School Enrollment and Attendance in Central South Somalia

A civil war has raged in Somalia for the past 20 years. The civil war fragmented the country into three zones: the Central South region, Somaliland, and Puntland. Puntland and Somaliland are relatively stable; however, Central South Somalia remains unstable. How has the ongoing civil war affected educational access in the Central South region of Somalia? This article examines 3,100 households, presents the extent of the education access in this increasingly unstable region of Somalia


Introduction
The push for universal primary education by 2015 has significantly increased educational enrollment in poor countries. However, the progress has been slow and uneven because significant obstacles still exist (Cohen, Bloom, & Malin, 2006;Lewin, 2009;United Nations Educational, Scientific andCultural Organization [UNESCO], 2010, 2011). Research shows that the most marginalized children are those facing poverty, geographic isolation, violent conflict, HIV/AIDS, corruption, and discrimination (Caillods, Phillips, Poisson, & Talbot, 2006;UNESCO, 2010). One barrier that has not received adequate attention is conflict. Conflict-affected countries are some of the farthest away from achieving universal education (UNESCO, 2010(UNESCO, , 2011. UNESCO (2011) reported that 18% of the children of primary school age are found in conflict countries, yet these countries account for 48% of out-of-school children. Reaching children in conflict countries is now a major challenge facing the international community.
Sub-Saharan Africa (SSA) has the highest number of violent conflicts in the world (UNESCO, 2011). One of the region's longest conflicts is the civil war in Somalia. Somalia has been at war since 1991. The civil war fragmented the country into three zones: the Central South region under the Transitional Federal Government (TFG), Somaliland in the northwest, and Puntland in the northeast. Somalia has received little research and policy attention (Abdi, 1998;Moyi, 2010). The little we know about education in Somalia is not encouraging. UNESCO (2005) reported that, "Somalia has the dismal distinction of having the world's highest proportion of primary school-age children not in school" (p. 24). Somaliland and Puntland have been able to establish functioning democracies and have also made gains in educational access (UNESCO, 2011). However, the Central South region remains unstable; it "has been devastated by clan rivalries, disputes over government and foreign invasions" (UNESCO, 2011 p. 243). The instability in Central South Somalia is undermining economic and social development.
Therefore, using the 2006 Somali Multiple Indicator Cluster Survey (MICS) data, this article seeks to examine the current state of education, extend our knowledge of the patterns of educational access in Central South Somalia, and identify the major challenges of expanding education access in this region of Somalia.

Education in Central South Somalia
There are significant challenges facing schooling in Central South Somalia. Despite the existence of the TFG of Somalia, Central South Somalia continues to experience violent conflict. Furthermore, unlike the other regions, it does not have a functioning ministry of education.
As a result of the conflict, parents are concerned for the safety of the children on the way to school and in school. Teachers are also concerned for their own safety because schools have been frequently targeted for attack-Students and teachers have been killed. UNESCO (2011) reported that "Transitional Federal Government and the Al-Shabaab militia have been cited by the Secretary-General for violence directed at schools" (p. 158). Parents also keep their children, especially the girls, at home to protect them from abduction; most of the abductions take place in school (Bekalo, Brophy, & Welford, 2003). Boys are used as combatants, and girls are used to cook and clean (O'Malley, 2010).
Most trained teachers fled the civil war in Somalia; female teachers are especially scarce. The gender balance among teachers affects whether or not girls attend school (Bekalo et al., 2003;Colclough, Al-Samarrai, Rose, & Tembon, 2003;Kirk 2003). This gender balance among teachers is especially important in a conservative Islamic country, like Somalia. Furthermore, a 1997 survey found that 30% of schools had no toilets, and about 90% had no running water in the school compound (Bekalo et al., 2003). The limited number of female teachers combined with the limited sanitation in schools is a significant barrier to schooling for girls.
A unique feature of schooling in Somalia is the Koranic school. Koranic schools are owned and operated by the community. They play a significant role in education especially early childhood education (Cassanelli & Abdikadir, 2007). They are nonformal schools 1 that seek to spread Islamic principles and lifestyle; however, they also teach arithmetic, Somali language, and Arabic (Bekalo et al., 2003;Cassanelli & Abdikadir, 2007;Morah, 2000). Despite their value in the community, Koranic schools face significant challenges. Most schools lack basic facilities like desks, chairs, and latrines. Researchers have highlighted the quality issues facing Koranic schools (Bekalo et al., 2003;Cassanelli & Abdikadir, 2007;Morah, 2000).

