- Email: [email protected]
Delayed primary school enrolment among boys and girls in Ghana
International Journal of Educational Development 49 (2016) 107–114
Contents lists available at ScienceDirect
International Journal of Educational Development journal homepage: www.elsevier.com/locate/ijedudev
Delayed primary school enrolment among boys and girls in Ghana Hellen A. Seshie-Nasser 1,*, Abena D. Oduro University of Ghana, Department of Economics, P.O. Box LG57, Legon, Accra, Ghana
A R T I C L E I N F O
A B S T R A C T
Article history: Received 12 January 2015 Received in revised form 19 December 2015 Accepted 25 December 2015
Using the fifth Ghana Living Standards Survey, this paper examines whether delayed entry is as likely among girls as it is among boys and the importance of the household’s poverty status in understanding delayed entry. No gender difference is found in the likelihood of delayed entry. However boys are much older for their grades than girls suggesting that they may experience a more extended period of delayed entry. The poverty status of the household is not an important correlate of delayed entry among girls or children in urban households but is important for boys and rural children. ß 2016 Published by Elsevier Ltd.
JEL classification: I21 J24 Keywords: Delayed enrolment Age-appropriate enrolment Primary education Ghana
1. Introduction Ghana has achieved significant strides in the provision of basic education for all children, however, the sector is still bedeviled with drawbacks and limitations (Akyeampong et al., 2007). Although gross primary enrolment rates exceed 100 percent, net enrolment rates were about 84 percent in 2013 (Ministry of Education, 2013). Net primary enrolment rates of less than 100 percent indicate that the goal of universal primary education has not been achieved whilst gross enrolment rates of more than 100 percent suggest that there are overaged children in primary school. The presence of overaged children in primary and secondary school can be due to one or a combination of reasons. These are failure of children to be enrolled in primary school at the official entry age which is usually six years old, class repetition or interrupted enrolment in school. Age-grade inappropriateness increases the risk of a child dropping out of school. Wils (2004) found that in Mozambique children who enter school late had substantially higher drop-out rates than younger entrants. In Ghana, Akyeampong et al. (2007) found that whilst enrolment
rates increase rapidly for children between the ages of 8 and 13, there is a decline in enrolment rates among children between the ages of 13 and 16. When girls enroll in primary school later than the official entry age, they are more likely to drop out, especially as they approach adolescence (Rose and Al-Samarrai, 2001; Wils, 2004). Delayed entry into primary school can make the attainment of the goal of universal primary education difficult for two reasons. First, not all children at the official entry age for primary school will be enrolled in school. Second, delayed entry can increase the likelihood of non-completion of primary education. Not many studies on education in Ghana have examined the incidence of age-appropriate grade enrolment or the determinants of age-appropriate grade enrolment. This paper intends to contribute to the literature on this important subject. It has three objectives. The first is to find out whether girls are just as likely as boys to experience delayed entry. The second is to identify the correlates of delayed entry for girls and boys and the importance of socio-economic circumstances of the household. The third objective is to examine the factors associated with the extent to which a child is overaged for his or her class. 2. Evidence on the determinants of age-appropriate enrolment
* Corresponding author. Current address: 237 Pennwood Street, Newark, DE 19713, USA. E-mail addresses: [email protected], [email protected] (H.A. Seshie-Nasser), [email protected] (A.D. Oduro). 1 Current address: This author is currently with University of Delaware. http://dx.doi.org/10.1016/j.ijedudev.2015.12.004 0738-0593/ß 2016 Published by Elsevier Ltd.
A few studies have examined the incidence of age-appropriate grade enrolment in the African context (Burke and Beegle, 2004; Moyi, 2011, 2010; Wells, 2009) and more particularly in Ghana (Akyeampong et al., 2007; Glewwe and Jacoby, 1995). Using data
108
H.A. Seshie-Nasser, A.D. Oduro / International Journal of Educational Development 49 (2016) 107–114
from the Core Welfare Indicators Survey of 2003, Akyeampong et al. (2007) found that the mean ages of pupils enrolled in primary 1 and primary 6 were 7.5 years and 13.3 years respectively. In some instances children aged ten years or more were in primary 1 or 2. They attributed this to delayed entry into primary school. Generally, enrolment for boys is higher than for girls in the African context (Glick and Sahn, 2000; Lincove, 2009). There is however, mixed evidence on the relationship between the sex of the child and whether or not the child would be enrolled at the right age. In Tanzania, an interesting revelation by Bommier and Lambert (2000) is that, girls start school earlier than boys. A possible reason assigned to this is the lump-sum gains made after receiving education, such as the bride-price, when the young lady is ready for marriage. These gains are modeled as a lump-sum gain and it was found that when these gains increase girls enroll earlier and stay in school for a shorter period. The intuition is that the present value of such a lump-sum gain is greater when obtained sooner. The parents’ interest is in having their daughters ready for the marriage market as soon as possible. Moyi (2010) on the other hand did not find a significant difference in the incidence of delayed entry to primary school between boys and girls in Malawi. Grogan (2009) found that fee elimination in Uganda increases the probability that girls begin school before age nine. Pradhan (1998) found that boys from wealthy households in Indonesia have a higher probability of being enrolled at the appropriate age, though the difference was minimal. The income or wealth of the household has been found to be significant determinants of age at enrolment. The probability of inappropriate age-grade enrolment declines as the income or wealth of the household increases (Glewwe and Jacoby, 1995; Moyi, 2010; Pradhan, 1998). In Indonesia, consumption per capita is found to increase the probability of enrolling in secondary school at the right age (Pradhan, 1998). Facing credit constraints, parents may choose to send children to work instead of enrolling them in school (especially in rural agricultural communities) in order to build up savings (Glewwe and Jacoby, 1995). Children in poor households may experience delayed entry for a longer period because their parents or guardians may take a longer time to generate the required savings. There are mixed findings regarding the relationship between the education of the household head and the age at enrolment of children in the household. While Moyi (2010) found a negative association in Malawi between the education of the head of the household and the likelihood of a child enrolling in school late, Pradhan (1998) in a model that included both parents’ education and literacy found that parents’ education was insignificant but literacy of the mother increased the probability of enrolling at the appropriate age. The number of children currently in school from a given household or family have been found to be negatively associated with the likelihood of age-appropriate enrolment (Wells, 2009). There are mixed findings concerning the relationship a child’s birth order and the likelihood of starting school on time. Pradhan (1998) found that being the eldest child increases age at secondary enrolment in urban areas and being the youngest child in the household reduces the age at enrolment. However, the existence of liquidity constraints can change this relationship. Bommier and Lambert (2000) found that in Tanzania being the eldest in a liquidity-constrained family could reduce the chances of being enrolled on time because these children have to help accumulate funds before they enroll. On the other hand, the youngest child is more likely to enroll early because funds are more easily available. Besides, the younger ones would have access to school materials and logistics handed down to them by their older siblings. Moyi (2010) has shown that in the Malawian context, children living in households headed by women are more likely to enroll
early. In addition, the death of a mother is less injurious than the death of a father to the likelihood of being enrolled in school on time. Ainsworth et al. (2005) also found a similar relationship between mother’s survival and delayed enrolment among children in north-western Tanzania. In addition to the characteristics of the household and the family the child belongs to other factors have been found to be associated with delayed enrolment. Bommier and Lambert (2000) found that the age at the time of school enrolment and schooling duration in Tanzania are both functions of a set of variables describing schooling costs and school quality. Glewwe and Jacoby (1995) focused attention on the effect of childhood nutrition on age of enrolment and school attainment in Ghana, while controlling for other community, household and individual characteristics. Strong evidence was found for nutritional deficiencies in early childhood; children with stunted growth start primary school later. Using more recent data (2005/06 Ghana Living Standards Survey), this paper investigates whether girls and boys are just as likely to start primary school late and the importance of socioeconomic correlates of delayed enrolment at the entry level. A revisit of the issue on delayed entry is pertinent given the introduction of policies to improve access to basic education in the intervening period since the Glewwe and Jacoby study and the implications delayed entry has for the attainment of the goal of universal primary education. 3. Ghana’s education system The education reforms of 1987 introduced fundamental changes to Ghana’s education system. The reforms were introduced as part of broader economic and structural reforms implemented under the aegis of the World Bank and IMF beginning in 1983 (see contributions in Aryeetey et al., 2000 for a discussion of the reforms). The structure of the education system was streamlined and the number of years of pre-tertiary education was reduced. The new structure introduced the basic education system which comprised six years of primary school and three years of junior high school. Basic education was subsequently redefined to include two years of pre-school. The official age of entry into primary 1 is six years. The completion of basic education is the first official transition point in the education system. All students sit for the official terminal examination that either marks the end of formal schooling or is the transition to senior high school education. The reforms of 1987 aimed at increasing access to education for both boys and girls and across income groups, changing the structure of the school system, improving the quality of education through improvements in pedagogy and the introduction of a more relevant curriculum, attaining financial sustainability and improving the management of resources within the sector. User fees were introduced to bring about financial sustainability. This however had an almost immediate detrimental effect on enrolment which was contrary to the objective of improving access. In the face of this the user fee policy had to be revised. The policy of free, compulsory and universal basic education was introduced in 1995 and had as one of its objectives the provision of free quality education at the basic level to all children by 2005. The components of education expenses that were to be provided free were tuition, textbooks, equipment and tools. At the junior high level, pupils were expected to pay a textbook user fee equivalent to not more than 10 percent of the average total cost of the textbook supplied per pupil. In 2004/05 the capitation grant scheme was introduced. The grant covers the extra cost and levies (such as examination, facilities management, security charges, games and sports) that parents usually pay as ‘‘school fees’’ in public schools. The National
H.A. Seshie-Nasser, A.D. Oduro / International Journal of Educational Development 49 (2016) 107–114
