Annotated Bibliography Project

late_entry_into_primary_school.asp_.pdf

Home >Social Science homework help >Annotated Bibliography Project

LATE ENTRY INTO PRIMARY SCHOOL IN DEVELOPING SOCIETIES:

FINDINGS FROM CROSS-NATIONAL HOUSEHOLD SURVEYS

YUKO NONOYAMA-TARUMI, EDILBERTO LOAIZA and PATRICE L. ENGLE

Abstract – Late entry into primary school is a widespread phenomenon in developing countries. Students who enter school late are more likely to repeat grades, drop out and perform more poorly. Yet the phenomenon has received little scholarly attention, and there is a dearth of cross-national data. In this paper, we draw on data from the Multiple Indicator Cluster Survey (MICS2), a cross-national household survey conducted in developing countries. We first estimate the percentage of students entering primary school late across 38 countries in order to identify the countries in which the issue of late entry is most common. Secondly, we describe the background characteristics of students who are more likely to enter school late. We then employ multinominal logistic regres- sion equations to predict the probability of late entry. Our findings highlight the need for policies to reduce late entry for children from disadvantaged backgrounds.

Résumé – LA SCOLARISATION TARDIVE ÉTUDIÉE DANS 38 PAYS – L’entrée tardive à l’école primaire est un phénomène très répandu dans les pays en développe- ment. Les élèves scolarisés tardivement risquent davantage le redoublement, la désco- larisation ou des résultats plus faibles. Le phénomène n’a pourtant suscité qu’un faible intérêt au sein du monde scientifique, et l’on peut parler d’une pénurie de données transnationales à ce sujet. Dans cet article, nous nous appuyons sur l’Enquête à in- dicateurs multiples (MICS2), une enquête ménages transnationale réalisée dans des pays en développement. Nous estimons tout d’abord le pourcentage d’élèves scolarisés tar- divement dans 38 pays, afin d’identifier les pays pour lesquels la scolarisation tardive est un problème plus fréquent. Dans un second temps, nous décrivons les origines ca- ractéristiques des élèves davantage susceptibles d’entrer plus tard à l’école. Enfin, nous utilisons les équations de régression logistique multinominale pour évaluer la probabilité d’une scolarisation tardive. Nos résultats éclairent la nécessité de formuler des politiques aptes à réduire la scolarisation tardive chez les enfants issus de milieux défavorisés.

Zusammenfassung – SPÄTE EINSCHULUNG IN 38 LÄNDERN – Der verspätete Eintritt in die Grundschule ist ein weit verbreitetes Phänomen in Entwicklungsländern. Schülerinnen und Schüler, die verspätet eingeschult werden, müssen häufiger Klassen wiederholen, brechen mit höherer Wahrscheinlichkeit ihre Schullaufbahn ab und zeigen schlechtere Leistungen. Dennoch wurde dieses Phänomen von der Wissenschaft wenig beachtet und es mangelt an länderübergreifenden Daten. In diesem Beitrag beziehen wir uns auf Daten aus der Multiple Indicator Cluster Survey (MICS2), einer länderüber- greifenden Haushaltsumfrage, die in Entwicklungsländern durchgeführt wurde. Als Erstes schätzen wir den Prozentsatz der verspätet eingeschulten Schülerinnen und Schüler in 38 Ländern, um herauszufinden, in welchen Ländern das Problem der verspäteten Einschulung am weitesten verbreitet ist. Als Nächstes beschreiben wir den Hintergrund der Schülerinnen und Schüler, die mit größerer Wahrscheinlichkeit vers- pätet eingeschult werden. Mittels multinominaler logistischer Regressionsgleichungen

International Review of Education (2010) 56:103–125 ! Springer 2010 DOI 10.1007/s11159-010-9151-2

errechnen wir sodann die Wahrscheinlichkeit der verspäteten Einschulung. Unsere Ergebnisse machen den Bedarf an politischen Maßnahmen zur früheren Einschulung benachteiligter Kinder deutlich.

Resumen – ESCOLARIZACIÓN TARDÍA EN 38 PAÍSES – El ingreso tardı́o en la escuela primaria es un fenómeno muy difundido entre los paı́ses en vı́as de desarrollo. Los estudiantes que ingresan tardı́amente en la escuela presentan una mayor proba- bilidad de repetir el grado, de abandonar los estudios y de presentar un desarrollo más deficiente. Hasta ahora, este fenómeno ha recibido poca atención por parte del mundo académico, y hay una falta de datos que abarquen diferentes paı́ses. En este trabajo, nos basamos en datos obtenidos por la metodologı́a de la Encuesta de Indicadores Múltiples por Conglomerados (MICS2), una encuesta realizada en los hogares de diferentes paı́ses en vı́as de desarrollo. En primer lugar, estimamos el porcentaje de estudiantes que ingresan tardı́amente en la escuela primaria en 38 paı́ses, para identificar de esta manera aquellos paı́ses donde el problema de la escolarización tardı́a es más común. En segundo lugar, describimos los orı́genes caracterı́sticos de los estudiantes que presentan una mayor probabilidad de ingresar tardı́amente en la escuela. Luego, empleamos ecuaci- ones de regresión logı́stica multinominal para predecir la probabilidad de escolarización tardı́a. Nuestros hallazgos ponen de relieve la necesidad de que se desarrollen e im- plementen polı́ticas que reduzcan la escolarización tardı́a de niños provenientes de entornos desaventajados.

Introduction

Late entry into primary school is a widespread phenomenon in many developing countries, and it is a concern because it may have detrimental consequences at various levels. At the student level, research has shown

– 38 – - -

. , - , ,

. , ,

. - ( -2) – ,

. , -

38 . – , -

. , , , .

- . -

, - , .

104 Yoko Nonoyama-Tarumi et al.

that students who enter school late are more likely to repeat a grade, have lower performance level, or drop out of school (Wils 2004). At the class- room level, a large number of older students and a wide age range in a classroom, in addition to the common situation of an over-crowded class- room, present a pedagogical challenge to the teachers (Fentiman et al. 1999). At the system level, late entry leads to ine!ciency, as it undermines the e"orts to improve school enrolment and school retention at the same time. Although these negative consequences of late entry are acknowledged, the prevalence of late entry and possible causes have received insu!cient scholarly attention, and there is a dearth of cross-national data (Lloyd and Blanc 1996).

This paper uses a cross-national household survey to determine the age children are entering the first grade of school. The percentage of early entrants, on-time entrants and late entrants in grade 1 are used to identify the countries in which the rates of late entry are highest. Second, the paper looks at factors associated with late entry to better understand why families in developing countries often end up delaying their children’s enrolment. By doing so, the paper suggests strategies to increase on-time entry and the population that needs to be targeted in developing policies and programmes to reduce late entry into primary school.

Prior research

Consequences of late entry

Late entry into primary school can have negative consequences at multiple levels. At the student level, late entry is associated with higher grade repe- tition and higher drop-out rate. Wils (2004) hypothesised that children in Mozambique who enter late have higher drop-out rates than those who enter early, based on the contradictory results from census data which indicated very low school drop out in the pre-teen years, and the enrol- ment data from the Ministry of Education which showed high drop-out rates in grades that are largely populated by pre-teen pupils. Through reconstructing the cohort-specific school experience of children in a 15 year period (1982–1997), she estimates that Mozambican children who enter school at age five to seven are expected to finish almost eight grades, while children who enter school at age 11–14 are expected to finish only two grades.

Wils (2004) also examines data from 26 countries, to see whether this association holds in other developing countries, and finds a negative cor- relation of r = -0.24 (p = 0.002) between late entry and survival ratios to grade 4, using the UNESCO database (1998/1999 and 1999/2000). She speculates that there are three processes explaining this relationship. First,

105Late Entry Into Primary School In Developing Societies

parents who send their children to school as early as permitted might be more motivated about education in general and may be able to a"ord the time and financial resources to help their children progress through schooling for a longer period than those who send their children to school later. Second, older children may be more likely than younger chil- dren to have conflicting responsibilities within the household or at work, leading to drop-out. Third, the long distance to school, which is also a cause of late entry, might cause early drop-out (Wils 2004). Data from African countries suggest that the relationship may be stronger for girls; girls who are overage for their grade may be more likely than boys who are overage to drop out before completing primary school (Uganda Bureau of Statistics and ORC Macro 2002; Zambia Central Statistical O!ce and ORC Macro 2003). Colclough and Lewin (1993) further points out that delay in school entry have a negative e"ect on the child’s achievement, and highlights the importance for children to start learning early.