Data
To examine the school enrollment patterns, this study uses the Somalia 2006 MICS. The Somalia 2006 MICS was implemented by UNICEF Somalia in collaboration with the Pan-Arab Project for Family Health (PAPFAM) project of the League of Arab States. The Somalia MICS 2006 is the third MICS survey. The first MICS, which covered only Somaliland, was conducted in 1995, and the second MICS in 1999. The surveys aim to assess the situation of children and women in Somalia.
The Somalia 2006 MICS was a nationally representative sample survey that covered 5,969 households. This study uses only data from the Central South region. The sample from this zone had about 6,226 children in 3,100 households. The sample was selected in four stages. First, a predetermined number of clusters were selected in each zone-Somaliland (60 clusters), Puntland (60 clusters), and Central South Somalia (130 clusters). Second, districts were selected in each zone using proportional probability to size; within the districts, permanent and temporary settlements were selected. The temporary settlements were included to ensure they included nomads. Third, clusters were selected within settlements. Finally, households were randomly selected.
The questionnaires collected information on the household, the parent or guardian, and the eligible children (6-18 years). The questionnaires were used to provide current information on education among school-age children, with a focus on factors influencing household decisions about schooling.

Descriptive Statistics
To better understand Central South Somalia, it is important to briefly compare it with the other relatively stable regions of Somaliland and Puntland. Somaliland was described as the "quiet success story" because it lowered child mortality rates and made sustained gains in primary education (UNESCO, 2011). Table 1 presents household characteristics by region, rural/urban residence, wealth quintiles, mother's education, and father's education. About 63% of the children in Central South resided in rural areas compared with about 55% in Somaliland and 67% in Puntland. About 20% of Central South residents were rural nomads compared with about 4% in Somaliland and 15% in Puntland. The Central South Somalia had the largest proportion of rural nomads in Somalia. The challenges created by the frequent movement in remote areas combined with the problem of violent conflict are likely to create significant barriers to schooling. Table 1 also shows the socioeconomic status as measured by parents' education and wealth. About 47% of the children in Central South were in the poorest 40% compared with about 30% in Somaliland and 28% in Puntland. The households in the Central South zone were poorer than those in the other regions.
About 66% of the mothers in Central South reported they had no education compared with about 71% in Somaliland and 61% in Puntland. About 30% of fathers in Central South reported no education, compared with about 49% in Somaliland and 43% in Puntland. Less than 5% of the mothers reported they had at least a secondary education in all three regions. About 43% of fathers in Central South reported they had Koranic education compared with 9% in Somaliland and 28% in Puntland. It is also important to note that mothers and fathers in the Central South region had the lowest levels of primary and secondary education in Somalia. So, while it is important to highlight the gap between regions, we cannot ignore the overall low levels of educational attainment among parents in the three regions.
Children in the Central South region face conflict, poverty, inequality, and very low levels of parental education, especially mothers' education; these are formidable obstacles to schooling. Furthermore, due to their livelihood, nomads face extreme educational disadvantage.
To generate school enrollment data, the household respondents were asked the following question: "Has (name) ever attended school or preschool?" Those who reported they had enrolled in school were further asked if it was a formal or Koranic school. Table 2 presents the enrollment status of girls and boys in the three regions of Somalia. The statistics show a large gender and regional gap in school enrollment. About 47% of girls, compared with 29% of boys, had never enrolled in school. This gap was evident in all three zones in Somalia, but the gap is largest in the Central South zone. From Table 2, we can see that the Central South region had lowest proportion of children enrolled in formal school. For example, about 17% of girls reported they were enrolled in formal school compared with about 42% and 30% in Somaliland and Puntland, respectively. Conversely, the Central South region had about the highest proportion of children enrolled in Koranic schools. Table 3 presents the enrollment data, by gender and place of residence, for children in Central South Somalia. The gaps were not only between rural and urban households, but also between girls and boys. Table 3 also shows there is very low formal-school enrollment. Less than 20% of girls and boys in rural areas reported they had enrolled in formal school. Less than 46% of girls and boys in urban areas reported they had enrolled in school. This is not surprising given region's ongoing conflict. It is likely that rural areas had few functioning formal schools left; this may explain the higher enrollment in the nonformal, community-supported Koranic schools. Furthermore, Bekalo et al. (2003) also found, "many children do not attend formal school not only because it is unavailable, but also it is not sufficiently flexible to meet their particular needs" (p. 470).