School Feeding Program was introduced in the 2005/06 academic year. Children in schools in deprived districts are provided with one meal a day. Osei-Fosu (2011) found that the capitation grant caused an immediate increase in enrolment and the school feeding program had a large positive and significant impact on school enrolment, attendance and retention. 4. Methodology A binary choice model is utilised to estimate the determinants of delayed enrolment and the ordinary least squares method is employed to explain the factors associated with the number of years a child is overaged. The dependent variable for the delayed enrolment model takes the value of 1 if a child in the sample of six and seven year olds is either not in school or is in pre-school. It takes the value of 0 if a child is in primary 1 or a higher grade. In a variant of binary estimation, an index function formulation explains an underlying unobserved continuous random variable y*, such as the desire to delay enrolment in school. This desire is unobservable. What is observed is the binary variable y, which takes the value 1 or 0 depending on whether or not y* crosses a threshold, say a, as follows:
Yi ¼
1 ðenrolment in school is delayedÞ if y > a 0 ðenrolment is not delayedÞ if y a
Employing a logistic distribution, the probability of delayed enrolment is expressed as: Pr½yi ¼ 1jxi ¼
exp ðx0i bÞ 1 þ exp ðx0i bÞ
(1)
where xi represents the vector of explanatory variables and b represents the vector of parameters. The empirical model in an odds ratio format is given as: pi ¼ exp ðx0i bÞ 1pi
(2)
where pi is the probability of delayed enrolment. To obtain a profile of girls and boys who are likely to experience delayed entry separate logistic regressions are run for the sample of boys and girls. Separate regressions are also run for the urban and rural samples. The second model estimates the correlates of the number of years a child’s age exceeds the appropriate age for the grade s/he is enrolled in. This second model is intended to shed some more light on the phenomenon of delayed entry. Ideally, longitudinal data is preferable to cross-sectional data for a study on the likelihood of delayed enrolment. This is because conditions of the household at the time a child is supposed to be enrolled into the first year of primary school may be different from conditions prevailing at the time the data is collected. In the absence of longitudinal data, this difficulty may be overcome by restricting the analysis to the sample of young children of entry level age. Hence, this study concentrates on the sample of six and seven year olds. Another reason for investigating the likelihood of inappropriate age-grade enrolment among this sample of children is because once a child does not enter primary school at the appropriate age the phenomenon of being overaged for one’s grade is likely to persist throughout the trajectory from basic to secondary education. We make the assumption that six and seven year olds who are not currently enrolled in primary school will eventually be enrolled. This assumption can safely be made because of the upward trends in gross primary enrolments and a gross primary
109
enrolment rate which exceeded 100 percent in the 2012/13 academic year (Ministry of Education, 2013). 4.1. Data This study utilises data from the fifth round of the Ghana Living Standards Survey (GLSS) conducted between September 2005 and September 2006. This is a nationally representative household survey that collected data from over 8500 households and about 37,000 household members. The fifth round of the GLSS has a sample of 2290 children aged six or seven years of whom 1140 are boys and 1150 are girls. About 56 percent of the sample of six and seven year olds comprises six year olds and the remaining 44.2 percent are seven year olds. Even though children are expected to start primary school when they are six years old, in some cases they enter at age seven due to date of birth considerations. Most seven year olds however are expected to be in primary 2. This then implies that, six and some seven year olds who are not currently in primary 1 are experiencing delay in enrolment and some seven year olds who are currently not in primary 2 are also enrolled late. In order to reduce the likelihood of misclassification the time of interview was taken into consideration when determining whether or not a child was enrolled in school late. For example a primary 1 seven year-old child surveyed in August 2005 may be wrongly considered to be experiencing delayed enrolment at the time of interview when in actual fact s/he would be entering primary 2 in September (just one month away). All seven year olds whose households were surveyed in September and were reported to be in primary 1 or pre-school were classified as being in an inappropriate grade for their age. About 29.5 percent of the six and seven year olds are in primary 1 and about 12.3 percent are in primary 2. About 40.8 percent are either in preschool or not enrolled, that is they are delayed entrants in primary 1 based on our working definition. A few children (3.6%) are in grades higher than primary 2 or in non-formal education, such as Islamic schools. 4.2. Explanatory variables The variables of interest are the sex of the child and the poverty status of the household the child is a member of. We hypothesise that girls are more likely to start primary school late because they may be considered more vulnerable. Parents would want to send them to school at an age when they are more articulate and less vulnerable. When there are scarce resources, parents will more likely delay girls’ entry. We expect the likelihood of delayed entry to be higher among the very poor compared to poor and non-poor households and to be higher among the poor compared to nonpoor households. The choice of additional explanatory control variables is informed by the literature and comprises the characteristics of the child’s parents or guardians and the household the child lives in. Household characteristics include the sex, age and religion of the household head, the number of siblings in school, location of the household, administrative region of residence of the household, whether or not the child lives with both, either or neither of its parents. The characteristics of the parents include employment status and level of education completed. There are other variables that are expected to influence age at enrolment, but are not included in the model because they are not found in the data set. For instance, the model does not include disability or other health-related factors, the nutritional status of the child or distance to the nearest school. The description of the variables and summary statistics are presented in Table 1. The sample is almost evenly distributed between girls and boys. The majority of children reside in rural households (71%) and about 65 percent live with both parents
110
H.A. Seshie-Nasser, A.D. Oduro / International Journal of Educational Development 49 (2016) 107–114
Table 1 Variable definitions and summary statistics. Variable
Description
Mean
Std. dev.
Min
Max
Male Female head Age_head Rural dummy Number of siblings
Sex of the child = 1 if male Sex of the household head = 1 if female Age of the household head Household located in a rural area Number of siblings in school
0.496 0.209 45.710 0.712 1.867
0.500 0.407 12.983 0.453 1.991
0 0 20 0 0
1 1 99 1 15
0.652 0.159 0.046 0.142
0.476 0.365 0.210 0.349
0 0 0 0
1 1 1 1
0.280
0.449
0
1
0.126
0.332
0
1
0.594
0.491
0
1
0.492
0.500
0
1
0.366
0.482
0
1
0.634
0.482
0
1
0.269
0.444
0
1
0.546 0.224
0.498 0.417
0 0
1 1
0.179 0.051
0.384 0.220
0 0
1 1
0.444 0.175 0.271 0.110
0.497 0.380 0.444 0.313
0 0 0 0
1 1 1 1
0.213
0.410
0
1
0.542
0.498
0
1
0.174
0.379
0
1
0.071
0.257
0
1
0.087 0.067 0.083 0.080 0.091 0.168 0.097 0.137 0.091 0.100
0.282 0.250 0.275 0.272 0.288 0.374 0.296 0.344 0.288 0.300
0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 1 1 1
Children living with parents: Reference = both parents Child living with both parents Both parents Mother only Child living with single mother Father only Child living with single father Neither Child living with non-parents Poverty status: Reference = extremely poora: Extremely poor Income/expenditure below the lower poverty line Income/expenditure between the upper and Poor lower poverty lines. Income/expenditure above the poverty line Non Poor Employment status of parents/guardians: Reference = not working Mother/female guardian works i Mother in agriculture n agricultural sector Mother/female guardian works Mother in other sector in other sector Father/male guardian works Father in agriculture in agricultural sector Father/male guardian works in other sector Father in other sector Mother: Reference = no education No formal education Some basic education Completed JHS/M4 Above JHS
No formal education At least primary but not completed junior high/middle school Completed JHS/M4 Education above JHS/M4
Father: Reference = no education No formal education Some basic education Completed JHS/M4 Above JHS Religion dummies: Reference = traditional religion Muslim Equals 1 if the household head is a Muslim Equals 1 if the household head Christian is a Christian Equals 1 if household head Traditional is of traditional religion Equals 1 if household head No religion adheres to no religion Administrative region dummies: Reference = greater Accra Western Central Greater Accra Volta Eastern Ashanti Brong Ahafo Northern Upper East Upper West
Source: GLSS5. a The poverty lines are constructed using household consumption expenditure per adult equivalent. For a discussion on the construction of the poverty lines see Ghana Statistical Service (2007).
whilst about 5 percent live with their fathers only. About 21 percent of children live in a household headed by a woman. The proportion of children whose fathers are employed in agriculture (63%) exceeds the proportion of children whose mothers are employed in that sector (49%). About 55 percent of children have mothers who do not have formal education whilst 44 percent of children have fathers who fall in this education category.
5. Incidence of age-grade appropriateness in Ghana Table 2 presents evidence on the proportions of children aged 6–15 years in appropriate and inappropriate grades using data from the fifth GLSS. The proportion of children in the appropriate age for their grade declines as the age cohort increases. Among six year olds, only 75 percent are enrolled; and out of that number 40 percent are not in the appropriate grade, i.e. primary 1. The
H.A. Seshie-Nasser, A.D. Oduro / International Journal of Educational Development 49 (2016) 107–114
111
Table 2 Incidence of age-grade inappropriateness for children aged 6–15 (2005/06). Age cohorts
Never enrolled
Right grade
Inappropriate grade
Total
% in Inappropriate Grade
% Right grade
6 7 8 9 10 11 12 13 14 15
307 179 160 148 149 93 146 106 120 159
434 608 438 361 350 257 256 253 205 168
496 215 381 426 549 464 583 579 521 570
1240 1003 980 939 1048 815 987 941 869 927
40.0% 21.4% 38.9% 45.4% 52.4% 56.9% 59.1% 61.5% 60.0% 61.5%
35.0% 60.6% 44.7% 38.4% 33.4% 31.5% 25.9% 26.9% 23.6% 18.1%
Total
9749
Source: Authors’ calculations from GLSS5.
percentage (21.4%) of 7 year olds who are overage for their grade is less than that of the six year olds. However, the incidence increases from the 8 year olds reaching 61.5 percent for 15 year olds. Even though part of the problem of inappropriate age-grade enrolment may be due to grade repetition or temporary withdrawals along the educational ladder, class repetition is quite low in Ghana, especially in the urban schools. It is also not common for Ghanaian students to temporarily withdraw from school (Glewwe and Jacoby, 1995). Another way to present the evidence on delayed entry is to calculate the share of children in a particular grade who are older than the appropriate age. Chart 1 provides evidence using data from the third (1991/92) and fifth (2005/06) Ghana Living Standard Surveys. Two interesting observations can be made. First, in both periods the proportion of children in a grade who are older than the appropriate age increases from primary 1 to the second year of junior high school. This suggests that in more recent years there has been an improvement in the proportion of children who are enrolled in school when they are six years old. Second, the incidence of age-inappropriateness is higher in 2005/06. This would seem to contradict the previous observation. The higher incidence of overage children in each grade in 2005/06 compared to 1991/92 is probably due to the introduction of the capitation grant and school feeding program which created an incentive for enrolment of out of school children may of whom were overage (Akyeampong, 2011).