In addition, it is plausible that late school entry may have more detrimen- tal consequences when the language of instruction in primary school is not in the child’s mother tongue. Children who do not begin school until 10–14 years will most likely learn a second language more slowly if at all compared to 6–7 year olds. This issue is critical for countries in which schooling is not in the mother tongue, as in many Francophone and Luso- phone African countries.

At the classroom level, a large number of over-age students and a wide age range in a classroom present a challenge for teachers, as they need to teach a diverse group with di"ering levels of maturity. First, the di"erences in cognitive and physical development at this age are large. Incorporating 6 year olds with teenage children can create problems in developing age- appropriate learning methods. For example, the pedagogy needed to keep the attention or the discipline of 6 year olds will be di"erent from those nee- ded for 12 year olds. Secondly, part of the socialisation skills learned is in response to age. For example, younger children may be expected to serve the older children, which can create an imbalance in the learning situation with- in the classroom. Although sex stereotyped tasks, such as sweeping and fetching water, which are perceived as female tasks, have been well docu- mented in the classroom, age-related tasks in the classroom have not been well explored and documented (Fentiman et al. 1999). It should be noted, however, that there has been much e"ort by schools and teachers in develop- ing countries to deal with students of heterogeneous age in one classroom through innovative pedagogies, such as multi-grade schools (Benveniste and McEwan 2000).

At the system level, late entry is associated with ine!ciency. Although late entry is not the only cause of repetition and drop-out, it undermines e"orts to increase school enrolment and retention.

106 Yoko Nonoyama-Tarumi et al.

Factors associated with late entry

The next question of interest is why families in poor countries often delay their children’s enrolment to older age, when the value of the child’s time is higher.1 Past studies have identified several factors that are associated with late entry: distance to school, school quality, place of residence, household wealth, nutritional and health status of the child, and parent’s perception of the child’s school readiness.

Families may often delay their children’s entry to primary school because they need to save money for schooling. However, completing school at an early stage in life may increase the individual’s total future income. Bommier and Lambert (2000) use a human capital investment model to estimate the age of enrolment and years of schooling in Tanzania. In other words, they take into account a trade-o" between late entry and early completion. Using the Tanzania Human Resource Development Survey, they find that longer distance to school, poor school quality, defined by the quality of teachers and supplies at the school, living in poor households and rural areas are associated with later enrolment and shorter years of schooling. Some chil- dren may not be mature enough to walk the distance to school at their legal enrolment age and in addition, the opportunity cost would be higher. If the quality of school is low, children would learn less and hence the costs of schooling would be higher, and thus, parents may find fewer incentives to start sending their children to school.

Glewwe and Jacoby (1993), using the 1988–1989 Ghana Living Standards Survey (GLSS), find strong evidence that delayed primary enrolment is the consequence of nutritional deficiencies in early childhood. They argue that malnutrition lowers young children’s learning abilities and reduces the rate of return to schooling. Thus, parents may choose to keep their children out of school. However, as children grow and the severity of malnutrition decreases, parents may come to feel that investment in schooling is worth its cost. The authors find that child height-for-age is negatively associated with the duration of delays in entering school, and furthermore, that when controlling for height-for-age, neither family income nor school fees account for late entry into primary school. Thus, they conclude that it is not that parents cannot a"ord to borrow money to send their children to school on-time, but rather, parents deliberately and rationally make the decision not to send their children on-time.

Fentiman et al. (1999), through their own data collection in Ghana, find that in addition to child’s poor health status and distance to school, parents’ perception of school readiness is an important determinant of late enrol- ment. Through focus group discussions and interviews, they find that some parents were not sending their children to school, because they thought the child was too young even when the child was of school age. The authors suggest that parents’ perception of child’s school readiness is shaped not only by child’s poor physical development, due to lack of nutrition, but also by lack of certain social and cognitive skills. With an increase in age-related

107Late Entry Into Primary School In Developing Societies

tasks assigned by parents in rural Ghana, when the child is not capable of performing these activities, parents may not believe the child is ready to be sent to school (Fentiman et al. 1999).

The EdData, which directly asked parents/guardians about the reasons why children started school at an older-than-average age, found that the top three reasons in Zambia are: financial need, distance to school and child labour. When asked for other reasons why children started school overage, the most cited reason was that child was not ready to start attending school, which is consistent with Fentiman et al.’s finding (Zambia Central Statistical O!ce and ORC Macro 2003).

The above mentioned studies have contributed to identifying factors that are determinants of late school entry, and framing the issue of age of entry as family’s decision. Building upon this, we use the framework of families’ readiness for school in our paper. However, it is important to note that chil- dren’s readiness for school and schools’ readiness for children are two other pillars of school readiness and should not be neglected in considering the issue of on-time entry.2

We contribute to existing literature by noting three limitations in prior research. First, all studies we examined are single-country studies.3 We are not able to compare the rate of late entry across countries, because the studies use di"erent sources of data and di"erent definitions of late entry. Secondly, the studies are all from Africa, begging the question whether the phenomenon is widespread and whether its causes are similar in other regions. Lastly, although the studies acknowledge that children often start schooling later than the o!cial age of entry because they need to work in family activity to save money for the schooling before enrolment, none of the studies examine the relationship between child labour and late entry empirically. Fourthly, most of the studies confound the issue of repetition and late entry in estimating the rate of late entrants (Glewwe and Jacoby 1993; Fentiman et al. 1999; Wils 2004).

Issue of early entry

Less is known about the consequences of early entry into primary school, and thus, we do not investigate factors associated with it in this paper. How- ever, Fentiman et al. 1999, through their focus groups in Ghana, find that younger siblings followed their older enrolled siblings to school, some teach- ers wanted to give their children a head start in education, and in some cases, the classroom was used as a ‘‘creche’’, where older siblings could com- bine schooling with childcare. Bommier and Lambert (2000) find that girls enter school earlier than boys, despite the fact that a greater proportion of girls never attend school. They suggest that this may be due to higher returns for family economic activities and hence higher opportunity cost for boys, or to the need for preparing girls for marriage at a younger age. One may hypothesise that in systems where school places are limited, the presence of early entrants may push older children, who are likely to assume

108 Yoko Nonoyama-Tarumi et al.

adult productive and reproductive roles sooner, out of the system (Zambia Central Statistical O!ce and ORC Macro 2003). As will be shown later, the rate of early entry is not negligible in some countries, and both its factors and consequences warrant further research.

Measurements of late entry

Researchers need to be cautious in comparing across countries on even com- monly used indicators, such as net enrolment or net intake rate, because dif- ferent countries often use di"erent definitions, and di"erent sources of data are used within countries (Urquiola and Calderon 2006). Researchers who have focused more specifically on the issue of age of entry have also used dif- ferent definitions. For example, age of enrolment and age of entry are often used interchangeably, but these two concepts should be distinguished. The former refers to the age of students that are attending grade 1, and includes students who have entered a year (or years) before and have repeated grades. The latter refers to the age of students that entered grade 1. It is important to distinguish these concepts, as policies needed to tackle these issues may be di"erent.4 Policies to reduce late entry need primarily to target families, as they are the primary decision makers of when to start sending their children to school. On the other hand, policies to reduce over-age enrolment will not only need to target families, but also schools, as what and how students learn in the classrooms, will a"ect whether students repeat grades or not.

