social devel- approaches, costs, management, and financing strate- opment, ill health, and lack of school readiness, and- gies within the African context. Yet, as Africa falls fur- in the longer-term-in poor academic performance, ther behind in the global race towards universal low economic productivity, and anti-social behaviors primary education,2 the potential of ECCD to con- as children grow to adulthood. tribute to the goals of education-for-all becomes more Numerous studies from both rich and poor coun- significant. There is a pressing need to identify and tries show that early childhood care and develop- develop models of implementation and finance that ment (ECCD) services can make a difference in the are efficient, affordable and likely to be sustained on physical, intellectual, social, and emotional develop- a large scale in sub-Saharan Africa, where over 40% of ment of this vulnerable age group.1 Moreover, these the population subsist on less than $1/day (World early interventions not only improve the immediate Bank 2000a, p. 10). well-being of young children, but may manifest As part of the World Bank's effort to amass the themselves at later stages in their lives with both building blocks needed to determine both the feasi- social and economic benefits for the individual, his bility of and best practices for investing in early child- family, and society. The potential positive rates of hood care and development, the objective of this return and cost-savings in terms of better school per- study is to contribute to the knowledge base about formance and health associated with ECCD invest- ECCD programs in Africa. This study examines the ments are of particular importance to the public preschool programs (which are the predominant sector and its ministries of education, health, and model of early childhood care and development) in a social affairs. comparatively wealthy country (by African stan- The governments of several developing countries dards), Cape Verde, and a very poor country, Guinea. -particularly in Latin America-have acted on this The study is based on the understanding that there is knowledge and in the past two decades, have incor- no blueprint for the implementation and support of porated national early childhood care and develop- early childhood activities, particularly in low-income ment policies and relatively low-cost ECCD African countries, and that the viability of different programs into their social sector portfolios. This is not approaches may be largely country-specific. As it is 6 AFRICA REGION HUMAN DEVELOPMENT WORKING PAPER SERIES the first study of its kind to be conducted in the sub- 5. What are the policy implications of these pro- ject countries, its findings are preliminary, and it grams in terms of beneficiaries, costs, financing, raises many questions that it cannot answer com- and implementation? pletely or definitively. At the same time, this study attempts to extract This study is organized into eight sections. Section some general guidelines that can inform educational 2 presents a brief background picture of early child- planning efforts. The study reviews the country con- hood care and development and its relevance to text in which early childhood care and development Africa. Section 3 provides a summary description of takes place, describes the range of preschool pro- the methodology used for the study. Sections 4, 5, 6, grams in operation, and begins to explore the impact and 7 examine the findings and policy implications of these programs on the children and the implica- from the Cape Verde and Guinea surveys, including a tions for their adaptation and implementation. description of the ECCD sector and an analysis of the The study addresses five main research questions: determinants of preschool attendance and impact on child development. Section 8 concludes with a sum- 1. What is the overall status and profile of preschool mary of findings and principal policy implications of or early childhood development programs in the the research, identifies additional issues, and makes targeted countries? recommendations for future action. Annex 1 presents a detailed description of the study's methodology. 2. Whiat factors determine child enrollment in pre- school programs? 3. To what extent does preschool attendance influ- ence child development? 1. The term "Early Childhood Care and Development (ECCD)", which encompasses all the supports necessary for a child's sur- vival, protection and care to ensure optimal development from 4. What programs or program characteristics birth to age eight (Evans et al. 2000), is used throughout this appear to be most effective in terms of child paper in order to place child care programs and interventions appear to be most effective in terms of child into a broader early childhood framework and to distinguish development? In other words, what types or the range of early childhood programs from preschools, characteristics of the surveyed preschool pro- although the latter are the focus of the study's research. grams are most assoce wt o2. Projections based on 1990 enrollment trends indicate that grams are most associated with or explain child Africa will account for 75% of the world's out-of-school chil- development gains? dren, if no reforms are undertaken (World Bank 2000c, p. 8). 2 Early Childhood Care and Development Background What is early childhood development two through five, a child develops language skills, and why is it important? fundamental social skills, and the base for "learning to learn" that translates into school readiness. Char- he period of early childhood has been identi- acter and personality are largely formed, and major T fied as the most formative in a child's devel- social and moral values are transmitted. From ages opment, one that will have long-lasting-even six through eight, a child consolidates earlier learn- permanent-effects on his/her adult life. From birth ing, begins to learn conceptually and manipulate to age eight, a child gradually masters increasingly ideas, and enters the "age of reason" (Eming Young complex levels of moving, thinking, feeling, and 1996; Evans et al. 2000). interacting with people and the world around him If the child's body and brain develop well, his (Evans et al. 2000). Although the physical develop- learning potential increases. Conversely, neglect of a ment and custodial requirements of young children child's biological and mental needs at this critical and are universally acknowledged, less is known about formative stage can have a negative impact, resulting the requirements for supporting the development of in delayed or debilitated cognitive development, intelligence in young children during this period. By stunted growth, and physical impairment. When a age six, the brain has reached 90% of adult size. child's inherent physical, social, and psychological Abundant evidence from diverse fields-physiology, capacities are not nurtured, they wither. The quality nutrition, health, sociology, psychology, and educa- of care during this period greatly affects the develop- tion-shows how these early years (beginning with ment of the child, his health, his psyche, and his conception) are crucial to the development of intelli- capacity for future learning. gence, personality, and social behavior. Roughly three stages of early childhood develop- What are early childhood care ment lay the foundation for future growth. The most and development programs? rapid period of brain development (cell growth and neural connections) takes place in the first two years The development process is affected by nutrition of life. Although the structure of the brain is deter- and health as well as by interaction with people and mined biologically and is considerably developed in things. The provision of an enabling environment- the prenatal stage, a child's interaction with his/her through an active feeding process, good nutrition, environment will develop the critical brain connec- healthy and safe environment, basic health care, tions that set the pathways for intellectual, emotional, encouragement and affection, stimulation, protection, physical, immunological, and social functions. The and modeling-can help create a healthy child ready ability to walk, manual dexterity, and other motor and able to learn and develop fully. However, many skills are developed in the first two years. From ages families, particularly poor ones, are unable to provide 8 AFRICA REGION HUMAN DEVELOPMENT WORKING PAPER SERIES all of this without aid, whether in the form of oppor- are that child development begins in utero, making tunity, time, information, or money. Early childhood prenatal care indispensable; that evolving develop- programs and approaches can help to provide both mental needs require different curricula; that young the care and development elements, not only to children learn by doing and thus learn to learn; and increase the chances of the survival of the child but that family well-being is essential to child develop- also to optimize his development. ment. ECCD programs recognize that school, home, Various terms are used, often with imprecision, to and out-of-home settings are mutually influential and describe the services provided to children before they should include parents, families and communities to enter primary school. These terms, such as "daycare," optimize child development. ECCD programs aim to "preschool," and "early child development," mask build on existing positive childcare practices, struc- the wide diversity in targeted beneficiaries, tures, and organizations so that they are culturally approaches, content, structure and organization of appropriate, affordable, and sustainable (Arnold the intervention. Myers (2000b) offer a simple typol- 1998; Evans et al. 2000). ogy for examining early childhood programs: The range of ECCD programs is not defined by a single model. Early childhood programs differ * Child care consists of custodial actions by care according to the age of the targeted beneficiaries, as givers to ensure protection, biologic survival and well as the culture, context, and locus of intervention. growth, and-less frequently-development. They can encompass support given to parents or childcare providers in the home or as well as in * Child development focuses on the child's mastery of formal settings such as a nursery or kindergarten increasingly complex levels of mobility, thought, with a defined curriculum and professionally trained emotion, and social relationships. teachers. ECCD programming can include strength- ening parenting and care-giver skills, working with * Early education promotes the learning process of other household and community members to address acquiring knowledge, skills, habits and values specific child development needs, providing better and is often narrowly interpreted as uniquely daycare options, and providing economic support to preparing children for primary school. women and families. Current understanding of the process of child Why invest in ECCD? development argues for combining all three aspects into early childhood programs. What distinguishes The early childhood years offer an unparalleled today's ECCD approach from earlier conceptualiza- window of opportunity to exercise a positive and tions is that it is not limited either to "survival" or to lasting influence on the health, intelligence, and "academic learning," but is holistic and integrated. It future development of the individual child. The long- is aimed at meeting the multiple needs of the child at term effect on development is greatest when nutri- various stages of early childhood "by taking into tional, health, and early education interventions are account health, nutrition, and psycho-social stimula- combined. ECCD programs target children in the tion while at the same time strengthening the envi- preschool years, generally seven and under, not only ronment in which children live" (Evans et al. 2000; because this is when a child is most susceptible to Myer3 2000b; Eming Young 1996). A move from a intervention, but because this is a period when a child system of largely separate, vertical programs to an is relatively neglected, prior to the attention that goes integrated program approach to health, nutrition, with school and a child's ability to do for him/ and early education will create efficiencies as well as herself. require significant institutional change (Heaver and A variety of rationales for investment in ECCD Hunt 1995). programs has been presented in the past decade that Earlv childhood care and development is the pro- can be condensed into three main arguments (Myers vision of services that respond to young children's 1995 and 2000a; Eming Young 1996; Van Der Gaag basic needs for cognitive stimulation, early learning, and Tan 1998). They are: nutrition, and basic health care. Basic tenets of ECCD EARLY CHILDHOOD DEVELOPMENT IN AFRICA: CAN WE DO MORE FOR LESS? 9 * Human Rights argument: That "children have a ECCD benefits can also be summarized using a right to live and develop to full potential" has hypothetical beneficiary's life cycle. ECCD programs been championed by several international agen- may yield immediate benefits to children, but they cies and human rights groups, declared in a may also produce social and economic returns over a variety of official forums, and codified in interna- lifetime in terms of ability to contribute to family, tional agreements such as the Convention of community, and a nation. Rights of the Child. ECCD programs enable effec- In the short term, during early childhood, the eco- tive child advocacy. nomic benefits due to a child's increased psycho- social development may be realized through the * Basic Needs argument: Scientific evidence shows reduced costs of caring for a healthier child, the that neglect of children's physical and mental increased family income contributed by a mother or needs during this critical period can result in family member freed frorn child care duties to find delayed or impaired physical and psycho-social employment, the improved earning potential of an development. Through poverty and the destruc- older sister or sibling able to attend school, and the tive chain of events it activates, many families are employment provided hired care-givers. unable to provide the care required for optimal In the medium term, during later childhood and child development and transmit the core cultural adolescence, the child's improved school readiness, and moral values prized by society and essential ability to learn, and self-confidence may increase the to its promulgation. ECCD programs can fill the likelihood of primary school enrollment, swifter pro- deficits in developmental inputs created by gression through the cycle, and transition to higher poverty and other adverse conditions. levels, resulting in cost-saving from decreased school wastage (repetition and drop-out). The child also is * Economic and Social Benefits argument: Although less likely to participate in destructive behaviors (e.g., the tragedy of child mortality and burden of care delinquency and drug-use), thus reducing the need may fall most heavily on individual families, soci- for costly remedial programs. eties also suffer economic and social conse- In the long-term, during adulthood, the ECCD ben- quences from the neglect of their young children eficiary may be more likely to find remunerative and reap the benefits of their healthy develop- employment and be more economically productive, ment. ECCD programs can help create more able, as signified by a higher lifetime earnings stream. He educated, and productive individuals who will may engage in more socially responsible behaviors- find employment, enjoy higher incomes, suffer such as preventative health/hygiene practices, family fewer health problems, and participate construc- planning, better parenting, leadership, less domestic tively in society. The ability to earn higher violence, and reduced criminality-that will require incomes will alleviate poverty, contribute to more fewer public (and private) resources for support, cor- equitable distribution of wealth, and promote rection, or control. social equity. Cost-savings will be generated Numerous claims have been made about the bene- through the reduced need for remedial and social fits that early childhood care and development gen- welfare programs. Early childhood care and erates. Not all have been substantiated through development interventions can boost the effi- longitudinal studies or reaffirmed through economic ciency of other programs through synergies and analysis. Most have been conducted in wealthy, positive interactive effects with health, nutrition, OECD countries, and their applicability to develop- education, and women's programs. ing countries has been questioned (Van Der Gaag and Tan 1998). But a growing number of studies have This last argument has received considerable attempted to isolate the impacts of ECCD programs, attention. It is central to the debate and discussion of measure the benefits, and determine the rate of return early childhood development as a tool for social and to the investment in early childhood