NiCE Working Paper 09-116
December 2009
Effects of Reproductive Health Outcomes on Primary School
Attendance: A Sub-Saharan Africa Perspective
Abiba Longwe
Jeroen Smits
Nijmegen Center for Economics (NiCE)
Institute for Management Research
Radboud University Nijmegen
P.O. Box 9108, 6500 HK Nijmegen, The Netherlands
http://www.ru.nl/nice/workingpapers
1
Abstract
To what extent and in which ways does poor RH at the household level negatively
influence educational attendance of young children in Sub-Saharan Africa? This paper
sets out to answer this question by analyzing household and district level data on school
attendance of 103,000 primary-school aged children living in 287 districts of 30 Sub-
Saharan African countries. Our multilevel analyses reveal substantially decreased
attendance rates of boys and girls with short preceding and succeeding birth intervals,
with more siblings, with a young sibling present and with a pregnant mother. These
findings remain intact when controlling for socioeconomic and demographic household
characteristics and economic and health-related context factors. Interaction analysis
shows that many effects of RH outcomes depend on the context in which the household is
living. Thus highlighting the importance of a situation-specific approach.
This research is part of the WOTRO-Hewlett PopDev programme “Impact of
reproductive health services on socio-economic development in sub-Saharan Africa:
Connecting evidence at macro, meso and micro-level” which is funded by a grant from
The William and Flora Hewlett Foundation through the Netherlands Organisation for
Scientific Research (NWO), WOTRO Science for Global Development.
Correspondence address:
Abiba Longwe & Jeroen Smits, Nijmegen Center for Economics (NiCE), Institute for
Management Research, Radboud University Nijmegen, P.O. Box 9108, 6500 HK
Nijmegen, The Netherlands, [email protected]; [email protected]
2
Introduction
Educating children is one of the main forms of human capital formation and is an
important instrument for sustainable development and poverty reduction. Education is
generally regarded as a powerful means for reducing poverty and achieving economic
growth. It empowers people, improves individuals’ earning potential, promotes a healthy
population, and builds a competitive economy (Hanushek and Wossmann 2007;
UNESCO 2007; Word Bank 2006). Available evidence shows that there are several
channels through which such effects may arise. For instance education raises labor
productivity (Welch 1970), increases technological innovation and adaptation (Rodriguez
and Wilson 2000), contributes to better health (World Bank 1993) and gives greater
ability to deal with shocks (World Bank 2001).
Universal free primary education has been at the center of most governments’
policies in the developing world. However, in spite of substantial progress made as part
of the Education for All campaign, millions of young children in Sub-Saharan Africa are
still not in primary education. Worldwide, primary school net enrollment exceeds 90% in
most of the worlds’ sub regions except for sub-Saharan Africa, where it was around 70%
in 2007 (Glewwe and Miguel 2008; UNESCO-UIS 2009). To improve the current
situation in this region, it is of fundamental importance to gain a better understanding of
the factors that influence school attendance of children there.
Most studies on school enrollment focus on the influence of socio-economic and
demographic factors and availability of education facilities (Bainbridge et al. 2003; Song
et al. 2006; Toor and Parveen 2004; Huisman and Smits 2009). Some focus has also been
on understanding the effects of education on family planning (Schultz 1981, 1997;
Cleland and Rodriguez 1988; Drèze and Murthi 2001; Basu 2002; Campbell et al. 2006).
There are few studies which have evaluated long-run consequences of family planning
programs on labor supply, savings or investment in children’s human capital. However,
short-run association of such programs with adoption of contraception or age-specific
fertility has been assessed (Schultz 2008).
This topic has been highly researched in developing countries as general, Asian
regions, mixing other regions and sub-Sahara African countries and/or as specific
countries from specific world regions. Very little has been done on sub-Saharan Africa as
3
a region. This paper concentrates on the effects of SRH on primary school attendance in
sub-Saharan Africa. Using a database with individual and household characteristics on
103,000 children of 73,000 mothers in 30 sub-Saharan African countries, we aim to
answer the following research question: “To what extent does poor RH at the household
level influence primary school attendance of young children?”
The RH factors on which we focus are number of siblings, preceding and
succeeding birth interval, presence of young children in the household, and pregnancy of
the mother. To study the role of these factors in children’s schooling, advanced multilevel
logistic regression models were used that allowed us to estimate the effects of factors at
the household, district and national-level simultaneously.
In the next sections, we first develop hypotheses regarding the effects of the RH
factors on educational attendance and regarding the way in which their effects may
depend on the context. After that the datasets and the operationalisation of the variables is
described. In the results section, we first present the bivariate effects of the explanatory
variables on the likelihood of going to school, followed by the outcomes of the
multivariate analyses. In the final section, conclusions are drawn and some policy
recommendations are suggested.
Theoretical background
Sexual and Reproductive Health (SRH) is a human right, essential for human
development and achievement of the Millennium Development Goals (MDG). Promotion
of family planning in countries with high birth rates has the potential to reduce poverty
and hunger, avert over 30% of all maternal deaths and nearly 10% of childhood deaths,
and contributes substantially to women’s empowerment, achievement of universal
primary schooling, and long-term environmental sustainability (Cleland et al. 2006).
Smaller families and wider birth intervals resulting from use of contraceptives allow
families to invest more of their resources in each child’s nutrition, health, education, and
can reduce poverty and hunger for all household members (Singh et al. 2004).
According to ICPD Programme of Action (UNFPA 1995, p 40), “RH is a state of
complete physical, mental and social well-being and not merely the absence of disease or
infirmity, in all matters relating to the reproductive system and to its functions and
4
processes. RH implies that people are able to have a satisfying and safe sex life and that
they have the capability to reproduce and the freedom to decide if, when and how often to
do so…”.
In developing countries, complications of pregnancy and childbirth (resulting
from poor RH conditions) are the principal cause of female mortality in the reproductive
ages (Maine et al. 1994) and have important long and short-term implications for
women’s health and their productivity and investments in children. Investments in
general and RH have been found to contribute to economic growth by reducing
production losses, increasing school enrollment and children's ability to learn, and freeing
up resources that would otherwise have been spent on treating illnesses (Seligman et. al.
1997; USAID 2005). For example, Sinha (2005) and Joshi and Schultz (2007) found for
Bangladesh that family planning may reduced fertility by one child and that there are
substantial gains associated with family planning in child survival and male child
schooling. Improvements in family planning - and health in general - are linked to
economic and social development and must be addressed to achieve sustainable
reductions in poverty. Moreover, the current status of a woman’s RH has a profound
impact on the human capital development of her children and, by consequence, on future
social and economic development outcomes (Seligman et al. 1997).
The direct and opportunity costs of investing in children’s education and the value
parents attach to education are influenced by many factors, both at the level of the
household and of the context in which the household lives. Because these different
factors are not independent of each other and may exert their influence at the same time,
insight into their relative importance can only be obtained if they are studied
simultaneously. This is what is done in this study. Figure 1 shows the different groups of
factors that are included in our analytical model and their expected direction of influence.
