EDUCATIONAL OUTCOMES OF SCHOOL FEEDING INTERVENTION: EVIDENCE FROM RURAL NORTHERN GHANA
Oloruntoba Abayomi, Musah, Bukari, Researchers Department of Agricultural Extension, Rural Development and Gender Studies University for Development Studies, Tamale, Ghana E-mail: yomitoba@hotmail.com; oabayomi@uds.edu.gh
ABSTRACT
The study investigated how policy intervention could have significant impact on beneficiaries. Using quasi-experimental design, 360 pupils from participating and non-participating schools in a feeding program were selected from a rural setting. Instrument for data collection was validated, pre-tested and administered on cross-section of respondents. Data were analyzed using descriptive statistics, t-test and regression analyses. Findings show that significant differences existed in educational outcomes as participating pupils performed better in core subjects of English Language, Mathematics and Integrated Science. In terms of socioeconomic determinants, findings show that selected variables such as sex of pupils and number of dependents in the family had directly impacted on the performance of pupils. The study also found causal link between the school feeding intervention and others as one of the multiplier effects was not only an increased enrolments by almost a quarter, but reduced dropout and absenteeism rates. This implied that the policy environments was sufficiently pro-poor and effective since it has strengthened the standard of foundational primary school and completion rate envisaged in the educational policy. It is recommended that further policy options that would facilitate the scaling-up of the program in the entire intervention area be formulated.
KEY WORDS
Educational outcomes; Academic performance; School feeding program; Ghana.
Over the last decade the high rates of pupil's dropout, absenteeism and low enrolments has been the bane of problems hindering the completion of primary schooling in vastly food insecure and disadvantaged communities in developing countries. Food security is the access by all people at all times to enough food for an active, healthy life, while on the contrary, the food-insecure households do not consume an adequate diet to maintain a healthy life. Therefore, when viewed against access to primary education, schooling of children of poor households could lead to unsavoury consequences. More worrisome is the trend in enrolment for girls which has continued to reduce unabated relative to boys which seems to be the order of the day especially among children from poorer households. The scenario painted a gloomy picture necessitating policy intervention from regional and national governments. The United Nations Millennium Development Goals (MDGs) and the Comprehensive African Agricultural Development Program (CAADP) Pillar-3 of the New Partnership for Africa Development (NEPAD) Program formulated policies for achieving universal primary education, food security and the reduction of hunger among the poor and vulnerable children in African countries. The Goal 2 of the UN-MDG seeks to achieve universal primary education with a target of ensuring that children, boys and girls alike will be able to complete a full course of primary schooling.
In Ghana, formal education remains the core pillar of human development with 18, 579 primary schools (GoG, 2010). The Universal Basic Education (UBE) which has become a right, mandatory and free for all pupils culminated into a dramatic improved access to primary education without corresponding attention to quality. For instance, Okyerefo et al., (2011) citing GoG (2008) and Etsey (2005) reported that there has been evidence of significant drop in the academic performance of pupils during the Basic Education Certificate Examinations in public schools between 2007 and 2010. In 2007, approximately 39 per cent that sat for the examination had aggregate 31-60 and the overall pass rate of 63 per cent. Again, Sarah et al., (2010) posited that despite the numerous efforts to improve the quality of primary
education the reality has been that the academic performance of pupils has even worsened. Furthermore, socio-economic backgrounds have also contributed to reducing differences in educational outcomes which are the achievements and benefits anticipated through the implementation of a policy intervention.
To stem the tide, recent events have rekindled stakeholders' interest in educational outcomes through the introduction of the Ghana School Feeding Program (GSFP) intervention in 2005 to stem the tide in poor academic performance among vulnerable children in food insecure area. The concept of SFP has gained ground not only in Ghana but in many countries. The intervention which has lately received renewed attention as a policy instrument was prompted by the NEPAD and MDG policies which allow poor and vulnerable children have access to a nutritious meal per day from locally available food crops for poor and disadvantaged pupils while in school. Access to school feeding and the quality of education is key and interventions aimed at making sure that hunger, poverty and universal primary education are achieved. This is because, getting children in school is one thing, but keeping them in school and making sure that they learn is another (Adamu-Issah et al., 2007). Consequently, the pilot phase of the school feeding program was initiated in Ghana in 10 selected schools across 10 Regions in Ghana. By 2008, it incorporated 170 districts and by 2011 it had covered 713,590 children in targeted schools across the entire country (GoG, 2006). The intervention has the objectives of rapidly improving the nutritional growth, and lead to educational outcomes by reduction dropout and absenteeism rates, increasing enrolment and academic performance in participating schools across the country.
The feeding program which has run from its inception in 2005 is still on-going. It has become highly auspicious and desirable at this time to assess its effectiveness as a policy intervention. Therefore it was presumed that measuring the effect of an intervention should provide answers to the following research questions:
Is the program implementable? If so, does the program achieve a set of desirable outcomes?
Does the school feeding intervention led to notable significant positive effect in educational outcomes of pupils in public primary schools?
What specific socio-economic variables determine significant educational outcomes of pupils in public schools?
Do parental socioeconomic characteristics have significant effect on pupil's academic performance?
Purpose of the Study. The purpose of the study was to investigate the effects of the school feeding program intervention on the educational outcomes of targeted pupils. Specifically, the objectives are to:
• ascertain if significant differences exist between participants and non-participants academic performance in core subjects of English Language, Mathematics and Integrated Science;
• determine the socio-economic variables that have effect on enrolments, absenteeism and academic performance;
• describe the parental / guardian socio-economic characteristics of pupils;
• ascertain if the school feeding program intervention has achieved the objective of setting it up as an educational policy.