programs. economic development. Myers (1996) has classified Several studies have linked early childhood care the benefits of ECCD programs by beneficiary and development interventions to potentially cost- groups. saving effects. For example, de-worming has had sig- 10 AFRICA REGION HUMAN DEVELOPMENT WORKING PAPER SERIES Table 2.1 Benefits of ECCD programs by beneficiary groups Beneficiary Group Area of Change Indicators of Change Children Psycho-social development Improved cognitive development (thinking, reasoning); improved social development (relationships to others); improved emotional development (self-image, security); improved language skills Health and nutrition Increased survival chances; reduced morbidity; improved hygiene; improved weight & height for age; improved micronutrient balances Progress and performance in Higher chance of entering; less chance of prmary school repeating; higher learning and better performance Adults (program staff, parents) and siblings General health knowledge, general Improved health and hygiene; improved nutrition health attitudes and practices (own status); preventative medical monitoring and attention; timely treatment; improved diet Self-confidence Improved relationships between husband and Relationships wife, between parents and older children; Employment caregivers freed to seek or improve employment; new employment opportunities created by program; increased market for program related goods Communities Physical environment Improved sanitation; more spaces for play; new Social participation facilities; greater female participation; greater Solidarty demand for existing services, community projects benefiting all Schools and health service facilities Efficiency Better attention to health; changed user practices; reduced school repetition and dropout Effectiveness Greater coverage; improved ability, confidence, or Capacity organization; methods and curriculum content Practice and content Society Health and education status Fewer days lost to sickness; a healthier Participation population; a more literate, education population; Productivity greater social participation; a more productive Delinquency labor force; reduced delinquency; reduced fertility; Fertility reduced social inequality Source: Myers 1996 Equality Source: Ntyers 1996 nificant effects on children's school performance However, ECCD interventions are often costly, and (Bundv 1997). School feeding programs have investment decisions by policy-makers require an increased both enrollment and achievement (Pollitt et analysis of the costs compared to the benefits.3 al. 1993). Low-cost ECCD programs boost the enroll- Economic analysis of the High/Scope Perry ment in school of older girls and increase the number Preschool Program, which was initiated in Michigan of mothers who work, which augments the incomes more than three decades ago, has set the standard of poor households (Lokshin et al. 2000). ECCD pro- both for quantifying a wide range of outcomes and grams have been found to alleviate disadvantages for cost-benefit analysis: The program followed 123 experienced by young children burdened by poverty, children, randomly divided into a program group hunger, and malnutrition (Haveman and Wolfe 1995). and control group, from age three to age 27, collecting EARLY CHILDHOOD DEVELOPMENT IN AFRICA: CAN WE DO MORE FOR LESS? 11 information on IQ score, school performance, grams only use the national (not maternal) language employment, and other aspects of well-being and and focus on "socialization". Edwards and Liang social behavior. Schweinhart et al. (1993) reported that found that Mexico's national preschool program has at age 19, the preschool participants had higher liter- no impact on primary school efficiency or on mater- acy test scores, and at age 27, they enjoyed higher nal labor force participation, and that the high unit monthly earnings, higher percentage of home owner- costs-at 132% of primary school student cost-do ship, higher levels of schooling, and lower percent- not justify the program as currently conceived. The ages receiving social services, fewer arrests, and authors suggest that the often neglected aspect of pre- fewer out-of-wedlock births. Returns to the public school as a "consumption good," which offers enter- over the lifetime of the participants were estimated at tainment to children and leisure time to their $7.16 for every dollar invested in the programs. mothers, accounts for their popularity. Also in Van Der Gaag and Tan (1998) conducted a similar Mexico, but using a different sample, Velez and analysis on a preschool program in Bolivia, although Prawda found that preschool does have a positive without the extended tracer study data. The program, impact, but mainly among students from a low socio- known as PIDI, provides non-formal home-based economic background. daycare centers. Van Der Gaag and Tan calculated Most studies point to the methodological chal- that the value of the productivity-related (i.e., earn- lenges of isolating impact, paucity of tracer studies, ings) benefits exceed the initial investment by 126%, lack of robust samples, and generally inadequate and that the cost savings of reduced fertility (cal- data. Myers (in Evans et al. 2000) cautions that lack of culated in births averted) is $190 per enrolled child, a common framework for estimating costs and calcu- resulting in a total cost-benefit ratio between 2.38 lating benefits limits the utility of ECCD program and 3.06. comparison. Further complicating benefit analysis is The benefits of the Preschool Feeding Program that child development is determined by "a child's (PROAPE), a non-formal integrated preschool pro- temperament, neurological integrity and impairment; gram serving marginal urban areas in Brazil, were immediate environment, relationships and interac- defined in terms of reduced primary school repeti- tion, and factors in larger social environment" which tion. The cost of producing a first grade graduate was interact in complex fashions to influence develop- less for PROAPE children, and the savings exceeded mental outcomes (Edwards and Liang 1998). Costs the per-child cost of the program (Evans et al. 2000). and benefits are specific to the political and economic Similarly, the one-year preschool Shishu Kaksha Cen- contexts and the social conditions; no single best solu- ters program in Nepal was found to generate $4 mil- tion can be predetermined (Evans et al. 2000). lion in cost savings to the government, due to reduced primary school repetition of the participants Does Africa need early childhood (Meyers 1998, in Evans et al. 2000). development programs? These positive effects and positive benefit-cost ratios are not universally proven, however. Several Despite improvements in many African countries analyses of the U.S. Head Start program-promoting over the past decade, Sub-Saharan Africa contains 28 school readiness, health and social skills of disadvan- of the 35 countries classified by the United Nations taged children-have concluded that the impact on Development Program as having low human devel- academic performance diminishes as the child pro- opment. This region accounts for a growing share of gresses through primary school and that there is "vir- the world's poor, and an increasing proportion of its tually no evidence" of long-term impact on measures citizens cannot meet their basic needs (e.g., 220 mil- of life success (Currie and Thomas 1995; Haskins in lion Africans live below the internationally recog- Edwards and Liang 1998). In a meta-analysis of the nized poverty line of $1/day).4 The increase of effects of preschool in West Africa (1998), Mingat and poverty, drought and natural disasters, war and polit- Suchaut observe that the positive influence of pre- ical conflict, and more recently, AIDS, have con- school on primary school performance is limited to tributed to the spread of disease and malnutrition, the early primary grades and disappears at higher and have created millions of refugees, orphans, and levels, particularly in countries where preschool pro- displaced persons (most often women and children). 12 AFRICA REGION HUMAN DEVELOPMENT WORKING PAPER SERIES The inability of governments to provide basic water, will not complete the primary cycle. Calculations sanitation, and adequate health and education sys- show that 11 of 33 SSA countries spend 50% or more tems has done little to mitigate the misery of their resources to produce a primary school graduate than populations. in an ideal system. African students also have the Africa's problems have taken their toll on its weakest learning levels. A 1991 survey of fourth grade youngest citizens. They are most vulnerable to risk, reading aptitude attributed the lowest test scores to whether demographic, income, environmental, geo- the three participating African countries, significantly graphic, political, or cultural. One-third of the 40,000 lower than other developing countries included in the children who die each day are African, although they study. More recently, a test revealed that average read- account for only one-tenth of the world's children. The ing scores of sixth graders in four African countries prevalence of malnutrition among African children is ranged between 38% and 58%, showing that about high; Africa is the only region where malnutrition rates half of those tested had not achieved minimum profi- did not decline in the 1980s. Inoculation rates continue ciency levels. to lag behind the rest of the developing world; only Today, most children in Africa-more than 98%- about 60% of one-year-old children are fully immu- grow up without benefit of ECCD programs or inter- nized against major childhood diseases. The onslaught ventions. And the quality of many of the ECCD pro- of HIV/AIDS affects one in ten adults, with devastat- grams is low. Although a wide range of ECCD ing effects on children as parents and care-givers suc- programs exists, the most prevalent model is a center- cumb to the disease and they themselves become based approach for children three to six years old, infected. In Namibia alone, it is estimated that 17.4% of employing traditional teaching methods and empha- pregnant women are HlV-positive (Washington Post sizing academic learning to serve as "grade zero" of September 2000). Ninety-five percent (about 8 million) primary school. Few programs relate to the home of all AIDS orphans worldwide are found in Sub-Saha- environment of children; rather, they tend to "isolate" ran Africa. Physical and emotional trauma caused by the child from the parents. There is very limited cov- violence, enforced migration, and separation of fami- erage of the birth-to-three-year-olds. The tendency of lies has a disturbing effect on children, undermining preschool services to be consumed by wealthier fam- their opportunities for development and growth in a ilies in urban areas results in increased disparities stable and safe environment. Currently, 60% of Sub- between the rich and the poor. Saharan Africa is involved in some level of violent con- Nevertheless, several innovative, alternative flict, and 28% of the world's 12 million refugees are models have been documented (Arnold 1998). In now in Africa. Mozambique, groups of children aged three to seven As African states struggle to provide educational years meet for two to three hours, led by "animators" opportunities to their burgeoning school-aged popu- who have been chosen by parents and the commu- lation, the extreme poverty of many families prevents nity. Parents determine the educational and social them from enrolling their children in school or sup- activities, which take place without the shelter of a porting them adequately once enrolled. In 1996, 23 of permanent school or fixed location. In Nigeria, rural 44 Sub-Saharan countries enrolled fewer than 60% of market women have organized childcare facilities for their primary school-aged population, and only 10 themselves on market days, run on a voluntary basis countries had 100% enrollment. By 2015, 75% of out- with women taking turns caring for children. In of-school children will be African compared with 33% Botswana, 8,000 "little teachers"-pupils in grades today. 1-3 of primary school-mentored preschool children Children suffering from developmental delays or in a child-to-child program. In Kenya, 70% of nearly physical debilitation are less able to cope with and 24,000 preschools are owned and operated by com- compensate for the inadequate learning environments munities (Norwegian Education Trust Fund report of many African schools. More than 40% of African August 2000). children are stunted and a third are underweight. Par- These examples and others suggest that character- ticularly vulnerable to illness, children miss school istics unique to Africa must be identified and factored and are forced to drop out or repeat grades. In nearly into the design of ECCD programs and interventions. half of 32 countries studied, one-third of the students Most importantly, ECCD programs must fit within EARLY CHILDHOOD DEVELOPMENT IN AFRICA: CAN WE DO MORE FOR LESS? 13 the restricted social sector investment framework in Interest is growing in ECCD in Africa. More coun- order to be affordable and sustainable. ECCD pro- tries are discussing ECCD and related school health grams must be less expensive (annual per child costs and nutrition programs with the World Bank and are estimated at less than $30), building on and opti- other donors. The World Bank currently supports 14 mizing what exists. This suggests that greater ECCD operations in Sub-Saharan Africa. Neverthe- emphasis be placed on supporting attitudinal and less, several questions have not yet been fully behavioral adjustments in African child-rearing and addressed, such as: how to ensure that the poorest, care-giving practices through strategic communica- most vulnerable children benefit from ECCD; what is tion and outreach training, rather than on direct the desired outcome of ECCD; what is meant by low services, such as supplemental feeding and infra- cost; what constitutes effective and appropriate pro- structure. The high degree of community involve- gram content; what is the responsibility and role of ment in the Latin American and Asian ECCD government; what are the optimal models of partner- programs that had positive results suggests that ship; and what is affordable and sustainable. While future efforts in Africa should be based on parental this study cannot answer many of these questions, it and community choice and involvement, and should attempts to address these and related issues in order blend traditional child-care practices with modern to extract the implications for policy and provide ones so that the programs reflect local cultural prac- some guidance to education-sector strategists in deal- tices and priorities. Such non-traditional approaches ing with ECCD, particularly preschool, interventions. mean a different, less exclusive role for the public sector and should incorporate multi-sector and decentralized implementation, including local levels of government, the private sector, NGOs, and com- munities. Finally, ECCD programs should build on 3. See Van Der Gaag and Tan 1998, for a discussion of estimat- and link to programs that are planned in comple- ing the economic benefits of early childhood care and develop- mentay aras, uch s scool ealt andnutrtion ment programs. mentary areas, such as school health and nutrition 4. Unless otherwise noted, all data derives from World Bank programs as well as women's programs. (2000a) and World Bank (2000c). 