Starting point for this model is a model presented by Huisman and Smits (2009) that, for
this paper, is extended with the RH variables on which we focus: the number of children,
their spacing, and the presence of a young sibling and pregnancy of the mother. In the
next sections we expound on these factors.
5
Preceding and succeeding birth interval
Child spacing benefits the mother, the child, the father and other family members. It
allows for time between births so that the mother can rest between pregnancies and
maintain good health, be less tired and have more energy. She therefore can give more
attention to her children and help ensure good health for the family. A well spaced child
grows up stronger and healthier and will be more likely to attend school when s/he is of
school going age (Nigerian government 2003).
The effect of short birth intervals is expected to have negative impact on
education participation of children. A Swedish study established that there is negative
association between very close child spacing (less than 2 years) and long-term outcomes
of children as measured by educational attainment (Pettersson-Lidbom and Thoursie
2009). According to Hill and Upchurch (1995), girls are more affected by short birth
intervals because of the practice of gender preference, whereby less priority is given to a
girl, particularly if resources are scarce. Hayes et al (2006) found that birth interval is a
significant predictor of school readiness in South Carolina, even after controlling for
various socio-demographic factors. Children born with inadequate birth intervals (less
than 24 months) are more likely to fail the Cognitive Skills Assessment Battery compared
to those with adequate birth intervals (pp 426-27).
Longer birth intervals allow parents to devote more time to each child in the early
years, ease pressures on the family’s finances and give parents more time for activities
other than child rearing (USAID 2006). In Nepal, focus group discussions revealed that
people in rural communities were in strong support of long child spacing because it
would help ease family problems, such as health and economic difficulties resulting from
too-frequent births. “Babies are forced from the breast too early,” said one. “Mothers and
newborns are weak and parents cannot send their children to school when there are too
many or they come too soon” (Shrestha and Manandhar 2003).
Number of children
Family size and composition may influence learning achievement. The lower the number
of children, the more education each individual child is likely to receive, because
available resources have to be shared among fewer children. However, not all linkages
6
between family size and children’s education are as clear-cut as this dilution effect.
Theory suggests a trade-off between child quantity and quality. Family size might
adversely affect the production of child quality within a family (Booth and Kee 2005).
Controlling for parental education, parental age at birth and family level attributes, Booth
and Kee found that in Britain children from larger families had lower levels of education
and that this family size effect did not vanish when they controlled for birth order.
Similarly, Li, Zhang and Zhu (2008) confirmed the existence of a negative family size
effects on children’s education for China using instrumental variable techniques.
Echevarria and Merlo (1999) restricted their focus to the role played by fertility
and discovered that both in the steady state and along the transition path, investments in
education of both males and females decreased with the number of siblings. In the same
light, DeGraff et al. (1993) examined how the number of siblings affected children’s time
use in Bicol Province, Philippines. They found that the greater the number of younger
siblings, the lower the likelihood that the child would be enrolled in school and the more
time the child would devote to domestic work. However, Kirdar et al. (2007) found no
causal impact of sibship size on school enrollment in Turkey.
Knodel et al (1990) showed that the decline in fertility and cohort sizes in
Thailand allowed existing primary school facilities to be adapted to accommodate
growing enrollment at the secondary level. These studies are important because of the
cross-generational effects they illuminate. Recent research confirms that children with
fewer siblings tend to outperform those with more siblings (Dronkers and Robert 2008;
Park, 2008). Eloundou-Enyegue (2000) found that African children with more siblings
tend to enroll later, repeat grades more often, and drop out of school earlier than children
with fewer siblings.
A number of studies suggest that siblings are unlikely to receive equal shares of
the resources devoted by parents to their children’s education. Booth and Kee (2005), for
example, found that in Britain the shares in the family educational resources decreased
with birth order and Jejeebhoy (1992) found for India that in families in which girls have
fewer siblings, they are disadvantaged because they are more likely to assume tasks
traditionally assigned to boys so that their brothers can pursue additional education. Such
sex differences disadvantaging girls have also been found for Turkey, where Dayioglu et
7
al. (2009) found sex composition of siblings to matter for girls’ (but not for boys) school
enrollment. The amount of education that girls receive is inversely related to the number
of children parents have, hence, although it is still true that boys receive more education
than girls, the gap increases with higher fertility (Echevarria and Merlo 1999). Adding to
this, Filmer et al. (2009) noted that as long as quantity-quality trade-offs exist that result
in fewer material and emotional resources allocated to children in larger families, son-
preference in fertility decisions can have important indirect implications for investments
and for the well being of girls relative to boys.
Mother’s pregnancy and presence of young children
Women in developing countries are more than 45 times more likely to die from
pregnancy related complications than women in the developed world. As a result, girls in
developing countries are often pulled out of school to care for their siblings. Access to --
as well as correct and consistent use of -- contraceptives can therefore significantly
increase girl’s schooling (UNFPA 2005). There are also indications that the presence of a
baby in the household has a negative effect on other children’s schooling time because of
increasing care needs of the household (Chernichovsky 1995). Using data from Ghana,
Lloyd and Gage-Brandon (1993) showed that teenage girls are relatively more likely to
be withdrawn from school as new siblings are added to the family.
It is possible that adverse effects are stronger if the pregnancy is unwanted.
Unintended pregnancy is an important public health concern, because of its association
with negative social and health outcomes for mothers, children and society as a whole.
Women with mistimed pregnancies tend to delay prenatal care, are associated with
maternally risk behavior, including cigarette smoking, alcohol and illicit drug use and
with postpartum depression, and are more likely to consume inadequate folic acid
(Weller et al. 1987; Altfeld et al. 1997; Than et al. 2005; Cheng et al. 2009).
Additionally, there is the potential for deleterious health outcomes of unintended
pregnancies for the index children, their siblings, and their parents (Gipson et al. 2008).
8
Control factors
Besides the RH variables discussed so far, our model contains a number of other factors
that are known or expected to influence educational enrollment. Most of them are
discussed in detail by Huisman and Smits (2009). They are included in our model as
control factors, so as to be able to estimate the independent effects of the RH outcomes
on educational enrollment. At the household level, parental education, father’s
occupation, and household wealth have been known for long to be important
determinants of educational participation in both the developed and developing world
countries (Coleman et al., 1966; Jencks, 1972; Shavit & Blossfeld, 1993; Tansel, 2002;
Glewwe & Jacoby, 2004; Mingat, 2007; Evangelista de Carvalho Filho, 2008).
Demographic characteristics of the child -- age, gender and birth order (Chiswick and
DebBurman 2004; Ejrnaes and Portner 2004; Kirdar et al. 2007; Dayioglu et al. 2009) --
as well as of the household they live in – presence of the parents, extended or nuclear
family, and age at which the mother got her first child (Bradbury 2007; Wichman et al.