Hypotheses for the Study. The general supposition is that pupils in participating schools would perform better than those from non-participating who did not in some educational outcomes. Similarly, boys would perform better than girls. The hypotheses for the study were set and in null forms as follows:
H01: There is no significant difference between pupils' academic performance in participating and non-participating schools.
H02: There is no significant difference between pupils' academic performance in core subjects of English Language, Mathematics and Integrated Science between participating and non-participating schools.
H03: There is no significant difference in enrolments between participating and non-participating schools.
H04: There is no significant difference between enrolments of boys and girls in participating schools
LITERATURE REVIEW
Educational outcomes could be divided into poverty reduction, nutritional and health status, food security and improved academic performance. Documentary evidences globally have reported educational outcomes with the school feeding program intervention. The variables of interests are socio-economic, enrolments, absenteeism, dropout rate and academic performance in core subjects of English Language, Mathematics and Integrated Science.
Adelman et al. (2008) have showed that in terms of performance, diet and nutrition play a critical role in the physical and intellectual development of pupils, this study was premised on the assumption that SFP would lead to improved educational outcomes particularly academic performance. Vermeersch (2003) showed a 30 per cent higher participation of children in the treatment group due to the effects of subsidized school meals, which also led to higher test scores. Kamlongera (2009) found that schools with feeding program in Malawi had higher enrolment, improved student's performance in class, a gender ratio of more than one in favor of girls, reduced absenteeism and parents encouraged to send their children than non-participating schools under the same circumstances. Contrary to this, Swart (2009) revealed that school feeding program was not effective and more should be done to improve its delivery. But, Cecilia and Maria (2011) found that school meals improved the academic performance of the pupils in English Language test scores though that could not be said of Mathematics. Etsy et al., (2009) reported that basic enrolment rate increased from 87-94 per cent between 2003 and 2006, while the completion rate stood at 85 per cent due to the Ghana School Feeding Program. Arkorful (2010) on the contrary posited that access to education for all school-aged children in Northern Ghana has not been met notwithstanding the number of state interventions for the marginalized and disadvantaged rural population. Rolleston et al., (2010) reported that Ghana is at the verge of achieving access to Universal Primary Education by 2015 due to exponential growth in enrolment figures backed by robust economic growth and increases in budgetary allocation in support of primary education. Ravallion (2011) observed that randomized control trials are sometimes used to estimate the aggregate benefit from some policy or programs.
On the impact of parenthood and birth order on academic achievement in primary school, Edun and Oguntola (2011) found no significant difference in the academic achievement of pupils with parents, single parents and no parents. Again, there was no significant difference in the academic achievement of 'first born' 'middle born' and 'last born. Stephanie (2012) posited that pupil's family and community background counts because the position of the child in relation to other siblings in the household may affect the schooling decision. On one hand, pupils born to the family early when resources are stretched over fewer numbers of household may be more likely to go to school. On the other hand, a child born into family later may have lower opportunity costs than an earlier born sibling because the need to look after other siblings within the family would be reduced. Graham-McGregor et al., (1998) reported that sex and type of residence / domicile could influence academic performance despite the school feeding program. Moreover, religious inequalities are also strong with Muslim girls being particularly disadvantaged. Girls from Moslem parents relative to Christian families have lower probability of attending secondary schools. Fogam (2010) imagined a Moslem daughter at age 10, 12, or 15 years, with so much of innocence, should be needing direction from parents not marriage and asserted that early marriage is capable of terminating the education of a girl child. On the part of the community, partnering schools to offer a wide range of services including safe and positive opportunities from which to choose can enhance students' achievement and well-being; and assist families as well (Sanders, 2003). However, limited community support and complex family characteristics compound these elements to create vulnerability in school-age children.
UNESCO (2012) reveal that Ghana has made progress over the past decade in reducing overall poverty from 52% to 40% between 1991/92 and 1998/99 (World Bank, 2004). It was also observed that extreme poverty declined from 37% to 27% over the same period (IMF, 2003). Again, poverty further fell to 28.5% in 2005/06 (Harold and Quentin, 2007). Ghana is on track to halve poverty by 2015, in line with the UN-MDG of eradicating extreme poverty and hunger (UNICEF Ghana, 2009). Ghana's Human Development Index (HDI) also rose by 0.9% annually from 0.391 to 0.558, between 1980 and 2012, which gives the country a rank of 135 out of 187 countries with comparable data. The HDI of sub-Saharan Africa as a region increased from 0.366 in 1980 to 0.475, placing Ghana above the regional average. However, despite these gains, gender vulnerabilities and risks remain significant (Christiana et al., 2010); income inequality has increased as shown by the worsening Gini coefficient (ISSER, 2007) and women are generally poorer than their men (Wrigley-Asante, 2008).