The place of residence is a strong indicator of school enrollment of girls and boys. About 3% of nomadic girls reported they had enrolled in formal school compared with 39% in urban areas. For boys, about 4% of those from nomadic households reported they enrolled in school compared with 45% of those from urban households. About 72% of nomadic girls and 53% of nomadic boys reported they had never enrolled in school compared with 28% of urban girls and 16% of urban boys.
Majority of children reported they had enrolled in Koranic schools. However, there is also a gender gap in Koranic schools. For example, about 24% of rural sedentary girls reported they had enrolled in Koranic schools compared with 42% of rural sedentary boys. Among the nomads, about 25% of girls reported they had enrolled compared with 44% of boys. In summary, Table 3 shows there is limited school enrollment in the Central South region of Somalia. Furthermore, the situation was more acute when we examined the data by gender and place of residence. The most disadvantaged group was the rural nomad, especially girls. Table 4 presents the enrollment data, by gender and wealth quintiles, for children in Central South Somalia. There was a large gap in school enrollment between the poor and wealthy households. About 76% of girls and 51% of boys in the poorest quintiles reported they had not enrolled in school compared with 16% of girls and 8% of boys in the wealthiest quintile. It is important to note that the wealth gap was larger for girls. This gap was also evident, but smaller, when we looked at the enrollment in Koranic schools. How many of these children actually attend school? To generate school attendance data, the household respondents' were asked the following question: "During the (2005)(2006) school year, did (name) attend school at any time?" Table 5 presents the school attendance rates for Central South children by gender and place of residence. It shows that between 3% and 50% reported they had, in the 2005-2006 school year, attended formal school compared with between 4% and 53% of boys. Rural nomads had the highest proportion of children who reported they had not attended school during the year. About 48% of nomadic girls and 40% of nomadic boys were no longer attending school. Attendance rates in Koranic schools are higher, between 38% and 60%. The highest attendance rates in Koranic schools were found among the rural sedentary households. Table 6 presents the school attendance rates for Central South children by gender and wealth quintiles. The table shows a wealth gap in school attendance. About 28% of girls and boys in the poorest quintile were no longer in school compared with 8% girls and 6% boys in the wealthiest quintiles. Furthermore, less than 4% of girls and boys in the poorest quintiles were attending formal school compared with about 40% in the wealthiest quintile.
How do the school attendance patterns compare with the other regions? Table 7 presents the level attended for Central South Somalia and Somaliland. About 12% of Central South children reported they attended primary school compared with about 57% in Somaliland. About 87% of 5-to 9-year-olds in Central South started school in Koranic schools compared with about 36% in Somaliland. This may be due to lack of formal school and higher number of nomads in the Central South region. Since the community organizes Koranic schools, it is likely nomads start schools that follow their movements. Among 10-to 14-year-old children, about 32% of Central South children attended primary school compared with about 79% in Somaliland. In all age groups, children in Central South Somalia lagged behind their peers in Somaliland in terms of formal school. The descriptive statistics have presented a picture of limited school enrollment and attendance in Central South Somalia. Girls were more disadvantaged in enrollment and attendance, especially rural nomadic girls. Koranic schools serve a large proportion of school-going children, especially younger children. Schooling in Central South Somalia was characterized by low enrollment and attendance in formal school, gender, and rural/urban inequality. Finally, poverty and inequality-gender, wealth, rural/urban, and nomadic/ sedentary-emerged as obstacles to universal education in Central South Somalia.