3 Study Methodology Approach and sample selection Verde, for each of the centers examined, a random sample of 15 five-year-old children enrolled in the his study's purpose was to develop a prelimi- preschool and their families were surveyed.5 As a T nary picture of the impacts and effectiveness basis of comparison, 10 five-year-old children from of early childhood programs in two contrast- the same community and sharing similar socio-eco- ing African countries-Cape Verde and Guinea. It ini- nomic characteristics but not attending preschool tially attempted to develop an overall profile of the formed the control group. early childhood development sector. However, since Because it was not feasible to administer a pretest preschools were the predominant model of interven- in order to establish a baseline to assess the magni- tion in early childhood in the two countries studied, tude of the relative impact or "value-added" of pre- as well as the most readily accessible, this study nec- school on child development, attention was paid to essarily focused on the preschool sector. Moreover, ensuring that the control group shared the same because of actual (Cape Verde) and potential (Guinea) background characteristics as the experimental pre- public-sector support of preschool as an early child- school group in order to eliminate a selection bias in hood development strategy, understanding the pre- the sample. Using the Heckman procedure, a low, school sub-sector-its programs and its impacts-is non-significant inverse Mills ratio was obtained, indi- essential to sound policy development. cating that no bias marred the sample and compen- The study used a survey methodology to identify sating for the lack of cross-sectional data. different programs and preschool models and to detail their characteristics. A sample of preschools Table 3.1 was selected to identify the impact that preschool Country samples attendance, different preschool programs/models, and various preschool characteristics had on cogni- Guinea Cape Verde tive and physical development of the children who Centers 36 34 attend them. The range of existing programs in each of the countries was examined to determine the pro- grams and characteristics that appeared to be most Preschool Group 529 490 effective-and least costly-in supporting children's Control Group 348 313 development. A sample of preschool centers was selected in each SES1 (% in the sample) 42% 49% of the countries, taking into account variability in SES2 (% in the sample) 36% 36% terms of socio-economic levels, geographic distribu- SES3 (% in the sample) 22% 15% tion, and types of programs. In both Guinea and Cape EARLY CHILDHOOD DEVELOPMENT IN AFRICA: CAN WE DO MORE FOR LESS? 15 The sample households in both Guinea and Cape documented in Cape Verde, very little informa- Verde were divided in to three socio-economic group- tion existed in Guinea. It is likely, therefore, that ings, according to the profession of the father or male more preschools exist in Guinea than are docu- head, with SES 1 being the poorest and SES 3 the mented. Moreover, because the study focused on wealthiest. Because profession is an inexact proxy for the more readily identified preschool sector, it is wealth and socio-economic status, occasionally SES 1 likely that in both countries other interventions or and SES 2 will display similar behaviors, or they will community-led programs supportive of early not be ranked in a strictly hierarchical order. childhood development are operating. The dependent variables used in the analysis were children's cognitive and physical development. The * Although robust, the samples used in the study independent variables-the factors influencing are not necessary representative of the ECCD or children's development-were divided into three cat- preschool sectors. Extrapolation to the larger pop- egories. The first group comprised individual socio- ulation must be done with caution. For example, demographic characteristics of the children such as although 22% of the families in the Guinea sample age, gender, health status, socio-economic status, and belonged to the poorest SES 3, this does not nec- family characteristics. The second group comprised essarily mean that this percentage would remain environmental characteristics, such as whether par- constant if the entire population of preschools ents read to the preschool-age child. The third group were examined, or even if additional preschools comprised variables related to the preschool/ECCD were opened. programs, such as student/teacher ratio, teachers' qualifications and experience, type of program * The descriptive data on preschool characteristics, offered (daycare, preschool, etc.) time dedicated to while broad, is not detailed enough to draw children's activities, urban or rural settings, if pro- definitive conclusions about the effectiveness of grams were public or private, and overall costs. certain factors. For example, while the survey Cross-tabs and multivariate analysis were used to documented whether a preschool had a canteen, compare the two groups and the influence of vari- it did not document to what extent it was opera- ables in the cognitive and physical development of tional. In addition, several of the teacher charac- children. These analyses were done separately for teristics were self-reported and, therefore, subject each country. Results and discussions on the findings to the usual cautions. are presented in Sections 5, 6 and 7. Annex 1 presents a more detailed discussion of the * The assessment of preschool's impact on chil- study methodology. dren's cognitive and physical development was limited to tests of five-year-olds in the sample. Limitations of the study The study did not attempt to assess its impact on children in primary school. This study is an initial attempt to describe and examine the early childhood programs and preschool * The cost data presented is notional. The study did sectors in two countries where little previous infor- not examine the full range of resources required to mation existed. Consequently, the reader must bear in operate the preschools and their cost implications. mind several cautionary notes about the data and its interpretation, many of which are flagged throughout the text. In general, they are: * The study is not a census of the early childhood Theostms nor preschool sector in either country. 5. In a few centers in Guinea, there were fewer than 15 five-year- programs or preschool sector in either country-* olds born in 1995. In order to complete the sample, children Although the preschool sector was relatively well bom in December 1994 were included. 4 Supply: Characteristics of the Preschool Sector in Guinea and Cape Verde his section provides an overview of the pre- are three public preschools in the country, and all T school sectors in Guinea and Cape Verde. other ECCD existing programs are either private or Although the subsequent analytical sections of NGO initiatives. According to a government effort to this study treat them together (as it is attempting to estimate the number of centers, there are approxi- discern cross-national trends), basic descriptions of mately 500 preschool centers in the country. The vast each will be presented separately. Moreover, sector majority of them are found in urban areas, of which description will occur on two levels. A general more than 50% are in Conakry, the capital city. There overview will be provided based on either govern- are several types of ECCD services, ranging from day- ment statistics or general survey information. Greater care to preschool programs. Although the provision of detail will be provided based on detailed sample this service is regulated by the Social Protection Min- information. istry, the number of unregistered centers is unknown. The most common type is the "ecole maternelle" that The preschool sector in Guinea follows the traditional French preschool model and targets children aged three to six. The next most Preschool access and distribution common is the "jardin d'enfants", whose main differ- ence from the eco1e maternelle is that it enrolls children In Guinea, about 2% of the 1.4 million children from the age of two. The Community Education Cen- under the age of six have access to preschool. There ters (CEC) are a recent initiative, financed by UNICEF and Action-Aid, concentrated in rural areas of the two Table 4.1 poorest regions (Haute Guinee and Basse Guinee). Basic social indicators The 60 CECs that have been established are operated by local community organizations. Guinea Cape Verde Total Population 7.2 M 430,000 Preschool provision GNP/Capita in US 510 1330 As noted earlier, the provision of preschool serv- Total Gov't expend. on ed. 11% 23% ices is mainly private. Of the 460 centers surveyed in Adult literacy rate 35% 85% the first stage of this study more than 50% are &oles maternelles, 26% are jardin d'enfants and 4% are Com- Gross Primary Enrollment 54% 118% munity Education Centers (CEC). The ecoles mater- Net primary boys enrollment 49% 100% ineles and jardin d'enfants are privately funded and run initiatives, and the CECs are a recent UNICEF Net primary girls enrollment 30% 97% and NGO supported program. In all cases, parents EARLY CHILDHOOD DEVELOPMENT IN AFRICA: CAN WE DO MORE FOR LESS? 17 pay fees, which range from $US3 to $US15 a month. 10 islands comprising the country. Nearly 60% of As a point of reference, the average primary school preschools are concentrated on the island of Santiago teacher salary in Guinea is approximately $US150 (and in the zone of Praia), corresponding to popula- and the GNP per capita is US$510. tion density. Preschool personnel Preschool provision There are approximately 1100 "preschool teachers" Preschools are operated and financed by a variety in Guinea, with a wide variety of training/experi- of organizations in Cape Verde. Half of them (53%) ence, ranging from none to high school with a few, receive public funds and are run by municipal gov- only 8%, having a higher education level. The vast ernments. Thirteen percent of all public ones are run majority has not received ECCD-related training; by the National Social Protection Institution (ICS). Of about 13% have been trained as "monitrices" in the those privately funded, 20% are run by the Red Cross, National Teacher Training Colleges. There are no 10% by religious institutions, 8% by private sector ECCD-related training institutions in the country, entities, 7% by NGOs, and 2% by local communities. with the exception of isolated NGOs that provide short courses for their personnel. Preschool personnel Preschool students Preschool personnel number less than 600 in the country. Preschools are staffed by three categories of About 75% of centers take children from ages three personnel: instructors (monitoras), educators (educa- to six years, and a small number of them take chil- tores) and assistants (orientadoras). Assistants, with no dren from the age of two. Some take even infants. special training in early childhood education, pre- Most of the centers take both boys and girls, and the dominate in the preschools, accounting for 88% of average for the preschool enrolled population is 52% personnel. Instructors, accounting for 11%, have boys. Nevertheless, the study found geographical dif- received training from the ICS. The six educators, ferences between girls and boys enrollment rates. In representing only 1% of personnel, have been trained the Haute Guinee region, where access to preschool is as professionals. considerably lower than for the other regions, the percentage of girls enrolled was 6-11 points lower Preschool students than that of boys. Table 4.2 shows the characteristics of the sample in Guinea. Children from two to six years of age are eligible for preschool, although it appears that some The preschool sector in Cape Verde preschools accept younger children and even infants. However, children aged five to six years constitute Preschool access and distribution more than 46% of the preschool students. Section 2 provided definitions of ECCD and other Preschool coverage in Cape Verde compares favor- related terms. In order to simplify the discussion that ably with other African countries, and with Guinea in follows, the term "preschool" will be used in a particular. The preschool gross enrollment rate is esti- generic way, to refer to the centers surveyed. Table 4.3 mated at around 40% or 24,000 children. According to summarizes the average characteristics of the sample government data, a total of 316 preschools are found in Cape Verde. in the nine administrative zones (conselhos) and on all 18 AFRICA REGION HUMAN DEVELOPMENT WORKING PAPER SERIES Table 4.2 Characteristics of preschools in Guinea Characteristics of the Centers EcolesMaternelles Community Centers Jardins d'enfants ORGANIZATION Average number of children 90.0 62.6 104.6 % of children less than 3 years of age 23 10 12 % of children less than 5 years of age 33 43 34 % of girls 46 50 50 Average number of teachers 2.9 1.4 3.6 Average of non-teaching personnel 1.1 0.1 1.1 Student/teacher ratio 1 /31 1 /48 1 / 29 % of male teachers 56 75 60 EQUIPMENT % with cafeteria 39 0 60 % with a courtyard 91 100 100 % with play yard and games available 52 37 60 CHARACTERISTICS OF TEACHERS Average age 33 35 35 Average years of experience 4 3 8 Average hours/week of work 27 25 24 Average years of schooling 12.9 11.2 14.8 % of teachers that lived in the school area 52 100 40 % of teachers chosen teaching carrier 69 37 40 EDUCATION ASPECTS % teaching in French 30 0 20 % teaching in local language 0 12 20 % bilingual teaching 70 88 60 % time for concept learning activities 31 35 32 % time for expression/play activities 46 45 50 % time for social activities 23 20 17 COSTS Source of funds Families Communities N.G.O. Families Average monthly fee (Guinean Francs)* 6860 875 4780 Average teacher salary (Guinean Francs) 66750 52500 72150 Average unit monthly cost (Guinean Francs) 9163 2049 8240 *1 $US= 1300 GF EARLY CHILDHOOD DEVELOPMENT IN AFRICA: CAN WE DO MORE FOR LESS? 19 Table 4.3 Characteristics of preschools in Cape Verde Characteristics of Centers Public Private N.G.O Community Religious ORGANIZATION Average number of children 78 114 62 97 60 Average number of teachers 2.9 5.3 2.6 2.8 4.0 Average of non-teaching personnel 1.8 2.7 1.0 0.8 1.0 Studentteacher ratio 26.9 21.6 24 34 17.9 EQUIPMENT % with a courtyard 62 67 47 50 100 % with kitchen 56 33 14 83 50 % with cafeteria 75 0 43 33 50 CHARACTERISTICS OF TEACHERS Average age 35.0 31.3 30.4 37.5 51.5 Average years of experience 9.5 9.3 8.2 11.5 21.0 Average hours/week of work 26.7 25.6 31.3 24.8 32.5 EDUCATION ASPECTS % main objective school readiness 87 33 57 17 0 % bilingual teaching 75 67 86 68 100 % time for concept learning activities 31 34 31 28 31 % time for expression/play activities 41 46 43 39 42 % time for social activities 22 20 21 22 20 COSTS % of parents that pay school fees 68 94 76 79 97 Average teacher salary (ESC. CV)* 12200 6600 10200 14400 24900 Average unit monthly cost (ESC. CV) 599 341 633 362 1025 1$US=1100 ESC.CV 5 Demand: Who Goes to Preschool and Factors Influencing Attendance The story SES level-are higher for preschool students. In Cape Verde, female-headed households, most likely poorer, I n the previous section, we saw that access to pre- are less likely to enroll their children in preschool. And school is extremely low in Guinea and extraordi- households in Cape Verde that require their young chil- narily high in Cape Verde (compared with dren to perform domestic tasks are 8% less likely to en- African averages). Among children aged 3-6, less roll their children in preschool. A similar relationship than 2% in Guinea and around 40% in Cape Verde was found in Guinea, although it was not statistically attend preschool. Despite this wide disparity in pre- significant. school enrollment between the two countries, there are notable similarities in terms of the characteristics Wealthier children attend "better" preschools. of those children that attend preschool and those that do not. This section explores these similarities and Socio-economic status affects household selection of discusses some of the policy implications associated preschool. In both countries, a greater percentage of the with them. Table 5.1 summarizes some of the data wealthiest SES 3 group children attended the more discussed below. expensive schools. (Whether this resulted in greater effectiveness will be discussed in a later section.) In Wealthier children attend preschool. Cape Verde, 83% of SES 3 children were enrolled in the more expensive schools, compared with 44% of SES 2 In both Guinea and Cape Verde, children from the children, and 61% of SES 1 children. The same pattern highest SES group were more likely to be enrolled in was found in Guinea: 52% of SES 3 children attended preschool. In Guinea, the probability of SES 3 children the pricier preschools, contrasted with 26% of SES 2 attending preschool was 20% more likely than SES 1 children and 41% of SES 1 children. Oddly, the poorest children. In Cape Verde, the probability of SES 3 was SES was more likely to enroll in more expensive schools 22%. Moreover, greater percentages of SES 3 children than the better-off middle SES group. To what extent have attended more than one year of preschool. In this is an artifact of availability is unclear. Guinea, 25% of the SES 3 children attended multiple Wealthy households were also more likely to opt for years of preschool compared with 18% in SES 1 and 8% preschools that had an academic orientation, rather in SES 2. In Cape Verde, 45% of SES 3 children attended than a play-base or daycare orientation. The former multiple years compared with 23% in SES 1 and 19% in institutions emphasize the acquisition of academic SES 2. skills, such as reading, writing, language, and math, Other variables also underscore the positive rela- while the latter institutions focus on games, child tionship between family SES and preschool attendance, entertainment and, quite simply, babysitting. In Parental education and literacy, contributing to higher Guinea, 62% of the SES 3 group children attended such EARLY CHILDHOOD DEVELOPMENT IN AFRICA: CAN WE DO MORE FOR LESS? 21 Table 5.1 Probability of preschool attendance Selected Variables' Guinea (n=877) Cape Verde (n=803) Reference Variable Active Variable Coefficient t Coefficient t Boy Girl +0.09 n.s. +0.14 n.s. Family with <5 children Family with >5 children -0.16 ** -0.64 Mother is housewife Mother works outside home +0.06 n.s +0.44 ** Illiterate mother Literate mother n.d. +0.42 n.s. Parents are a couple Mother is alone n.d. -0.47 ** No/few household tasks by child Household tasks by child -0.12 n.s. -0.38 ** Low household SES (SES 1) Middle SES (SES 