METHODOLOGY
Study Location. Garu-Tempane District is located in the Upper East Region of Ghana. It has an area of 1,230 km2 and population density of 99 persons per km2 lies between latitude 10° 38'N and 110N and longitude 00 06'E and 23'E in the south east corner of the Upper East Region with Garu as the capital. It shares boundaries with Bawku Municipal to the north, Bunkpurugu-Yuuyoo District to the south, Bawku West to the west and Republic of Togo to the east. The Northern Region is made up of the Northern, Upper West and Upper East. Of these, the Upper East was the poorest with most food -insecure households of approximately 20 per cent (The Daily Dispatch, 2013). Indeed the largest proportions of peasants who cultivate two hectares or less and are asset poor are in this region (Roger, 2006). According to Kazianga et al., (2012), poverty still has a firm grip in rural areas, especially in Northern Ghana. Since pupils / parental socio-economic conditions could also have affected the educational outcomes. This had on toward effects as many parents and guardians are unable to meet the basic educational needs of their children and wards. The area was selected for the study because poverty is still endemic and highly food insecure. The area has low access to primary education coupled with socio-economic and cultural factors which work against enrolment of pupils in public primary schools. This had on-toward effects as many parents and guardians are unable to meet the basic educational needs of their children and wards, despite the fact that primary education was expanded so as to achieve the Universal Primary Education (MOESS, 2008). In terms of performance in primary leaving examinations and gaining access to junior secondary schools, the area was also worst-off. Since inception in 2005, the District has been participating in the feeding program.
Study Design and Sample size. The study used a quasi-experimental design which was acceptable in the circumstance, given the human nature of the subjects and several other factors that may bear on the performance beyond the school feeding program intervention. Again, due to the absence of baseline data on the program which had ran for seven years. According to Duflo (2004), randomized design could be appropriate in order to obtain credible and transparent estimates of program impact that overcome the problems often encountered when using other evaluation practices. The critical objective of any evaluation of this nature was to establish that a non-participating group, who in the absence of the school feeding intervention would have had outcomes similar to the participants. In support of this assertion, Oloruntoba (2000) posited that the comparison group (non-participants) gives us an idea of what would have happened to the participating group if it had not been exposed, and thus allows us to obtain an estimate of the average effect on the treatment group (participants). The study therefore involved selection of 360 sample size using Krejcie and Morgan Table (1970) by randomized pairing of 180 participating and 180 non-participating pupils based on similar socio-economic characteristics. The pupils from grade four to six (aged 12-14 years) attending 12 schools were purposively selected because they were assumed to be matured enough and able to respond adequately to questionnaire. Moreover, these pupils would be most likely to have participated in the school feeding
program in the last few years. Stratified random sampling was used in selecting 10 pupils each from the class, while parents of each index pupil provided data on parental / guardian socio-economic characteristics. Head teachers / class teachers also assisted in the selection exercise. Consistent with national data, the average size of primary school class is small, at 35 pupils on the average per school (GoG, 2008).
Method of Data Collection and Analysis. Both primary and secondary data were used for the study. Primary data was obtained from the instrument developed to elicit data from respondents while secondary data from District Secretariat of GSFP and Ghana Education Service. The instrument was subjected to both face and content validity. Thereafter, a reliability of the instrument was conducted using the test-and-retest method at an interval of two weeks at a location outside the study area. A high Cronbach alpha coefficient, r=.72 was obtained which indicated that the instrument was reliable enough to be used for data collection. The instrument was administered on the respondents in various schools during the lunch break periods. Data collected was sorted, coded and analyzed using SPSS version 16.0 Special Edition for Window for descriptive statistics, t-test and multiple regressions analyses with decision taken a priori at 5%, level of significant.
RESULTS AND DISCUSSION
Socio-Economic Characteristics. The parental and guardian socio-economic status pupils pointed to peasantry as majority of the pupils (73%) reside in the rural area with their parents and guardian whose main occupation is farming (78.7%).
In support of this finding, FAO (2009) reported that in sub-Saharan Africa, nearly 75% of the extreme poor reside in rural areas, and 91% of the rural extremes poor are estimated to participate in agriculture. The level of illiteracy were high (60.1%) and with little or no education. The mean age of pupils was 13.2 years and the schooling age was between 4-14 years. More than half of the homestead was constructed with mud and zinc roof (54.5%). In terms of religion and type of marriage, most of them are Moslem (59.5%) and polygamous (63.3%) in nature. Grace et al., (2012) have reported that that parents / guardians have a vital role to play in the life of a child especially when schools work together with families to support learning.
Increased Enrolments. Findings show that the feeding program had succeeded in significant improvement in gross enrolment by 24 per in participating schools (Fig. 1). But sadly, the enrolment rate in non-participating schools plummeted to 7%. In addition, more than half (51.3%) of boys were enrolled compared to that of girls (48.7%). The positive signs offer a glimmer of hope for the program in the near future. According to Rolleston et al. (2010), Ghana is at the verge of achieving access to Universal Primary Education (UPE) by 2015 due to the exponential growth in enrolment figures backed by robust economic growth and increases in budgetary allocation in support of primary education.
Reduced Dropouts and Absenteeism. There was a sharp decline in dropouts and absenteeism rates which was significant in participating schools as against non-participating schools (Figs 2 and 3). Findings show that, gross drop-out reduced by almost half (48%) in participating schools but increased almost by the same proportion (49%) in non-participating schools. Interestingly, the drop-out rate among girls (54.2%) was slightly better than boys (45.8%). In terms of absenteeism, the school feeding program had succeeded in reducing gross absenteeism too by 35% in participating schools but a quarter of pupils (25%) in non-participating schools were truants; by sex, boys absenteeism (53.7%) was higher than girls (46.3%).
Improved Academic Performance. There were significant improvement in academic performance of primary six pupils in the First School Leaving Examinations in core subjects among participating school pupils The gross pass rate increased by 35% in English, 21% in Mathematics and 18% in Integrated Science as against a decrease of 26% in English, 34% in Mathematics and 30% in Integrated Science in non-participating schools respectively.