6. The factors associated with delayed entry Table 3 presents the results of the logistic regression that investigates the correlates of delayed entry for the full sample, the 80.00% 70.00% 60.00% 50.00% 40.00% 30.00% 20.00%
1991/92 2005/06
10.00% 0.00%
Chart 1. Incidence of overage children in basic education (1991/92 and 2005/06).
urban and rural samples and the samples of girls and boys. Odds ratios are reported and the standard errors are corrected for clustering of children in the household. In the model with the full sample of six and seven year olds there is no significant difference between boys and girls in the likelihood of delayed entry. The odds of a non-poor child enrolling late is about 35% lower, compared to children from extremely poor households. Similar results are found in age-grade appropriateness literature (Pradhan, 1998; Grira, 2001; Moyi, 2010; Wells, 2009). The more siblings a child has in school the lower the likelihood of starting school late. The number of siblings in school reduces the odds of delayed enrolment by 10 percent. Thus, the more children there are in school from the household the less likely a six or seven year old would experience delay in enrolment. Children living in households headed by Muslims or Christians have a lower likelihood of delayed entry compared to those living in households headed by adherents to traditional religion. The higher the mother’s and father’s education level the lower the odds of delayed entry. Compared to children whose mothers have no education, the odds of a child experiencing delayed entry reduces by 30% if the mother has basic education. In order to control for contextual differences, locality dummies are included in the estimation. Sometimes the dominant economic activity of the ecological zone may influence children’s educational outcomes. Ghana is made up of three major ecological zones: coastal, forest and savannah. Fishing is the main economic activity of especially rural coastal areas whilst farming is the main stay of forest areas; and animal rearing and farming are the main economic activities of the savannah. In the estimation, dummies are generated for each of the localities while accounting for the rural-urban divide. None of the locality or regional dummies is statistically significant. This suggests that delayed enrolment is not different across regions or localities. There are similarities and differences in the correlates of delayed enrolment among urban and rural children. As in the model with the full sample, in neither the urban nor rural samples is there a significant difference between girls and boys in the likelihood of delayed entry. Among urban children the poverty status of the household is not significantly correlated with delayed entry. The poverty status of households matter for rural children, however, where those living in non-poor households have lower odds of delayed entry compared to children living in extremely poor households. Urban children whose mothers have some basic education have twice the odds of not starting school on time compared to children whose mothers have no education whilst urban children whose fathers have completed junior high/middle school education or have completed a level higher than junior high/middle school have about 60 percent lower odds respectively
H.A. Seshie-Nasser, A.D. Oduro / International Journal of Educational Development 49 (2016) 107–114
112
Table 3 Determinants of delayed entry into primary 1. Delayed enrolment
Full sample
Urban model
Rural model
Male model
Female model
I
II
III
IV
V
0.956
1.277
0.877
0.839 1.050 0.868***
0.492 0.606 0.887
1.457 1.223 0.844***
0.772 0.900 0.867***
1.005 1.032 0.867***
Children living with Mother only Father only Neither parent
0.688 0.688 1.297
0.622 0.551 2.960**
0.777 0.797 0.905
0.697 0.481 1.472
0.649 0.910 1.308
Poverty status Poor Non-poor
0.880 0.657**
0.899 0.821
0.887 0.645**
0.697 0.540***
1.227 0.843
Religion of Household head Muslim Christian No religion
0.475*** 0.596*** 0.756
1.151 1.691 1.325
0.500*** 0.535*** 0.777
0.582** 0.659 0.900
0.394*** 0.523** 0.639
Characteristics of Parents/Guardians Employment status Mother in agriculture Mother in other sector Father in agriculture Father in other sector
0.961 1.191 0.878 0.710
0.633 0.851 1.229 0.677
1.256 1.665** 0.832 0.728
1.307 1.191 0.628 0.818
0.656 1.143 1.265 0.631
1.248 0.698* 0.353***
2.199** 1.706 0.489
1.065 0.467*** 0.400*
0.981 0.982 0.112***
1.675** 0.526** 0.906
0.794 0.730* 0.591**
0.925 0.400** 0.369**
0.773 0.895 0.680
1.095 0.856 0.959
0.538** 0.569** 0.323***
Locality Urban Coastal Urban Forest Urban Savannah Rural Coastal Rural Forest Rural Savannah
0.532 0.886 1.857 0.609 1.417 1.385
0.177 0.357 2.319 2.476*** 2.101**
0.689 1.499 2.046 0.588 2.225 1.506
0.314 0.520 1.667 0.548 0.874 1.208
Region Western Central Volta Eastern Ashanti Brong Ahafo Northern Upper East Upper West
1.400 1.773 1.186 1.004 0.854 2.012 1.048 0.687 0.729
3.297 4.774 1.716 3.660 1.535 5.261 1.368 0.186 1.116
1.160 1.360 1.092 0.744 0.760 1.541 0.821 0.723 0.638
1.436 1.270 1.672 0.745 0.663 2.160 1.375 0.643 0.852
1.631 3.400 1.040 1.491 1.108 2.482 0.927 0.811 0.725
Pseudo R2 chi2 Observations
0.101 183.9 1720
0.163 62.74 427
0.084 115.4 1293
0.105 96.18 850
0.147 145.0 870
Male Household characteristics Female head Age of head Number of siblings in school
Education Mother: Some basic education Completed JHS/M4 Above JHS Father: Some basic education Completed JHS/M4 Above JHS
* **
p < 0.1. p < 0.05. p < 0.01.
***
of delayed entry. Among rural children on the other hand having a mother who has completed junior high school or a higher level reduces the odds of delayed entry. Fathers’ education is not a significant correlate of delayed entry. The number of siblings in school and living in a household with a Muslim or Christian headreduces the odds of delayed enrolment among rural children only. Rural children whose mothers are employed in sectors other than agriculture are more likely to start school late compared to children whose mothers are not working as are children who live in the forest or savannah zone compared to
those living in the coastal zone. Dummies were created indicating whether or not a child is from a single parent home and for children living with others other than their own parents. The reference scenario is a child living with both parents and the single parent scenarios are; the child lives with ‘‘only the mother’’ or ‘‘only the father’’. The results show that urban children living with people other than their own parents have higher odds of experiencing delayed entry compared to children living with both parents. The implication is that orphan children are more likely to experience delays in enrolment.
H.A. Seshie-Nasser, A.D. Oduro / International Journal of Educational Development 49 (2016) 107–114 Table 4 Mean number of years children’s ages exceed the appropriate age (2005/06). Variable
Mean
Std. dev.
Min
Max
All children Rural Urban Boys Girls
1.23 1.42 0.88 1.36 1.09
1.807 1.921 1.521 1.916 1.667
0 0 0 0 0
20 20 13 13 20
Even though there is no significant difference in the odds of delayed entry between girls and boys the factors associated with the likelihood of delayed entry into primary school are not identical. The relationship between the child’s household’s poverty status and delayed entry is not the same for girls and boys. The child’s household’s poverty status is not significant in differentiating between girls who are likely to be enrolled in school late and those who will be enrolled on time. Among boys, however, those living in non-poor households are significantly less likely than those living in very poor households to be enrolled in school at an inappropriate age. Among both girls and boys an increase in the number of siblings in school reduces the odds of delayed entry. Compared to living in a household headed by an adherent to traditional religion, both boys and girls living in households headed by Muslims have lower odds of delayed entry. Boys are less likely to start school late when their mothers have attained more than junior high school education. The importance of mother’s education plays out differently for girls. Girls whose mothers have some basic education have 67 percent higher odds of starting school late. On the other hand girls whose mothers have completed junior high school education are less likely to experience delayed enrolment compared to girls whose mothers have no education. Fathers’ education does not matter for boys. For girls, on the other hand, having an educated father reduces the odds of delayed entry and the odds are much lower when their fathers have completed an education level higher than junior high school. 7. Correlates of number of years children’s age exceeds the appropriate age for their grade
113
Table 5 Factors associated with the number of years the ages of children exceed the appropriate age.
Male Number of siblings
Full sample
Urban model
Rural model
Male model
Female model
0.25*** 0.01
0.12** 0.03
0.32*** 0.01
0.00
0.02
0.10 0.29*** Yes
0.23 0.22 Yes
0.07 0.30*** Yes
0.22 0.44*** Yes
0.07 0.10 Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
a
Poverty status Poor Non-poor Household characteristics Religion of household head Parents’/guardians’ Characteristics Urban, rural locality Region Constant
Yes
Yes
Yes
Yes
Yes
Yes 3.37***
Yes 2.26***
Yes 3.18***
Yes 3.42***
Yes 3.12***
F R-squared Observations
19.44 0.11 7798
9.02 0.14 2514
10.91 0.09 5284
14.91 0.13 4129
9.06 0.10 3669
a
The reference category is ‘‘extremely poor’’ *p < 0.1, **p < 0.05, ***p < 0.01.