Multivariate Analysis
The multivariate analysis estimates the household characteristics associated with the different school attendance choices. Children in Central South Somalia face three basic choices: attend Koranic school, attend formal school, or not attend any school. Therefore, we used multinomial logistic regression to analyze the data. A multinomial logistic model is an extension of logistic regression. It is extended when response variable has more than two categories. A response variable with n categories will therefore have n − 1 equations. These equations are binary logistic regressions that compare one category with the reference category (not attending school). The n − 1 multinomial logit equations contrast each of the Categories 1, 2, . . . n − 1 with Category n.
The reference group in the models were the children who were not currently attending school. Therefore, the estimates indicate the effect of the explanatory variables on the probability the child was either attending formal school or Koranic school, relative to the probability the child was not attending school. The models include the following independent variables: children's age and gender, number of children under age 5, number of 5-to 14-year-old children, gender of the head of household, place of residence (urban, rural sedentary, rural nomad), wealth quintiles, education of mother and father (none, Koranic education, at least primary education). The analysis is conducted for the whole sample and then separately for girls and boys.
The models of school attendance are presented in Tables  8 and 9. The tables with the multivariate analysis report the relative risk ratios (RRRs) for each variable in the model. The RRR is the ratio of the probability of choosing one outcome category over the probability of choosing the reference category (not attending school). A RRR that is greater than 1 indicates that the predictor variable will lead to an increase in the child being involved in that activity relative to the child not attending school. Conversely, a RRR that is less than 1 indicates that the predictor variable will lead to a decrease in the child being involved in that activity relative to not attending school. For example, in Table 8, the RRR of 0.686 for female variable means that girls were less likely to attend formal school compared with boys. The RRR of 3.695 for the child of the head of household means that the child of the head of household was more likely to attend formal school.
The results indicated an age gap in school attendance among Central South children. Children who were aged 10 to 14 had 2 times greater odds of formal-school attendance than children aged 5 to 9 years; they were 2 times more likely to be in formal school than be out of school. It is interesting to note there was no statistical difference in the odds of formal-school attendance between 5-to 9-year-olds and 15-to 17-year-olds. In the case of Koranic schooling, the odds of Koranic school attendance reduce with age-As children got older they were less likely to be in Koranic schools and more likely to be out of school. These findings hold across both models. Table 8 also indicates a gender gap in current formal and Koranic school attendance among children in Central South Somalia. Girls were consistently less likely to attend formal and Koranic schools; this finding was statistically significant in both models. Model 1 indicates that the odds for girls to attend formal school were about 32% lower than boys' odds; the odds for girls to attend Koranic school were 44% lower than boys' odds. Table 8 also indicates that children living in households with a greater number of children aged 5 to 14 increased the odds of formal-school attendance by 15%. Children of the head of household were about 4 times more likely to attend school. The number of children in the household also had an impact on school attendance. The larger the number of children aged 5 to 14, the greater the odds of formal-school attendance. Children whose mothers have at least primary education were about 2.5 times more likely to attend formal school than those whose mothers reported no education. The effects of fathers' education were similar but weaker and not statistically significant. Furthermore, children whose fathers' reported they only had Koranic education were about 53% less likely to attend formal school. Children in urban areas were 28 times more likely than rural nomads to attend formal school; rural sedentary children were 13 times more likely than rural nomads to attend formal school. The effects were similar but weaker when we look at rural/urban residence and Koranic schooling.
In Model 2, we add the household wealth quintiles. Household wealth has a large effect and is statistically significant on formal-school attendance but has no statistically significant effect on Koranic school attendance. Central South children in the wealthiest quintile had 16 times greater odds of attending formal school than those from the poorest quintile. The children in the second richest quintile were about 7 times more likely to attend formal school than those in the poorest quintile. These findings indicate that, in Central South Somalia, wealth inequality had a greater impact on school attendance than all other variables in the model.
When we controlled for household wealth, the effect of place of residence reduced substantially but remained statistically significant. This suggests that difficult conditions in rural areas combine with poverty to hinder school attendance; this means poor rural residents face double the disadvantage. The inclusion of wealth quintiles had little impact on the effect of age, gender, relationship to the head of household, and the number of children aged 5 to 14. However, controlling for wealth inequality reduced the effect of mothers' education on formal-school attendance; the effect declined from 2.454 to 1.712. This may suggest that some of the effects of educated mothers were due to their higher economic status.
The inclusion of wealth quintiles slightly increased the gender gap. In Model 1, the odds of girls attending school were 32% lower than boys' odds. In Model 2, the girls' odds were 37% lower than the boys' odds. To examine the gender effects more deeply, we estimated models separately for girls and boys. From Table 9, we see that differences in the probability of formal-school attendance between poor and wealthy girls are greater than the differences between poor and wealthy boys. Girls from the wealthiest household were 69 times more likely to attend formal school than those in the poorest quintile. In comparison, boys' from the wealthiest household were 6 times more likely to attend formal school than those in the poorest quintile. However, the rural/urban differences are greater for boys than girls. The age differences between girls were greater than the differences between boys. Girls in households with a larger number of children aged 5 to 14 had greater odds of school attendance; the differences for boys are not statistically significant. This suggests that girls benefit from larger households because they can share domestic chores to enable schooling to take place. Surprisingly, boys from households with educated mothers were 6 times more likely to attend school than those with uneducated mothers; the effect for girls with educated mothers is not statistically significant.
In summary, wealth, rural/urban residence, age, and mothers' education determine school attendance. We do find that wealth differences had a greater influence on girls, but rural/urban differences have a greater effect on boys. In Central South Somalia, differences in wealth were a significant barrier to school attendance; residing in a household in the wealthiest quintile substantially increases the odds of school attendance. Wealth combined with the place of residence and mothers' education to create layers of exclusion.