1400
1200
c 1000 v
E 800 2 600 £ 400 200 0
Girls
Boys
2008
Girls
Boys
2009
Girls
Boys
2010 Years
Girls
Boys
Girls
Boys
2011
Participant -Non-Participant
Figure 1 - Enrolment rates (2008-2012)
2012
Participant -Non-Participant
Figure 2 - Drop-out rates (2008-2012)
•—•Participant -Non-Participant
Figure 3 - Absenteeism rates (2008-2012)
Hypotheses Testing. H01: There is no significant difference between pupils' academic performance in participating and non-participating schools.
The results of t-test in Table 1 show that significant difference existed between the participating and non-participating pupils they performed better as reflected in their mean scores of 62.46 and 58.57 respectively. The difference in their mean scores is also statistically significant at p=0.05. The implication was that the GSFP has a positive relationship on the academic performance of pupils in participating schools. Consequently, one of the objectives of the SFP aimed at strengthening the academic performance of pupils in the GSFP schools has been achieved. This agrees with the findings by Kamlongera (2009) and Adelman et al., (2008) who also reported that school feeding improves academic performance of pupils. The null hypothesis is thus rejected. Hence, there is a significant difference in the academic performance of pupils in participating and non-participating schools.
Table 1 - t-test of Pupils' Academic Performance in Participating and Non-Participating Schools
Variables Mean N Std. Deviation Std. Error Mean Difference t -test df P Decision
Participating 62.46 174 16.40 1.24 3.89 2.24 173 0.03 S
Non- Participating 58.57 174 17.19 1.30
*p = 0.05
H02: There is no significant difference between pupils' academic performance in core subjects of English Language, Mathematics and Integrated Science between participating and non-participating schools.
Academic performance was one of the principal objectives of the School Feeding Program. Hence, it is likely that pupils who participated in it would be different from those who did not. In this study, the possible difference between academic performances in English Language, Mathematics and Integrated Science were determined using t-test statistic.
For English Language, Table 2 shows that participating schools performed more academically than the non-participating schools as reflected in their mean scores of 64.37 and 58.71 respectively. The difference between their mean scores is also statistically significant showing that the school feeding program intervention affected academic performance in English Language. This implied that the participating and non-participating schools were not equivalent in terms of their academic performance in English Language after the GSFP. This is because, while the academic performance in English Language of the former is on the increase, that of the later group is not. Therefore, the difference in means was not by chance. The null hypothesis is thus rejected.
For Mathematics, again, the possible difference between academic performances in Mathematics was determined using t-test statistic. Table 2 shows that participating schools performed more academically than the non-participating schools as reflected in their mean scores of 60.78 and 56.71 respectively. The difference between their mean scores is also statistically significant showing that the school feeding program intervention affected academic performance in Mathematics. This implied that the participating and non-participating schools were not comparable in terms of their academic performance in Mathematics after the GSFP. While the academic performance in Mathematics of the former is on the increase, that of the later group is not. Therefore, the difference in means was not by chance. The null hypothesis is thus rejected.
For Integrated Science, the possible difference between academic performances in Integrated Science was determined using t-test statistic. Table 2 shows that participating schools performed better than non-participating schools as reflected in their mean scores of 62.22 and 60.50 respectively. However, the difference between the mean scores is not statistically significant showing that the school feeding program intervention did not affect academic performance in Integrated Science substantially. Hence, the null hypothesis is thus accepted.
Table 2 - t-test of Performances in Core subjects of Pupils in Participating and
non-Participating Schools
Variables Mean N Std. Deviation Std. Error Mean Difference t -test df P Decision
English Language:
Participating 64.37 178 19.79 1.48 5.66 3.12 177 0.00 Significant
Non- Participating 58.71 178 18.99 1.42
Mathematics:
Participating 60.78 178 18.86 1.41 4.07 2.10 177 0.04 Significant
Non- Participating 56.71 178 18.74 1.40
Integrated Science:
Participating 62.22 177 19.30 1.45 1.72 0.77 176 0.44 Not Signif
Non- Participating 60.50 177 21.50 1.62
*p = 0.05
H03: There is no significant difference in enrolments between participating and non-participating schools.
Table 3 shows that enrolments of pupils in participating schools were much improved than non-participating as reflected in their mean scores of 302.13 and 290.13 respectively. However, the difference in their mean scores was statistically significant indicating that the school feeding program intervention affects enrolments of pupils in participating schools substantially.
Table 3: t-test Statistics between Pupils' Enrolments in Participating and Non-Participating Schools
Variables Mean N SD SE Mean Difference t -test df P Decision
Participating 302.13 55 182.55 24.62 12.00 0.44 54 0.04 Significant
Non- Participating 290.13 55 122.21 16.48
*p = 0.05
H04: There is no significant difference between enrolments of boys and girls in participating Schools.
Table 4 presents the t-test statistics of difference in enrolments of boys and girls in participating schools.
Results show that boys enrolled more in schools than girls as reflected in their mean scores of 159.47 and 143.65 respectively. The difference between their mean scores is also statistically significant indicating that the school feeding program intervention affected enrolments of boys and girls substantially.