As Bommier and Lambert (2000) point out, when researchers have to reconstruct the age at school enrolment from the current level of education and the age of the child, they need to assume that there is no grade repeti- tion. Therefore, researchers often estimate the age of enrolment or number of over-age children, which includes grade repeaters (Glewwe and Jacoby 1993; Fentiman et al. 1999; Wils 2004). Although this measurement option is often due to the limitation of data in analyses using this concept, one cannot rigorously distinguish the determinants of delayed school enrolment from those of grade repetition.5

Both concepts are important and are realities that policy makers and school administrators need to consider and deal with. This paper focuses on late entry rather than over-age enrolment to keep the issue narrowly defined, that is, to disentangle the issue of early repetition from late entry, and to explore policy implications for this specific issue.

Methods

Data

In this paper, we use the Multiple Indicator Cluster Surveys (MICS2), which is a household survey collected only in economically less-developed countries.

109Late Entry Into Primary School In Developing Societies

MICS is currently administered every five years, and MICS2 was collected during the period 1999–2003. UNICEF coordinates the implementation of MICS at the country level and provides technical and financial support to government agencies implementing the survey. Data were available for 40 countries. We exclude two countries, Equatorial Guinea and Comoros, due to the high percentage of grade 1 repeaters – and consequently small sizes for our analyses – in these countries. MICS uses nationally representative sam- ples of children living in households, and the questions cover a large array of issues ranging from nutrition, health and education, birth registration, family environment, child labour, to knowledge and attitudes about HIV/AIDS.

Using cross-national household surveys is suitable for our analyses for three reasons. First, the data provides a comparable framework across coun- tries, because the questions we use to estimate the number of late entrants are consistent across countries. Most studies on late entry up to now have been single-country studies, thus, not allowing us to compare the rate inter- nationally. Data on net enrolment and net intake rate are reported interna- tionally, which partially captures the issue of age of entry, but they are often constructed from administrative data, which varies widely across countries. Second, using a household survey allows us to construct a measure based on age, which is the crux of the issue. Although we do acknowledge that house- hold’s self-declared reports of children’s age are not without limitations. Lastly, by using a household survey, we are able to investigate various fam- ily background characteristics of late entrants to suggest whom policies may need to target in reducing late entry into primary school.

Procedures

First, we examine the percentage of children entering primary school on- time, early and late for each country. Second, we examine the cross-tabula- tion to see which factors are consistently associated with late entry across a range of countries. Third, we run multivariate analysis for one country, in which late entry is most prominent, and examine the net e"ect of each vari- able. Multinominal logistic regression equation is used, because the depen- dent variable has three values. Multinominal logistic regression assesses the odds of late entry versus on-time entry and the odds of early entrants versus on-time entry. It should be noted that we do not investigate the consequence of late entry, for example the relationship between late school entry and poor learning outcome or drop-out, because MICS is a cross-sectional data and do not have any data on learning outcomes.

Measures

We frame our model as households considering whether or not to send their children to school on-time. The variables included in the final analyses are described in Table 1.

110 Yoko Nonoyama-Tarumi et al.

On-time entry is defined as children who entered primary school at the o!cial entry age or one year above the o!cial entry age. For example, if the country’s o!cial entry age is six, children who are age six and seven would be considered as on-time entrants. Children aged five would be considered early entrants. It should be noted that the questions used to construct this variable were only asked to children aged five years and older, and therefore would not include children who are younger. Therefore, these data are likely to underestimate the number of early entrants. Children that are age eight and above are considered as late entrants.6

Because the age of entry is not directly asked in the MICS, we use sev- eral questions to construct this variable: (1) the level and grade the child attends current year, (2) the level and grade the child attended last year, and (3) the age of the child. Of those children currently attending grade 1, if the child had attended preschool or had not attended school last year [the year before the survey was conducted], it is assumed that the child entered grade 1 for the first time in current year. If the child had attended school besides preschool the year before, it is assumed that the child has

Table 1. Description of variables

Variables Definition

Dependent variable Age of entry An ordinal variable denoting whether

the child entered primary school on-time (defined as entering at the o!cial age or one year above the o!cal age); Early entry, late entry, with on-time entry as the reference group

Independent variables Gender Dummy coded variable; 1 = female Place of residence Dummy coded variable; 1 = urban Mother’s education Dummy coded variable; 1 = some

primary education and above Household wealth Dummy coded variable; 1 = top 40%

in the household wealth distribution Child labour Dummy coded variable; 1 = involved

in child labour (defined as being involved in either at least one hour of economic activity, paid and conducted outside the household, or at least 28 hours of domestic work, such as housekeeping and family business, per week)

Perception of child’s school readiness

Dummy coded variable; 1 = younger sibling attends preschool

Older sibling in primary school Dummy coded variable; 1 = older sibling attends primary school

111Late Entry Into Primary School In Developing Societies

repeated grades, and thus is excluded from our sample. It should be noted that this method is useful in distinguishing the repeaters from late entrants, but reduces the sample size in countries that have a large number of repeaters in grade 1.

In this paper, we include seven independent variables. We dichotomise all variables in order to not lose degrees of freedom.7 Research has shown that students from disadvantaged family backgrounds are less likely to start primary school on-time (Bommier and Lambert 2000). For family background, we use mother’s education and household wealth. We use mo- ther’s education and not father’s education to prevent co-linearity, and based on LeVine’s (1980) theory on how mother’s own experience of schooling may have an independent e"ect on their attitudes towards child rearing independently from wealth or income. LeVine (1980) suggests that the experience of schooling provides mothers with expanded awareness of means-ends relationship, such as early learning experience influencing later school attainment, and reinforces their self-esteem, such as providing them with a stronger sense of personal responsibility for the welfare of their children.

Household wealth is constructed using principal component analysis and includes several items, such as main material of dwelling floor; number of rooms in dwelling; main source of drinking water; toilet facility used; household has electricity, radio, television, refrigerator; member of house- hold owns bicycle, motorcycle, car; and main cooking fuel used by house- hold. While the levels of wealth index are not directly comparable across countries, for example the poorest 20% in Venezuela cannot be compared to the poorest 20% in Laos, they are derived using an identical methodol- ogy. Filmer and Pritchett (1998, 1999) have shown that wealth indices are better than traditional consumption expenditures in predicting school enrol- ment. Whereas measures of income or consumption expenditures only re- flect the family economic environment at a particular time point, wealth reflects lifetime earnings and purchasing power as well as the economic environment in which the child developed (Orr 2003). Wealth is coded as one if the family is in the top 40% of the wealth distribution.

We also include gender and place of residence in our model. According to the EdData from Uganda and Zambia, a larger percentage of boys in grade 1 are over-age compared to girls (Uganda Bureau of Statistics and ORC Macro 2002; Zambia Central Statistical O!ce and ORC Macro 2003), but it is of interest to see whether this pattern holds across coun- tries. Parents may feel that girls need to finish their schooling earlier than boys to get married, and hence, may be more ready to send their female child on-time compared to male child. Gender is coded as one if the child is female. There is also evidence that children in urban areas are more like- ly to start school on-time (Bommier and Lambert 2000), partly due to

112 Yoko Nonoyama-Tarumi et al.

shorter distance to school.8 Parents in urban areas may feel safer to send their young child to school or they may have more information about school compared to parents in rural areas, due to proximity to school and to other parents sending their child to school. Place of residence is coded as one if the family lives in urban area.

In addition to these four background variables, three variables that are specific to the issue of on-time entry are included in the analyses: child labour, parent’s perception of child’s school readiness, and whether the older sibling is in school.

Child labour is defined as being involved in either at least one hour of economic activity, paid and conducted outside the household, or at least 28 hours of domestic work, such as housekeeping and family business, per week. The variable is coded as one if the child is involved in the above de- fined child labour. Families may decide to withhold sending their children to school when they are involved in some kind of child labour.

We use preschool attendance as a proxy of parent’s perception of child’s school readiness. It can be hypothesised that if the child has attended pre- school, the child will be more developmentally ready, or the parent may per- ceive the child to be more developmentally ready to start primary school. In the MICS data, the child’s experience of preschool is not directly asked for children of five and above. Thus, we construct a variable to indicate whether the child’s younger sibling is currently attending preschool. This variable is not without limitation, as first it reduces the sample size, because it is only applicable to children who have younger sibling of age three and four; sec- ond, because we have to assume that the child has had the experience of pre- school if the younger sibling is currently attending preschool; and third, we are assuming that a child would be more developmentally ready if the child had attended preschool.