In the model with the full sample and in both the urban and rural models boys are significantly older than the appropriate age for their grade than girls. Children living in non-poor households are fewer years above the appropriate age than children from very poor households. This relationship between poverty status and number of years above the age appropriate for the grade is driven by rural households. The coefficient of the dummy variable for non-poor households is significant and negative in the rural model and insignificant in the urban model. The poverty status of the household is not significant in explaining the number of years above the age appropriate for the grade among girls but matters for boys where boys from non-poor households have fewer years above the appropriate age than boys living in very poor households. 8. Discussion
The previous section has identified the correlates associated with the likelihood of delayed entry into primary school. This section explores the factors associated with the number of years the ages of children in school exceeds the appropriate age for their grade. Table 4 presents information on the mean number of years the ages of children in primary 1 through to the third year of junior high school exceeds the age appropriate for their grade. Children are on average 1.23 years older than the appropriate age for their grade with the maximum number of years by which their age exceeds the appropriate age being 20. The mean for urban children is lower than that for rural children and the mean for girls is lower than that for boys. Table 5 presents the findings for the model that explores the factors associated with the number of years a child’s age exceeds the appropriate age. It presents the results for the full sample, urban and rural samples and the samples for girls and boys.2 The dependent variable is the difference between the age of the pupil and the appropriate age for the grade. The sample comprises pupils in primary 1 to JHS 3 (i.e. grades 1–9). The same set of explanatory variables employed in the delayed entry model are utilised here. The discussion will concentrate on two issues of interest. These are whether the number of years above the appropriate age for the grade is the same for boys and girls and the importance of the poverty status of the households the children live in. 2
The results for the full model can be obtained from the authors on request.
An important finding of this paper is that girls are no more likely than boys to be sent to school late. Glewwe and Jacoby (1995) in their earlier study on Ghana obtained a similar result. This finding is not entirely unexpected since primary gross enrolment rates of girls and boys have been increasing in tandem over time and gender parity in primary education has almost been achieved. A second important finding is that the poverty status of the household matters for whether a child will be sent to school on time. The insignificance of the sex dummy in the delayed entry model notwithstanding, the phenomenon of delayed entry in Ghana does have some gender dimensions. First is the finding that the mean of the difference between actual age and appropriate age for the grade is higher for boys than for girls. Thus even though both boys and girls are just as likely not to start primary school at the official age, boys are probably kept out of school longer, or are more likely to repeat a class or spend some time out of school and then reenroll at a later period. Data collected during the 2008 Demographic and Health Survey shows that boys are more likely to be stunted (they do not have the required height for their age) than girls (Ghana Statistical Service et al., 2009). Thus if parents’ decision on whether a child is ready to be sent to school is informed by the child’s physical development (see Glewwe and Jacoby, 1995) this may be one reason why the mean of the difference between actual age and the appropriate age for the grade is higher
114
H.A. Seshie-Nasser, A.D. Oduro / International Journal of Educational Development 49 (2016) 107–114
for boys. Sensitisation messages that contain information on the benefits of sending all children to school at the appropriate age is required. If indeed the child’s physical development is an important factor in parents’ considerations this suggests that policy measures such as bussing young children to school and sitting primary schools closer to communities will be required to ensure that all children irrespective of their gender and urban or rural location are sent to school when they are six years old. The second gender dimension is the differences in the correlates of delayed entry between girls and boys. Of particular interest is nature of the relationship between the poverty status of the child’s household and delayed entry of girls and boys. The significance of the household poverty status variable for boys and not for girls suggests that the relationship between the poverty status of the household and children’s enrolment depends on the sex of the child, the age of the child and/or the stage of the child in education system. This study finds that the poverty status of the household may not be important for whether girls start school on time, however it matters for whether they complete primary school and transition to junior and senior high school (Nguyen and Wodon, 2014). Further research is required to unearth the reasons why the poverty status of the households matters for the enrolment of boys on time but is not important for girls. In addition to the variables of interest other variables have been found to be associated with delayed entry. Parent’s education has emerged as an important correlate of delayed entry. This study confirms that some level of formal education on the part of the parents reduces the odds of them sending their children to school late. For both boys and girls, mothers’ education is important and the more educated mothers are the better. Father’s education is important and the higher the level of education achieved the lower the odds of delayed entry, particularly among girls. Thus policies that succeed in increasing junior high school completion rates of both boys and girls have positive spillover effects for future generations. These graduates are more likely to ensure that the next generation will start school on time. Adult learning programs that are targeted at school drop outs, and are designed to make it easy to enroll for classes to obtain the Basic Education Certificate and the Senior High School Certificate whilst working may reduce the time frame within which the problem of delayed entry will be overcome. Even though enrolling a large number of children in school can spread a household’s resources thinly, the number of children enrolled may be indicative of the premium placed on education by the household. The premium a household places on education as measured by the number of children in school significantly reduces the likelihood of delayed entry in all the models except for the urban model. Thus, a sensitisation policy that educates parents, especially in rural areas on the benefits of sending their children to school would have a positive ripple effect for the younger children that follow, since it would increase the likelihood that the six and seven year olds who follow are enrolled on time. Urban children who do not live with their parents are more at risk of delayed entry. Further research is required to understand why this category of children are not sent to school at the appropriate age. 9. Conclusion Despite the high gross and net primary enrolment rates, a not insignificant proportion of children do not start primary school
when they are six years old. Girls aged six or seven years old are no more likely to experience delayed entry than boys of the same age. However, there is evidence that boys are much older than the appropriate age for their grade than girls suggesting that even though they are just as likely as girls to experience delayed entry they may be sent to school at an older age than girls. The poverty status of the household matters for whether boys will be sent to school on time but is not important for girls. Policy to address gender inequality within education has tended to focus on the girl child. The findings of this paper reveal that gender inequality within education is quite nuanced. Therefore policies must be designed that take cognisance of the different needs of boys and girls and the different constraints they face if both boys and girls are to start school on time and complete primary school.
References Ainsworth, M., Beegle, K., Koda, G., 2005. The impact of adult mortality and parental deaths on primary schooling in North-Western Tanzania. J. Dev. Stud. 41, 412–439, http://dx.doi.org/10.1080/0022038042000313318. Akyeampong, K., 2011. (Re) Assessing the Impact of School Capitation Grants on Education Access in Ghana (CREATE Pathways to Access Research Monograph No. 71) Center for International Education, University of Sussex, Brighton, UK. Akyeampong, K., Djangmah, J., Oduro, A., Seidu, A., Hunt, F., 2007. Access to Basic Education in Ghana: The Evidence and the Issues. Aryeetey, E., Harrigan, J., Machinko, N. (Eds.), 2000. Economic Reforms in Ghana: The Miracle and Mirage. James Currey and Woeli Publishers, Oxford. Bommier, A., Lambert, S., 2000. Education demand and age at school enrollment in Tanzania. J. Hum. Resour. 35, 177–203, http://dx.doi.org/10.2307/146360. Burke, K., Beegle, K., 2004. Why children aren’t attending school: the case of Northwestern Tanzania. J. Afr. Econ. 13, 333–355, http://dx.doi.org/10.1093/ jae/ejh011. Ghana Statistical Service, 2007. Patterns and Trends of Poverty in Ghana: 1991– 2006. Ghana Statistical Service. Glewwe, P., Jacoby, H.G.H.G., 1995. An economic analysis of delayed primary school enrollment in a low income country: the role of early childhood nutrition. Rev. Econ. Stat. 77, 156–169, http://dx.doi.org/10.2307/2110001. Glick, P., Sahn, D.E., 2000. Schooling of girls and boys in a West African country: the effects of parental education, income, and household structure. Econ. Educ. Rev., http://dx.doi.org/10.1016/S0272-7757(99)00029-1. Grira, H., 2001. Delayed School Enrolment: Evidence from Rural Bangladesh [WWW Document]. http://dse.univr.it/espe/documents/Papers/E/2/E2_1.pdf (accessed 08.04.15). Grogan, L., 2009. Universal primary education and school entry in Uganda. J. Afr. Econ. 18, 183–211, http://dx.doi.org/10.1093/jae/ejn015. Lincove, J.A., 2009. Determinants of schooling for boys and girls in Nigeria under a policy of free primary education. Econ. Educ. Rev. 28, 474–484, http:// dx.doi.org/10.1016/j.econedurev.2008.10.001. Ministry of Education, 2013. 2012/2013 Education Profile. Moyi, P., 2011. Delayed School Entry in Uganda. Res. Comp. Int. Educ. 6, 222, http://dx.doi.org/10.2304/rcie.2011.6.2.222. Moyi, P., 2010. Household characteristics and delayed school enrollment in Malawi. Int. J. Educ. Dev. 30, 236–242, http://dx.doi.org/10.1016/ j.ijedudev.2009.11.008. Nguyen, M.C., Wodon, Q., 2014. Analysing the gender gap in education attainment: a simple framework with application to Ghana. J. Int. Dev. 26, 59–76, http://dx.doi.org/10.1002/jid. Osei-Fosu, A., 2011. Evaluating the impact of the capitation grant and the school feeding programme on enrollment, attendance and retention in schools: the case of Weweso circuit. J. Sci. Technol. 31 , http://dx.doi.org/10.4314/ just.v31i1.64886. Pradhan, M., 1998. Enrolment and delayed enrolment of secondary school age children in Indonesia. Oxf. Bull. Econ. Stat. 60, 413–430, http://dx.doi.org/ 10.1111/1468-0084.00107. Rose, P., Al-Samarrai, S., 2001. Household constraints on schooling by gender: empirical evidence from Ethiopia. Comp. Educ. Rev. 45, 36–63, http:// dx.doi.org/10.1086/447644. Wells, R., 2009. Gender and age-appropriate enrolment in Uganda. Int. J. Educ. Res. 48, 40–50, http://dx.doi.org/10.1016/j.ijer.2009.03.002. Wils, A., 2004. Late entrants leave school earlier: evidence from Mozambique. Int. Rev. Educ. 50, 17–37, http://dx.doi.org/10.1023/B:REVI.0000018201.53675.4b.