Table 4 - t-test statistics between pupils' enrolments in participating schools
Variables Mean N Std. Deviation Std. Error Mean Difference t -test df P Decision
Boys 159.47 55 92.51 12.47 16.82 4.34 54 0.00 Significant
Girls 143.65 55 92.29 12.45
*p = 0.05
Socio-Economic Determinants of Performance in Core Subjects. Regression was used to assess the effect of certain socio-economic factors on performance in core subjects of English Language, Mathematics and Integrated Science of pupils. The linear regression equation (1) is represented thus:
Y = A + P1X1 + P2X2 + P3X3 + P4X4 + P5X5 + PaXa + P7X7 + P8X8 + P9X9 + € (1),
where:
Y = Core subject;
A = Constant; X1 = Sex;
X2 = Age (in years);
X3 = Type of Marriage by Parents/Guardian;
X4 = Religion of Pupils' Parent/Guardian;
X5 = Number of Dependents;
X6 = Type of Residence/Domicile;
X7 = Type of Dwelling;
X8 = Occupation of Parent/Guardian;
X9 = Highest Qualification of Parent/Guardian;
€ = Stochastic error term.
English Language (Participating Schools). Table 5 showed the results of fitted regression for English Language (Y) estimated from equations 2 and 3 were obtained:
Therefore,
Y = 8.01 + -0.17 (X!) + -0.02 (X2) + 0.04 (X3) + -0.11 (X4) + 0.15 (X5) + 0.16 (X6) + 0.04 (X7) + -0.07 (X8)
+ 0.03 (X9) + R2 = 0.10 (2)
Y = 8.01 + -0.17 ((-2.67) + -0.02 (0.13) + 0.04 (0.56) + -0.11 (-1.54) + 0.15 (2.32) + 0.16 (1.78) + 0.04
(0.49) + -0.07 ((-0.73) + 0.03 (0.64) + R2 = 0.10 (3)
The coefficient of the three explanatory variables, that is, sex of the pupils -0.17(X1), number of dependents 0.15(X5) and type of domicile 0.16(X6) were found to be significant at 1%, 5% and 10%. The Coefficient of Multiple Determination, R2 was found to be 0.10 meaning that 10 per cent variation in academic performance in English Language is explained by the independent variables identified.
Mathematics (Participating Schools). Table 6 showed the results of fitted regression for Mathematics (Y) estimated from equations 2 and 3 were obtained:
Therefore,
Y = 9.30 + -0.25 (X1) + -0.22 (X2) + 0.12 (X3) + -0.01 (X4) + 0.29 (X5) + 0.15 (X6) + 0.00 (X7) + -0.06 (X8)
+ -0.00 (X9) + R2 = 0.22 (2)
Y = 9.30 + -0.25 (-4.04) + -0.22 (-1.64) + 0.12 (1.91) + -0.01 (-0.22) + 0.29 (4.42) + 0.15 (1.79) + 0.00
(0.06) + -0.06 (-0.65) + -0.00 (-0.08) + R2 = 0.22 (3)
The coefficient of the three explanatory variables, that is, sex of the pupils -0.25(X1), age -0.22(X2), type of marriage by parents / guardians 0.12(X3), religion of parents / guardians -0.1(X4) and number of dependents 0.29(X5) were found to be significant at sex and religion of parents / parents at 1%, age of pupils, type of marriage and number of dependents at 10%. The Coefficient of Multiple Determination, R2 was found to be 0.22 meaning that 22 per cent variation in academic performance in Mathematics is explained by the independent variables identified.
Integrated Science (Participating Schools). Table 7 showed the results of fitted regression for Integrated Science (Y) estimated from equations 2 and 3 were obtained:
Therefore,
Y = 8.04 + -0.13 (X1) + 0.05 (X2) + 0.05 (X3) + -0.04 (X4) + 0.14 (X5) + 0.19 (X6) + -0.06 (X7) + -0.05 (X8) + 0.03 (X9) + R2 = 0.11 (2)
Y = 8.04 + -0.13 (-2.31) + 0.05 (0.44) + 0.05 (0.88) + -0.04 (-0.67) + 0.14 (2.33) + 0.19 (2.44) + -0.06 (-
0.89) + -0.05 (-0.69) + 0.03 (0.65) + R2 = 0.11 (3)
The coefficient of the three explanatory variables, that is, sex of the pupils -0.13(X1), number of dependents -0.14(X5), and type of residence 0.19(X6) were found to be significant at 5%. The Coefficient of Multiple Determination, R2 was found to be 0.11 meaning that 11 per cent variation in academic performance in Integrated Science is explained by the independent variables identified.
Generally the regression results showed that the determinants socioeconomic variables were sex of pupils and number of dependents which had directly impacted on the performance of pupils in core subjects of English Language (R2=0.10), Mathematics (R2=0.22) and Integrated Science (R2=-0.14). This suggests that improved performance in core subjects could be achieved by giving consideration to those significant variables. Stephanie (2012) reported that sex of a child could influence performance despite the feeding program. Again, the type of domicile / residence was also significant and could influence performance. Grantham-McGregor et al., (1998) have also reported that the type of residence / domicile could have influence on pupil's performance.