The third variable is related to school attendance of an older sibling. Research has shown that if the older sibling is in school, the younger child is more likely to start school on-time or even early, as parents feel safer and sometimes more convenient to send them to school together (Fentiman et al. 1999). Thus, one can expect that a child whose older sibling is currently in primary school is less likely to start school late. This variable also reduces the sample size, as it restricts the sample to children who have older siblings under the age of 17.

It should be noted that we exclude some variables that have been found to be important determinants of late entry in past research from our model. We are not able to capture constructs related to school, such as school qual- ity and distance to school, because we use a household survey. However, as mentioned earlier, place of residence may partially capture distance to school, as the distance to school is likely to be larger in rural area in most countries. We are also not able to include a variable for child’s nutritional status with our data.9

113Late Entry Into Primary School In Developing Societies

Findings

Late entry across countries

Table 2 shows the percentage of students that entered early, on-time, and late across countries. Figure 1 shows the same values graphically, with the countries sorted by the percentage of late entrants.

One can see that in some countries, the percentage of late entrants is quite large. For example, in seven countries, Myanmar, Lesotho, Bosnia and Herzegovina, Angola, Rwanda, Guinea-Bissau and Kenya, over half of the students that enter primary school are 2 years or above the o!cial entry age. The median across the 38 countries is 28.3%. Table 2 also reveals that early entry, although it is not the focus of this paper, is not a negligible issue in some countries. The median is not so high, 8.4%; however, in six countries, Dominican Republic, Sierra Leone, Senegal, Cameroon, Sao Tome, and Zam- bia, over 15% of students enter primary school below the o!cial entry age. It should be noted that these values are likely to be under-estimated, due to the fact that the questions were only asked to children of age five and older. What is more striking is the wide age range, when we look at the age distribution within late entrants. For example, not only are 8-year-olds starting primary school for the first time, but as Figure 2 reveals, nearly 30% of students in Lesotho entering grade 1 are already age nine and 10, and nearly 10% are as old as age 11 and 12. This pattern of wide age range is found across countries.

Factors associated with late entry across countries

As summarised earlier, research has identified several factors that are associ- ated with age of entry. We use cross-tabulations to distinguish the factors associated with late entry across a range of countries. Table 3 shows the per- centage of late entrants among each sub-group.

For example, in Albania, 47.9% of males entering grade 1 are late entrants, whereas only 36.2% of females entering grade 1 are late entrants. The ratio shows the extent of disparity among the sub-groups; the further away the value is from one indicates that the percentage of late entrants is unequally large in one sub-group compared to the other sub-group. One can find that place of residence, mother’s education, household wealth, child labour, and older sibling attending primary school tend to be associated with late entry across countries. In other words, a child living in rural area, with a mother with no education, from a poor household, being involved in child labour, and with an older sibling not attending primary school is more likely to start school late. Although we hypothesised that boys are likely to start school late compared to girls, a significant gender di"erence was found in only about half of the countries. Similarly, although we expected that a child whose parent perceives the child to be developmentally ready is less likely to start primary school late, this pattern was not detected universally across

114 Yoko Nonoyama-Tarumi et al.

T a b le

2 . P er ce n ta g e o f ea rl y en tr y , o n -t im

e en tr y , a n d la te

en tr y in to

p ri m a ry

sc h o o l

C o u n tr y

(1 ) P ri m a ry

sc h o o l

o ! ci a l en tr y a g e

(2 ) %

E a rl y

en tr y a

(3 ) %

O n -t im

e en tr y b

(4 ) %

L a te

en tr y c

(5 ) S a m p le

si ze

d (6 ) %

R ep ea te r

A lb a n ia

6 N /A

5 6 .9

4 3 .1

2 1 1

1 9 .5

A n g o la

6 8 .4

3 7 .8

5 3 .8

1 ,4 0 3

1 4 .9

A ze rb a ij a n

6 0 .4

6 8 .6

3 1 .0

5 4 5

1 .1

B o li v ia

6 1 2 .6

7 8 .5

9 .0

5 8 1

4 .3

B o sn ia

a n d H er ze g o v in a

6 0 .2

4 1 .3

5 8 .4

4 6 2

1 .1

B u ru n d i

7 9 .2

4 7 .7

4 3 .0

4 8 8

2 8 .6

C a m er o o n

6 1 6 .7

5 2 .0

3 1 .3

5 0 2

4 6 .1

C en tr a l A fr ic a n R ep u b li c

6 1 0 .6

5 1 .3

3 8 .2

1 ,3 7 3

3 9 .5

C o te

d ’ Iv o ir e

6 1 2 .8

6 3 .3

2 3 .9

1 ,0 7 8

2 4 .5

D o m in ic a n R ep u b li c

6 2 5 .6

5 9 .5

1 5 .0

5 2 8

9 .4

D R

C o n g o

6 6 .8

4 5 .3

4 7 .9

1 ,4 6 2

3 5 .6

G a m b ia

7 1 2 .9

6 1 .6

2 5 .5

5 2 1

1 3 .0

G u in ea -B is sa u

7 1 4 .9

3 4 .4

5 0 .7

1 ,0 0 9

1 7 .9

G u y a n a

6 1 0 .2

8 5 .0

4 .8

4 2 1

4 .3

Ir a q

6 0 .4

7 0 .6

2 9 .0

2 ,5 1 7

1 0 .6

K en y a

6 N /A

4 9 .5

5 0 .5

1 ,4 1 7

2 1 .0

L a o

6 7 .2

5 4 .6

3 8 .2

1 ,0 3 8

4 1 .7

L es o th o

6 1 .3

3 5 .2

6 3 .4

9 0 8

2 1 .3

M o ld o v a

7 1 .1

9 0 .1

8 .8

4 6 7

0 .6

M o n g o li a

8 8 .0

8 2 .2

9 .9

5 3 8

7 .2

M y a n m a r

5 N /A

2 9 .0

7 1 .0

3 ,2 5 4

3 .6

N ig er

7 1 4 .3

7 0 .8

1 4 .9

3 7 0

8 .0

P h il ip p in es

6 1 .3

7 9 .2

1 9 .5

9 8 2

3 .3

R w a n d a

7 3 .5

4 5 .3

5 1 .2

8 6 9

3 1 .7

S a o T o m e

7 1 6 .2

6 9 .0

1 4 .8

2 1 6

4 2 .1

S en eg a l

7 2 0 .0

6 3 .5

1 6 .5

1 ,2 3 1

1 3 .2

S ie rr a L eo n e

6 2 0 .6

4 4 .8

3 4 .5

5 0 4

2 9 .9

S u d a n (N

o rt h )

6 5 .7

6 1 .8

3 2 .5

2 ,8 0 1

5 .2

115Late Entry Into Primary School In Developing Societies

T a b le

2 . C o n ti n u ed

C o u n tr y

(1 ) P ri m a ry

sc h o o l

o ! ci a l en tr y a g e

(2 ) %

E a rl y

en tr y a

(3 ) %

O n -t im

e en tr y b

(4 ) %

L a te

en tr y c

(5 ) S a m p le

si ze

d (6 ) %

R ep ea te r

S u d a n (S o u th )

6 9 .4

4 4 .7

4 5 .9

3 4 2

1 3 .6

S u ri n a m e

6 2 .5

9 2 .4

5 .1

3 1 6

3 7 .8

S w a zi la n d

6 6 .1

6 6 .3

2 7 .6

5 5 5

2 1 .1

T a ji k is ta n

7 N /A

9 5 .2

4 .8

6 4 3

0 .3

T o g o

6 1 2 .9

4 8 .5

3 8 .5

7 1 1

3 2 .5

T ri n id a d T o b a g o

5 N /A

9 9 .5

0 .5

1 9 9

2 2 .3

U zb ek is ta n

7 3 .2

9 0 .8

6 .1

6 6 0

5 .7

V en ez u el a

6 1 .3

7 8 .5

2 0 .2

4 5 6

5 .0

V ie t N a m

6 3 .0

8 3 .9

1 3 .1

8 9 8

6 .6

Z a m b ia

7 1 5 .1

5 7 .4

2 7 .5

8 8 1

5 .8

M ed ia n

8 .4

6 1 .7

2 8 .3

1 3 .4

A u th o r’ s es ti m a te

u si n g M IC

S 2 .

a T h e d en o m in a to r is

th e n u m b er

o f ch il d re n

cu rr en tl y in

g ra d e 1 , w h o

w er e n o t in

sc h o o l o r w er e in

p re sc h o o l th e p re v io u s y ea r.