Contents lists available at ScienceDirect
International Journal of Educational Development journal homepage: www.elsevier.com/locate/ijedudev
Delayed primary school enrolment among boys and girls in Ghana Hellen A. Seshie-Nasser 1,*, Abena D. Oduro University of Ghana, Department of Economics, P.O. Box LG57, Legon, Accra, Ghana
A R T I C L E I N F O
A B S T R A C T
Article history: Received 12 January 2015 Received in revised form 19 December 2015 Accepted 25 December 2015
Using the fifth Ghana Living Standards Survey, this paper examines whether delayed entry is as likely among girls as it is among boys and the importance of the household’s poverty status in understanding delayed entry. No gender difference is found in the likelihood of delayed entry. However boys are much older for their grades than girls suggesting that they may experience a more extended period of delayed entry. The poverty status of the household is not an important correlate of delayed entry among girls or children in urban households but is important for boys and rural children. ß 2016 Published by Elsevier Ltd.
JEL classification: I21 J24 Keywords: Delayed enrolment Age-appropriate enrolment Primary education Ghana
1. Introduction Ghana has achieved significant strides in the provision of basic education for all children, however, the sector is still bedeviled with drawbacks and limitations (Akyeampong et al., 2007). Although gross primary enrolment rates exceed 100 percent, net enrolment rates were about 84 percent in 2013 (Ministry of Education, 2013). Net primary enrolment rates of less than 100 percent indicate that the goal of universal primary education has not been achieved whilst gross enrolment rates of more than 100 percent suggest that there are overaged children in primary school. The presence of overaged children in primary and secondary school can be due to one or a combination of reasons. These are failure of children to be enrolled in primary school at the official entry age which is usually six years old, class repetition or interrupted enrolment in school. Age-grade inappropriateness increases the risk of a child dropping out of school. Wils (2004) found that in Mozambique children who enter school late had substantially higher drop-out rates than younger entrants. In Ghana, Akyeampong et al. (2007) found that whilst enrolment
rates increase rapidly for children between the ages of 8 and 13, there is a decline in enrolment rates among children between the ages of 13 and 16. When girls enroll in primary school later than the official entry age, they are more likely to drop out, especially as they approach adolescence (Rose and Al-Samarrai, 2001; Wils, 2004). Delayed entry into primary school can make the attainment of the goal of universal primary education difficult for two reasons. First, not all children at the official entry age for primary school will be enrolled in school. Second, delayed entry can increase the likelihood of non-completion of primary education. Not many studies on education in Ghana have examined the incidence of age-appropriate grade enrolment or the determinants of age-appropriate grade enrolment. This paper intends to contribute to the literature on this important subject. It has three objectives. The first is to find out whether girls are just as likely as boys to experience delayed entry. The second is to identify the correlates of delayed entry for girls and boys and the importance of socio-economic circumstances of the household. The third objective is to examine the factors associated with the extent to which a child is overaged for his or her class. 2. Evidence on the determinants of age-appropriate enrolment
* Corresponding author. Current address: 237 Pennwood Street, Newark, DE 19713, USA. E-mail addresses: [email protected], [email protected] (H.A. Seshie-Nasser), [email protected] (A.D. Oduro). 1 Current address: This author is currently with University of Delaware. http://dx.doi.org/10.1016/j.ijedudev.2015.12.004 0738-0593/ß 2016 Published by Elsevier Ltd.
A few studies have examined the incidence of age-appropriate grade enrolment in the African context (Burke and Beegle, 2004; Moyi, 2011, 2010; Wells, 2009) and more particularly in Ghana (Akyeampong et al., 2007; Glewwe and Jacoby, 1995). Using data
108
H.A. Seshie-Nasser, A.D. Oduro / International Journal of Educational Development 49 (2016) 107–114
from the Core Welfare Indicators Survey of 2003, Akyeampong et al. (2007) found that the mean ages of pupils enrolled in primary 1 and primary 6 were 7.5 years and 13.3 years respectively. In some instances children aged ten years or more were in primary 1 or 2. They attributed this to delayed entry into primary school. Generally, enrolment for boys is higher than for girls in the African context (Glick and Sahn, 2000; Lincove, 2009). There is however, mixed evidence on the relationship between the sex of the child and whether or not the child would be enrolled at the right age. In Tanzania, an interesting revelation by Bommier and Lambert (2000) is that, girls start school earlier than boys. A possible reason assigned to this is the lump-sum gains made after receiving education, such as the bride-price, when the young lady is ready for marriage. These gains are modeled as a lump-sum gain and it was found that when these gains increase girls enroll earlier and stay in school for a shorter period. The intuition is that the present value of such a lump-sum gain is greater when obtained sooner. The parents’ interest is in having their daughters ready for the marriage market as soon as possible. Moyi (2010) on the other hand did not find a significant difference in the incidence of delayed entry to primary school between boys and girls in Malawi. Grogan (2009) found that fee elimination in Uganda increases the probability that girls begin school before age nine. Pradhan (1998) found that boys from wealthy households in Indonesia have a higher probability of being enrolled at the appropriate age, though the difference was minimal. The income or wealth of the household has been found to be significant determinants of age at enrolment. The probability of inappropriate age-grade enrolment declines as the income or wealth of the household increases (Glewwe and Jacoby, 1995; Moyi, 2010; Pradhan, 1998). In Indonesia, consumption per capita is found to increase the probability of enrolling in secondary school at the right age (Pradhan, 1998). Facing credit constraints, parents may choose to send children to work instead of enrolling them in school (especially in rural agricultural communities) in order to build up savings (Glewwe and Jacoby, 1995). Children in poor households may experience delayed entry for a longer period because their parents or guardians may take a longer time to generate the required savings. There are mixed findings regarding the relationship between the education of the household head and the age at enrolment of children in the household. While Moyi (2010) found a negative association in Malawi between the education of the head of the household and the likelihood of a child enrolling in school late, Pradhan (1998) in a model that included both parents’ education and literacy found that parents’ education was insignificant but literacy of the mother increased the probability of enrolling at the appropriate age. The number of children currently in school from a given household or family have been found to be negatively associated with the likelihood of age-appropriate enrolment (Wells, 2009). There are mixed findings concerning the relationship a child’s birth order and the likelihood of starting school on time. Pradhan (1998) found that being the eldest child increases age at secondary enrolment in urban areas and being the youngest child in the household reduces the age at enrolment. However, the existence of liquidity constraints can change this relationship. Bommier and Lambert (2000) found that in Tanzania being the eldest in a liquidity-constrained family could reduce the chances of being enrolled on time because these children have to help accumulate funds before they enroll. On the other hand, the youngest child is more likely to enroll early because funds are more easily available. Besides, the younger ones would have access to school materials and logistics handed down to them by their older siblings. Moyi (2010) has shown that in the Malawian context, children living in households headed by women are more likely to enroll
early. In addition, the death of a mother is less injurious than the death of a father to the likelihood of being enrolled in school on time. Ainsworth et al. (2005) also found a similar relationship between mother’s survival and delayed enrolment among children in north-western Tanzania. In addition to the characteristics of the household and the family the child belongs to other factors have been found to be associated with delayed enrolment. Bommier and Lambert (2000) found that the age at the time of school enrolment and schooling duration in Tanzania are both functions of a set of variables describing schooling costs and school quality. Glewwe and Jacoby (1995) focused attention on the effect of childhood nutrition on age of enrolment and school attainment in Ghana, while controlling for other community, household and individual characteristics. Strong evidence was found for nutritional deficiencies in early childhood; children with stunted growth start primary school later. Using more recent data (2005/06 Ghana Living Standards Survey), this paper investigates whether girls and boys are just as likely to start primary school late and the importance of socioeconomic correlates of delayed enrolment at the entry level. A revisit of the issue on delayed entry is pertinent given the introduction of policies to improve access to basic education in the intervening period since the Glewwe and Jacoby study and the implications delayed entry has for the attainment of the goal of universal primary education. 3. Ghana’s education system The education reforms of 1987 introduced fundamental changes to Ghana’s education system. The reforms were introduced as part of broader economic and structural reforms implemented under the aegis of the World Bank and IMF beginning in 1983 (see contributions in Aryeetey et al., 2000 for a discussion of the reforms). The structure of the education system was streamlined and the number of years of pre-tertiary education was reduced. The new structure introduced the basic education system which comprised six years of primary school and three years of junior high school. Basic education was subsequently redefined to include two years of pre-school. The official age of entry into primary 1 is six years. The completion of basic education is the first official transition point in the education system. All students sit for the official terminal examination that either marks the end of formal schooling or is the transition to senior high school education. The reforms of 1987 aimed at increasing access to education for both boys and girls and across income groups, changing the structure of the school system, improving the quality of education through improvements in pedagogy and the introduction of a more relevant curriculum, attaining financial sustainability and improving the management of resources within the sector. User fees were introduced to bring about financial sustainability. This however had an almost immediate detrimental effect on enrolment which was contrary to the objective of improving access. In the face of this the user fee policy had to be revised. The policy of free, compulsory and universal basic education was introduced in 1995 and had as one of its objectives the provision of free quality education at the basic level to all children by 2005. The components of education expenses that were to be provided free were tuition, textbooks, equipment and tools. At the junior high level, pupils were expected to pay a textbook user fee equivalent to not more than 10 percent of the average total cost of the textbook supplied per pupil. In 2004/05 the capitation grant scheme was introduced. The grant covers the extra cost and levies (such as examination, facilities management, security charges, games and sports) that parents usually pay as ‘‘school fees’’ in public schools. The National
H.A. Seshie-Nasser, A.D. Oduro / International Journal of Educational Development 49 (2016) 107–114