2006) -- are also known to influence education enrolment
Besides these household-level factors, characteristics of the context the household
lives in may affect the ability of children to enroll in education. Previous research has
shown that, for example, the relationship between family size and educational attainment
is likely related to a society’s level of development, modes of production, and access to
schooling, which in turn shapes the relative influence of the family on the schooling of
children (Desai, 1995; King, 1987; Lloyd, 1994). Depending on the context or level of
development, having more siblings to share household and labor market work may
provide children with more resources for schooling. The macro-socioeconomic
mechanisms relating family size and children’s schooling include availability of schools,
transportation and communication infrastructure, and participation in labor-intensive
production like agriculture.
In less developed regions of a country and in rural areas, schooling may entail
higher production costs due to more limited availability and accessibility of schools. For
instance, in rural areas of Turkey, especially in the eastern part of the country students
have to be bussed to schools that are sometimes at significant distance to their village or
they have to walk long distances which may be a challenging task in winter (Kirdar 2007;
9
Hazarika 2001; Smits and Gunduz-Hosgor 2006). Higher schooling costs would imply a
lower marginal rate of return and, therefore, a lower demand for schooling. Even when
schools are available in less developed or rural regions, they may be of lower quality
because of poor facilities and teachers who are less motivated as these are less popular
places for them to live and to work.
The urban child has for long been considered to be more likely to realize the
dream of fully participating in school than his/her rural counterpart. This ‘‘urban
advantage’’ is associated with increased access to facilities such as schools in urban areas
(Mugisha 2006). Returns to education are higher in cities because there are better chances
to find a white collar job (Hazarika 2001; Huisman and Smits 2009). In more modern and
urban areas, there is also more impact of globalization, including the diffusion of value
patterns that stress the importance of education and equality among sexes. However,
children living in urban or better to do regions of a country also have better chances of
finding market work (piece work), which increases the opportunity cost of schooling for
them (Kirdar 2007). This relationship between the RH factors and children’s school
attendance can differ within the same country and change over time as contextual factors
evolve with socioeconomic development. This is evidenced by a study in Indonesia
(Maralani, 2008), which showed that in urban areas, the association between family size
and children’s schooling was positive for older cohorts but negative for more recent
cohorts while there was no significant association for any cohort in rural areas.
Potential relevant context factors that have not yet been studied very much are
characteristics of the available health facilities in the area where a household lives. The
presence of good and accessible health facilities can be expected to lead to a better
general health status in the area of mothers and children, which may translate into a
higher educational enrollment. RH facilities are also important for educational
participation. However, as they will have their major effect through the household-level
RH variables, we do not include them in our model.
Interactions with the context
The effects of RH factors on educational participation of children need not be everywhere
the same. For example, Heyneman and Loxley (1983) found that in less developed areas
10
the importance of school and teacher quality is higher and of resources at the household
level is lower than in more developed areas in explaining educational achievement (see
also Fuller and Clarke 1994). If this is also true for effects of RH related resources, we
would expect the relationships between the RH factors studied in this paper and
educational participation to be stronger in the more developed SSA regions.
On the other hand, it seems not unreasonable to suppose that the benefits children
can reap from a favorable situation at the household level – small family size, long birth
intervals, no pregnant mother or young siblings – are higher under more difficult
circumstances. From this perspective we would expect to find stronger effects of the RH
factors in less developed areas, in rural areas, and in areas where the quality of the
educational facilities is lower and the general health situation is worse.
Data and methods
To answer the research question, we have combined Demographic and Health Surveys
(DHS) for 30 Sub-Saharan African countries. Our dataset contains information on all
children aged 8-11 born from the women interviewed in the women’s surveys. In total,
102,638 children (52,520 boys and 50,118 girls) born from 73,337 women were included.
The household-level data on the children, their mothers and households was
supplemented with context information at the district and nation-level. District-level
information for 287 districts was derived by aggregating from the household surveys. The
aggregation was possible because the surveys involved large samples and had a variable
indicating the district.
Methods and variables
The data are analyzed with multilevel logistic regression models, including explanatory
variables at the household, district and national level. With multilevel analysis it is
possible to include explanatory variables at different levels simultaneously and to study
interactions among levels (Hox 2002; Snijders and Bosker 1999). The analyses are
performed separately for boys and girls. In all analyses robust standard errors (sandwich
estimators) are used.
11
The dependent variable is a dummy variable indicating whether (1) or not (0)
children aged 8-11 were attending school at the time of the interview. The upper age limit
of 11 was chosen to restrict the analysis to primary education. The lower age limit was
put at 8, because compulsory entry ages differ per country and not all children start
schooling at the compulsory age (compare Huisman and Smits 2009).
Independent variables include RH variables, other household-level factors, district
characteristics and national characteristics. Regarding the RH variables, we measure child
spacing by two dummy variables indicating whether the preceding and succeeding birth
intervals were less (1) or more (0) than two years. Number of siblings is indicated by an
interval variable. The presence of a young child in family is measured by a dummy
variable indicating whether (1) or not (0) a child aged less than 3 years was present in the
household. Pregnancy of the mother is measured by a dummy with categories (0) not
pregnant and (1) pregnant.
Of the household characteristics, father’s occupation is measured as (1) farm, (2)
lower non-farm and (3) upper non-farm. Employment of the mother is measured by two
categories indicating whether (1) or not (0) the mother is gainfully employed. Father’s
education is measured by three categories: (1) none, (2) at least some primary, and (3) at
least some secondary. Due to low levels of education for women in sub-Saharan Africa,
mother’s education is measured by a dummy indicating whether (1) or not (0) she has at
least some primary education. Household wealth is used as a proxy for income and is
measured by an index constructed on the basis of household assets. Using a method
developed by Filmer and Pritchett (1999), all households within the country were ranked
on the basis of the available characteristics and divided into wealth index deciles, with 1
representing the poorest 10% and 10 representing the richest 10%. Whether the
household is an extended family is measured with two categories: (0) nuclear family, and
(1) extended family (more than two adults in the household). Age of the child and age of
the mother at first birth are measured in years. For the mother’s age at first birth also a
quadratic term is included in the model, because both starting getting children at a very
young age and starting at a relatively old age point to an exceptional position of the
mother. Birth order of the child is measured as an interval variable.
12
Of the context factors is level of urbanization measured by a dummy indicating
whether (1) or not (0) the household lives in an urban area. District-level of development
is measured by an index constructed on the basis of six variables aggregated from our
household datasets: the percentages of households in the district with a fridge, car,
telephone, television, electricity, or running water. Of these characteristics the mean was
taken of the standardized values. Availability of health facilities in the district is indicated
by the percentage of women in the district who delivered their last baby in a hospital and
by the percentage of the last-born children of the women in our dataset who received a
DTP vaccination. Public expenditure on education and national GDP per capita in
Purchasing Power Parity (constant 2000 international dollar) are derived from World
Bank (2007).