T h e n u m er a to r in cl u d es

ch il d re n w h o a re

y o u n g er

th a n en tr y a g e.

b T h e d en o m in a to r is

th e n u m b er

o f ch il d re n

cu rr en tl y in

g ra d e 1 , w h o

w er e n o t in

sc h o o l o r w er e in

p re sc h o o l th e p re v io u s y ea r.

T h e n u m er a to r in cl u d es

ch il d re n w h o a re

en tr y a g e o r 1 y ea r a b o v e en tr y a g e.

c T h e

d en o m in a to r

is th e

n u m b er

o f ch il d re n

cu rr en tl y

in g ra d e

1 , a n d

w er e

n o t in

sc h o o l o r

w er e

in p re sc h o o l th e

p re v io u s

y ea r. T h e n u m er a to r in cl u d es

ch il d re n w h o a re

2 y ea rs

o ld er

th a n en tr y a g e.

d T h e sa m p le

co n si st s o f ch il d re n th a t a re

cu rr en tl y in

g ra d e 1 a n d w er e n o t in

sc h o o l o r w er e in

p re sc h o o l th e p re v io u s y ea r. S o m e o f th e

sm a ll sa m p le

si ze

is d u e to

th e la rg e n u m b er

o f re p ea te rs .

N /A

: U n a b le

to es ti m a te

d u e to

th e a g e ra n g e co v er ed

b y th e su rv ey .

D a ta

is w ei g h te d b y h h w ei g h t.

116 Yoko Nonoyama-Tarumi et al.

countries. This may have been due to the limitation of the measure. As noted earlier, we used the younger sibling’s preschool attendance as a proxy of that child’s preschool attendance, assuming that if the parent is currently sending the younger child to preschool, the parent is more likely to have sent the child of our interest (the older child) to preschool as well. But if preschool is still seen as a luxury in that country, families may be allocating their limited resources within a family in certain ways, and sending only one of their children to preschool or only certain gender child to preschool.10

20%

40%

60%

80%

100%

T ri n id

a d T

o b a g o

G u ya

n a

T a jik

is ta

n S

u ri n a m

e U

zb e ki

st a n

M o ld

o va

B o liv

ia M

o n g o lia

V ie

t N

a m

S a o T

o m

e N

ig e r

D o m

in ic

a n R

e p u b lic

S e n e g a l

P h ili

p p in

e s

V e n e zu

e la

C o te

d ' I

vo ir e

G a m

b ia

Z a m

b ia

S w

a zi

la n d

Ir a q

A ze

rb a ija

n C

a m

e ro

o n

S u d a n (

N o rt

h )

S ie

rr a L

e o n e

L a o

C e n tr

a l A

fr ic

a n R

e p u b lic

T o g o

B u ru

n d i

A lb

a n ia

S u d a n (

S o u th

) D

R C

o n g o

K e n ya

G u in

e a -B

is sa

u R

w a n d a

A n g o la

B o sn

ia a

n d H

e rz

e g o vi

n a

L e so

th o

M ya

n m

a r

Late-Entry On-Time-Entry Early-Entry

Figure 1. Percentage of early entry, on-time entry and late entry into primary school

5 6 7 8 9 10 11 12 13 14 15 16 17 Age

P e rc

e n ta

g e

Figure 2. Age distribution of primary school entrants in Lesotho

117Late Entry Into Primary School In Developing Societies

T a b le

3 .

P er ce n t o f la te

en tr y ch il d re n b y b a ck g ro u n d ch a ra ct er is ti cs

C o u n tr y

(1 ) G en d er

(2 ) R es id en ce

(3 )