2) +0.31 -0.21 n.s. High SES (SES 3) +0.82 +0.98 Child seldom ill Sometimes ill -0.03 n.s. n.d. Usually ill -0.08 n.s. n.d. Parents do NOT read to child Parents read to child +0.10 n.s. n.d. Constant +0.05 n.s. +0.32 D of Sommers 0.28 0.33 nd.: no data n.s.:not significant *:significant to 10% *:significant to 5% -**significant to 1% *Year and month of birth variable not presented in table. schools, followed by 59% of SES 1 and 49% of SES 2. In In Cape Verde, the greatest number of children are Cape Verde, 73% of SES 3 children were enrolled in enrolled in public preschools. While NGO-run and academic program preschools, compared with 15% of community preschools also provide services to the SES 1 and 2. lowest and middle groups, the wealthiest SES group is Preschool selection according to quality variables much more likely to send its children to private-and presents a mixed picture, although some clear patterns more costly-preschools. stand out. SES 3 children in Cape Verde were twice as likely (30%) to have more highly qualified teachers Childcare may also be a factor than the lower SES groups (15%). However, the mid- in preschool enrollment. level SES 2 appeared to benefit marginally more from "better equipped" schools (with playgrounds and The poorest SES group exceeds the middle SES dining halls) and bilingual instruction. In Guinea, both group in multiple years of preschool enrollment, the poorest and the wealthiest children had formally probably because both parents must work. A greater qualified teachers; a notably greater percentage of SES percentage of preschool students have mothers who 2 children were enrolled in schools with lower teacher work outside of their homes or engage in professional qualifications. (See Section 7 for a discussion of teacher activity. In Cape Verde, children whose mothers work qualifications.) This is largely due to the prevalence of outside of home are 9% more likely to attend pre- the CEC model that serves the rural communities school, although in Guinea the positive correlation where SES 2 predominates and accounts for nearly half was not statistically significant. Nonetheless in both of this group's enrollments. In contrast, about three- countries, the percentage of students who attend fourths of the poorest and wealthiest groups are multiple years of preschool increases notably over enrolled in daycare programs. those attending only one year when their mothers 22 AFRICA REGION HUMAN DEVELOPMENT WORKING PAPER SERIES Table 5.2 Preschool attendance by SES (%) Guinea (n=529) Cape Verde (n=490) Preschool Attributes SES 1 SES 2 SES 3 SES 1 SES 2 SES 3 SCHOOL COST -Low 59 74 48 56 39 17 -High 41 26 52 44 61 83 EDUCATIONAL MODEL -Academic 59 49 62 53 47 74 -Games/nursery - - - 47 53 26 QUALITY FACTORS -Low pupil: teacher ratio 35 31 33 30 30 30 -Highly qualified teachers 42 27 40 15 15 30 -Well-equipped school 11 03 11 22 28 24 -Bilingual instruction 75 83 74 75 78 73 PRESCHOOL S'RUCTURE n.a. n.a. n.a. -Ecole maternelle 75 45 71 -CEC 09 49 07 -Jardin d'enlants 16 06 22 PRESCHOOL INSTITUTION n.a. n.a. n.a. -Public 51 41 41 -Private 03 02 35 -NGO 23 25 09 -Community 20 20 11 -Religious 03 12 04 work. In Cape Verde, 53% of children who attended pation rate of girls in preschool is slightly more than more than one year of preschool had a working boys at 56% for both single-year and multiple-year mother, compared with 37 who attended only one attendance. In Cape Verde, girls represent 53% of chil- year. Likewise in Guinea, 54 of children with working dren enrolled in one year of preschool and 55% of chil- mothers attended multiple years of preschool, com- dren enrolled in more than one year. The lack of gender pared with 32% who attended for only one year. Chil- dren from larger families are less likely to attend preschool, possibly because older siblings are tasked with childcare. Girls' preschool participation in Guinea The survey of 460 preschools (comprising 26,652 children) Girls have an equal chance at revealed that co-educaton is the predominate model and near preschool attendance. gender parity prevails in most schools. Overall, girls account for 48% of the students and the per schooi gender distribution is strongly correlated (Brave-Pearson r = +0.86). Why this is so is Unlike higher levels of education, gender does not not dear. Analysis found little relaton between the proporbon of appear to be a major determinant of preschool enroll- girls enrolled in preschools and preschool predominately staffed ment. In contrast to primary school, where the gross with female teachers, contrary to findings at the primary school primary enrollment of boys exceeds that of girls by 19% level in Guinea and other African countries. There is, however, in Guinea and by 3% in Cape Verde, the percentage of some variation according to region, with the lowest female partic- in Guinea and by 3% in Cape Verde, the percentage ot ipation rate in Haute Guinee and the highest in Moyenne Guinee. girls enrolled in the preschool samples investigated modestly exceeds that of boys. In Guinea, the partici- EARLY CHILDHOOD DEVELOPMENT IN AFRICA: CAN WE DO MORE FOR LESS? 23 disparity at this level has two possible explanations. While there is abundant evidence that preschool pro- First, the small percentage of parents (in Guinea) who grams must combine physical, mental, and emotional enroll their children in preschool may already be highly stimulation and that emphasis on primary school aca- motivated to educate their daughters, thus biasing the demic competencies is premature for young children, sample. Second, preschool enrollment may represent a programs that provide basic conceptual and pre-read- consumption decision by parents to obtain childcare to ing and writing skills can enhance school readiness. free them for work or leisure, rather than an investment Poor children may not receive the same preparation decision about the benefits and the positive returns for primary school, putting them at a disadvantage. captured later in terms of increased educational attain- [Note that this study did not look at the impact of dif- ment. Girl children, as well as boy children, equally ferent programs on primary school student success.] require care, although they may not be seen to equally The fact that there is little gender bias evidenced in require schooling. enrollments also suggests that efforts to ensure afford- able, quality preschool for poor children will benefit Some policy implications girls as well as boys. The provision of preschool education can Public policy can alleviate these disparities exacerbate disparities between rich and poor through progressive policies aimed at lowering the children, as the wealthy are its biggest users price barriers to preschool entry for poor families and beneficiaries. and thereby improving household welfare. Because children from wealthier households are Since children from poorer households do attend more likely to attend preschool, they will reap the preschool in significant proportion, there is little evi- developmental benefits it provides (discussed in later dence to suggest that constraints other than economic section). The interactive effect between (1) higher ones would prevent lower SES households from socio-economic status and related variables, such as enrolling their children in preschool, if the service was parental education and literacy, which is well-docu- available and the price affordable. [Given the designs of mented in the literature as a major explanatory factor the two studies, it is impossible to show to what extent in child development and educational attainment, poorer parents decided not to enroll their children in and (2) preschool attendance, which improves cogni- preschool due to the poor quality options available.] In tive development can widen the gap between wealth- light of the indications that preschool is used by work- ier and poor children, making the former more ready ing mothers as a means of child care, it is reasonable to for primary school and contributing to their chances assume that lowering preschool fees-possibly of eventual academic success. Meanwhile, the poorer through government subsidies for poorer families- children struggle with the inherent disadvantages would free mothers to find employment, thus improv- associated with poverty and without the added and ing household welfare. Lokshin et al. have found in remedial benefits of preschool, because their parents Kenya, Romania, and Russia that government subsi- are less able or inclined to enroll them. dies to reduce cost and improve the quality of child care are effective in boosting the use of ECCD programs, The gap may widen due to the type and quality of resulting in increased levels of women's labor force preschools affordable to different SES-level participation. households. Without government intervention, it is almost inevitable that preschool can have a regressive effect in Even if children from poorer households have the which the rich benefit disproportionately from both the opportunity to attend preschool, they may not benefit benefits of preschool and any public funds that support as much as their wealthier peers because of the (osten- it, as in Cape Verde where over 40% of the better-off sibly) lower quality schools they are more likely to children from SES groups 2 and 3 are enrolled in public attend. Preschool for poorer children appears to focus preschools, compared with the 51% of the poorest more on childcare and less on academic preparation. SES 1. There are several options for public policy to alle- 24 AFRICA REGION HUMAN DEVELOPMENT WORKING PAPER SERIES viate these disparities, depending on the country con- resources for preschool, while at the same time encour- text and resource envelop. Public preschools could aging private-sector provision through supportive and uniquely target poor children for admission. Public non-restrictive policies, demand for private sector pre- subsidies could be provided only to those private- school services by the wealthy could in effect free up sector preschools that serve a poor clientele. By exclud- public resources to be used for the poor. ing or minimizing wealthy families' access to public 6 Impact: The Effects of Preschool on Child Development The story difference in the total score; in Cape Verde it is even more pronounced, with a 10-point difference between A lmost universally, studies on the impact of preschool and control groups. AA preschool have found that enrolled children Multivariate analysis-used to identify the rela- benefit by gains in both cognitive and physi- tive impact of a number of socio-demographic and cal development. This study obtained similar results environmental variables in addition to preschool in both Guinea and Cape Verde in terms of cognitive attendance that are not captured in the raw scores- development. The evidence is less straightforward reveals that overall, preschool will account for about and compelling on the impact of preschool on physi- two percentage points of the total variation explained cal development. This section examines the effects of in scores in Guinea and Cape Verde. Controlling for preschool on child development and the other factors several socio-demographic and environmental vari- that come into play and discusses some of the policy ables (see Table 6.2), preschool attendance will sub- implications associated with them. stantially increase a child's score by 4.41 points in Guinea and 7.27 points in Cape Verde. Few other fac- The effect of preschool on cognitive tors will have as large an effect as preschool. development is a positive one. Preschool will increase a girl's score in Guinea by more then three points, but the small increase in Preschool students in each SES grouping in both Cape Verde is not statistically significant. Parents Guinea and Cape Verde attain higher razv test scores who read stories at home to their children will than the control group children who have not increase the scores by 2.5 points in Guinea and nearly attended preschool. In Guinea, there is a five-point four points in Cape Verde. Table 6.1 Cognitive development scores Selected Variables Guinea (n=877) Cape Verde (n=803) Sample SES 1 SES 2 SES 3 Sample SES 1 SES 2 SES 3 School Con Schl Ctrl Schl Ctr Schl CtrI Schl Ct Schl Ctr Schl Ctr Schl Ctr Total Score 102 97 98 93 104 99 105 103 104 94 101 92 105 95 109 95 Language Score 102 97 100 95 101 98 105 100 103 95 100 93 104 96 110 95 Pre-reading Score 102 96 99 95 103 97 105 99 103 95 100 95 104 96 109 94 Note: Cognitive score average of 100 with standard deviation of 15 in Guinea and Cape Verde. 26 AFRICA REGION HUMAN DEVELOPMENT WORKING PAPER SERIES Table 6.2 Impact of preschool, socio-demographic, and environmental factors on cognitive development Selected Variables Guinea (n=877) Cape Verde (n=803) Reference Variable Active Variable Coefficient t Coefficient t Boy Girl +3.08 +0.57 n.s. Family with <5 children Family with >5 children +2.23 -1.92 *t Mother is housewife Mother works outside home -3.15 ... +2.43 Illiterate mother Literate mother n.d. +2.51 ** Parents are a couple Mother is alone n.d. +2.71 No/few household tasks by child Household tasks by child +3.72 -2.49 Low household SES (SES 1) Middle SES (SES 2) +4.43 +3.45 High SES (SES 3) +6.34 +2.86 ii Child seldom ill Sometimes ill -5.77 n.d. Usually ill -7.60 n.d. No Preschocl Preschool Attendance +4.41 +7.27 Parents do NOT read to child Parents read to child +2.46 +3.90 Constant 97.49 i* 88.8 R2 (percentage of variance explained with all socio-demographic, environmental, and preschool variables) 16.04 20.69 R2 (percentage of variance explained with all preschool and socio-demographic variables only) 13.2 14.0 R2 (percentage of variance explained with all socio-demographic variables only) 11.2 12.75 n.d.: no data r.s.: not significant *:significant to 10% -*:significant to 5% ***:significant to 1% *Year and month of birth omitted from table but included in R2. Despite this agreement, it is interesting to note that found in Cape Verde, where single mothers also seem several variables have opposite effects on student to contribute positively to a child's cognitive score. scores. In Guinea, a working mother will have a neg- (Cape Verde's demographic pattern presents a high ative effect on a child's score, whereas in Cape Verde incidence of single women as a result of male immi- a working mother will result in higher scores. In the gration overseas). Not surprisingly, the scores of early 1960s the High/Scope Perry Preschool study Guinean children who are occasionally and fre- found that children in both the experimental and con- quently ill will decrease significantly, with the drop trol groups did better during their school years than being greater than the point benefits derived from their peers if their mother was working. This out- preschool attendance. come was attributed to the fact that the working mother exhibited more initiative, organization, and Preschool will have different effects on goal focus behavior than the non-working mother, various aspects of cognitive development. especially among the poor. In Guinea, large family size and child performance of household tasks does Guinean preschools have the greatest success in not appear to handicap, while the opposite was developing pre-reading or visual discrimination EARLY CHILDHOOD DEVELOPMENT IN AFRICA: CAN WE DO MORE FOR LESS? 27 Table 6.3 Domains of cognitive development Variable Guinea (n=877) Cape Verde (n=803) Basic Concepts Language Pre-reading Basic Concepts Language Pre-reading Coeff t Coeff t Coeff t Coeff t Coeff t Coeff t Preschool Impact +2.75 *** +4.08 +4.83 *** +6.5 *** +5.2 *** +4.7 1 year only n.d. n.d. n.d. +6.6 +3.2 +3.4 1+ years n.d. n.d. n.d. +7.2 *** +9.9 +8.0 n.s.: not significant ':significant to 10% ^':significant to 5% *-:sinificant to 1% skills (five points) and oral language skills (four while the wealthiest group-SES 3-differs only 1.04 points), both essential to school readiness. Preschool points. This indicates that preschool can help narrow adds less than three points for basic concepts, possi- the gap between disadvantaged and advantaged bly because these skills can be more easily gained at groups. home. There are no gender differences in pre-reading This equalizing trend is reversed in Cape Verde, scores. In contrast, Cape Verde preschools enjoy the where the 14-point gap between the wealthiest SES 3 most success in developing basic concepts (6.5 points) preschool and control group exceeds the 10- and and language skills (5.2 points), followed by pre- nine-point differences in the SES 2 and SES 3 groups, reading or visual discrimination skills (four points). respectively. The most favored SES 3 group appears Moreover, multiple years in preschool will increase a to benefit most from preschool (9.9 points), followed child's total score by three points, with language by SES 2 (7.4 points) and SES 1 (6.7) points. skills improving most with time in preschool and Girls, in general, seem to benefit more from pre- basic concepts improving most with only one year of school. While there is no statistically significant dif- preschool (see Section 7 for a discussion on language ference in the point difference accruing to boys and of instruction and cognitive development). girls in Guinea in the poorest SES 1 group, the bene- fit of preschool to girls becomes apparent in SES 2 The impact of preschool attendance on and is strongest in SES 3, accounting for a six-point cognitive development varies according gain. In Cape Verde, there is a three-point difference to SES level and gender. for girls. The several variables exert different and occasion- A child's socio-economic status will affect the ally contradictory influences according to SES level. impact of preschool. In both Guinea and Cape Verde, Little pattern can be discerned amongst the groups. the cognitive scores (Table 6.1) of the preschool and For example, doing household tasks has no significant control groups follow the expected socio-economic impact on children in the lowest SES, but strongly and hierarchy-scores increase with socio-economic positively affects the score of SES 2 and SES 3 children. status. Children in the wealthiest SES 3 group (both Mothers working will significantly and negatively control and preschool) demonstrate the highest impact SES 1 and SES 3 children, but have no signifi- scores and the poorest SES 1 group the lowest, aver- cant impact on SES 2 children, possibly because this aging an eight-point span in Guinea and a nine-point group comprises rural, agrarian families where most span in Cape Verde. work is performed close by on the farm. However, in Guinea, the lower SES groups 1 and 2 In Guinea, SES level was also found to influence seem to derive greater benefit from preschool, with the point difference in the three areas of cognitive raw point gains of around five compared with a two- development. Although children in SES 1 and SES 2 point difference for SES 3. These patterns are con- tended to derive a constant or equal benefit in all firmed using multivariate analysis, with SES 1 and three areas, the wealthier SES 3 children appeared to SES 2 differences of 5.07 and 4.86 points, respectively, derive no benefit in basic concepts. 