Children with a missing father were given the mean score of the other children in
the database on fathers’ education and occupation. Because there are dummies for
missing father in the model, this procedure leads to unbiased estimates of these variables
(Allison, 2001, p. 87). To address the fact that the effects of the RH factors may differ
depending on the situation of the household, also models with interactions between
control factors and the RH outcomes are estimated. To compute these interaction terms,
centered versions of the involved variables are used. Given the large number of possible
interactions, only significant interaction effects are included in these models.
Results
Bivariate analysis
The coefficients of the bivariate logistic regression models are shown in Table1. Both the
logistic and multiplicative versions (within brackets) of the coefficients are presented.
The multiplicative versions can be interpreted more easily. For example, the value of
0.852 for the effect of preceding birth interval on boys school attendance means that the
odds of attending school is 0.852 times (or 14.8%) lower for boys with a preceding birth
interval of less than 2 years compared to boys with a long preceding birth interval. The
value of 1.278 for girls whose mothers are gainfully employed indicates that these girls
have 1.278 times (or 27.8%) higher odds of attending school than girls whose mothers are
not employed.
13
Table1 shows that the effects of most of the RH variables are in line with our
expectations. The likelihood to attend school for both boys and girls is significantly
reduced by short preceding and succeeding birth intervals. Having more siblings shows
the expected negative effect on school attendance. If a mother is currently pregnant, her
children have less chance of attending school. The presence of a child below age three in
a family has a negative influence on older children’s ability to participate in school.
Most of the effects of the household-level control factors are in line with
expectations. Age of child, household wealth, parental education and occupation have
significant positive effects on school attendance of both boys and girls. There is a
significant negative relationship between missing father, living in rural area and attending
school. Belonging to an extended family has a significant positive influence on school
attendance for girls while it has a negative non-significant effect on boys. Birth order is
negatively associated with boys’ school attendance while it positively affects girls’
likelihood to attend school. Mothers’ age at first birth has a parabolic effect. Both
children born to relatively young and to relatively old mothers are less likely to attend
primary school.
All the context factors show the expected positive effects on children’s
educational attendance. Attendance levels for boys and girls are significantly higher in
more developed districts and countries, in countries with higher public expenditure on
education, and in districts where more women give birth in a hospital and where more
children received a DTP vaccination.
The coefficients of the bivariate analyses show how school attendance of boys
and girls varies among households and districts with different RH characteristics. These
coefficients thus represent the observable reality in the countries and districts under
study. However, they give no insight into the relative importance of the various
characteristics explaining school attendance, because these characteristics may be related
to each other (e.g. mothers with many children tend to have them within short birth
intervals; being the fifth child means having at least four siblings). To gain insight into
the relative importance of the RH and other explanatory variables, we now turn to the
multivariate results.
14
Multivariate analysis
The results of the multivariate logistic regression analyses are presented in Tables 2 to 4.
Table 2 presents results for boys and Table 3 for girls. Models 1 in these tables contains
only direct effects of the explanatory variables. Models 2 contain beside direct effects
also interaction effects between RH factors and other factors. To keep the tables readable,
the interaction effects for both boys and girls are presented separately in table 4.
Comparing the coefficients in Tables 2 and 3 with the bivariate coefficients, we
see that the effects of socio-economic as well as the RH factors do not differ much. The
effect of number of siblings on school attendance is strong in both the bivariate and
multivariate models. Boys’ as well as girls’ school attendance is negatively affected by
number of siblings present in the household. School attendance is reduced for the
children with short preceding birth interval in all models. The effect of the succeeding
birth interval becomes insignificant in multivariate Model 1, but retains its significance
for girls with the addition of interactions in Model 2. Overall, this effect is weaker than
that of the preceding birth interval. The presence of a young child of less than 3 years in a
household reduces school attendance for both boys and girls. Pregnancy of the mother
has a substantial negative effect in the bivariate analysis, but loses its significance at first
in the multivariate models. However, after adding the significant interaction effects in
Models 2, it becomes significant again for both boys and girls. This indicates that
mother’s pregnancy is important under specific conditions.
The effects of the control factors at the household level differ little between the
bivariate and multivariate analyses. The effect of birth order loses its significance in the
multivariate models for girls but not for boys. Belonging to an extended family, which
was not significant for girls and positive for boys bivariately, shows significant negative
effects in the multivariate models. All other household-level control factors keep their
sign and significance.
Regarding the context factors, we see that district level of development, which
was positively significant in the bivariate model, becomes significantly negative in the
multivariate models. This difference seems surprising, but additional analyses (not
presented) showed that it is due to the presence of wealth at the household level in the
multivariate models; hence the positive effect in the bivariate analyses was due to the fact
15
that wealthier households are living in more developed districts. All other context factors
keep their significant positive effects on educational attendance in the multivariate
models.
Interaction effects
Tables 2 and 3 show that, apart from some small changes in the magnitude of the
coefficients, there are no substantial differences between Models 1 and 2 for boys and
girls. All household-level and context control factors with significant effects in model 1
remain significant and keep their sign in Model 2. The only substantial changes were the
ones discussed above, regarding the effects of succeeding birth interval and mothers’
pregnancy.
Table 4 shows that all RH variables interact significantly with some of the
household and context factors for both boys and girls. We are therefore led to the
conclusion that their effects depend on the situation of the household and the context in
which the household lives.
With regard to the number of siblings, we see negative interaction effects with
mother’s education and employment and with public expenditure on education. Hence,
children with more siblings seem to be less able to fully enjoy the positive benefits of
having an educated and working mother, or to profit from educational investments. On
the other hand, there is a positive interaction effect with district level of development,
which suggests that better infrastructure or the influence of more modern values
compensates to a certain extent the family size effect. For girls, there is also a positive
interaction effect of number of siblings with father’s education (secondary), which might
mean that educated fathers discriminate less against girls when sending their children to
school. However, at the same time we see that girls with a short preceding birth interval
can profit less than other children from having a high educated father. So when there is
direct competition with a closely spaced earlier born sibling, even having a highly
educated father is less profitable.
We also see that a negative effect of a short succeeding birth interval is enhanced
for boys if the father is missing from the family. This might mean that when the husband
is missing, mothers are less able to cope with short spacing and older sons might have to
16
take over some of the missing father’s tasks. This is also suggested by the negative
interaction effect of mother’s pregnancy with missing father on boys’ school attendance.
Both boys and girls are less able to enjoy the benefits of good school facilities provided
by government if they are succeeded by a sibling in less than two years. Boys are also
less able to profit from a higher household wealth. However, when they have a mother
with some education having a short succeeding birth interval is less problematic.