M o th er ’s

ed u ca -

ti o n

(4 ) H o u se h o ld

w ea lt h

(5 ) C h il d la b o u r

(6 ) P er ce p ti o n

o f ch il d ’s

sc h o o l re a d i-

n es s

(7 ) O ld er

si b li n g s

in p ri m a ry

sc h o o l

M a le

F em

a le

R a ti o R u ra l U rb a n R a ti o N o

ed u ca -

ti o n

P ri m a ry

a n d

a b o v e

R a ti o B o tt o m

6 0 %

T o p

4 0 %

R a ti o Y es

N o

R a ti o N o

Y es

R a ti o N o

Y es

R a ti o

A lb a n ia

4 7 .9

3 6 .2

1 .3

4 5 .8

3 6 .2

1 .3

– 4 2 .6

– 4 4 .5

3 7 .5

1 .2

5 7 .1

4 2 .1

1 .4

6 6 .7

– –

4 4 .4

4 0 .6

1 .1

A n g o la

5 1 .7

5 5 .8

0 .9

6 5 .7

4 8 .0

1 .4

6 5 .4

4 7 .1

1 .4

6 4 .9

4 0 .0

1 .6

8 3 .2

5 0 .5

1 .6

6 1 .9

6 9 .5

0 .9

6 5 .1

4 9 .0

1 .3

A ze rb a ij a n

2 9 .9

3 2 .3

0 .9

3 0 .6

3 1 .4

1 .0

5 7 .1

3 0 .5

1 .9

2 9 .7

3 3 .9

0 .9

2 5 .0

3 0 .9

0 .8

3 6 .3

2 7 .3

1 .3

4 7 .1

2 4 .4

1 .9

B o li v ia

9 .9

7 .9

1 .3

1 3 .5

5 .5

2 .5

2 7 .1

5 .9

4 .6

1 2 .1

1 .2

1 0 .1

2 8 .6

8 .4

3 .4

2 9 .7

1 0 .0

3 .0

1 6 .0

1 0 .1

1 .6

B o sn ia

a n d H er ze g o v in a

6 0 .7

5 6 .1

1 .1

6 1 .6

5 2 .2

1 .2

7 2 .7

5 7 .9

1 .3

N /A

– 5 8 .6

– 6 4 .1

6 6 .7

1 .0

6 0 .0

5 5 .2

1 .1

B u ru n d i

4 3 .0

4 3 .1

1 .0

4 5 .1

2 6 .4

1 .7

4 4 .9

3 5 .7

1 .3

5 0 .5

3 7 .4

1 .4

7 3 .5

3 9 .5

1 .9

4 3 .8

2 7 .3

1 .6

5 3 .8

3 8 .1

1 .4

C a m er o o n

3 2 .4

2 9 .7

1 .1

3 7 .6

1 4 .1

2 .7

4 9 .1

1 6 .2

3 .0

3 6 .0

2 1 .2

1 .7

4 4 .1

2 5 .7

1 .7

4 2 .3

3 1 .0

1 .4

5 5 .9

2 7 .0

2 .1

C en tr a l A fr ic a n R ep u b li c 4 0 .2

3 5 .9

1 .1

4 6 .0

2 6 .0

1 .8

4 3 .5

3 2 .2

1 .4

4 6 .8

2 5 .7

1 .8

5 6 .9

3 4 .4

1 .7

4 4 .5

1 5 .8

2 .8

6 3 .1

3 1 .7

2 .0

C o te

d ’ Iv o ir e

2 4 .4

2 3 .3

1 .0

2 9 .5

1 7 .1

1 .7

2 8 .2

1 5 .8

1 .8

3 0 .2

1 3 .3

2 .3

4 6 .2

2 0 .9

2 .2

3 4 .7

2 4 .4

1 .4

3 6 .8

2 0 .8

1 .8

D o m in ic a n R ep u b li c

1 8 .9

1 0 .9

1 .7

1 7 .5

1 2 .8

1 .4

4 0 .7

1 0 .4

3 .9

2 0 .3

3 .0

6 .8

5 0 .0

1 3 .4

3 .7

5 2 .4

3 0 .6

1 .7

1 5 .8

1 5 .9

1 .0

D R

C o n g o

4 9 .3

4 6 .4

1 .1

5 9 .1

2 4 .9

2 .4

6 8 .4

4 0 .2

1 .7

5 9 .1

3 3 .4

1 .8

1 0 0 .0

4 4 .5

2 .2

7 8 .3

3 3 .3

2 .4

6 0 .6

4 3 .6

1 .4

G a m b ia

2 1 .8

3 0 .0

0 .7

2 7 .3

2 2 .7

1 .2

2 1 .1

3 1 .6

0 .7

3 0 .2

1 8 .8

1 .6

4 4 .4

2 4 .5

1 .8

2 8 .4

3 6 .4

0 .8

3 3 .8

1 9 .4

1 .7

G u in ea – B is sa u

5 0 .1

5 1 .5

1 .0

6 4 .6

4 0 .3

1 .6

5 6 .6

3 2 .0

1 .8

6 3 .6

4 0 .0

1 .6

7 1 .7

4 7 .0

1 .5

6 7 .7

6 6 .0

1 .0

6 6 .7

4 2 .5

1 .6

G u y a n a

6 .6

3 .1

2 .1

5 .5

2 .7

2 .0

– 4 .6

– 4 .8

4 .6

1 .0

1 2 .5

4 .6

2 .7

3 .8

4 .6

0 .8

5 .7

3 .5

1 .6

Ir a q

3 0 .1

2 7 .7

1 .1

3 8 .5

2 4 .5

1 .6

4 2 .1

2 1 .1

2 .0

3 3 .8

2 0 .2

1 .7

7 1 .7

2 8 .0

2 .6

3 1 .0

4 0 .0

0 .8

6 1 .2

2 5 .5

2 .4

K en y a

5 1 .3

4 9 .7

1 .0

5 5 .1

2 2 .2

2 .5

6 1 .3

4 8 .4

1 .3

5 8 .5

3 1 .0

1 .9

6 8 .7

4 7 .7

1 .4

5 6 .5

7 1 .4

0 .8

5 8 .2

5 2 .0

1 .1

L a o

4 0 .1

3 6 .1

1 .1

4 5 .5

2 0 .7

2 .2

5 0 .5

3 0 .8

1 .6

4 1 .4

3 3 .4

1 .2

6 8 .4

3 6 .2

1 .9

4 9 .2

3 5 .4

1 .4

5 2 .9

3 3 .7

1 .6

L es o th o

6 8 .7

5 8 .0

1 .2

6 4 .0

6 1 .1

1 .0

6 5 .8

6 3 .2

1 .0

6 9 .8

5 4 .0

1 .3

7 9 .6

6 1 .5

1 .3

8 8 .4

8 6 .4

1 .0

6 9 .6

5 9 .7

1 .2

M o ld o v a

9 .7

7 .9

1 .2

9 .9

6 .6

1 .5

7 5 .0

8 .2

9 .1

9 .9

6 .4

1 .5

1 6 .7

8 .0

2 .1

8 .6

6 .7

1 .3

1 3 .2

6 .7

2 .0

M o n g o li a

1 0 .9

8 .9

1 .2

1 2 .8

6 .2

2 .1

– 1 0 .0

– 1 3 .5

4 .8

2 .8

1 2 .5

8 .5

1 .5

1 4 .9

9 .6

1 .6

1 4 .7

1 0 .9

1 .3

M y a n m a r

7 1 .6

7 0 .3

1 .0

7 4 .1

5 8 .4

1 .3

1 0 0 .0

7 0 .9

1 .4

7 5 .0

6 2 .4

1 .2

N /A

9 7 .5

9 6 .1

1 .0

6 9 .2

6 8 .1

1 .0

N ig er

1 8 .9

9 .2

2 .1

1 8 .8

6 .1

3 .1

1 5 .7

1 3 .0

1 .2

2 0 .7

1 1 .4

1 .8

2 7 .6

1 1 .0

2 .5

3 2 .5

– –

2 8 .7

6 .7

4 .3

P h il ip p in es

2 1 .1

1 7 .6

1 .2

2 1 .0

1 7 .6

1 .2

N /A

2 3 .4

8 .6

2 .7

4 1 .2

1 9 .1

2 .2

2 6 .6

1 9 .4

1 .4

2 6 .8

2 0 .8

1 .3

118 Yoko Nonoyama-Tarumi et al.

T a b le

3 .

C o n ti n u ed

C o u n tr y

(1 ) G en d er

(2 ) R es id en ce

(3 ) M o th er ’s ed u ca ti o n

(4 ) H o u se h o ld

w ea lt h

(5 ) C h il d la b o u r

(6 ) P er ce p ti o n

o f ch il d ’s

sc h o o l re a d i-

n es s

(7 )

O ld er

si b li n g s

in p ri m a ry

sc h o o l

M a le

F em

a le

R a ti o

R u ra l U rb a n

R a ti o

N o

ed u ca -

ti o n

P ri m a ry

a n d

a b o v e

R a ti o

B o tt o m

6 0 %

T o p

4 0 %

R a ti o

Y es

N o

R a ti o

N o

Y es

R a ti o

N o

Y es

R a ti o

R w a n d a

5 1 .0

5 1 .4

1 .0

5 2 .8

3 9 .6

1 .3

5 8 .0

4 4 .2

1 .3

5 4 .9

4 2 .4

1 .3

6 6 .2

4 5 .7

1 .4

5 7 .2

2 7 .3

2 .1

6 4 .9

4 6 .7

1 .4

S a o T o m e

1 4 .2

1 5 .5

0 .9

1 8 .5

9 .8

1 .9

1 2 .2

1 5 .9

0 .8

1 6 .2

1 3 .2

1 .2

3 5 .7

1 3 .6

2 .6

2 6 .7

– –

1 2 .1

1 3 .9

0 .9

S en eg a l

1 6 .2

1 6 .8

1 .0

2 2 .5

8 .6

2 .6

1 9 .4

7 .7

2 .5

2 2 .7

8 .9

2 .6

3 3 .7

1 5 .0

2 .2

2 1 .4

1 2 .0

1 .8

3 1 .6

1 1 .6

2 .7

S ie rr a L eo n e

3 3 .8

3 5 .2

1 .0

3 9 .3

2 2 .0

1 .8

3 5 .7

2 6 .0

1 .4

3 9 .9

2 5 .5

1 .6

4 3 .6

3 2 .8

1 .3

6 0 .1

5 0 .0

1 .2

4 3 .2

2 7 .2

1 .6

S u d a n (N

o rt h )

3 1 .4

3 3 .7

0 .9

4 2 .7

2 6 .3

1 .6

4 2 .8

2 1 .1

2 .0

4 6 .5

2 0 .0

2 .3

7 7 .4

3 1 .3

2 .5

5 6 .6

4 0 .4

1 .4

6 5 .2

3 0 .5

2 .1

S u d a n (S o u th )