28 AFRICA REGION HUMAN DEVELOPMENT WORKING PAPER SERIES Table 6.4 Impact of preschool on cognitive development by socio-economic level Selected Variables* Guinea (n=877) Cape Verde (n=803) SES 1 SES2 SES 3 SES I SES2 SES 3 Active Variable Coeff t Coeff t Coeff t Coeff t Coeff t Coeff t Girl +1.79 n.s. +2.49 * +6.1 *- +1.09 n.s -0.48 n.s +1.27 Family with >5 children +1.95 n.s. +3.56 -0.34 n.s. -0.47 n.s 4.17 +1.39 n.s Mother works outside home -3.99 -0.83 n.s. -3.50 * +0.4 n.s +3.74 ** +6.23 - Literate mother n.d. n.d. n.d. +2.37 n.s +3.74 +2.00 n.s Mother is alone n.d. n.d. n.d. +2.25 n.s -2.79 n.s +0.72 n.s Household tasks by child +1.45 n.s. +6.25 *** +4.3 -2.6 * -2.59 n.s -3.10 n.s Sometimes ill -3.25 n.s. -5.90 * 6* -.0 n.s. -0.06 n.s +1.88 n.s -2.52 n.s Usually ill -3.36 n.s. -11.05 -7.0 n.s. +1.29 n.s +1.78 n.s -1.72 n.s Preschool Attendance +4.86 *** +5.07 *t +1.04 n.s. +7.24 *** +7.34 +9.89 Parents read to child +3.34 ** +0.55 n.s. +4.24 ** +4.29 +4.54 *' +2.04 Con.s.tant 101.6 103.0 *** 100.2 89.03 92.61 93.85 R2 (percentage of variance explained with all variables) 10.8 15.6 15.8 13.3 21.7 25.8 n.s.: not significant ':significant to 10% **:significant to 5% -i:significant to 1% *Year and month of birth omitted from table but included in R2. Preschool attendance did not have a significant In Guinea, although the preschool children from impact on weight/height ratio. the poorest SES 1 and the wealthiest SES 3 groups have higher ratios than the control groups, the Improved physical development is often associ- inverse is true for SES 2 (Table 6.5). Even more oddly, ated with preschool attendance. However, in both both the wealthiest SES 3 preschool and control Guinea and Cape Verde preschool was not found to groups have the least robust weight:height ratio. have a significant or appreciable positive impact on Multivariate analysis using the standard indicators physical development. found in Table 6.5 explains only 3% of the variance. Preschool had a significant and positive effect on the Figure 1 physical development of children from the highest Impact of preschool on cognitive development SES level, and children who attended preschool for according to SES in Guinea more than one year have a slightly more favorable | SESI ESES2 ES3E weight:height ratio. More straightforward and easily interpreted is the relationship between region of resi- dence and weight:height ratio, which accounts for 6---- 37% of the variation in the weight:height ratio of 4 --- - -- ---- Guinean preschoolers. On average, the more robust 2 2- - . children live in the urban capital of Conakry rather o -_ than in other regions, which are predominantly rural. -2 The Forest Region (Guin&e forestiere) and Lower -2 - _ - - Guinea (Base Guinee) suffer from the weakest language concepts pre-reading weight:height ratios, corresponding to governmental Cognitive Development estimates of prevailing malnutrition. EARLY CHILDHOOD DEVELOPMENT IN AFRICA: CAN WE DO MORE FOR LESS? 29 Table 6.5 Physical development scores Selected Variables Guinea (n=877) Cape Verde (n=803) Sample SES I SES 2 SES 3 Sample SES 1 SES 2 SES 3 School Control Schl Ctrl Schl Ctrl Schl Ctrl Schl Ctrl Schl Ctrl Schl CtrI Schl Ctrl Weight:Height Ratio 14.5 14.7 15.0 14.6 14.7 15.4 13.7 13.3 15.2 15.3 15.1 15.3 15.0 15.3 15.8 15.2 Vaccination (average # of vaccinations) 8.6 7.4 9.2 7.9 7.4 6.4 9.2 8.8 7.7 7.3 7.6 7.3 7.6 7.2 8.0 7.8 Infantile Illness Incidence (average # of illness) 1.5 1.6 1.4 1.6 1.6 1.7 1.5 1.3 1.04 1.1 1.0 1.1 1.0 1.1 1.0 1.2 Note: Weight:Height ratio average of 14.6 with standard deviation of 3.9 in Guinea; and 15.2 with standard deviation of 2 in Cape Verde. In Cape Verde, SES 3 preschool and control groups Some policy implications have the most robust weight:height ratio, and, like Guinea, those who attend multiple years of preschool The gains in cognitive development will be more robust physically. associated with preschool suggest that the There are several explanations for the negligible provision of preschool should be supported impact of preschool on the weight:height ratio. First, and/or encouraged by government as an approximately half of the preschools surveyed did not investment in human capital. have a cafeteria or canteen, possibly signaling that there was no nutrition or feeding program. Second, the study The studies in Guinea and Cape Verde confirm the did not verify whether the reported canteens were results of a growing body of research. Although operational; nor did it look into nutrition inputs of family background exercises the most influence, pre- these programs. Third, and most important, is the school increases the explanatory power to approxi- appropriateness the weight:height measure for the age mately 16 to 20%. group examined by the study. Since stunting occurs in children under age three and cannot be remedied by Preschool can have an equalizing effect on better nutrition at a later age, the weight:height ratio cognitive development. may have limited validity as a measure of the develop- ment of five-year-olds. Not only does preschool improve all children's cognitive development scores, it appears to have a Preschool students are slightly more likely to be disproportionate effect on children from lower SES vaccinated and less prone to childhood illnesses. groups in Guinea who experience relatively greater point difference from the control group, although it In both Guinea and Cape Verde, vaccination cover- does not close the gap between wealthier and poorer age for all SES levels is better for preschool students children. In Cape Verde, it appears to reverse the dis- than for control group children. There are two possible advantage of children from large families and whose explanations for this: one is that parents who enroll mothers work. And since the wealthiest children in their children in preschool may be better informed or Cape Verde appear to derive the greatest benefit in more conscientious about vaccinating their children; terms of point gains on cognitive development tests, another is that preschools are both more accessible to failure to enroll lower SES group children in pre- and targeted by health campaigns and inoculation serv- school may result in even wider disparities in cogni- ices. In total, the incidence of childhood disease is tive development and school readiness skills. slightly mitigated by preschool attendance in both Another disadvantaged group-girls-also ap- countries, but in Guinea SES 3 preschool student suffer pears to benefit significantly from preschool. slightly more illness than the unschooled control group. Preschool has a greater impact on girls' cognitive 30 AFRICA REGION HUMAN DEVELOPMENT WORKING PAPER SERIES development than that of boys. This suggests that ment process, not a "one-off" intervention that essen- preschooling girls may have a disproportionately tially serves as the introductory year to primary positive influence on primary school participation at school. Public policy could facilitate multiple-year a future date, as girls are generally at a disadvantage attendance by targeting subsidies at and encouraging academically even in the early primary grades. For the development of preschools that serve poor chil- example, student performance data in Guinea show dren, who are less likely to attend multiple years, and that as early as the second grade, girls are doing less by offering programs that are gauged to serve the well on math and language tests (TietJen in Sutton et development needs of different age groups. al. 1999). As countries intensify their pursuit of universal pri- Preschools-particularly public ones serving mary education to meet Education-For-All goals estab- all SES levels of children-should tailor their lished for 2015, the policy implications of the equalizing programs to suit the needs of their students effect of preschool are evident. Vulnerable and at-risk or the most vulnerable population. children-whether the urban or rural poor, disadvan- taged minorities, or girls-are likely both to need and Preschool programs may need to change emphasis to profit from the "value-added" of preschool in order among basic concepts, language, and reading, to overcome their deprived environments and-based depending on the SES group being served, given on findings of several ECCD studies-to enter primary their different performances. school on a more equal footing with their more advan- taged peers. Governments should consider how they The impact of preschool and better child health can support ECCD programs, encourage their growth, and physical development should be further and target their resources, efforts and policy at the investigated. needy children who will benefit disproportionately from them (as in Guinea) or will fall further behind if Although there was no significant relationship they are not enrolled or are enrolled in ineffective pro- found between preschool programs and physical grams (as in Cape Verde). development in either Guinea or Cape Verde, the rea- sons for this are unclear and not adequately addressed Public policy could encourage the by the study. Preschools may not be providing sup- development of multiple-year programs. plementary feeding programs, or these may be insuf- ficient to compensate for the conditions contributing Children who participate in more than one year of to stunted growth and malnutrition suffered by the preschool score higher on cognitive development children of impoverished families. Moreover, the tests and gain greater language skills. Ideally, pre- weight:height ratio used may not be valid as a meas- school should be seen as part of the child develop- ure of the physical development of five year olds. 7 Effectiveness: Programs or Program Characteristics That Appear to be Most Effective in Terms of Child Development The story This section examines the effect of various preschool characteristics-including type of school, program W hile there is little question that preschool and organization, teachers, and costs-on preschool ev w has a positive impact on a child's develop- student cognitive and physical development, and dis- ment as shown in the previous section, it is cusses some of the policy implications of the research. not so evident which preschool/ECCD program Table 7.1 presents some descriptive information that attributes have the greatest influence on these results. will be further elaborated in the text. Table 7.1 Child development scores, school characteristics, and SES level School Characteristics Guinea (n=529) Cape Verde (n=490) Sample SES I SES 2 SES 3 Sample SES I SES 2 SES 3 Cog W:H Cog W:H Cog W:H Cog W:H Cog W:H Cog W:H Cog WAH Cog W:H SCHOOL TYPE -Ecole maternelle 100 15.1 96 15.4 100 15.9 104 13.9 -CEC 109 13.7 109 13.6 109 13.7 113* 13.7 -Jardins d'enfants 102 13.5 102 14.0 98 13.4 103 13.2 -Public 99 15.5 97 15.3 101 15.2 103 16.4 -Private 105 15.5 102 14.0* 101 15.1* 109 15.8 -NGO 97 14.4 95 14.2 100 14.4 102 14.7 -Community 98 15.5 97 15.7 99 15.3 104 15.0 -Religious 111 14.6 118* 14.4 108 14.7 110 14.4 EDUCATIONAL MODEL -academic 102 14.7 98 15.4 102 15.2 105 13.4 102 15.0 98 14.9 103 14.4 106 15.9 -games/nursery 105 14.2 98 14.6 107 14.0 111 13.9 101 15.2 99 14.9 102 15.3 106 15.4 QUALITY FACTORS -low pupil:teacher ratio 99 15.7 96 15.9 100 15.6 101 15.6 101 14.8 97 14.9 102 14.7 111 15.0 -highly qualified teachers 102 15.7 100 16.2 102 16.5 106 14.6 101 15.4 99 14.8 97 15.1 107 16.5 -well-equipped school 109 12.9 102 13.5 105 13.3 118 12.1 98 14.7 94 14.6 100 14.5 101 15.2 -bilingual instruction 102 14.8 97 15.5 104 14.9 105 13.7 100 15.1 97 14.9 101 14.9 106 15.8 *indirates fewer than 10 in the sample 32 AFRICA REGION HUMAN DEVELOPMENT WORKING PAPER SERIES Table 7.2 10% (21-11) indicates that particular school character- Differentiated impact of preschools on istics are likely to exert significant influence on stu- cognitive development dent development, whereas the 6% (20-14) marginal Guinea (n=877) Cape Verde (n=803) gain in Cape Verde suggests that school effectiveness factors may be difficult to identify with precision and Variance Variance explained (%) explained (%)elude measurement. Child characteristics 11.4 14.0 Community education centers in Guinea and Preschool 42.1 55.0 religious preschools in Cape Verde emerge as the most effective schools in raising cognitive Child + Preschool 45.0 58.0 development scores. Child + Aggregated School Characteristics 21.0 20.0 The CEC preschools in Guinea and the religious- run preschools in Cape Verde exhibit the highest cog- nitive development scores. In Guinea, the raw test scores for the total sample average 109, compared Various preschool programs, organizational with 100 points for ecoles maternelles and 102 points structures and attributes will affect child for jardins d'enfants (Table 7.1). This is borne out by development differently. multivariate analysis, which shows a ten-point advantage over &oles maternelles and jardins. A closer Variations in preschools do matter in terms of cog- look at the score breakdown shows that CEC pre- nitive development, and some preschools are more school scores are mainly due to a strong performance effective than others (Tables 7.1 and 7.2). While child on the basic concepts portion of the test, accounting characteristics explain 14% (Guinea) and 11% (Cape for a 13-point advantage. This is particularly notable Verde) of variance in preschool student cognitive because CEC preschools primarily serve rural and development, the preschools themselves account for often poor communities. In fact, no less than 70% of 55% in Guinea and 42% in Cape Verde, with scores students in any CEC preschool came from the lower varying as much as 30 points in Guinea. Even when SES levels, primarily SES 2. However, many ques- socio-demographic and household environment are tions remain about why this model has proved effec- taken into account, the impact of the individual tive. On the plus side, CEC preschools have higher school remains important. However, when aggregate community involvement, a higher percentage of five- school characteristics are considered, the picture and six-year-olds, and their teachers tend to live in becomes less clear. In Guinea, the marginal gain of the village where the preschool is located. On the Table 7.3 Impact of school type on cognitive score Guinea (n=529) Cape Verde (n=490) Reference Variable Active Variable Coefficient t Coefficient t Ecoles maternelles CEC +10.05 NA Jardin d'enfants +1.12 n.s. NA Public Centers Private Preschools NA -0.4 n.s. NGOs NA -1.1 n.s. Communal Centers NA -1.7 n.s. Religious Preschools NA +10.7 n.s.: not significant *:significant to 10% **:significant to 5% ***:significant to 1% EARLY CHILDHOOD DEVELOPMENT IN AFRICA: CAN WE DO MORE FOR LESS? 33 bilingual instruction-had different effects. In What are CEC preschools? Guinea, children receiving preschool instruction in The community education center preschools (centre d'den- French and a local language experienced the largest cadrement communautaire de la petite enfance") are non-formal negative effect (-5.80) produced by any school char- institutions, created (with some extemal assistance) and man- aged by the local community or women's association. Serving children aged three to six years, their initial objective was to into consideration, this large negative effect remains. relieve working mothers of some childcare burdens, anc there- Also, there appears to be no inherent value to the lan- fore they are organized around the local agricultural calendar. Ideally (not necessarily reflective of the reality presented above), guage of struction: children taught solely in French each CEC is staffed by two monitrices, with a Grade 6 education or solely in a local language scored better. One supplemented by special training in early childhood care and methodological explanation is that households self- education. The CEC preschool is housed in a modest three- room structure, built or rehabilitated with local resources. The select the preschool which uses the language em- program focuses on child development and nutritional care, and ployed at home, thus ensuring compatibility between the teachers are responsible for fabricating toys and teaming child and language of instruction. A developmental aids. The community is responsible for teacher salaries (in-kind or in cash) and the provision of local construction. Village alumni explanaton iS that the child's first language has to be associations (associations de ressortissants) have provided developed to the level where he/she has the concep- funds for both construction and equipment. The government (the tual and linguistic prerequisites for the acquisition of Ministry of Social Affairs or the Prefectoral Education Offce) may provide vacant building and some modest supplies. Parents pay literacy and pre-literacy skills. These findings are a monthly fee of 1000FG per child, with half of this going to consistent with other research (Cummins 1979; teacher compensation. Development support-teacher training, Tucker 1977; Dutcher 1982). When children have to program creation, organization, etc.