If there is a young sibling in the family, boys are less able to profit from their
father’s education. The negative effect of a young sibling is stronger in urban and more
developed areas, thus indicating that under more traditional circumstances it is easier to
arrange care (or children don’t go to school anyway, so that the presence of a young
sibling cannot make much of a difference anyway). Girls with a pregnant mother are less
able to benefit from the positive impact of household wealth on their school attendance.
Conclusions
We have studied the effects of RH outcomes (number of siblings, birth interval, mother’s
pregnancy, and presence of a young sibling) alongside other household- and district-level
factors on primary school attendance of over 100,000 children aged 8-11 in 30 Sub-
Saharan African countries. We used multilevel logistic regression analysis in order to
include explanatory variables at different levels simultaneously and to study interactions
among levels. Poor RH outcomes were expected to reduce investment in human capital of
children, because with less children there are more resources for each individual child,
longer birth intervals are better for women’s health and allow mothers to give children
more attention in their vulnerable young years, and pregnancy of the mother or presence
of a young sibling might pull children, particularly daughters, out of school to help at
home.
The findings of our study are largely in line with these expectations. Our analyses
revealed that, in these countries, children living in larger families have a significantly
lower probability of being in school than children living in smaller families. This is also
true for children who were born shortly after their preceding sibling, for girls who were
succeeded shortly by a younger sibling, for children with a pregnant mother or a young
sibling when they reach at school-going age.
17
For each additional sibling, the odds of being in school decreases by 4% for boys
and 5% for girls. A preceding birth interval of less than two years decreases these odds
for boys and girls by about 15%. Having a pregnant mother decreases it by about 9% and
presence of a sibling below age three in the family by 13% for boys and by 18% for girls.
These effects are independent of each other; which means they add up. A girl with five
siblings, one of which is below age three, which has a pregnant mother and a short
preceding birth interval has a 50% lower odds of being in school than a girl with no
siblings and whose mother is not pregnant. A boy in this situation would have a 45%
reduced odds of being in school.
Regarding the effects of the control factors, our results are mostly in line with
earlier findings that socio-economic and demographic characteristics of a household
contribute significantly to a child’s chance of attending primary school (e.g. Huisman &
Smits 2009; Tansel 2002; Mingat 2007; Evangelista de Carvalho Filho 2008). If the
parents have a higher educational level, if the household is wealthier, and if the father is
working in a non-farm job the chances that children are in school are substantially
increased. Having an employed mother also has a positive effect on children’s schooling.
Effects of demographic factors are less pronounced, but still important. Living in an
extended family reduces the chance of boys attending school while it does not affect
girls. For girls, birth order increases the likelihood of later-born girls to be in school
while the older girls stay at home. If a father is missing in a household, both boys’ and
girls’ odds of attending school is reduced. Primary school attendance is also influenced
by characteristics of the context. Living in a rural area reduces the odds of attending
school by almost 30% for both boys and girls. National level of development and public
investment in education enhances primary school participation. As could be expected,
also a good health infrastructure (measured by availability of health facilities and access
to vaccination) tends to increase children’s ability to attend school.
Besides models with direct effect of the explanatory variables, we estimated
models with interaction effects between RH variables and household- and district-level
factors. This interaction analysis was meant to increase our understanding of how the
effects of RH factors differ among households and contexts with different characteristics
and to make the outcomes more situation-specific. Results reveal a substantial number of
18
significant interactions. The idea that the effects of RH factors differ among contexts and
that a situation-specific approach is important is thus confirmed by our data. However,
the results do not clearly lead into one direction. Parental education is a good example.
Children with more siblings profit less of the education of their mother, but at the same
time girls in that situation profit more of having a well-educated father. Similarly, we see
that girls with a short preceding birth interval profit less of the education of their father,
while boys with a short succeeding birth interval profit more of the education of their
mother. Another example is the effects of the context factors. For children with a young
sibling in the family it is negative for their educational attendance to live in a more
developed or urban area. At the same time we find that children with more siblings tend
to be more in school if they live in more developed areas. These seemingly contrasting
findings suggest that each interaction effect should be carefully considered on its own
merits. It might for example be more difficult to arrange childcare in the more developed
areas, while at the same time the negative effect of large family size might be
compensated by the shorter distances to school and more modern values there.
Other interaction effects are more straightforward. Having an employed mother is
negative for children with more siblings and having a missing father is negative for boys
with a pregnant mother; both are situations in which children may have to take over tasks
at home. Also the finding that children in more difficult situations can profit less of
household wealth or of higher public expenditure on education is not so surprising in
light of earlier findings indicating that investments in education are not enough to bring
the children in the weakest position into school (Huisman and Smits, 2009).
This research contributes evidence to the ongoing debates on linkage between
SRH investments and poverty reduction. It also contributes new knowledge to research
on factors influencing education participation by demonstrating that poor RH at
household level can be detrimental to the achievement of Millennium Development Goal
2: Universal primary education. The substantial negative effects of short birth intervals,
number of siblings, presence of young sibling and mothers’ pregnancy on school
attendance of children in sub-Saharan Africa stresses the importance of good and
accessible RH facilities. Since RH directly relates to the achievement of universal
19
primary education, countries should focus their efforts on actual usage of the RH services
by all women of child bearing age.
References
Allison, Paul D. 2001. Missing Data: Series/Number 07-136. Quantitative Applications
in the Social Sciences. London: Sage University.
Altfeld, Susan, Arden Handler, Dee Burton, and Leatrice Berman. 1997. “Wantedness of
pregnancy and prenatal health behaviors”, Women & Health 26(4): 29–43.
Basu, Alaka Malwade. 2002. “Why does education lead to lower fertility? A critical
review of some of the possibilities”, World Development 30 (10). 1779-1790.
Booth, A and Kee, H. J. 2005. “Birth Order Matters: The Effect of Family Size and Birth
Order on Educational Attainment”. Discussion Paper No. 506. The Australian
National University Centre for Economic Policy Research.
Bradbury, B. 2007. “Why do children of young mothers have poorer outcomes?”
Presentation at the 2007 LSAC research conference.
Burney, N. A. and Irfan, M. 1995. “Determinants of child school enrolment: evidence
from LDCs using choice-theoretic approach”, International Journal of Social
Economics, Vol. 22 No. 1, pp. 24-40
Campbell, Martha, Sahin-Hodoglugil, Nuriye Nalan, and Potts, Malcolm. 2006. “Barriers
to Fertility Regulation: A Review of the Literature”, Studies in Family Planning,
37, 87-98.
Cheng, D., Schwarz, E. B., Douglas E., and Horona, I. 2009. “Unintended pregnancy and
associated maternal preconception, prenatal and postpartum behaviors”,
Contraception 79, 194–198
Chernichovsky, D. 1995. “Socioeconomic and Demographic Aspects of School
Enrollment and Attendance in Rural Botswana”, Economic Development and
Cultural Change, Vol. 33, No. 2 pp. 319-332.