4 4 .2

4 7 .6

0 .9

– 4 5 .9

– 4 6 .9

4 3 .5

1 .1

4 2 .5

5 1 .1

0 .8

5 0 .0

4 0 .9

1 .2

7 8 .6

4 5 .7

1 .7

7 8 .6

4 0 .6

1 .9

S u ri n a m e

7 .3

2 .6

2 .8

6 .7

3 .3

2 .0

4 .9

4 .3

1 .1

6 .3

2 .2

2 .9

N /A

7 .6

9 .2

0 .8

1 5 .2

3 .8

4 .0

S w a zi la n d

3 5 .5

1 8 .6

1 .9

2 9 .7

1 3 .7

2 .2

3 7 .3

2 4 .1

1 .5

3 2 .0

1 4 .2

2 .3

5 0 .0

2 6 .4

1 .9

3 4 .9

2 9 .7

1 .2

4 8 .0

2 5 .3

1 .9

T a ji k is ta n

5 .3

4 .3

1 .2

4 .4

6 .5

0 .7

– 5 .0

4 .3

1 .2

1 4 .8

3 .9

3 .8

7 .8

– –

9 .1

2 .6

3 .5

T o g o

3 8 .1

3 9 .1

1 .0

4 4 .3

2 6 .1

1 .7

4 5 .5

2 6 .6

1 .7

4 5 .7

2 6 .8

1 .7

5 9 .7

3 0 .5

2 .0

5 4 .0

1 5 .0

3 .6

5 6 .4

3 6 .0

1 .6

T ri n id a d T o b a g o

0 .0

1 .1

0 .0

N /A

– 0 .5

– 0 .7

– –

– 0 .5

– –

– 0 .9

–

U zb ek is ta n

4 .7

7 .5

0 .6

7 .5

2 .9

2 .6

– 6 .1

– N /A

N /A

1 3 .3

5 .6

2 .4

8 .3

8 .5

1 .0

1 1 .8

4 .9

2 .4

V en ez u el a

2 3 .2

1 6 .7

1 .4

N /A

4 1 .9

1 8 .6

2 .3

2 3 .4

1 2 .9

1 .8

4 0 .0

1 9 .8

2 .0

2 8 .2

2 3 .6

1 .2

– 1 9 .6

–

V ie t N a m

1 5 .6

1 0 .6

1 .5

1 4 .9

3 .5

4 .3

4 7 .0

7 .2

6 .5

1 6 .5

4 .4

3 .8

6 1 .8

1 1 .2

5 .5

2 5 .6

1 1 .1

2 .3

1 3 .7

1 4 .8

0 .9

Z a m b ia

3 1 .4

2 3 .6

1 .3

3 5 .7

1 3 .5

2 .6

N /A

3 7 .7

1 5 .3

2 .5

N /A

4 1 .2

– –

3 5 .8

2 2 .7

1 .6

N o te : N /A

in d ic a te s th a t th e co u n tr y d o es

n o t h a v e d a ta

fo r th e v a ri a b le . – in d ic a te s th a t th er e is n o re le v a n t ca se

in th e ce ll (s u b -g ro u p ). D a ta

is w ei g h te d b y h h w ei g h t.

119Late Entry Into Primary School In Developing Societies

Determinants of late entry in the case of Angola

Table 4 shows the estimates from multinominal logistic regression equations for Angola, where the probability of entering primary school late, on-time and early is predicted by seven variables: gender, place or residence, mother’s education, household wealth, child labour, parent’s perception of school readiness, and older sibling in primary school. Angola was chosen, because it is one of the countries with large percentage of late entrants, and also because the reduction in sample size was not severe for the variables of preschool attendance and older sibling in primary school to conduct multi- variate analyses. We show the odds ratio, in which one represents no e"ect, while a ratio greater than one indicates that being in the non-reference cate- gory in the independent variable (for example, female) increases the odds of entering primary school late, and a ratio less than one indicates that it diminishes the odds of entering primary school late compared to the refer- ence category of the independent variable (for example, male).

As past research has shown, family background, such as household wealth and mother’s education, have significant e"ects on whether the child starts primary school on time. Children from rural areas are also more likely to enter late, controlling for household wealth and mother’s education. This suggests that distance to school may be an important factor for parents’ decision to send their children to school. In addition, two factors that were not tested in previous studies, child labour and older sibling attending pri- mary school, have independent e"ects on the age of entry from the e"ects of family background characteristics and place of residence. For example, the odds of entering school late is almost two times greater if the child is involved in child labour compared to a child not involved in child labour. Similarly, the odds of entering school late for a child whose older sibling is attending primary school is almost half of that of a child whose older sibling is not attending primary school. Parents may feel more comfortable about

Table 4. Odds ratio for late entry in primary school for Angola

(Multinominal Logistic Regression) n=1,403

Odds ratio SE

Female 1.05 (0.041) Urban 0.78*** (0.047) Mother with some education 0.71*** (0.041) Rich household 0.74*** (0.048) Child labour 1.92*** (0.070) Perception of child’s school readiness 0.94 (0.068) Older sibling in primary school 0.55*** (0.059)

* < 0.05, ** < 0.01, *** < 0.001. Reference group: on-time entry. Parameter estimates for early entry not shown.

120 Yoko Nonoyama-Tarumi et al.

sending a young child to school when the older sibling is in primary school, because the older sibling can walk the younger child to the school, the older sibling may be able to provide help in school work, or the older sibling may share his or her textbook with the younger child.

Conclusions

This paper has examined the issue of late school entry by using household surveys. The advantage of using household surveys is that it is information collected directly from households. Using aggregate data from secondary source, such as administrative data, pose questions in terms of the validity of cross-country comparisons. By using microdata, namely household surveys, we were first able to distinguish the children who simply entered late, from those who had grade repetition, and second estimate the percentage of late entrants universally across countries, which was not possible in past studies.

First, the high percentage of late entrants across developing countries in general, and the extremely high percentage in a number of countries warrant attention. The median of almost 30% across the 38 countries we examined suggests that the phenomenon of entering school at least 2 years past the o!cial entry age is common in many developing countries. Furthermore, in seven countries (Myanmar, Lesotho, Bosnia and Herzegovina, Angola, Rwanda, Guinea-Bissau and Kenya), over half of the students that enter pri- mary school are 2 years or more above the o!cial entry age. Late entry is particularly common during a period of rapid school expansion, such as after the disruption of a long war (Wils 2004).

Second, the data suggest that children entering school late are from disad- vantaged families. Children entering school late tend to be children with mothers with no education and from poor households. Although we did not investigate the consequence of late entry in this paper, past research has shown that late entry is associated with poor learning outcome and higher rate of drop-out. This suggests that children from disadvantaged families are at double disadvantage at the start of primary school, which has strong implica- tions on issues of inequity. Furthermore, the finding that child labour has an e"ect on the probability of late entry over and beyond family wealth and parental education suggest that families deliberately make the decision to delay their children’s schooling in order to engage their children in immediate income generation for the family. Families living in poverty need to be informed of the possible consequences of late entry. However, in advocating for on-time entry, one needs to be aware that going to school at the correct age will not fix the problem of poor school performance or drop out per se. It could also be argued that the reasons causing late entry, poor school performance and early dropout are the same and further research with longitu- dinal data is necessary to establish any link between them. The fact that an old- er sibling attends primary school has a positive e"ect on the probability of child

121Late Entry Into Primary School In Developing Societies

entering school on-time suggests that some of the advocacy for on-time entry might be implemented through older siblings. For example, schools may send the message of ‘‘starting school on-time’’ through older siblings to parents.

In this paper, we have focused on family’s readiness for school, as we relied on household surveys and did not have data on school’s readiness, such as school quality, distance to school, or school safety. The findings in this paper should not result in merely blaming the families for sending their children to primary school late. Rather, the purpose of the paper is to shed light on the background characteristics and the family context of children entering late so that it can inform educational policies that aim to reduce late school entry. Moreover, the evidence that children living in rural areas are more likely to enter school late suggests that school factors, such as dis- tance to school, also a"ect parents’ decision on when to start sending their children to school. Further research is necessary on the characteristics of schools to which parents are more likely to send their children to school on- time, and feel at ease with the transition from early learning programmes or from home, as in many cases, to school.

Acknowledgments

We would like to thank Yumiko Ota for the production of data files used for the analyses and Anna Smeby for her comments on earlier drafts. The views expressed herein are those of the authors and do not necessarily reflect the opinions or policies of the UNICEF or any of its affiliated organisations.

Notes

1. Glewwe and Jacoby (1993) suggest that, in Ghana, the cost of the average delay of two years later than the o!cial entry age, is about six percent of an individual life-time wealth. This is based on a simple ‘‘back-of-the-envelope calculation,’’ assuming constant post-school earnings, zero earnings prior to and during school enrolment, zero school fees, a constant three percent real interest rate, and infi- nite horizon, and that those who do delay complete the same number of grades and earn the same amount as delayers.