-have been provided by d UNICEF and Aide et Action, an international NGO. deal in a bilingual environment, their cognihve and linguistic proficiency in both languages are interde- pendent, and for these proficiencies to be developed in a second language, a certain level of cognitive and linguistic development in the first language has to be minus side, these teachers spend somewhat less time accomplished. ' ~~~~~~~~~~The inverse situation was found in Cape Verde, teaching, and a higher proportion of them are male, wher cildrse sinubilngual p o od s Cemployin unrie,adieprene. oevr h E where children in bilingual preschools employing uraed,hoos an hin erienced Mrerr, the CEC both Creole and Portuguese enjoy a 2.5 point differ- pnges3cpupls hav each hig her.pupil:teacher ratio, ence in their overall cognitive scores, over those chil- i .g 53ppl.o ahtahr . dren who are taught uniquely in Creole. To Religious schools in Cape Verde take the lead in cognitive development score, also enjoying a 10 point understand the different findings in both countries, it cognitve development score, also enjoying a 10 point i . ' ) > O r ~~~~~~~iS important to note that there are fundamental dif- advantage. In particular, poor SES 1 children enrolled motn ont htteeaefnaetldf idvanthges tyes par schlar,poos scoES Ihigrerin thesled ferences in the use of the European languages in both schools thananyothesety f res.Thels sores arei tepered countries. In Guinea, French is not widely spoken by schools than any others. w ere only ltw re schools the majority of the population and is a second lan- by the fact that there were only two religilous schools gu etomn exsiglclagae.InCp ,2 , ,, , D , ~~~~~~guage to many existing local languages. In Cape included in the sample, throwing into question any Ver meanngfl cocluionsabot th suerioityof Vei- de, Portuguese is widely spoken by the majority geainugshl ionclst ions , curric ulum,or managemet of the population, and the only local language is Hioweve,u seeralo schoolsnstruction,cu, in C napVerden Creole, which has roots in ancient Portuguese. As However, several other schools in Cape Verde also noebyDthrteprctinftewdeom ' . ,, r ~~~~~~~noted by Dutcher, the perception of the wide corn- produced high cognitive development scores ' munity of the status of both the local and national demonstrating that effective schools do not necessar- y U . . ~~~~~~~languages is key in understanding the learning ily adhere to a particular type or organization, process in bilingual environments. Bilingual pro- grams do not retard the development of children in chogicive ofvelanguageofinstruction.affecttheir native language if either the school or the com- cognitive development. munity provides support and stimulation for the In both Guinea and Cape Verde, the language of continued development of the native language (Dutcher 1982). instruction had a significant impact on cognitive ' development scores. Even more, the same variable- 34 AFRICA REGION HUMAN DEVELOPMENT WORKING PAPER SERIES Table 7.4 Impact of school (and child) characteristics on cognitive score Selected Variables* Guinea (n=529) Cape Verde (n=490) Reference Variable Active Variable Coefficient t Coefficient t CHILD CHARACTERISTICS Boy Girl +3.68 +0.57 n.s. Family with <5 children Family with >5 children +2.2 ** -1.92 ** Mother is housewife Mother works outside home n.d. +2A43 .. Illiterate mother Literate mother n.d. +2.51 ** Parents are a couple Mother is alone n.d. +2.71 ... No/few household tasks by child Household tasks by child +6.55 -2.49 Low household SES (SES 1) Middle SES (SES 2) +4.34 ** +3.45 High SES (SES 3) +4.57 ** +2.86 ** Child seldom ill Sometimes ill -4.12 ** n.d. Usually ill 4.48 ** n.d. Parents do NOT read to child Parents read to child +2.80 +3.90 SCHOOL CHARACTERISTICS Single language instruction Bilingual instruction -5.80 +2.48 * Non-academic program Academic program +1.01 n.s. n.d. Female teacher Male teacher -2.49 n.d. Professional experience (number of working years) +0.72 n.d. Teacher does NOT reside in village Teacher is resident +3.46 n.d. Number of declared weekly teaching hours +0.21 n.d. Number of pupils in class +0.11 *-0.30 No dining hall in school Dining Hall n.d. +3.64 Percentage of paying families n.d. +0.21 ** Constant 85.8 91.0 ... R2 (percentage of variance explained) 20.5 19.8 n.s.: not significant *:significant to 10% **:significant to 5% ***:significant to 10% *Year and month of birth omitted from table but included in R2. Class size can vary somewhat without serving fewer than 35 children and a quarter serving negative consequences. fewer than 26 children. The situation is different in Cape Verde where it appears that exceeding a thresh- In both Guinea and Cape Verde, class size proved old level of 30 children per class will result in cogni- to have a significant effect on cognitive scores. How- tive score point losses. For every 10 additional ever, it exerted opposite influences in each country. In children, the cognitive score will be reduced by three Guinea, for every child added to the class, there was points. a fraction of a point difference. However, this should not be interpreted to mean that increasing class size Teacher characteristics present a mixed bag. will improve cognitive scores. This anti-intuitive result can be explained by the fact that in general Several teacher variables were examined: sex, pro- class sizes are relatively low in Guinea, with most fessional experience, training, age, and place of resi- EARLY CHILDHOOD DEVELOPMENT IN AFRICA: CAN WE DO MORE FOR LESS? 35 dence. Of these, none exerted a significant influence emerged in terms of cognitive scores for any one ap- on cognitive scores in Cape Verde, although several proach. It is important to note that this study did not did in Guinea. In Guinea, female teachers appear to attempt to do classroom observation, which is a more be more effective in teaching preschool children than appropriate method to evaluate the interaction be- men, accounting for 2.5 point increase. (Unlike other tween teachers and children and to draw conclusions countries, female preschool teachers are in the minor- regarding key elements of the different types of cur- ity, comprising less than 40% of the teaching force. riculum. The tight job market makes any teaching post desir- able, and men are not reluctant to teach at this level.) High-cost preschools do not guarantee high child While the academic qualifications of teachers has no development scores, but parents paying school significant effect on student development, teacher fees has a positive effect. professional experience contributed positively to stu- dent scores: for every five years of experience, there High per-student expenditure does not necessarily will be a commensurate gain of 3.5 points. More result in high cognitive development scores (Table important was the influence of teacher residence. A 7.5). Although the average raw scores vary according teacher who lives in the same village as the school or to SES level, lower unit costs are consistent with in a nearby neighborhood will account for a 3.5 point higher scores in both countries, with preschool chil- gain. There are two possible reasons for this. First, a dren scoring two points higher in Guinea and four local teacher will bring a greater understanding of the points higher in Cape Verde. Indeed, this trend holds milieu to her classroom interactions. Secondly, a local true for all SES groups except SES 3 in Guinea, teacher is likely to have more regular attendance and although it should be noted that moderate unit costs spend more time at the school and/or with children's in Cape Verde are associated with higher scores for families. In fact, the more time a teacher spends on SES 1 and SES 3. In contrast, the weight:height ratio school activities, the greater the positive impact on appears to be positively related to higher unit costs in student scores. Guinea, although not in Cape Verde. These relations are confirmed by multivariate analysis in Guinea, Whether a school employs an academic, with unit cost showing a weak negative (-0.000135 ) developmental or play-based curriculum albeit non-significant influence on cognitive develop- has no impact on cognitive development. ment and a weak positive (0.00033) but significant relation to weight:height ratio. Preschools were classified into three types, accord- In Cape Verde, the percentage of parents who pay ing to time usage: academic, developmental or play- school fees correlates with a higher cognitive devel- based models. No clear-cut or significant advantage opment score and a lower weight:height score. Table 7.5 Child development and unit costs School Characteristics Guinea (n=529) Cape Verde (n490) Sample SES I SES 2 SES 3 Sample SES I SES 2 SES 3 Cog W:H Cog W:H Cog W:H Cog W:H Cog W:H Cog W:H Cog W:H Cog W:H % PARENT PAYING FEES NA -4ow 95 15.4 94 15.4 95 14.9 99 16.5 -high 104 15.0 100 14.6 105 15.0 107 15.6 UNIT COSTS -4ow 101 13.6 96 13.7 104 13.9 101 12.9 100 15.7 97 15.6 103 15.6 102 16.0 -moderate NA NA - - - - - - 101 14.8 99 14.7 99 14.5 110 15.7 -high 99 16.0 96 16.8 97 16.9 103 14.6 96 14.8 91 14.8 100 14.8 100* 14.7* 'indicates fewer than 10 in the sample 36 AFRICA REGION HUMAN DEVELOPMENT WORKING PAPER SERIES Although the reasons for this are unclear, it can be from the scatterplot, the most expensive schools pro- speculated that regular and consistent fee payment is duce moderate to poor results. a proxy for other parental support of the school and In Cape Verde, no one particular type of school commitment to children's education. Since all parents stands out for its efficiency, although the scatterplot must pay for preschool in Guinea, a similar analysis shows that many schools are efficient, producing was not feasible. high cognitive development scores at low unit cost. Interestingly, the one aspect that the highest scoring In Guinea, the most effective preschools are also schools have in common-whether they are public, the least expensive and provide affordable options communal, private, NGO or religious-is that over to the rural poor, while in Cape Verde the most 90% of parents pay fees to the school. cost-effective schools include all types. Some policy implications While not all low-cost preschools produce high cognitive development scores, the following two scat- Simply providing preschools is not enough. ter plots show that (i) a notable percentage of low- Effective programs must be adopted. cost schools are effective and (ii) many high-cost schools are not. In Guinea, the most effective Not all preschool programs are equal. Differentia- preschools-the CECs-are distinguished by the tion among individual preschools explains between lowest costs per student, ranking them as the most 42% and 55% of variation in cognitive scores. How- cost-effective or efficient option.7 Although other ever, there appears to be no clear pattern in terms of schools-particularly the &oles maternelles-enjoy the factors and/or models that make a clear differ- similar success in terms of cognitive development ence in cognitive development. In fact, frequently the scores, the CEC preschools have lower unit costs pri- school characteristics will have opposite effects in marily due to the high pupil:teacher ratio.8 The CEC Guinea and Cape Verde, indicating that preschool preschools primarily serve a rural population of effectiveness is highly contextualized. One relatively modest means (SES 2). They demonstrate that lower unambiguous finding is the positive impact of com- cost preschool options exist that poorer households munity education centers in Guinea, whose students can afford and are willing to pay for, and that low cost consistently score higher than the ecoles maternelles need not sacrifice educational quality. As can be seen and jardins d'enfants models there. Scatterplot for most efficient preschools Cape Verde Guinea 45 -- - - -- - - - - - -- - --- -------- ------- ------- ------ 40 EI4 0 - -- -- -33 i - 40' , .. .. ~~~~ E1&~~~~~~~~~ 3 F30 30 -E -- X,i -- - -- -- -- ----------- #;"--;*f 35 .E048 ----- 3---- -----0 ----- - 4l 25 @ @ E25 S~~~~~~~~~~~~~~~~~~3 _4 --- -- I ~~~~l~~-r---------- -- -------tt t; 0k:0 f;0t 0; 00;t 30 *El2l30 -0 ----- 25 E26E7 -- -- -- - -- -- - T E2--------s 040S * lE I- *---------I-------- 255 120 ----2- -0------------------- - ------------------ - 20 _ __ ------- -----_-- E1 0 ; f r 0 l 0;l Jardin ' ' ' 27 '' t tt 400000 0, 040titi ;t''0 t; 9 t 00t; f:;' 0 t M aternelle 20, . I E28~~~~~~E7 E26 E24 A~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~C .C 0 S000 10000 15000 20000 25000 30000 Monthly Unit Cost Monthly Unit Cost ( GF.) (ECV) EARLY CHILDHOOD DEVELOPMENT IN AFRICA: CAN WE DO MORE FOR LESS? 37 The teacher's cultural background and proximity and SES 1) households. In Cape Verde, it was found to the school are more likely to have an impact that many schools are efficient, producing high cog- than formal qualifications. nitive development scores at low unit cost. Teacher qualifications appeared to have little effect There is no single effective model of in either country, suggesting that the official training preschooling or "magic bullet" characteristic teachers have received is not effective or sufficient to that will lead to child development. make a discemable difference. However, having teach- ers who lived in the same village had significant No single variable stood out as important to cog- impact, presumably due to their familiarity with the nitive development, and the effectiveness of certain local environment and their having more time avail- types of preschools remains open to question. How- able to spend with the children. This has important ever, a combination of variables or threshold levels of cost-saving implications, as has been the case in Latin certain variables may prove to have greater explana- America and elsewhere, where local community lead- tory value. The attributes of the most effective reli- ers, if effectively trained on ECCD related matters, can gious schools in Cape Verde and the community provide low-cost, effective services. education centers in Guinea need to be examined closely to determine why they seem to produce better Language of instruction in preschool must be results than other models. This is particularly true of considered carefully and policy will differ the low-cost CECs in Guinea, as many of their indi- according to country. vidual characteristics appear to contradict some of the quantitative findings of this study. Further study/ Not surprisingly, the language of instruction was analysis must be conducted of what constitutes key revealed to be one of the most important factors in features of the effective programs. Specifically, direct cognitive development. Single language instruction observation is needed of teachers and children and produced higher scores in Guinea, but bilingual their interactions, as well as qualitative analysis of instruction was more effective in Cape Verde, sug- curriculum and teacher/care-giver training. gesting that context must inform policy. As has been found in previous research in African multilingual Preschools do not need full government subsidies, environments, the question of choosing the language but will probably need some support if equity for initial primary education does not have a single concerns are considered. best answer. The cognitive development of the child in his/her first language, the attitudes and support of Even poor communities are willing to support the parents for the language chosen in the school, and preschools. Parents are willing to pay fees to enroll the perception of the wider community of the status their children, as evidenced in both Guinea and Cape of both the local and second language are to be taken Verde.9 However, this finding must be approached into account in choosing the language of instruction. with some caution. In Guinea, only a small percentage of children are enrolled in preschool, and families who Low-cost preschools serving poor communities have opted to do so may not be representative of the can be effective. vast majority of households. Further, thresh-hold levels must also be identified as well as the trade-off Cost, as represented by fees and teacher salaries, with primary school enrollment. does not appear to be a factor in providing effective Finally, and most importantly, the equity implica- preschooling. Both high- and low-cost schools pro- tions must be considered. Although this study found duced admirable (and lamentable) results. In Guinea, evidence in Guinea that preschools have a strong pos- the seemingly lowest cost community education itive effect on the cognitive development of poor chil- center (CEC) model produced highest cognitive dren growing up in deprived environments, their scores. Moreover, over 70% (on average 90%) of the parents are the very people who cannot afford the children in these schools came from poorer (SES 2 fees. Consequently, these children are less likely to be 38 AFRICA REGION HUMAN DEVELOPMENT WORKING PAPER SERIES enrolled in preschool. Public policy should explore how best to encourage the creation and support the 6. Female teachers in Guinea appear to be more effective. operation of the low-cost, effective preschools that 7. The reader is cautioned that only cursory cost data was col- lected, and it is likely that the inevitable hidden costs omitted have shown success in attracting children from from the analysis will affect future cost-effectiveness analysis. poorer families, along the lines of the CECs. 8. No relation was found between teacher qualifications or sen- iority and salary level, and no one type of school was distin- guished by low teacher wages. 