School Feeding Program was introduced in the 2005/06 academic year. Children in schools in deprived districts are provided with one meal a day. Osei-Fosu (2011) found that the capitation grant caused an immediate increase in enrolment and the school feeding program had a large positive and significant impact on school enrolment, attendance and retention. 4. Methodology A binary choice model is utilised to estimate the determinants of delayed enrolment and the ordinary least squares method is employed to explain the factors associated with the number of years a child is overaged. The dependent variable for the delayed enrolment model takes the value of 1 if a child in the sample of six and seven year olds is either not in school or is in pre-school. It takes the value of 0 if a child is in primary 1 or a higher grade. In a variant of binary estimation, an index function formulation explains an underlying unobserved continuous random variable y*, such as the desire to delay enrolment in school. This desire is unobservable. What is observed is the binary variable y, which takes the value 1 or 0 depending on whether or not y* crosses a threshold, say a, as follows:
Yi ¼
1 ðenrolment in school is delayedÞ if y > a 0 ðenrolment is not delayedÞ if y a
Employing a logistic distribution, the probability of delayed enrolment is expressed as: Pr½yi ¼ 1jxi ¼
exp ðx0i bÞ 1 þ exp ðx0i bÞ
(1)
where xi represents the vector of explanatory variables and b represents the vector of parameters. The empirical model in an odds ratio format is given as: pi ¼ exp ðx0i bÞ 1pi
(2)
where pi is the probability of delayed enrolment. To obtain a profile of girls and boys who are likely to experience delayed entry separate logistic regressions are run for the sample of boys and girls. Separate regressions are also run for the urban and rural samples. The second model estimates the correlates of the number of years a child’s age exceeds the appropriate age for the grade s/he is enrolled in. This second model is intended to shed some more light on the phenomenon of delayed entry. Ideally, longitudinal data is preferable to cross-sectional data for a study on the likelihood of delayed enrolment. This is because conditions of the household at the time a child is supposed to be enrolled into the first year of primary school may be different from conditions prevailing at the time the data is collected. In the absence of longitudinal data, this difficulty may be overcome by restricting the analysis to the sample of young children of entry level age. Hence, this study concentrates on the sample of six and seven year olds. Another reason for investigating the likelihood of inappropriate age-grade enrolment among this sample of children is because once a child does not enter primary school at the appropriate age the phenomenon of being overaged for one’s grade is likely to persist throughout the trajectory from basic to secondary education. We make the assumption that six and seven year olds who are not currently enrolled in primary school will eventually be enrolled. This assumption can safely be made because of the upward trends in gross primary enrolments and a gross primary
109
enrolment rate which exceeded 100 percent in the 2012/13 academic year (Ministry of Education, 2013). 4.1. Data This study utilises data from the fifth round of the Ghana Living Standards Survey (GLSS) conducted between September 2005 and September 2006. This is a nationally representative household survey that collected data from over 8500 households and about 37,000 household members. The fifth round of the GLSS has a sample of 2290 children aged six or seven years of whom 1140 are boys and 1150 are girls. About 56 percent of the sample of six and seven year olds comprises six year olds and the remaining 44.2 percent are seven year olds. Even though children are expected to start primary school when they are six years old, in some cases they enter at age seven due to date of birth considerations. Most seven year olds however are expected to be in primary 2. This then implies that, six and some seven year olds who are not currently in primary 1 are experiencing delay in enrolment and some seven year olds who are currently not in primary 2 are also enrolled late. In order to reduce the likelihood of misclassification the time of interview was taken into consideration when determining whether or not a child was enrolled in school late. For example a primary 1 seven year-old child surveyed in August 2005 may be wrongly considered to be experiencing delayed enrolment at the time of interview when in actual fact s/he would be entering primary 2 in September (just one month away). All seven year olds whose households were surveyed in September and were reported to be in primary 1 or pre-school were classified as being in an inappropriate grade for their age. About 29.5 percent of the six and seven year olds are in primary 1 and about 12.3 percent are in primary 2. About 40.8 percent are either in preschool or not enrolled, that is they are delayed entrants in primary 1 based on our working definition. A few children (3.6%) are in grades higher than primary 2 or in non-formal education, such as Islamic schools. 4.2. Explanatory variables The variables of interest are the sex of the child and the poverty status of the household the child is a member of. We hypothesise that girls are more likely to start primary school late because they may be considered more vulnerable. Parents would want to send them to school at an age when they are more articulate and less vulnerable. When there are scarce resources, parents will more likely delay girls’ entry. We expect the likelihood of delayed entry to be higher among the very poor compared to poor and non-poor households and to be higher among the poor compared to nonpoor households. The choice of additional explanatory control variables is informed by the literature and comprises the characteristics of the child’s parents or guardians and the household the child lives in. Household characteristics include the sex, age and religion of the household head, the number of siblings in school, location of the household, administrative region of residence of the household, whether or not the child lives with both, either or neither of its parents. The characteristics of the parents include employment status and level of education completed. There are other variables that are expected to influence age at enrolment, but are not included in the model because they are not found in the data set. For instance, the model does not include disability or other health-related factors, the nutritional status of the child or distance to the nearest school. The description of the variables and summary statistics are presented in Table 1. The sample is almost evenly distributed between girls and boys. The majority of children reside in rural households (71%) and about 65 percent live with both parents
110
H.A. Seshie-Nasser, A.D. Oduro / International Journal of Educational Development 49 (2016) 107–114
Table 1 Variable definitions and summary statistics. Variable
Description
Mean
Std. dev.
Min
Max
Male Female head Age_head Rural dummy Number of siblings
Sex of the child = 1 if male Sex of the household head = 1 if female Age of the household head Household located in a rural area Number of siblings in school
0.496 0.209 45.710 0.712 1.867
0.500 0.407 12.983 0.453 1.991
0 0 20 0 0
1 1 99 1 15
0.652 0.159 0.046 0.142
0.476 0.365 0.210 0.349
0 0 0 0
1 1 1 1
0.280
0.449
0
1
0.126
0.332
0
1
0.594
0.491
0
1
0.492
0.500
0
1
0.366
0.482
0
1
0.634
0.482
0
1
0.269
0.444
0
1
0.546 0.224
0.498 0.417
0 0
1 1
0.179 0.051
0.384 0.220
0 0
1 1
0.444 0.175 0.271 0.110
0.497 0.380 0.444 0.313
0 0 0 0
1 1 1 1
0.213
0.410
0
1
0.542
0.498
0
1
0.174
0.379
0
1
0.071
0.257
0
1
0.087 0.067 0.083 0.080 0.091 0.168 0.097 0.137 0.091 0.100
0.282 0.250 0.275 0.272 0.288 0.374 0.296 0.344 0.288 0.300
0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 1 1 1
Children living with parents: Reference = both parents Child living with both parents Both parents Mother only Child living with single mother Father only Child living with single father Neither Child living with non-parents Poverty status: Reference = extremely poora: Extremely poor Income/expenditure below the lower poverty line Income/expenditure between the upper and Poor lower poverty lines. Income/expenditure above the poverty line Non Poor Employment status of parents/guardians: Reference = not working Mother/female guardian works i Mother in agriculture n agricultural sector Mother/female guardian works Mother in other sector in other sector Father/male guardian works Father in agriculture in agricultural sector Father/male guardian works in other sector Father in other sector Mother: Reference = no education No formal education Some basic education Completed JHS/M4 Above JHS
No formal education At least primary but not completed junior high/middle school Completed JHS/M4 Education above JHS/M4
Father: Reference = no education No formal education Some basic education Completed JHS/M4 Above JHS Religion dummies: Reference = traditional religion Muslim Equals 1 if the household head is a Muslim Equals 1 if the household head Christian is a Christian Equals 1 if household head Traditional is of traditional religion Equals 1 if household head No religion adheres to no religion Administrative region dummies: Reference = greater Accra Western Central Greater Accra Volta Eastern Ashanti Brong Ahafo Northern Upper East Upper West
Source: GLSS5. a The poverty lines are constructed using household consumption expenditure per adult equivalent. For a discussion on the construction of the poverty lines see Ghana Statistical Service (2007).
whilst about 5 percent live with their fathers only. About 21 percent of children live in a household headed by a woman. The proportion of children whose fathers are employed in agriculture (63%) exceeds the proportion of children whose mothers are employed in that sector (49%). About 55 percent of children have mothers who do not have formal education whilst 44 percent of children have fathers who fall in this education category.
5. Incidence of age-grade appropriateness in Ghana Table 2 presents evidence on the proportions of children aged 6–15 years in appropriate and inappropriate grades using data from the fifth GLSS. The proportion of children in the appropriate age for their grade declines as the age cohort increases. Among six year olds, only 75 percent are enrolled; and out of that number 40 percent are not in the appropriate grade, i.e. primary 1. The
H.A. Seshie-Nasser, A.D. Oduro / International Journal of Educational Development 49 (2016) 107–114
111
Table 2 Incidence of age-grade inappropriateness for children aged 6–15 (2005/06). Age cohorts
Never enrolled
Right grade
Inappropriate grade
Total
% in Inappropriate Grade
% Right grade
6 7 8 9 10 11 12 13 14 15
307 179 160 148 149 93 146 106 120 159
434 608 438 361 350 257 256 253 205 168
496 215 381 426 549 464 583 579 521 570
1240 1003 980 939 1048 815 987 941 869 927
40.0% 21.4% 38.9% 45.4% 52.4% 56.9% 59.1% 61.5% 60.0% 61.5%
35.0% 60.6% 44.7% 38.4% 33.4% 31.5% 25.9% 26.9% 23.6% 18.1%
Total
9749
Source: Authors’ calculations from GLSS5.
percentage (21.4%) of 7 year olds who are overage for their grade is less than that of the six year olds. However, the incidence increases from the 8 year olds reaching 61.5 percent for 15 year olds. Even though part of the problem of inappropriate age-grade enrolment may be due to grade repetition or temporary withdrawals along the educational ladder, class repetition is quite low in Ghana, especially in the urban schools. It is also not common for Ghanaian students to temporarily withdraw from school (Glewwe and Jacoby, 1995). Another way to present the evidence on delayed entry is to calculate the share of children in a particular grade who are older than the appropriate age. Chart 1 provides evidence using data from the third (1991/92) and fifth (2005/06) Ghana Living Standard Surveys. Two interesting observations can be made. First, in both periods the proportion of children in a grade who are older than the appropriate age increases from primary 1 to the second year of junior high school. This suggests that in more recent years there has been an improvement in the proportion of children who are enrolled in school when they are six years old. Second, the incidence of age-inappropriateness is higher in 2005/06. This would seem to contradict the previous observation. The higher incidence of overage children in each grade in 2005/06 compared to 1991/92 is probably due to the introduction of the capitation grant and school feeding program which created an incentive for enrolment of out of school children may of whom were overage (Akyeampong, 2011).