Chiswick, B and DebBurman, N. 2004. “Pre-School Enrollment: An Analysis by
Immigrant Generation”, CReAM Discussion Paper No 04/04. Centre for Research
and Analysis of Migration Department of Economics, University College London.
20
Cleland, J., Bernstein, S., Ezeh, A., Faundes, A., Glasier, A. and Innis, J. 2006. “Family
Planning: the unfinished agenda”, In the Lancet: Sexual and Reproductive Health.
P 47-63.
Coleman, James S., Campbell, E.Q., Hobson, C.J., McPartland, J., Mood, A.M.,
Weinfeld, F.D. and York, R.L. 1966. Equality of educational opportunity.
Washington, DC: U.S. Government Printing Office.
Dayioğlu, M., Kirdar, M. and Tansel, A. 2009. “Impact of Sibship Size, Birth Order and
Sex Composition on School Enrolment in Urban Turkey”, Oxford Bulletin of
Economics and Statistics Vol. 71, Issue 3, pp. 399-426.
DeGraff, D., Billsborrow, R. and Herrin, A. 1993. “The Implications of High Fertility for
Children’s Time Use in the Philippines”, Pp. 297-329. In C. Lloyd, ed. Fertility,
Family Size and Structure: Consequences for Families and Children. New York:
Population Council.
Dronkers, J. and Robert, P. 2008. “Differences in scholastic achievement of public,
private government-dependent and private independent schools: a cross-national
analysis”, Educational Policy Vol. 22, No. 4, pp. 541–77.
Duesenberry, J.S. 1952. Income, Saving and the Theory of Consumer Behaviour. Harvard
University Press, Harvard, Cambridge, MA.
Ejrnaes Mette and Portner Claus C. 2004. “Birth Order and the Intra-household
Allocation of Time and Education”, Review of Economics and Statistics 86(4);
1008-1019.
Echevarria, C., and Merlo, A. 1999. “Gender differences in education in a dynamic
Household Bargaining Model”, International Economic Review Vol. 40, no. 2.
Eloundou-Enyegue, P. 2000. “Tradeoff between Family Size and Education. The Africa
Notes”, Rural Sociology. Cornell University.
Evangelista de Carvalho Filho, Irineu 2008. “Household income as a determinant of child
labor and school enrollment in Brazil.” IMF Working Paper, WP/08/241.
Filmer, D., and Pritchett, L. 1999. “The effect of household wealth on educational
attainment: Evidence from 35 countries”, Population and Development Review,
25(1), 85–120.
21
Fuller, Bruce, and Prema Clarke. 1994. “Raising School Effects while Ignoring Culture?”
Review of Educational Research 64: 119-157.
Gipson, J. D., Koenig, M. A. and Hindin, M. J. 2008. “The Effects of Unintended
Pregnancy on Infant, Child, and Parental Health: A Review of the Literature”,
Studies in Family Planning 39[1]: 18–38).
Glewwe, Paul, and Hanan G. Jacoby 2004. “Economic growth and the demand for
education: is there a wealth effect?” Journal of Development Economics 74: 33-
51.
Glewwe, P. and Miguel, E.A. 2008. “The impact of Child Health and Nutrition on
Education in Less Developed Countries”, In: Handbook of Development
Economics, Volume 4.
Hanushek, E. A., and Wossmann, L. 2007. Education quality and economic growth.
Washington: The World Bank.
Hayes, H., Luchok, K., Martin, A. B., McKeown R. E. and Evans, A. 2006. “Short birth
intervals and the risk of school unreadiness among a Medicaid population in
South Carolina”, Child: Care, Health & Development 32, 4, 423–430.
Hazarika, G. 2001. “The Sensitivity of Primary School Enrollment to the Costs of Post-
Primary Schooling in Rural Pakistan: a Gender Perspective”, Education
Economics Vol. 9, No. 3
Heyneman, Stephen P., and William A. Loxley 1983. “The Effect of Primary-School
Quality on Academic Achievement Across Twenty-nine High- and Low-Income
Countries.” American Journal of Sociology 88: 1162-1194.
Hill, K., and Upchurch, D. M. 1995. “Evidence of gender differentials in child health
from the demographic health surveys”. Population and Development Review, 21.
Hox, J. 2002. Multilevel analysis: Techniques and applications. New York: Erlbaum.
Huisman, J and Smits, J. 2009. “Effects of household- and district-level factors on
primary school enrollment in 30 developing countries”, World development
Vol.37, No. 1. pp 179-193.
Jejeebhoy, S. (1992). “Women’s Education, Fertility and the Proximate Determinants of
Fertility.” Presented at Expert Group Meeting on Population and Women,
Gaborone.
22
Jencks, Christopher 1972. Inequality; A Reassessment of the Effect of Family and
Schooling in America. New York: Basic Books.
Joshi, S. and Schultz, T. P. 2007. “Family Planning as an investment in development:
evaluation of a program’s consequences in Matlab, Bangladesh”, Economic
Growth Center Discussion Paper 951, Yale University, New Haven CT.
Kirdar, M. 2007. “Explaining Ethnic Disparities in School Enrollment in Turkey”. MPRA
Paper No. 2649, http://mpra.ub.uni-muenchen.de/2649/
Kırdar, Murat G., Dayıoglu, Meltem and Tansel, Aysıt. 2007. “Impact of sibship size,
birth order and sex composition on school enrollment in urban turkey”, Middle
East.
Knodel, J., Havanon, N., and Sittitrai, W. 1990. Family size and the education of children
in the context of rapid fertility decline. Population and Development Review
16(1): 31-62.
Li, Hongbin, Junsen Zhang and Yi Zhu. 2008. The quantity-quality trade-off of children
in a developing country: Identification using Chinese twins. Demography, 45,
223-243.
Lloyd, C. and A. Gage-Brandon. 1993. “Does Sibsize Matter? The Implications of
Family Size for Children’s Education in Ghana” Pp. 481-519. In C. Lloyd (ed.).
Fertility, Family Size and Structure: Consequences for Families and Children.
New York: Population Council.
Maine, D., L. Freedman, F. Shaheed and S. Frautschi. 1994. “Risk, Reproduction and
Rights: The Uses of RH Data.” Pp. 203 -228. In R. Cassen (ed.). Population and
Development: Old Debates, New Conclusions. Washington, DC: Overseas
Development Council.
Mingat, Alain 2007. “Social Disparities in Education in Sub-Saharan African Countries.”
in International Studies in Educational Inequality, Theory and Policy, vol. 1,
edited by R. Teese, S. Lamb, and M. Duru-Bellat. Dordrecht: Springer.
Mugisha, F. 2006. “School enrollment among urban non-slum, slum and rural children in
Kenya: Is the urban advantage eroding?”, International Journal of Educational
Development 26(5) 471–482.