2. A reviewer noted that in some countries, late school entry may be due to the school systems, such as double schooling. For example, in some African countries such as Senegal, a child having to first attend a qu’ranic school, which cannot be equated with preschool or kindergarten, before attending primary school, may be a reason for late entry into o!cial primary school.

3. Wils’ study is an exception, in which she further tests her findings from Mozam- bique with a cross-national data. However, this study does not di"erentiate late entrants from repeaters.

4. It should be noted that even with the use of cross-national data and cross-national definition; we need to be sensitive to unique context and policies of each country.

122 Yoko Nonoyama-Tarumi et al.

For example, a reviewer noted that in Burkina Faso, although data not included in this study, the o!cial school entry age is seven and a child who is nine or over can no longer be admitted to school even if the parents were willing to send their child (Décret N! 289, Article 30, PRES (Président de la République)/EN (Ministère de l’Education Nationale), 03 août 1965). In such cases, the percentage of late entrants is likely to be low and the percentage of on-time entrants is likely to be high, rela- tive to other countries. But this does not indicate that the absolute number of on-time entrants is high, because it may be the case that the policy leads to high number of out-of-school children (as late entrants are not admitted).

5. One exception to this is the DHS EdData which directly asks the question ‘‘How old was (NAME) when he/she first attended primary 1?’’ (Central Statistical Of- fice and ORC Macro, 2003).

6. It should be noted that one of the limitations of this study is that we use the age at the time of the survey. Thus, we are not able to fully account for the interac- tion between the month in which the survey was conducted and the month in which the school year starts in a given country.

7. For variables such as mother’s education and household wealth, the analyses were originally run with a number of dummy variables, such as four variables to repre- sent household wealth quintiles. But in the final analysis, we only use one dummy variable for each measure because of the small sample size in some countries. We acknowledge that these cut-o" points (like defining ‘‘low’’ levels of mother’s educa- tion as no education) are somewhat arbitrary and throws away information, and that having a uniform cut-o" point across societies is simplistic when a level of dis- advantage associated with no education may vary across societies.

8. Distance to school is identified as one of the major determinants of late entry in past research, but unfortunately, there is no question asked in MICS that directly cap- tures the concept. The closest proxy would be the place of residence (urban/rural).

9. The MICS data includes information on stunting, but these questions are only available for children under five years old. Although we use the sibling’s informa- tion as proxies of some of our variables, such as attendance in preschool or pri- mary school, we do not take the same procedure to construct a variable for nutritional status, because we believe it would introduce too much noise.

10. This analysis is based on the MICS question ‘‘Does the child attend any organ- ised learning or early childhood education programme, such as a private or government facility, including kindergarten or community child care?’’ Therefore, it takes a broad concept of early learning programme, not limiting to ‘‘pre- school’’ and including ECD programmes targeting the poor. The analyses of this question showed that access to ECD programmes, even with a broad definition, is still limited in some countries. For example, in 17 out of 27 African countries for which data was available, the percentage of three and four year olds attending such programme was below ten percent. (Nonoyama-Tarumi et al. 2009).

References

Benveniste, Luis, and Patrick McEwan. 2000. Constraints to Implementing Educational Innovations: The Case of Multigrade Schools. International Review of Education 46(1/2): 31–48.

Bommier, Antoine, and Sylvie Lambert. 2000. Education Demand and Age at School Enrolment in Tanzania. The Journal of Human Resources 35(1): 177–203.

123Late Entry Into Primary School In Developing Societies

Colclough, Christopher, and Keith Lewin. 1993. Educating All the Children: Strategies for Primary Schooling in the South. Oxford: Clarendon Press.

Fentiman, Alicia, Andrew Hall, and Donald Bundy. 1999. School Enrolment Patterns in Rural Ghana: A Comparative Study of the Impact of Location, Gender, and Health on Children’s Access to Basic Schooling. Comparative Education 35(3): 331–349.

Filmer, Deon, and Lant Pritchett. 1998. Estimating Wealth E!ects without Income or Expenditure data–or Tears: Educational Enrolment in India (World Bank Policy Research Working Paper No. 1994). Washington, D.C.: Development Economics Research Group, World Bank.

——. 1999. The E"ect of Household Wealth on Educational Attainment: Evidence from 35 countries. Population and Development Review 25(1): 85–120.

Glewwe, Paul, and Hanan Jacoby. 1993. Delayed Primary School Enrolment and Childhood Malnutrition in Ghana. Living Standards Measurement Study Working Paper No. 98. Washington, D.C.: World Bank.

LeVine, R. 1980. Influences of Women’s Schooling on Maternal Behavior in the Third World. Comparative Education Review 24(2): 78–105.

Lloyd, C., and A. Blanc. 1996. Children’s Schooling in Sub-Saharan Africa: The Role of Fathers, Mothers, and Others. Population and Development Review 22(2): 265–298.

Nonoyama-Tarumi, Yuko, Edilberto Loaiza, and Patrice Engle. 2009. Inequalities in Attendance in Organized Early Learning Programmes in Developing Societies: Findings from Households Surveys. Compare 39(3): 381–405.

Orr, Amy. 2003. Black-White Di"erences in Achievement: The Importance of Wealth. Sociology of Education 76(4): 281–304.

Uganda Bureau of Statistics and ORC Macro. 2002. Uganda DHS EdData Survey 2001: Education Data for Decision-making. Calverton: Uganda Bureau of Statistics and ORC Macro.

Urquiola, Miguel, and Valentina Calderon. 2006. Apples and Oranges: Educational enrolment and attainment across countries in Latin America and the Caribbean. International Journal of Educational Development 26: 572–590.

Wils, Annababette. 2004. Late Entrants Leave School Earlier: Evidence from Mozambique. International Review of Education 50(1): 17–37.

Zambia Central Statistical O!ce and ORC Macro. 2003. Zambia DHS EdData Survey 2002: Education Data for Decision-making. Calverton: Zambia Central Statistical O!ce and ORC Macro.

The authors

Yuko Nonoyama-Tarumi is Early Childhood Development Specialist in United Nations Children’s Fund (UNICEF), New York. She received her Ph.D. from Teachers College, Columbia University in Comparative and International Education. Her main research interests are in comparative education, sociology of education, family e"ects, and social stratification. She is currently conducting research on parenting practices in early childhood and children’s transition from families to schools in developing societies.

124 Yoko Nonoyama-Tarumi et al.

Contact address: Ochanomizu University, Research Center for Human Development and Education, 2-1-1, Otsuka, Bunkyo-ku, Tokyo 112-8610, Japan. E-mail: [email protected].

Edilberto Loaiza is Senior Programme O!cer for Statistics and Monitoring in Policy Planning in UNICEF, New York. He earned his Ph.D. in Social Demography. He focuses on the areas of child mortality, education and child protection. Previous to joining UNICEF, he worked as global coordinator of the Multiple Indicator Cluster Survey (MICS, 1999–2001) and as Demographic Expert at the Demographic and Health Surveys (DHS) Program of ORC Macro (1989–1999).

Contact address: Strategic Information Section (SIS), Division of Policy and Plan- ning (DPP), UNICEF, New York, USA. E-mail: [email protected].

Patrice L. Engle is Professor of Psychology and Child Development at California Polytechnic (Cal Poly) State University in California. She received her Ph. D. in Child Development from Stanford University. She was Senior Advisor for Early Childhood Development in UNICEF, New York, and Chief of Child Development and Nutrition for UNICEF India. Her major areas of work are in factors influencing early child development in developing countries, care for nutrition, the impact of HIV/AIDS on young children, and the role of the psychological environment on complementary feeding.

Contact address: Department of Psychology and Child Development, Cal Poly State University, 1 Grand Avenue, Building 47-24, San Luis Obispo, CA 93407-0387, USA. E-mail: [email protected].

125Late Entry Into Primary School In Developing Societies

Copyright of International Review of Education is the property of Springer Science & Business Media B.V. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.