6. The factors associated with delayed entry Table 3 presents the results of the logistic regression that investigates the correlates of delayed entry for the full sample, the 80.00% 70.00% 60.00% 50.00% 40.00% 30.00% 20.00%
1991/92 2005/06
10.00% 0.00%
Chart 1. Incidence of overage children in basic education (1991/92 and 2005/06).
urban and rural samples and the samples of girls and boys. Odds ratios are reported and the standard errors are corrected for clustering of children in the household. In the model with the full sample of six and seven year olds there is no significant difference between boys and girls in the likelihood of delayed entry. The odds of a non-poor child enrolling late is about 35% lower, compared to children from extremely poor households. Similar results are found in age-grade appropriateness literature (Pradhan, 1998; Grira, 2001; Moyi, 2010; Wells, 2009). The more siblings a child has in school the lower the likelihood of starting school late. The number of siblings in school reduces the odds of delayed enrolment by 10 percent. Thus, the more children there are in school from the household the less likely a six or seven year old would experience delay in enrolment. Children living in households headed by Muslims or Christians have a lower likelihood of delayed entry compared to those living in households headed by adherents to traditional religion. The higher the mother’s and father’s education level the lower the odds of delayed entry. Compared to children whose mothers have no education, the odds of a child experiencing delayed entry reduces by 30% if the mother has basic education. In order to control for contextual differences, locality dummies are included in the estimation. Sometimes the dominant economic activity of the ecological zone may influence children’s educational outcomes. Ghana is made up of three major ecological zones: coastal, forest and savannah. Fishing is the main economic activity of especially rural coastal areas whilst farming is the main stay of forest areas; and animal rearing and farming are the main economic activities of the savannah. In the estimation, dummies are generated for each of the localities while accounting for the rural-urban divide. None of the locality or regional dummies is statistically significant. This suggests that delayed enrolment is not different across regions or localities. There are similarities and differences in the correlates of delayed enrolment among urban and rural children. As in the model with the full sample, in neither the urban nor rural samples is there a significant difference between girls and boys in the likelihood of delayed entry. Among urban children the poverty status of the household is not significantly correlated with delayed entry. The poverty status of households matter for rural children, however, where those living in non-poor households have lower odds of delayed entry compared to children living in extremely poor households. Urban children whose mothers have some basic education have twice the odds of not starting school on time compared to children whose mothers have no education whilst urban children whose fathers have completed junior high/middle school education or have completed a level higher than junior high/middle school have about 60 percent lower odds respectively
H.A. Seshie-Nasser, A.D. Oduro / International Journal of Educational Development 49 (2016) 107–114
112
Table 3 Determinants of delayed entry into primary 1. Delayed enrolment
Full sample
Urban model
Rural model
Male model
Female model
I
II
III
IV
V
0.956
1.277
0.877
0.839 1.050 0.868***
0.492 0.606 0.887
1.457 1.223 0.844***
0.772 0.900 0.867***
1.005 1.032 0.867***
Children living with Mother only Father only Neither parent
0.688 0.688 1.297
0.622 0.551 2.960**
0.777 0.797 0.905
0.697 0.481 1.472
0.649 0.910 1.308
Poverty status Poor Non-poor
0.880 0.657**
0.899 0.821
0.887 0.645**
0.697 0.540***
1.227 0.843
Religion of Household head Muslim Christian No religion
0.475*** 0.596*** 0.756
1.151 1.691 1.325
0.500*** 0.535*** 0.777
0.582** 0.659 0.900
0.394*** 0.523** 0.639
Characteristics of Parents/Guardians Employment status Mother in agriculture Mother in other sector Father in agriculture Father in other sector
0.961 1.191 0.878 0.710
0.633 0.851 1.229 0.677
1.256 1.665** 0.832 0.728
1.307 1.191 0.628 0.818
0.656 1.143 1.265 0.631
1.248 0.698* 0.353***
2.199** 1.706 0.489
1.065 0.467*** 0.400*
0.981 0.982 0.112***
1.675** 0.526** 0.906
0.794 0.730* 0.591**
0.925 0.400** 0.369**
0.773 0.895 0.680
1.095 0.856 0.959
0.538** 0.569** 0.323***
Locality Urban Coastal Urban Forest Urban Savannah Rural Coastal Rural Forest Rural Savannah
0.532 0.886 1.857 0.609 1.417 1.385
0.177 0.357 2.319 2.476*** 2.101**
0.689 1.499 2.046 0.588 2.225 1.506
0.314 0.520 1.667 0.548 0.874 1.208
Region Western Central Volta Eastern Ashanti Brong Ahafo Northern Upper East Upper West
1.400 1.773 1.186 1.004 0.854 2.012 1.048 0.687 0.729
3.297 4.774 1.716 3.660 1.535 5.261 1.368 0.186 1.116
1.160 1.360 1.092 0.744 0.760 1.541 0.821 0.723 0.638
1.436 1.270 1.672 0.745 0.663 2.160 1.375 0.643 0.852
1.631 3.400 1.040 1.491 1.108 2.482 0.927 0.811 0.725
Pseudo R2 chi2 Observations
0.101 183.9 1720
0.163 62.74 427
0.084 115.4 1293
0.105 96.18 850
0.147 145.0 870
Male Household characteristics Female head Age of head Number of siblings in school
Education Mother: Some basic education Completed JHS/M4 Above JHS Father: Some basic education Completed JHS/M4 Above JHS
* **
p < 0.1. p < 0.05. p < 0.01.
***
of delayed entry. Among rural children on the other hand having a mother who has completed junior high school or a higher level reduces the odds of delayed entry. Fathers’ education is not a significant correlate of delayed entry. The number of siblings in school and living in a household with a Muslim or Christian headreduces the odds of delayed enrolment among rural children only. Rural children whose mothers are employed in sectors other than agriculture are more likely to start school late compared to children whose mothers are not working as are children who live in the forest or savannah zone compared to
those living in the coastal zone. Dummies were created indicating whether or not a child is from a single parent home and for children living with others other than their own parents. The reference scenario is a child living with both parents and the single parent scenarios are; the child lives with ‘‘only the mother’’ or ‘‘only the father’’. The results show that urban children living with people other than their own parents have higher odds of experiencing delayed entry compared to children living with both parents. The implication is that orphan children are more likely to experience delays in enrolment.
H.A. Seshie-Nasser, A.D. Oduro / International Journal of Educational Development 49 (2016) 107–114 Table 4 Mean number of years children’s ages exceed the appropriate age (2005/06). Variable
Mean
Std. dev.
Min
Max
All children Rural Urban Boys Girls
1.23 1.42 0.88 1.36 1.09
1.807 1.921 1.521 1.916 1.667
0 0 0 0 0
20 20 13 13 20
Even though there is no significant difference in the odds of delayed entry between girls and boys the factors associated with the likelihood of delayed entry into primary school are not identical. The relationship between the child’s household’s poverty status and delayed entry is not the same for girls and boys. The child’s household’s poverty status is not significant in differentiating between girls who are likely to be enrolled in school late and those who will be enrolled on time. Among boys, however, those living in non-poor households are significantly less likely than those living in very poor households to be enrolled in school at an inappropriate age. Among both girls and boys an increase in the number of siblings in school reduces the odds of delayed entry. Compared to living in a household headed by an adherent to traditional religion, both boys and girls living in households headed by Muslims have lower odds of delayed entry. Boys are less likely to start school late when their mothers have attained more than junior high school education. The importance of mother’s education plays out differently for girls. Girls whose mothers have some basic education have 67 percent higher odds of starting school late. On the other hand girls whose mothers have completed junior high school education are less likely to experience delayed enrolment compared to girls whose mothers have no education. Fathers’ education does not matter for boys. For girls, on the other hand, having an educated father reduces the odds of delayed entry and the odds are much lower when their fathers have completed an education level higher than junior high school. 7. Correlates of number of years children’s age exceeds the appropriate age for their grade
113
Table 5 Factors associated with the number of years the ages of children exceed the appropriate age.
Male Number of siblings
Full sample
Urban model
Rural model
Male model
Female model
0.25*** 0.01
0.12** 0.03
0.32*** 0.01
0.00
0.02
0.10 0.29*** Yes
0.23 0.22 Yes
0.07 0.30*** Yes
0.22 0.44*** Yes
0.07 0.10 Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
a
Poverty status Poor Non-poor Household characteristics Religion of household head Parents’/guardians’ Characteristics Urban, rural locality Region Constant
Yes
Yes
Yes
Yes
Yes
Yes 3.37***
Yes 2.26***
Yes 3.18***
Yes 3.42***
Yes 3.12***
F R-squared Observations
19.44 0.11 7798
9.02 0.14 2514
10.91 0.09 5284
14.91 0.13 4129
9.06 0.10 3669
a
The reference category is ‘‘extremely poor’’ *p < 0.1, **p < 0.05, ***p < 0.01.