23
Nigerian Government. 2003. “Maternal and child health in Kano state”: Workshop report
conducted by members of the State Council of Ulama’s and the Ministry of
Health. http://www.jhuccp.org/legacy/countries/nigeria/KanoAdvocacyBooklet-
English.pdf
Park, H. 2008. “The varied educational effects of parent-child communication: a
comparative study of fourteen countries”, Comparative Education Review Vol.
52, No. 2, pp. 219–43.
Pettersson-Lidbom, Per & Skogman Thoursie, Peter, 2009. "Does child spacing affect
children’s outcomes? Evidence from a Swedish reform," Working Paper Series
2009:7, IFAU - Institute for Labour Market Policy Evaluation.
Rodriguez, F. and Wilson, E.J. 2000. “Are Poor Countries Losing the Information
Revolution?” InfoDev Working Paper. Washington D.C. World Bank.
http://www.infodev.org/library/working.htm
Schultz, T. P. 1981. Economics of Population. Reading, MA: Addison-Wesley
Publishers.
Schultz, T. P. 1997. “The Demand for Children in Low-Income Countries”. In M. R.
Rosenzweig and O. Stark (eds.) Handbook of Population and Family Economics
Vol 1A, Chapter 8. Amsterdam: North Holland Pub. Co.
Schultz, T. P. 2008. “Population Policies, Fertility, Women’s Human Capital, and Child
Quality”, Handbook of Development Economics, Volume 4.
Seligman, B., Kress, D., Winfrey, W., Feranil, I,, and Agarwal, K. 1997. Reproductive
Health and human capital: a framework for expanding policy dialogue.
Washington, D.C., Futures Group International, POLICY Project, 1997 Apr. [4],
27 p. (POLICY Occasional Papers No. 1USAID Contract No. CCP-3078-C-00-
5023-00).
Shavit, Yossi, and Hans-Peter Blossfeld 1993. Persistent Inequality; Changing
Educational Attainment in Thirteen Countries. Boulder: Westview Press.
Shrestha, C., and Manandhar, B. 2003. “How to Say ‘Birth Spacing’ in Kathmandu: A
Public/Private Partnership to Promote Contraception for Family Health in Nepal”.
24
Nepal: World Education, Inc. (reproduced from the PATH Web site at
http://path.org).
Singh, S., Darroch, J. E., Vlassoff, M., and Nadeau, J. 2004. Adding it up: the Benefits of
Investing in Sexual and Reproductive Health Care. New York: The Alan
Guttmacher Institute/UNFPA.
Sinha, Nistha, 2005, “Fertility, Child Work, and Schooling Consequences of Family
Planning Programs”, Economic Development and Cultural Change, 54(1): 97-
128.
Smits, J. & Gündüz- Hoşgör, A. 2006. “Effects of family background characteristics on
educational participation in Turkey”. International Journal of Educational
Development 26, 545-560.
Snijders, T., and Bosker, R. 1999. Multilevel analysis: An introduction to basic and
advanced multilevel modeling. London: Sage Publishers.
Song, L., Appleton, S., and Knight, J. 2006. “Why do girls in rural China have lower
school enrollment?” World Development 34(9), 1639–1653.
Tansel, Aysit 2002. “Determinants of school attainment of boys and girls in Turkey:
individual, household and community factors.” Economics of Education Review
21: 455-470.
Than, L.C., Honein, M.A., Watkins, M.L., Yoon, P.W., Daniel, K.L. and Correa, A.
2005. “Intent to become pregnant as a predictor of exposures during pregnancy: is
there a relation?” Journal of Reproductive Medicine 50, pp. 389–396.
Toor, I. A., and Parveen, R. 2004. “Factors Influencing Girls’ Primary Enrolment in
Pakistan”, The Lahore Journal of Economics. Volume 9, No.2
United Nations Educational, Scientific and Cultural Organization. 2007. Education for all
by 2015: Will we make it? Oxford: Oxford University Press.
UNESCO-UIS. 2009. Global education digest: Comparing Education Statistics across
the World.
UNFPA. 1995. Report of the International Conference on Population and Development.
United Nations, New York.
UNFPA. 2005. The Millennium Development Goals: Why Every Woman Counts?
25
USAID. 2005. Strengthening family planning policies and programs in developing
countries: An advocacy toolkit.
USAID. 2006. “Healthier Mothers and Children Through Birth Spacing”, Issue Brief.
http://www.usaid.gov/our_work/global_health/pop/news/issue_briefs/healthy_birt
hspacing.pdf
Weller, Robert H., Isaac W. Eberstein, and Mohamed Bailey. 1987. “Pregnancy
wantedness and maternal behavior during pregnancy”, Demography 24(3): 407–
412.
Wichman, A., Rodgers, J. L. and MacCallum, R. 2006. “A Multilevel Approach to the
Relationship between Birth Order and Intelligence”, Personality and Social
Psychology Bulletin 32 (1) 117-127.
World Bank. 2007. World development report 2007: Development and the next
generation. Washington: The World Bank.
World Bank. 2001. World Development Report 2000/01: Attacking Poverty. New York,
Oxford University Press.
World Bank. 1993. World Development Report 1993: Investing in Health. New York,
Oxford University Press.