Discussion
The impact of violent conflict has been described in the literature as "de-development" (Abdi, 1998) and "development in reverse" (World Bank, 2003). Research found that the impact of violent conflict falls disproportionately on the poor and marginalized in society (United Nations Development Programme [UNDP], 2005). There is ample evidence that armed conflict reinforces existing obstacles associated with poverty and gender.
The descriptive and multivariate results show differences in the access to schooling in Central South Somalia. The differences exist in both school enrollment and attendance. Girls, the poor, and nomads are the most disadvantaged. A girl's access to school depends on the household socioeconomic status and place of residence; poverty and rural livelihoods combine to create significant obstacles to schooling. Nomadic lifestyle requires travel in search of pasture for their livestock. Consequently, children never stay in the same location for long, which makes it difficult for them to fit into school. Furthermore, these mostly illiterate and conservative nomads do not attach much importance to formal education. Instead they require their children to help them take care of the livestock and/or do housework.
However, it is important to remember that these problems of unequal access, poor quality, and inadequate financing were evident in Somalia even before the conflict (Bekalo et al., 2003;Cassanelli & Abdikadir, 2007;Morah, 2000;UNDP, 1998;World Bank, 2006). Therefore, the problems of the education sector in Somalia may have been exacerbated by the civil war.
The challenges facing children in Central South Somalia are substantial. First, the level of poverty remains very high in this region. The wealth differences are a significant barrier to school enrollment and attendance. Yet the ongoing conflict makes it difficult for any meaningful development to take place; the conflict will make it difficult to reduce the inequality. The wealth inequality exacerbates the gender inequality in access to school; poor girls face some of the greatest obstacles to schooling. The unequal access to education for girls may not be new because the levels of education of the mothers were also very low. About 15% of mothers reported they had any form of formal schooling. Low education of mothers is likely to impact health and education outcomes in the household. For example, children born to uneducated mothers are more likely to die before age 5 than children born to mothers with at least secondary education (UNESCO, 2011). Given the ongoing conflict, it seems unlikely that Central South Somalia will be able to reduce the levels of poverty; therefore, poverty will continue to be a significant barrier to schooling in the region.
Second, the region remains unstable; international efforts to stabilize Somalia led to the establishment of the TFG in Mogadishu. However, sharp divisions within members of the TFG and the limited assistance from the international community has left the TFG weak and ineffective, even in Mogadishu. The conflict makes it difficult for any meaningful development to take place. The education system had virtually collapsed when the civil war broke out in 1991, but now there is the additional burden of child soldiers, orphans, and other conflict-affected children. The limited schooling opportunities due to the war may explain why a high proportion of children had either not enrolled or have dropped out of school. The ability of the Somaliland region to raise enrollment in lower levels may be due to the relative stability in the region.
Third, Koranic schools remain prevalent in Central South Somalia. For many children in the region, they may be the only schools they find accessible. The school enrollment and attendance rates were higher, and the gender and wealth inequalities smaller in Koranic schools. However, the quality of these schools remains low-lack of qualified teachers (especially female teachers), lack of physical facilities, and limited financial support (Cassanelli & Abdikadir, 2007;Morah, 2000).
Although the descriptive statistics showed higher rates of formal-school attendance and lower rates of Koranic school attendance among wealthier households, it is not entirely clear whether the households chose to send their children to Koranic schools because they were a low-cost alternative to formal schools or because they wanted their children to learn Islamic principles and culture, or both. However, in Somaliland, Bekalo et al. (2003) found that "many children do not enter formal school not only because it is unavailable, but also it is not sufficiently flexible to meet their particular circumstances" (p. 470). This means that Koranic schools are not only more widely accessible but also serve the needs of the households. The needs may be religious and cultural. Given their prevalence and the wide acceptance, they may be the most effective way to achieve the universal education in Somalia. However, it is not clear whether western donors would be willing to support schools that, in part, provide Islamic education.
Fourth, the international community has increased aid flows to Somalia, but large financing gaps remain (Othieno, 2008;UNESCO, 2011). Furthermore, there exists a skewed pattern of aid to conflict countries. "Strategic" countries like Afghanistan and Iraq receive the lion's share of aid while other "peripheral" conflict countries like Somalia are largely ignored (UNESCO, 2011). Given the number of out-of-school children shown in this study, substantial resources are needed to provide education to all children in Central South Somalia.

Declaration of Conflicting Interests
The author(s) declared that they had no conflicts of interest with respect to their authorship or the publication of this article.

Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article. Note 1. Koranic schools are considered "non-formal" schools in the study because they operate without any coordinated guidance from any formal government authority; they work outside the framework of the formal school system.