In the model with the full sample and in both the urban and rural models boys are significantly older than the appropriate age for their grade than girls. Children living in non-poor households are fewer years above the appropriate age than children from very poor households. This relationship between poverty status and number of years above the age appropriate for the grade is driven by rural households. The coefficient of the dummy variable for non-poor households is significant and negative in the rural model and insignificant in the urban model. The poverty status of the household is not significant in explaining the number of years above the age appropriate for the grade among girls but matters for boys where boys from non-poor households have fewer years above the appropriate age than boys living in very poor households. 8. Discussion
The previous section has identified the correlates associated with the likelihood of delayed entry into primary school. This section explores the factors associated with the number of years the ages of children in school exceeds the appropriate age for their grade. Table 4 presents information on the mean number of years the ages of children in primary 1 through to the third year of junior high school exceeds the age appropriate for their grade. Children are on average 1.23 years older than the appropriate age for their grade with the maximum number of years by which their age exceeds the appropriate age being 20. The mean for urban children is lower than that for rural children and the mean for girls is lower than that for boys. Table 5 presents the findings for the model that explores the factors associated with the number of years a child’s age exceeds the appropriate age. It presents the results for the full sample, urban and rural samples and the samples for girls and boys.2 The dependent variable is the difference between the age of the pupil and the appropriate age for the grade. The sample comprises pupils in primary 1 to JHS 3 (i.e. grades 1–9). The same set of explanatory variables employed in the delayed entry model are utilised here. The discussion will concentrate on two issues of interest. These are whether the number of years above the appropriate age for the grade is the same for boys and girls and the importance of the poverty status of the households the children live in. 2
The results for the full model can be obtained from the authors on request.
An important finding of this paper is that girls are no more likely than boys to be sent to school late. Glewwe and Jacoby (1995) in their earlier study on Ghana obtained a similar result. This finding is not entirely unexpected since primary gross enrolment rates of girls and boys have been increasing in tandem over time and gender parity in primary education has almost been achieved. A second important finding is that the poverty status of the household matters for whether a child will be sent to school on time. The insignificance of the sex dummy in the delayed entry model notwithstanding, the phenomenon of delayed entry in Ghana does have some gender dimensions. First is the finding that the mean of the difference between actual age and appropriate age for the grade is higher for boys than for girls. Thus even though both boys and girls are just as likely not to start primary school at the official age, boys are probably kept out of school longer, or are more likely to repeat a class or spend some time out of school and then reenroll at a later period. Data collected during the 2008 Demographic and Health Survey shows that boys are more likely to be stunted (they do not have the required height for their age) than girls (Ghana Statistical Service et al., 2009). Thus if parents’ decision on whether a child is ready to be sent to school is informed by the child’s physical development (see Glewwe and Jacoby, 1995) this may be one reason why the mean of the difference between actual age and the appropriate age for the grade is higher
114
H.A. Seshie-Nasser, A.D. Oduro / International Journal of Educational Development 49 (2016) 107–114
for boys. Sensitisation messages that contain information on the benefits of sending all children to school at the appropriate age is required. If indeed the child’s physical development is an important factor in parents’ considerations this suggests that policy measures such as bussing young children to school and sitting primary schools closer to communities will be required to ensure that all children irrespective of their gender and urban or rural location are sent to school when they are six years old. The second gender dimension is the differences in the correlates of delayed entry between girls and boys. Of particular interest is nature of the relationship between the poverty status of the child’s household and delayed entry of girls and boys. The significance of the household poverty status variable for boys and not for girls suggests that the relationship between the poverty status of the household and children’s enrolment depends on the sex of the child, the age of the child and/or the stage of the child in education system. This study finds that the poverty status of the household may not be important for whether girls start school on time, however it matters for whether they complete primary school and transition to junior and senior high school (Nguyen and Wodon, 2014). Further research is required to unearth the reasons why the poverty status of the households matters for the enrolment of boys on time but is not important for girls. In addition to the variables of interest other variables have been found to be associated with delayed entry. Parent’s education has emerged as an important correlate of delayed entry. This study confirms that some level of formal education on the part of the parents reduces the odds of them sending their children to school late. For both boys and girls, mothers’ education is important and the more educated mothers are the better. Father’s education is important and the higher the level of education achieved the lower the odds of delayed entry, particularly among girls. Thus policies that succeed in increasing junior high school completion rates of both boys and girls have positive spillover effects for future generations. These graduates are more likely to ensure that the next generation will start school on time. Adult learning programs that are targeted at school drop outs, and are designed to make it easy to enroll for classes to obtain the Basic Education Certificate and the Senior High School Certificate whilst working may reduce the time frame within which the problem of delayed entry will be overcome. Even though enrolling a large number of children in school can spread a household’s resources thinly, the number of children enrolled may be indicative of the premium placed on education by the household. The premium a household places on education as measured by the number of children in school significantly reduces the likelihood of delayed entry in all the models except for the urban model. Thus, a sensitisation policy that educates parents, especially in rural areas on the benefits of sending their children to school would have a positive ripple effect for the younger children that follow, since it would increase the likelihood that the six and seven year olds who follow are enrolled on time. Urban children who do not live with their parents are more at risk of delayed entry. Further research is required to understand why this category of children are not sent to school at the appropriate age. 9. Conclusion Despite the high gross and net primary enrolment rates, a not insignificant proportion of children do not start primary school
when they are six years old. Girls aged six or seven years old are no more likely to experience delayed entry than boys of the same age. However, there is evidence that boys are much older than the appropriate age for their grade than girls suggesting that even though they are just as likely as girls to experience delayed entry they may be sent to school at an older age than girls. The poverty status of the household matters for whether boys will be sent to school on time but is not important for girls. Policy to address gender inequality within education has tended to focus on the girl child. The findings of this paper reveal that gender inequality within education is quite nuanced. Therefore policies must be designed that take cognisance of the different needs of boys and girls and the different constraints they face if both boys and girls are to start school on time and complete primary school.
References Ainsworth, M., Beegle, K., Koda, G., 2005. The impact of adult mortality and parental deaths on primary schooling in North-Western Tanzania. J. Dev. Stud. 41, 412–439, http://dx.doi.org/10.1080/0022038042000313318. Akyeampong, K., 2011. (Re) Assessing the Impact of School Capitation Grants on Education Access in Ghana (CREATE Pathways to Access Research Monograph No. 71) Center for International Education, University of Sussex, Brighton, UK. Akyeampong, K., Djangmah, J., Oduro, A., Seidu, A., Hunt, F., 2007. Access to Basic Education in Ghana: The Evidence and the Issues. Aryeetey, E., Harrigan, J., Machinko, N. (Eds.), 2000. Economic Reforms in Ghana: The Miracle and Mirage. James Currey and Woeli Publishers, Oxford. Bommier, A., Lambert, S., 2000. Education demand and age at school enrollment in Tanzania. J. Hum. Resour. 35, 177–203, http://dx.doi.org/10.2307/146360. Burke, K., Beegle, K., 2004. Why children aren’t attending school: the case of Northwestern Tanzania. J. Afr. Econ. 13, 333–355, http://dx.doi.org/10.1093/ jae/ejh011. Ghana Statistical Service, 2007. Patterns and Trends of Poverty in Ghana: 1991– 2006. Ghana Statistical Service. Glewwe, P., Jacoby, H.G.H.G., 1995. An economic analysis of delayed primary school enrollment in a low income country: the role of early childhood nutrition. Rev. Econ. Stat. 77, 156–169, http://dx.doi.org/10.2307/2110001. Glick, P., Sahn, D.E., 2000. Schooling of girls and boys in a West African country: the effects of parental education, income, and household structure. Econ. Educ. Rev., http://dx.doi.org/10.1016/S0272-7757(99)00029-1. Grira, H., 2001. Delayed School Enrolment: Evidence from Rural Bangladesh [WWW Document]. http://dse.univr.it/espe/documents/Papers/E/2/E2_1.pdf (accessed 08.04.15). Grogan, L., 2009. Universal primary education and school entry in Uganda. J. Afr. Econ. 18, 183–211, http://dx.doi.org/10.1093/jae/ejn015. Lincove, J.A., 2009. Determinants of schooling for boys and girls in Nigeria under a policy of free primary education. Econ. Educ. Rev. 28, 474–484, http:// dx.doi.org/10.1016/j.econedurev.2008.10.001. Ministry of Education, 2013. 2012/2013 Education Profile. Moyi, P., 2011. Delayed School Entry in Uganda. Res. Comp. Int. Educ. 6, 222, http://dx.doi.org/10.2304/rcie.2011.6.2.222. Moyi, P., 2010. Household characteristics and delayed school enrollment in Malawi. Int. J. Educ. Dev. 30, 236–242, http://dx.doi.org/10.1016/ j.ijedudev.2009.11.008. Nguyen, M.C., Wodon, Q., 2014. Analysing the gender gap in education attainment: a simple framework with application to Ghana. J. Int. Dev. 26, 59–76, http://dx.doi.org/10.1002/jid. Osei-Fosu, A., 2011. Evaluating the impact of the capitation grant and the school feeding programme on enrollment, attendance and retention in schools: the case of Weweso circuit. J. Sci. Technol. 31 , http://dx.doi.org/10.4314/ just.v31i1.64886. Pradhan, M., 1998. Enrolment and delayed enrolment of secondary school age children in Indonesia. Oxf. Bull. Econ. Stat. 60, 413–430, http://dx.doi.org/ 10.1111/1468-0084.00107. Rose, P., Al-Samarrai, S., 2001. Household constraints on schooling by gender: empirical evidence from Ethiopia. Comp. Educ. Rev. 45, 36–63, http:// dx.doi.org/10.1086/447644. Wells, R., 2009. Gender and age-appropriate enrolment in Uganda. Int. J. Educ. Res. 48, 40–50, http://dx.doi.org/10.1016/j.ijer.2009.03.002. Wils, A., 2004. Late entrants leave school earlier: evidence from Mozambique. Int. Rev. Educ. 50, 17–37, http://dx.doi.org/10.1023/B:REVI.0000018201.53675.4b.