CONCLUSIONS AND RECOMMENDATIONS
Since educational outcomes were one of the principal objectives of the School Feeding Program intervention, the results of the study indicated that adoption of the program was high among the targeted public primary schools. Therefore, the School Feeding Program could well be said to be well-implemented as one of the yardsticks used by policymakers interested in the impact on the beneficiaries. This assertion is in line with that of Beatrice et al. (2004) who reported a significant rise in enrolment rates in intervention schools in rural Bangladesh between baseline and follow up, constant in the urban areas in intervention schools, but fell significantly in the control schools without feeding program. There seems to be a causal link between intervention and educational outcomes as evident in enrolment rates, absenteeism and drop-out rates and improved performance in core subjects of English Language, Mathematics and Integrated Science. Generally, the impact of the program suggested positive effect of the intervention on school children in targeted primary schools. The Ghana School Feeding Program intervention as a pro-poor policy has succeeded in achieving its objectives of increasing access to primary education and reduction in poverty by targeting the poor and vulnerable school pupils in food insecure rural area. Due to the program, there has been an upsurge in school enrolment by almost a quarter from inception leading to increased primary school completion rate. However, more still need to be done in re-targeting beneficiary schools and pupils to capture the real poor children. Girl's education has also been guaranteed and absenteeism among boys reduced. However, it is difficult to generalize and isolate the effect of educational outcomes to the feeding program intervention alone. This was because the study was limited to a district in food insecure region. Furthermore, the absence of data on the nutritional status of pupils, quality of food served, teachers' quality and work attitude, parental attitude towards supporting their children, absenteeism among teachers, quality and types of infrastructure in the school learning environment and other externalities such as government policies may all have profound influence. Nevertheless, the study has shown that the school feeding was a success and has benefited the poor and disadvantaged pupils.
The findings have implication for poverty reduction since the program had ensured that the less-privileged pupils have access to primary education and increased completion rate. Again, the dropout and absenteeism rates and improved performance in core subjects have implication for higher literacy. However, the in-equalizing affect both across and within participating and non-participating school pupils in the District may explain why in-depth socio-economic and vulnerability studies are imperative in scaling-up the program. Consequently, when these submissions are viewed against poverty reduction in the District, it could be infer that the noticeable increase in primary school enrolment and completion rate would have profound effects on poverty.
Table 5 - Regression Results for English Language
Equation N A X1 X2 X3 X4 X5 X6 X7 Xs X9 R2
Linear 160 5.18 -6.90** (-2.21) 0.00 (0.04) 1.83 (1.35) -2.93 (-1.51) 2.78* (1.82) 4.91* (1.94) 2.31 (0.92) -0.18 (-0.10) 0.48 (0.47) ++ 0.10
Semi Log 160 2.43 -6.94** (-2.23) 0.58 (0.09) 3.46 (1.08) -6.04* (-1.78) 6.48** (1.98) 8.22* (1.89) 2.69 (0.63) -2.39 (-0.53) 1.88 (0.71) + 0.10
Double Log 160 8.01 0 17*** (-2.67) -0.02 (0.13) 0.04 (0.56) -0.11 (-1.54) 0.15** (2.32) 0.16* (1.78) 0.04 (0.49) -0.07 (-0.73) 0.03 (0.64) + 0.10
Figures in parentheses are the t-value: *** t-value significant at 1%, ** t-value significant at 5%, * t-value significant at 10%, ++ F-value significant at 5%, + F-value significant at 10%. Legend: X1 = Sex; X2 = Age (in years); X3 = Type of Marriage by Parents/Guardian; X4 = Religion of Pupils' Parent/Guardian; X5 = Number of Dependent; Xe = Type of Residence/Domicile; X7 = Type of Dwelling; Xs = Occupation of Parent/Guardian; X9 = Highest Qualification of Parent/Guardian
Table 6 - Regression Results for Mathematics
Equation N A X1 X2 X3 X4 X5 X6 X7 X8 X9 R2
Linear 160 6.44 -10.79*** (-3.74) -0.23** (-2.07) 3.06** (2.43) -0.08 (-0.05) 6.19*** (4.39) 4.35* (1.86) 0.77 (0.33) -1.50 (-0.88) 0.148 (0.15) +++ 0.21
Semi Log 160 4.47 -11.38*** (-3.95) -11.62* (-1.91) 5.77* (1.95) 0.19 (0.06) 13.61*** (4.50) 7.98** (1.99) 1.37 (0.35) 0.33 (0.08) -0.42 (-0.17) +++ 0.22
Double Log 160 9.30 -0.25*** (-4.04) -0.22* (-1.64) 0.12* (1.91) -0.01*** (-0.22) 0.29* (4.42) 0.15 (1.79) 0.00 (0.06) -0.06 (-0.65) -0.00 (-0.08) +++ 0.22
Figures in parentheses are the t-value: *** t-value significant at 1%; ** t-value significant at 5%; * t-value significant at 10%; +++ F-value significant at 1%; ++ F-value significant at 5%; + F-value significant at 10%. Legend: X1 = Sex; X2 = Age (in years); X3 = Type of Marriage by Parents/Guardian; X4 = Religion of Pupils' Parent/Guardian; X5 = Number of Dependent; Xe = Type of Residence/Domicile; X7 = Type of Dwelling; Xs = Occupation of Parent/Guardian; X9 = Highest Qualification of Parent/Guardian.