26
Tables and Figures
TABLE 1 Logistic coefficients of the bivariate multilevel logistic regression analysis
BOYS GIRLS
Household level variables
Demographic factors
Age 0.237 (1.268)** 0.185 (1.203)**
Father missing -0.002 (0.998) 0.037 (1.038)
Birth order -0.017 (0.983)** 0.013 (1.013)*
Extended family -0.011 (0.989) 0.086 (1.089)**
Age of mother at first birth 0.212 (1.236)** 0.228 (1.256)**
Age of mother at first birth square -0.005 (0.995)** -0.005 (0.000)**
Socio-economic factors
Education father primary 0.506 (1.659)** 0.427 (1.533)**
Education father more than primary 1.483 (4.408)** 1.674 (5.331)**
Education mother at least some primary 1.302 (3.675)** 1.409 (4.093)**
Occupation father lower nonfarm 0.822 (2.276)** 0.761 (2.140)**
Occupation father upper nonfarm 1.255 (3.507)** 1.509 (4.523)**
Mother employed 0.122 (1.129)** 0.206 (1.229)**
Wealth index 0.281 (1.324)** 0.300 (1.350)**
Living in rural area -1.350 (0.259)** -1.400 (0.247)**
Contextual Variables
Economic factors
district level of development 0.675 (1.964)** 0.789 (2.201)**
National level of development -GDP 0.712 (2.038)** 0.837 (2.310)**
Educational facilities
Public expenditure on education 0.257 (1.293)** 0.346 (1.414)**
Health facilities
Hospital birth 0.852 (2.345)** 0.965 (2.625)**
DTP vaccination 0.759 (2.137)** 0.915 (2.496)**
RH factors
Number of siblings -0.091 (0.913)** -0.094 (0.911)**
Short preceding birth Interval -0.285 (0.752)** -0.231 (0.794)**
Short succeeding birth interval -0.194 (0.824)** -0.196 (0.822)**
Presence of young child in family -0.248 (0.781)** -0.334 (1.3970**
Mother currently pregnant -0.199 (0.819)** -0.245 (0.783)**
N 52469 50169
Children not in school 14600 15664
** P < 0.01; * P < 0.05
27
TABLE 2 Logistic and multiplicative (bracketed) coefficients of multilevel logistic
regression analysis for boys
Model 1 Model 2
Household level variables
Demographic variables
Age of child 0.229 (1.257)** 0.229 (1.257)**
Father missing -0.111 (0.895)** -0.126 (0.881)**
Birth order of child 0.019 (1.019) 0.019 (1.019)
Extended family -0.170 (0.843)** -0.173 (0.841)**
Age of mother at first birth 0.092 (1.096)** 0.092 (1.097)**
Age of mother at first birth square -0.002 (0.998)** -0.002 (0.998)**
Socio-economic factors
Education Father
None 0.000 (1.000) 0.000 (1.000)
At least some primary 0.824 (2.279)** 0.843 (2.323)**
At least some secondary 0.850 (2.339)** 0.850 (2.339)**
Education mother at least some primary 0.930 (2.536)** 0.921 (2.511)**
Occupation father
Farm 0.000 (1.000) 0.000 (1.000)
Lower nonfarm 0.426 (1.532)** 0.427 (1.533)**
Upper nonfarm 0.462 (1.5870** 0.450 (1.568)**
Mother employed 0.171 (1.186)** 0.175 (1.191)**
Household wealth 0.133 (1.142)** 0.133 (1.142)**
Living in a rural area -0.338 (0.713)** -0.333 (0.717)**
Contextual variables
Economic factors
District development index -0.112 (0.894)** -0.119 (0.888)**
National GDP per capita 0.387 (1.473)** 0.394 (1.483)**
Educational facilities
National public expenditure on Education 0.097 (1.102)** 0.094 (1.099)**
Health factors
Hospital birth 0.260 (1.297)** 0.260 (1.297)**
Vaccination for DPT 0.152 (1.1640** 0.151 (1.163)**
RH factors
Number of siblings -0.030 (0.970)** -0.038 (0.963)**
Short preceding birth interval -0.160 (0.852)** -0.162 (0.851)**
Short succeeding birth interval -0.019 (0.981) -0.032 (0.969)
Presence of young child in family -0.085 (0.918)** -0.142 (0.868)**
Mother currently pregnant -0.069 (0.933) -0.084 (0.920)*
** P < 0.01; * P < 0.05
28
TABLE 3 Logistic and multiplicative (bracketed) coefficients of multilevel logistic
regression analysis for girls
Model 1 Model 2
Household level variables
Demographic variables
Age of child 0.195 (1.215)** 0.195 (1.215)**
Father missing -0.096 (0.908)** -0.094 (0.910)**
Birth order of child 0.057 (1.058)** 0.053 (1.055)**
Extended family -0.071 (0.931)** -0.070 (0.932)**
Age of mother at first birth 0.079 (1.082)** 0.079 (1.082)**
Age of mother at first birth square -0.002 (0.998)** -0.002 (0.998)**
Socio-economic factors
Education Father
None 0.000 (1.000) 0.000 (1.000)
At least some primary 0.898 (2.455)** 0.901 (2.463)**
At least some secondary 1.047 (2.850)** 1.041 (2.832)**
Education mother at least some primary 1.124 (3.077)** 1.115 (3.050)**
Occupation father
Farm 0.000 (1.000) 0.000 (1.000)
Lower nonfarm 0.360 (1.433)** 0.350 (1.419)**
Upper nonfarm 0.565 (1.759)** 0.551 (1.735)**
Mother employed 0.245 (1.278)** 0.253 (1.288)**
Household wealth 0.131 (1.140)** 0.131 (1.140)**
Living in a rural area -0.329 (0.720)** -0.323 (0.7240**
Contextual variables
Economic factors
District development index -0.064 (0.938)* -0.068 (0.934)*
National GDP per capita 0.400 (1.492)** 0.408 (1.492)**
Educational facilities
National public expenditure on
Education
0.165 (1.179)** 0.163 (1.177)**
Health factors
Hospital birth 0.201 (1.223)** 0.202 (1.223)**
Vaccination for DPT 0.242 (1.274)** 0.242 (1.274)**
RH factors
Number of siblings -0.046 (0.955)** -0.048 (0.953)**
Short preceding birth interval -0.131 (0.877)** -0.155 (0.856)**
Short succeeding birth interval -0.056 (0.946) -0.071 (0.932)**
Presence of young child in family -0.167 (0.846)** -0.197 (0.821)**
Mother currently pregnant -0.071 (0.931) -0.100 (0.905)**
** P < 0.01; * P < 0.05
29
TABLE 4 Logistic and multiplicative (bracketed) interaction coefficients of
multilevel logistic regression analysis for boys and girls (Extension of Models 2 of
tables 2 and 3)
Boys Girls
Number of siblings
Education father at least secondary 0.063 (1.065)**
Education mother at least some primary -0.050 (0.951)** -0.051 (0.950)**
Mother employed -0.029 (0.971)** -0.027 (0.973)**
District level of development 0.022 (1.022)** 0.025 (1.025)**
National public expenditure on Education -0.013 (0.987)** -0.020 (0.980)**
Short preceding birth interval
Education father at least secondary -0.251 (0.7780**
Short succeeding birth interval
Father missing -0.189 (0.827)**
household wealth -0.026 (0.974)**
Education of mother (at least primary) 0.138 (1.148)*
National public expenditure on Education -0.036 (0.964)* -0.041 (0.960)*
Presence of young child in family
Education father at least some primary -0.159 (0.853)**
Living in rural area 0.210 (1.234)**
District level of development -0.083 (0.920)* -0.085 (0.918)*
Mother currently pregnant
Father missing -0.220 (0.802)*
Household wealth -0.031 (0.970)*
** P < 0.01; * P < 0.05
30
FIGURE 1 Theoretical model of school attendance
-father’s occupation (+)
-mother employment status (+/-)
-parent’s education (+)
-wealth (+)
-age of child (+)
-extended family (+)
-birth order (+) -missing father (-)
-age of mother at first birth (+/-)
Economic factors
Socio-economic factors
-modernization (+)
-urbanization (+)
-public expenditure on
education (+)
Education facilities
-availability of hospitals (+)
-vaccinations (+)
Health status
Primary School
Attendance
Demographic factors
Context level
-number of siblings (-) -preceding birth interval (+)
-succeeding birth interval (+) -mother’s pregnancy (-)
-presence of young child (-)
Family planning factors
Household
level
Top Related