9. Paying fees even exerts a positive influence on cognitive scores in Cape Verde. 8 Conclusions and Recommendations Preschool education can be a double-edged be further disadvantaging its poor children, by allow- sword that can both alleviate and exacerbate ing the gap between the rich and poor to grow through disparity the regressive effect of subsidizing preschool services that serve wealthy families, as in Cape Verde. Public There is little question that preschool, one facet of policy should beformulated to emphasize the support of pre- ECCD, can have a positive impact on the cognitive school or ECCD programs for poor and disadvantaged chil- development of young children. This study has pro- dren. There is little evidence to suggest that constraints vided evidence that children from all socio-economic other than economic ones would prevent lower SES backgrounds can benefit from preschool. Moreover, it families from enrolling their children in ECCD pro- demonstrates that providing preschool services can grams. In addition to reducing disparities in school have an equalizing effect on cognitive development, readiness among children from different SES levels, the compensating for disadvantaged backgrounds. In a provision of subsidized ECCD programs for lower very poor country, Guinea, children from the lowest income families could free mothers to find employ- SES groups, as well as the traditionally disadvantaged ment, thus improving household welfare. girl-child, were found to benefit disproportionately from preschool attendance. But it has also shown that More and better can be done with less even if preschcol cannot entirely close the cognitive development gap between rich and poor children, a High cost programs may not necessarily produce lack of preschool can certainly widen the gap in a rel- the best results. Lower cost preschool programs exist atively well-off country, such as in Cape Verde, where that poorer households can both afford and are will- nearly half the population attends preschool. ing to pay for. Moreover, this study demonstrates that Although children from poorer households do low-cost preschools need not sacrifice quality, as attend preschool, children from wealthier families are many such preschools are effective, producing high disproportionately represented there. Not only are cognitive development scores at low unit costs. rich children more likely to be school-ready by virtue This study found that traditional approaches- of their privileged background, but they are also government-supported, highly paid teachers, formal- more likely to be enrolled in preschool, in better ized training, etc.-do not in themselves guarantee equipped preschools, and often for multiple years, high cognitive development scores. Teachers do not thus gaining even further advantage in terms of cog- need to be highly qualified; training for teachers nitive development and school readiness. and/or caregivers need not be formal. Short, focused, Therefore, the consequences of non-intervention by reliable, and contextualized training can be more efficient government are twofold. First, society may be losing and less costly than most availableformal programs. Com- the benefits of preschool for its children. Second, it may munity-based programs, such as the CECs in Guinea, 40 AFRICA REGION HUMAN DEVELOPMENT WORKING PAPER SERIES appear promising, due in part to parental involve- interest is critical. Both the services the government ment and financing, and in part to the proximity of provides and the policies it promulgates will depend the locally-resident teacher, who is able to spend on context and should work within a realistic invest- more time at the school. Elsewhere in Africa, similar ment framework. community-based programs have been effectively Much can be done in ternms of improving access, quality, taken to scale. Most of Kenya's 25,000 preschool cen- and equityfor ECCD in both Guinlea andil Cape Vcrde. For ters are managed by parents (Garcia March 2, 2001). example, in Guinea where preschool enrollment is Despite these findings, this study has only begun low and there are scant public resources for provid- to address the more complex questions of the type of ing ECCD programs, the government can establish a programs and the full range of features that are most supportive policy framework that creates an environ- efficient. The wide variation in effectiveness of ment in which private sector and community ECCD preschools at many different price points indicates initiatives can flourish. This could take many differ- that there are many factors that can influence a ent forms, such as supporting IEC and parental edu- preschool's success. This study has confirmed that in the cation campaigns about the importance of early provision of preschool services, context is important. There childhood education and care, piloting low- is no one-size-fits-all answer to effective and affordable cost/community-supported ECCD models based on ECCD. Context will vary not just by country but also by promising indigenous programs, creating ECCD SES, ethnicity, language used, and urban-rural residence. start-up credit funds, or developing non-onerous There should be closer examination of what make guidelines and guidance for community programs. indigenous models of ECCD successful, particularly In Cape Verde where the public sector is already in the case of the CEC programs in Guinea, as well as providing a significant proportion of the population in other francophone African countries where these with preschool services, the government role should types of initiatives are underway. focus on the more efficient use and distribution of Additional attention must be paid to both the costs public-sector resources. Efforts should be made to and financing of preschools and ECCD. A more accu- further study the preschools that have enjoyed the rate idea of costs is needed, as well as what motivates greatest success at the least cost and to replicate them. parents to enroll their children in ECCD/preschool. Focused, ECCD-related training should be provided While the CEC programs in Guinea are promising, to public-sector teachers. The redistribution of they serve only a small percentage of the population. resources to benefit the poor could be effected by levying user charges on wealthier households, limit- Government can play a critical and essential ing public preschool access to poor families, or pro- role in supporting early childhood care and viding them with subsidies (scholarships, uniforms, development programs nutrition or health services) to encourage enrollment. Government can act to improve the quality of Given the demands for primary education and the ECCD services available either through public- or pri- current national education budgets, few will argue vate-sector providers. In both countries, a notable with the contention that many African governments number of children from the poorest SES levels go to lack the resources to invest in the widespread provi- the most expensive preschools. Parents should be sion of ECCD programs. But this does not mean that offered the best for less. In some countries, such as there is no role or room for action and intervention by Cape Verde, it may be appropriate for government to government. regulate and inspect private-sector preschools or While debate continues among development ECCD facilities to ensure the safety and protection of experts about the efficacy of investing in ECCD, com- young children. In other countries, such as Guinea munities and families throughout Africa have made where fledgling private-sector and community ECCD their own decisions by founding, funding and/or initiatives could collapse under heavy-handed gov- enrolling their children in preschool and ECCD pro- ernment regulation, training modules, materials, and grams. As with all services affecting public welfare, guidance could be provided to NGOs and CBOs most particularly education, the role of government in involved either in community support or ECCD pro- formulating policy for ECCD to optimize the public gram management. In either case, government can EARLY CHILDHOOD DEVELOPMENT IN AFRICA: CAN WE DO MORE FOR LESS? 41 help equip local organizations and parents with the these community-based programs and about how knowledge of what to expect from ECCD providers in these programs can be implemented. order to select or monitor their children's preschool. In pursuit of these answers, the government of Finally, government and donors should think out- Guinea, with World Bank assistance, will initiate side the box of formal, traditional preschool pro- ECCD pilot programs in the two poorest regions of grams. This study suggests that this type of the country. These programs will provide a more in- instruction is not necessarily tailored to meet chil- depth and qualitative look at the features of success- dren's needs, and may be more expensive. They ful programs, as well as their costs, and to replicate should encourage and support communities to them within the framework suggested by this study, engage in the creation of ECCD services, and they i.e., community-based and parent-supported centers. should target whatever assistance is feasible to those A monitoring and evaluation component will test and communities that have acted to provide preschool or document the effectiveness of the promising commu- ECCD services. nity-based model. From this, the Bank will work with the government to formulate its policy objectives and What are the next steps in supporting framework for ECCD support. ECCD in Africa? In Cape Verde, several of the research findings- particularly the equitable allocation of preschool Despite the promising indications that low-cost, resources and the effectiveness of government pre- community-supported preschools offer a viable schools-are subjects of the policy dialogue currently option for expanding support of early childhood taking place between the government and the World development, initial studies of this type often tend to Bank. The results of these discussions will be incorpo- raise more questions rather than provide answers. rated into Cape Verde's education sector strategy. Several issues requiring further investigation are The capacity of human beings to learn and to highlighted by this research, ranging from the longer- develop is enormous and continues throughout life. term impact of preschool on primary student per- ECCD is only one mechanism, albeit a promising one, formance and the elusive factors that make a for ensuring that many of the factors that can con- preschool effective, to how best to expand access to tribute to the intellectual, physical and emotional preschools programs without raising public expendi- development of a child are made accessible to fami- ture burdens or pricing poor parents out of the pre- lies in need. ECCD initiatives throughout Africa pro- school market. None of the programs examined in vide fertile ground for study to determine how best to this study were distinguished by a rigorous monitor- mobilize and manage the resources of households, ing and evaluation component. More knowledge communities, governments, and donors to help must be gained, both about how governments and Africa's most vulnerable citizens-its children- funding agencies can support the development of grow to rewarding and productive adulthood. Annex 1 Study Methodology Approach veyed, a random sample of 15* five-year-old children who were enrolled in the preschool and their families his study's purpose was to develop a prelimi- were surveyed. As a basis of comparison, 10 five- T nary picture of the impacts and effectiveness year-old children from the same community and of early childhood programs, in two contrast- sharing similar socio-economic characteristics but not ing African countries-Cape Verde and Guinea. It ini- attending preschool formed the control group. tially attempted to develop an overall profile of the The dependent variables used in the analysis were early childhood development sector. However, since children's cognitive and physical development. The preschools were the predominant model of interven- independent variables-the factors influencing chil- tion in early childhood in the two countries studied, dren's development-were divided into three cate- as well as the most readily accessible, this study nec- gories. The first group comprised individual essarily focused on the preschool sector. Moreover, socio-demographic characteristics of the children because of actual (Cape Verde) and potential (Guinea) such as age, gender, health status, socio-economic public-sector support of preschool as an early child- status, and family characteristics. The second group hood development strategy, understanding the pro- comprised environmental characteristics, such as grams and impacts of the preschool sub-sector is whether parents read to the preschool-age child. The essential to sound policy development. third group comprised variables related to the pre- The study used a survey methodology to identify school/ECCD programs, such as student/teacher different programs and preschool models and to ratio, teachers' qualifications and experience, type of detail their characteristics. A sample of preschools program offered (daycare, preschool, etc.), time dedi- was then selected to identify the impact that pre- cated to children's activities, urban or rural settings, if school attendance, different preschool programs/ programs were public or private, and overall costs. models, and various preschool characteristics had on Cross-tabs and multivariate analysis were used to cognitive and physical development of the children compare the two groups and the influence of vari- who attend them. The range of existing programs in ables in the cognitive and physical development of each of the countries was examined to determine the children. These analyses were done separately for programs and characteristics that appeared to be each country. Results and discussions on the findings most effective and least costly in supporting chil- are presented in Sections 5, 6 and 7 in the body of the dren's development. study. The samples of preschool centers took into account variability in terms of socio-economic levels, geo- graphic distribution, and types of programs. In both * In a few centers in Guinea, there were fewer than 15 five-year- olds born in 1995. To complete the sample, children born in Guinea and Cape Verde, for each of the centers sur- December 1994 were included. 