Table 7 - Regression Results for Integrated Science (Participating Schools)
Equation N A X1 X2 X3 X4 X5 X6 X7 X8 X9 R2
Linear 159 0.00 -6.38** (-2.11) 0.02 (0.13) 1.19 (0.91) -0.48 (-0.26) 3.47** (2.35) 8.84*** (3.60) -2.39 (-0.98) -0.96 (-0.54) 1.15 (1.15) +++ 0.13
Semi Log 159 0.03 -6.64** (-2.19) 2.28 (0.36) 2.63 (0.85) -1.48 (-0.45) 7.85** (2.47) 13.35*** (3.14) -3.36 (-0.81) -1.89 (-0.42) 2.49 (0.97) ++ 0.12
Double Log 159 8.04 -0 13** (-2.31) 0.05 (0.44) 0.05 (0.88) -0.04 (-0.67) 0.14** (2.33) 0.19** (2.44) -0.06 (-0.89) -0.05 (-0.69) 0.03 (0.65) ++ 0.11
Figures in parentheses are the t-value: *** t-value significant at 1%; ** t-value significant at 5%; * t-value significant at 10%; +++ F-value significant at 1%; ++ F-value significant at 5%; + F-value significant at 10%. Legend: Xi = Sex; X2 = Age (in years); X3 = Type of Marriage by Parents/Guardian; X4 = Religion of Pupils' Parent/Guardian; X5 = Number of Dependent; X6 = Type of Residence/Domicile; X7 = Type of Dwelling; X8 = Occupation of Parent/Guardian; Xg = Highest Qualification of Parent/Guardian.
Policy Recommendations. For sustainability, the school feeding programs should be well-targeted not only on the basis of food insecurity but through a more rigorous in-depth socio-economic survey and vulnerability mapping with a view to scaling-up the program to incorporate more schools in food deficit areas. To this end, the following policy option was recommended:
Government should formulate new policy that would clearly differentiate the real poor and disadvantaged children from vulnerable homes from those better-offs. This is to prevent the program from being hijacked by unintended beneficiaries. The identified problem of low literacy rates which started with low enrolment in primary schools will be overcome with this policy.
Awareness creation among the communities in the area should be intensified. Consequently, more pupils especially girls need to be encouraged to enroll and reduce absenteeism / dropouts this would help increase the completion rate in primary school. Furthermore, the present enrolment policy of 20% annual increment in participating schools though laudable but still a far cry in relation to the total number of rural-based primary schools in the District. There is a need for an upward review based on available finance to the government. More funds would have to be committed and more teachers recruited in order to scale up the program in the entire food deficit rural area in the region. This will in real term means an increase in the budgetary allocation for personnel emoluments and incidental costs.
There is an urgent need to partner with Non-Governmental Organizations as well as the private sector to foster strong commitment to the program. The idea of Private Public Partnership (PPP) will boost the feeding program in schools and contribute to eradication of illiteracy in food insecure areas. To this end, government should encourage Corporate Social Responsibility (CSR)-driven initiative through mounting of strong advocacy. For instance, corporations could
be made to immediately embark on 'Adopt-A-School Initiative' and be granted tax holidays for their participation. In addition, aggressive advocacy should also be mounted on parents and guardians on the need for enrolments of school-age children as well as ensuring their regular attendance in schools.
There is a need to also strengthen the institutions responsible for the governance of the program such as the Ministry of Education, District Education Service and Primary Schools by enhancing their funding. It is also imperative that the school feeding program monitoring process should be localized over time by enhancing the capacities of rural communities whose children are beneficiaries to monitor the progress. Schools participating in the program need to learn from each other about what works and under what circumstances, through an exchange program.
ACKNOWLEDGEM ENT
We are most grateful to IFPRI office Ghana for the provision of a study grant. We also wish to thank the University for Development Studies for creating an enabling environment for the research to take place. We thank and appreciate the cooperation and assistance of the District Director, Garu-Tempane District Education Service, District Circuit Supervisors, Headmasters, parents / guardians and pupils during the field data collection.
REFERENCES
1. Adamu-Issah M., Elden, L., Forson, M., and Schrofer, T. (2007). 'Achieving Universal Primary Education in Ghana by 2015. A Reality or a Dream? Division of Policy and Planning, UNICEF.
2. Adelman, W. S., Daniel, O. G. and Kim, L. (2008). How Effective are Food for Education
3. Program? A Critical Assessment of the Evidence from Developing Countries. Food Policy Review 9. International Food Policy Research Institute. Washington, D.C. USA. DOI: 10 2499/0895095 FPREV. 9.
4. Arkorful, K. (2010). Complementary Education and Access to Primary Schooling in Northern Ghana, Create Ghana Policy Brief 2. Ghana.
5. Beatrice, L. R., Jannifer, C. and Akoto K. O. (2004). WFP Bangladesh School Feeding Program.
6. Mid Term Evaluation Final Report. Friedman School National Science and Policy, Tufts University 150 Harrison Avenue, Boston, Ma02111, USA.
7. Cecilia, A and Maria, E. O. (2011). Do In-school Feeding Programs have Impact on Academic Performance and Dropouts? The Case of Public Argentine Schools. UdeSA-CONICET, Argentina.
8. Christiana, G, Nicola, J. and Pabla, P. (2010). Project Briefing. Overseas Development Institute, No. 52.
9. Duflo, E, (2004). 'Using Randomization in Development Economics Research: A Toolkit'. Working Paper 138. Cambridge, Mass. USA: Harvard University, Center for International Development.
10. Edun, T. and Oguntola, S. (2011). The Impact of Parenthood and Birth Order in the Academic Achievement of Pupils in Abeokuta South Local Government Area, Ogun State, Nigeria. European Journal of Social Sciences, Vol. 20. No. 4.
11. Etsey, K (2005). Causes of Low Academic Performance of Primary School Pupils in the Shama Ahanta East Metropolitan Assembly (SAEMA) in Ghana. Regional Conference of Education in West Africa, Dakar, Senegal.
12. Etsy, K., Thomas, M. S., Emma, G., Joe K., Jenifer, D., Emmanuel, H. and Stephen, P. H. (2009).
13. Review of Basic Education Quality in Ghana: Basic Education in Ghana; Progress and Problems. Final Report. USAID, USA.
14. FAO (2009). 'How to Feed the World in 2050'. High level expert forum-the special challenge for sub-Saharan Africa. Rome, Italy: FAO. Retrieved from http://www.fao.org/fileadmin/templates/wsfs/docs/Issues_papers/HLEF2050_Africa.pdf.
15. Fogam, P. (2010). Early Marriage, Threat to Girl-Child Education. Vanguard Education Weekly, Lagos, Nigeria.
16. GoG, (2006). 'Ghana School Feeding Program Document 2007-2010', Government of Ghana, 77 pp.
17. GoG (2008). Preliminary Education Sector Performance Report. Ghana.
18. GoG (2010). Education Sector Performance Report 2010. Ministry of Education, Ghana
19. Grace, A. M., Jethro, O. O. and Aina, F. (2012). Roles of Parent on the Academic Performance of Pupils in Elementary Schools. International Journal of Academic Research in Business and Social Sciences. Vol. 2, No. 1. ISSN: 2222-6990. www.hrmars.com/journals.
20. Grantham-McGregor, S., Chang, S. and Walker, P. S. (1998). Evaluation of School Feeding Programs: Some Jamaican Examples. Jamaica.
21. Harold, C. and Quentin, W. (2007). Poverty, Livelihoods and Access to Basic Services in Ghana.
22. Ghana CEM: Meeting the Challenge of Accelerated and Shared Growth.
23. ISSER (2007). Shared and Inclusive Growth in Ghana. Report for DFID. Accra
24. International Monetary Fund (2003). Ghana: Poverty Reduction Strategy Paper. Washington DC. IMF Country Report No. 03/56. http://www.imf.org.
25. Kamlongera, A. F. (2009). School Feeding in Malawi: A Strategy for Equitable Access in Education, Cairo, June 15-18
26. Kazianga, H., Damien, de W and Harold, A. (2012). Educational and Child Labor Impacts of Two Food for Education Schemes: Evidence from a Randomized Trial in Rural Burkina Faso. JEL Codes: D04, 120,015.
27. Krejcie, R.V. and Morgan, D. W. (1970). Determining Sample Size for Research Activities, Educational and Psychological Measurements.
28. MOESS (2008). 'Preliminary Education Sector Performance Report' Ghana.
29. Okyerefo M. P. K., Fiaveh Y. D. and Lamtey S. N. L. (2011). Factors Prompting Pupils' Academic Performance in Privately owned Junior High Schools in Accra, Ghana. International Journal of Sociology and Anthropology vol.3 (8), pp. 280-289.
30. Oloruntoba, A. (2000). 'Evaluation of Management Training Program on Job Behaviour of Senior Agricultural Research Managers in Nigeria.' PhD Thesis Submitted to the Department of Agricultural Extension and Rural Development, University of Ibadan, Ibadan, Nigeria 227 pp.
31. Ravallion, M. (2011). 'On the Implications of Essential Heterogeneity for Estimating Causal Impacts Using Social Experiments. World Bank Policy Research Working Paper 5804. World Bank, Washington DC 13 pp.
32. Roger, B. (2006). 'Working Paper: Background conditions in Upper East Region, Northern Ghana, IFAD.
33. Rolleston, C., Akyeampong, K., Ampiah, G. J., and Lewin M. K. (2010). 'Educational Access in Ghana'; Country Research Summary, Consortium for Research on Educational Access, Transitions and Equity, Ghana.
34. Sanders, G. M. (2003). Community Involvement in Schools: From Concept to Practice. John Hopkins University. Education and Urban Society, vol.35 No. 2, 161-180, DOI: 10. 1177/00131 24502239390.
35. Sarah, N. D. A., Micheal, A. A., Harriet, A. M. A., Anita, A. and Isaac A. (2010). Dangme West District and Quality Education. Southern Portfolio, SNV, Ghana.
36. Stephanie, B. (2012). 'Food for Thought: Evaluating the Impact of Indian's Mid-Day Meal Program on Educational Attainment'. Department of Economics, University of California, Berkeley.
37. Swartz, A. M. (2009). An Evaluation of the School Feeding Programme as a Service Delivery Mechanism to Improve Academic Performance of Needy Learners in Bonteheuwel. CPUT Theses and Dissertations page 143 http:dk.cput.ac.za/td-cput/143.
38. The Daily Dispatch (2013). The Food Security Situation in Northern Ghana. The Daily Dispatch, March 26, 2013 pp. 9.
39. UNESCO (2012). Youth and Skills: Putting Education to Work. 10th Edition of Education for All Global Monitoring Report. UNESCO Publishing; Second Edition.
40. UNICEF Ghana (2009). Child Poverty and Disparities in Ghana: UNICEF Global Study on Child Poverty and Disparities, 2007-2008. Accra: UNICEF Ghana.
41. Vermeersch, C. (2003). School Meals, Educational Achievement and School Competition: Evidence from a Randomized Evaluation; University of Oxford.
42. World Bank (2004). Country Assistance Strategy for the Republic of Ghana. Country Development 10, The International Finance Corporation, Sub-Saharan Africa Department.
43. Wrigley-Asante, C. (2008). Men Are Poor but Women Are Poorer: Gendered Poverty and Survival.
44. Strategies in the Dangme West District of Ghana'. Norwegian Journal of Geography 62(3): 161-170.