44 AFRICA REGION HUMAN DEVELOPMENT WORKING PAPER SERIES Sample and sampling frame in Guinea Annex Table I and Cape Verde Guinea Cape Verde In Guinea the sample was designed in a two-phase Centers 36 34 approach. Because there were no existing data avail- Children and their families 877 803 able for the sector, a first stage was needed to identify the types and number of preschools available. In this P 529 490 first stage, 460 centers and 1089 teachers were initially Control Group 348 313 surveyed. Based on the information gathered, three SES 1 (% in the sample) 42% 49% types of centers were identified, individual centers were categorized according to type, and a total SES 2 (% in the sample) 36% 36% sample of 36 preschool centers was selected. The total SES 3 (% in the sample) 22% 15% number of children surveyed in the second stage was 900, although complete data are available for 877. The geographic distribution of centers was: wealth and socio-economic status, occasionally SES 1 Conakry 36% and SES 2 will display similar behaviors, or they will Moyenne Guinee 11% not be ranked in a strictly hierarchical order. Haute Guin6e 14% In Cape Verde, the lowest or poorest group (SES 1) Basse Guin6e 28% comprised the jobless or unskilled laborers, the Guinee Forestiere 11% middle group (SES 2) comprised skilled workers or small tradesmen, and the highest or wealthiest group Of the sample, 64% were Ecoles maternelles, 22% (SES 3) comprised professionals or managers. Forty- were Community Education Centers (CEC) and 14% nine percent of the sample was from SES 1, 36% from were Jardins d'enfants. (See Section 4 in the body of the SES 2, and 15% from SES 3. report for a description of the sector, including the In Guinea, additional information on parental lit- differences of the type of programs in Guinea and eracy, education, language usage, and household Cape Verde.) goods were also considered, albeit with similar In Cape Verde, the sample consisted of 34 centers, results. The lowest group (SES 1) primarily com- and complete data are available for 803 children and prised unskilled or jobless laborers, the middle group their families. The geographic distribution was: (SES 2) mainly comprised tradesmen or farmers, and Island of Santiago 20 centers the highest group (SES 3) included professionals and Island of Maio 4 centers mid-to-high level managers. Forty-two percent of the Island of Fogo 4 centers sample was from SES 1, 36% from SES 2, and 22% Island of Sal 3 centers from SES 3. Not surprisingly, SES 3 households were Island of Sao Nicolau 3 centers distinguished by high levels of literacy, education, Sixty percent of the centers were located in either and use of French langtage rather than local lan- urban or semi-urban areas, and 40% were from rural guages. More significantly, rural households almost areas. The distribution according to the type of cen- exclusively fell into SES 2. ters was: Because it was not feasible to administer a pretest Public 50% in order to establish a baseline to assess the magni- Private 18% tude of the relative impact or "value-added" of pre- NGO 11% school on child development, attention was paid to Community 12% ensuring that the control group shared the same Religious 9% background characteristics as the experimental pre- school group in order to eliminate a selection bias in The sample households in both Guinea and Cape the sample. Using the Heckman procedure, a non-sig- Verde were divided in to three socio-economic group- nificant inverse Mills ratio was obtained, indicating ings, according to the profession of the father or male that no bias marred the sample and compensating for head. Because profession is an inexact proxy for the lack of cross-sectional data. Annex Table 2 pres- EARLY CHILDHOOD DEVELOPMENT IN AFRICA: CAN WE DO MORE FOR LESS? 45 Annex Table 2 Comparison of preschool and control group characteristics Guinea Cape Verde Unit of Preschool Control Preschool Control Variable Analysis Group Group Group Group Number of children in sample 529 358 490 313 Girls % 55.4 52.1 54.9 51.1 Family with >5 children % 40.8 47.3 30.2 47.6 Mother works (outside home) % 36.7 31.7 43.7 31.3 Mother is literate % 49.9 31.1 81.2 70.0 Mother heads household % 13.6 13.6 28.8 35.8 Child performs household tasks % 29.7 36.1 24.1 27.2 Parents read to child % 56.5 50.0 56.8 28.3 Child is seldom ill % 9.5 8.0 36.7 35.1 Child is sometimes ill % 48.8 45.3 31.6 31.9 Child is usually ill % 41.8 46.2 31.6 32.9 Low household SES (SES 1) % 35.0 53.0 47.6 51.8 Middle household SES (SES 2) % 35.5 35.8 32.4 42.2 High household SES (SES 3) % 29.5 11.2 20.0 6.1 ents the preschool and control group characteristics dren's families; and (2) cognitive development tests for each country. to evaluate the children. Measuring Cognitive Development: Cognitive devel- Instruments opment tests were administered only to five-year-old children, both to control for natural development Two type of instruments were used: (1) question- associated with age and to avoid the difficulties of naires addressed to the teachers and personnel in the testing younger children. Instrument development centers and to the preschool and control group chil- had to respond to several challenges. First, the instru- Annex Table 3 Summary of instruments Instrument Target group Content Application Preschool Director of center, Descrption of the preschool program, type of activities developed, Direct administration to questionnaire teachers administration of the center, characteristics of personnel directors and teachers individually. Family Parents of the child Family characteristics, health status of the child Direct administration to Questionnaire the parents of each child, individually Cognitive and Physical Five-year-old children Cognitive development (language, basic concepts, Direct administration Development Test visual discrmination) individually to each child Physical development, (heighVweight) 46 AFRICA REGION HUMAN DEVELOPMENT WORKING PAPER SERIES ments had to measure the real skills of young chil- interaction with children began in Portuguese. If chil- dren who had limited or no experience with pre- dren failed to understand Portuguese, tests were school. Consequently, they were based on tests used administered in Creole. by early childhood psychologists to measure overall Measuring Physical Development: Identifying appro- cognitive development. Second, to ensure uniform priate and practical measures of child physical devel- application, they had to be easily administered by the opment and well-being were particularly study enumerators. Third, they had to take into challenging. Young children's physical development account both contextual and cultural differences, can be measured in several ways. In addition to departing from "western" assumptions that young recording the degree of vaccination and the number children will be familiar with pencils and writing of incidences of illness experienced by the child, an implements and ensuring that test contents, whether attempt was made to measure the degree of physical pictures, language, or descriptions, corresponded to development by comparing weight to height. This the child's environment. measure is most often used by researchers to assess A simplified version of the Boehm Basic Concept the level of stunting or wasting caused by malnutri- Test was used. The test was pre-tested in a small tion during early childhood. The resulting ratio can sample in each country (15 children) to ensure that then be compared with that for children from indus- the items selected (24 in all) were culturally relevant trialized countries. Since stunting occurs in children and permitted test administration in local languages under age three and cannot be remedied by better without ambiguities. Three domains of knowledge nutrition at a later age, the weight:height ratio may were tested: basic concepts, oral language, and pre- have limited validity as a measure of the develop- reading. Basic concepts were tested using drawings ment of five-year-olds. However, as other options to elicit the child's responses about relationships were too complicated or cumbersome for the field (space, time and quantity) and his ability to use math- enumerators to administer, the weight:height ratio ematical reasoning. Language skills were evaluated was used. (Enumerators were equipped with according to the child's ability to describe the situa- portable scales and tapes.) tions represented in simple pictures. Visual discrimi- Calculating Costs: Although the study did not col- nation was assessed by the child's identification of lect full cost information about the school, sufficient symbolic shapes that closely mirror written language. data was collected to estimate unit costs of the sample Figure 2 presents an example of the type of items preschools that provide for comparative analysis, used to evaluate basic concepts. All tests were admin- although not accurate cost projection. In Cape Verde, istered in local languages in Guinea. In Cape Verde monthly unit costs were derived from reported teacher monthly salary divided by the number of stu- Figure 2 dents per teacher. In Guinea, monthly unit costs were The child is asked to point to the smallest fish. based on four indicators: teacher wage, number of teachers per school, school fees paid by students, and the number of students per school. They were not adjusted for time spent teaching and do not include financial or in-kind resources provided by other insti- tutions, such as NGOs (although preschool operating budgets primarily consist of teacher salaries). They also do not take into consideration other personnel that may be associated with the school. References Arnold, C. 1998. "Early Childhood, Building Our Under- Evans, J. with R.G. Myers and E. Ilfeld 2000. Early Child- standing and Moving Toward the Best of Both Worlds." hood Counts: A Programming Guide on Early Child- Paper presented at the International Seminar, Ensuring a hood Care for Development. Washington, D.C.: World Strong Foundation: An Integrated Approach to early Bank. Childhood Care and Development, 23-27 March, Insti- Garcia, Marito H. March 2001. E-mail comments on Cape- tute for Educational Development, Aga Khan Univer- Verde and Guinea Studies. sity, Karachi, Pakistan. Haveman, R. and B. Wolfe. 1995. "The Determinants of Boehm, A. 1973. Test des concepts de base. Les editions du Children's Attainments: A Review of Methods and Find- Centre de Psychologie Appliqee, Paris. ings," Journal of Economic Literature, 33(4). Bundy, Donald A.P., 1997. "Health and Early Childhood Heaver, R., and J. Hunt. 1995. "Improving Early Childhood Development," in Young, Mary E., Early Child Develop- Development: An Integrated Program for the Philip- ment: Investing in Our Children's Future, Amsterdam: pines." A collaborative report by the World Bank and the Elsevier. Asian Development Bank for the Government of the Cummins, J. 1979. Linguistic Interdependence and the edu- Republic of the Philippines. Washington, D.C. cational development of bilingual children. Review of Lokshin, M.M., E. Glinskaya and M. Garcia. June 2000. "The Educational Research, 49, 222-251. Effect of Early Childhood Development Programs on Cummins, J. 1979. Cognitive-academic language profi- Women's Labor Force Participation and Older Chil- ciency linguistic interdependence, the optimal age ques- dren's Schooling in Kenya:" Policy Research Working tion, and some other matters. Working Papers in Paper 2376. Washington, D.C.: World Bank Develop- Bilingualism, No. 19, Toronto: Ontario Institute for Stud- ment Research Group. ies in Education. Meyers, C. "The Case for Pre-Primary Education: The Cost Currie, J. and D. Thomas 1995. "Does Head Start Make a Effectiveness of Shishu Kaksha Centers, Nepal" in Difference?" American Economic review, 85(3) Evans, J with R.G. Myers and E. Ilfeld 2000. Early Child- Delhaxhe, A. Godenir, A. 1990. L'acquisition du nombre a hood Counts: A Programming Guide on Early Child- l'ecole maternelle, Service de Psychologie experimen- hood Care for Development. Washington, D.C.: World tale, Universite de Liege. Bank. Dutcher, N. The Use of First and Second Languages in Pri- Mingat, A. and B. Suchaut. November 1998. "Une Analyse mary Education: Selected Case Studies. World Bank Staff Economique Comparative des Systemes Educativifs Working Paper No. 504, January 1982. Africains," A report prepared for the French Ministry of Edwards, J.H.Y and X. Liang, May 1998. "Mexico's Cooperation and Development. Preschools: Coverage, Equity and Impact," Human Myers, R.G. March 2000 "Thoughts on the Role of the "Pri- Development Departmnent LCSHD Paper Series No. 16, vate Sector" in Early Childhood Development, Paper Washington, D.C.: World Bank Latin America and the prepared for presentation at The Year 2000 Conference Caribbean Regional Office. on Early Childhood Development "Investing in Our Eining Young, M. 1996. Early Child Development: Invest- Children's Future-From Science to Public Policy," ing in the Future. Directions in Development. Washing- Washington, D.C. ton, D.C.: World Bank. 48 AFRICA REGION HUMAN DEVELOPMENT WORKING PAPER SERIES Myers, R.G. February 2000 "Summary: The EFA Global Tucker, G.R. " The Linguistic Perspective" in Parker, L. (ed.) Thematic Review of Early Childhood care and Develop- Education: Current Perspectives, vol. 2. Linguistics. ment, ' Prepared for the Consultative Group on Early Arlington, Va. Center for Applied Linguistics, 1977. Childhood Care and Development. Van Der Gaag, J., and J.P Tan, 1998. The Benefits of Early Myers, R.G. 1995. The Twelve Who Survive: Strengthening Childhood Development Programs: An Economic Programs of Early Childhood Development in the Third Analysis. Washington, D.C.: The World Bank. World. Ypsilanti, Michigan: High/Scope Press. Washington Post, September 17, 2000. "A Look at ... Africa Myers, R.G. 1988. "Effects of Early Childhood Intervention and the AIDS Loan." on Primary School Progress and Performance in Devel- World Bank. 2000a. Can Africa Claim the 21st Century? oping Countries: An Update," Paper presented at a sem- Washington, D.C. inar on The Importance of Nutrition and Early World Bank. August 2000b. "Norwegian Education Trust Stimulation for the Education of Children in the Third Fund Report Annex XX: Early Childhood Development, World, Stockholm. Regional Activities and Plans," prepared by the Africa Pollitt, Ernesto, K.S. Gorman, PL Engle, R. Martorell, and J. Region's Human Development Department. Rivera. 1993. "Early Supplemental Feeding and Cogni- World Bank. 2000c. "Une Chance Pour Apprendre: Le tion." Society for research in Child Development. 58 (7). Savoir et Le Financement pour l'Education en Afrique Scheinhart, L., H. Barnes and D.P Weikart. 1993 Significant Subsaharienne," Sector Assistance Strategy prepared by Benefits: The High/Scope Perry Preschool Study Africa Region's Human Development Department for Through Age 27. Ypsilanti, Michigan: High/Scope Press. the World Conference on Education for All, Dakar, 26-28 Sutton, P., K. Tietjen, A. Bah and P. Kamano (1999) Promot- April 2000. ing Primary Education for Girls in Guinea: A Case World Bank. 1999. Education Sector Strategy. Washington, Study, U.S. Agency for International Development, D.C.: The Human Development Network. Washington, D.C. THE WORLD BANK 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. Telephone: 202 477 1234 Facsimile: 202 477 6391 Internet: www.worldbank.org E-mail: afrhdseries@worldbank.org Many of the factors constraining human development in Africa are rooted in the early childhood of its impoverished and disadvantaged populations. Early childhood care and development (ECCD) programs, which respond to young children's basic needs for cognitive stimulation, early learning, nutrition and basic health care, offer an opportunity to exercise a positive and lasting influ- ence on the health, intelligence and future development of the individual child. Yet, for a variety of reasons today most children in Africa-almost 98 percent- grow up without the benefit of such programs. However, this is now changing and interest in ECCD is growing in Africa. The World Bank is supporting ECCD and related school health and nutrition programs in more than 14 countries. But, despite the recognized benefits of ECCD, many questions about quality, affordability and sustainability remain. Early Childhood Development in Africa: Can we do more for less? examines a variety of alternatives in terms of content, implementation and financing in two countries, Cape Verde and Guinea. The study suggests that there is demand for ECCD programs at all socio-economic levels, that the provision of such programs can reduce disparities in school readiness among children from dif- ferent socio-economic backgrounds and that high cost programs do not nec- essarily produce the best results. The objectives of the study are to contribute to the knowledge base about the status of ECCD in West Africa, to identify the building blocks of af- fordable and cost-effective ECCD programs within an African context, and to better understand the role that governments and donors can play to support such programs. Finally, the study aims at providing guidance to education pol- icymakers and staff in the World Bank and other partner agencies seeking ways to best serve the developmental needs of Africa's youngest citizens.