SlideShare a Scribd company logo
1 of 16
Download to read offline
0	
“Poor Economics WiSe 2015”
School Feeding Programs: RCTs as effective Evaluation Tools and their Implications
David Dingus
Göttingen, 11403109
October 26, 2015
Supervisors: Dr. Sebastian Vollmer & Ms. Jana Kuhnt
1	
Introduction
“Undernourished children are more likely to become short adults, to have lower
educational achievement, and to give birth to smaller infants. Under nutrition is also
associated with lower economic status in Adulthood.” (Victoria et al., 2008)
Education is widely considered a key component on the pathway to development, yet
approximately 59 million primary aged children are not in school, with preferential treatment
given to boys over girls in many regions (UNESCO, 2013, World Bank, 2012). In total there
are an estimated 300 million chronically hungry children in the world and of those who do
attend school, an estimated 66 million children go to school hungry (World Bank, 2004,
2012). This hunger prevents many children from reaching their full potential in school, with
life-long consequences. In a case study in Kenya, deworming was used as a proxy for
undernourishment and found that children who were given a deworming pill for two years,
stayed in school longer and earned up to 20% more than children who received the pill for
only 1 year (Baird et al., 2010). Clearly improved nutrition can have long-term effects on the
lives of children.
School feeding programs began as early as 1790 in Munich as a way to feed hungry
children, spreading to France in 1867 (FAO, 2005). Around the same time SFP were first seen
in the United States (Dwyer, 1995). Today, SFPs can been found in almost every country,
with many different organizational structures and goals. In the developing world SFPs are
used as a method to increase student enrollment, attendance, improve child health outcomes,
and often target girls. According to estimates from the World Bank (2012), SFPs have a
global turnover of approximately $100 billion USD, with the WFP alone reaching 22 million
children.
Many SFPs have come under criticism for having unclear, ambiguous effects. In this
paper we will focus on two Randomized-Controlled-Trials, to show that RCTs can be used to
more accurately isolate the effects of SFPs on children and establish that SFPs have clear
benefits, but that there are many key factors to be considered with an emphasis on nutrition
content. We will begin by discussing the Conceptual framework of SFPs, then discuss the
empirical evidence and other factors of consideration, before we conclude with closing
remarks.
Background
Conceptual Framework of SFP
SFPs seek to improve both the educational and nutritional outcomes of children and
thereby improve their lives in the long-run (Grantham-McGregor et al., 1998). The dynamics
of these effects can be seen in Figure 1. SFPs have positive effects by first attracting
School Feeding Programs: RCTs as effective Evaluation Tools and their Implications
	
	 2	
households to enroll their children in school, then by keeping them enrolled longer, and by
increasing their attendance because they essentially subsidize the costs of school, by reducing
food costs at home (Adelman et al., 2007). In addition, evidence has shown that long-term
and short-term child hunger has several adverse effects on children including: reduced
cognitive ability and the ability to partake in the learning environment, increased probability
of suffering from illness and a higher reate of absence, and result in lower academic
performance (Adelman et al., 2007, 2008, Ahmed, 2004). By reducing short and long term
hunger, SFPs seek to mitigate these effects.
Not all SFPs are the same, there can be many differences in their design such as
conditional, targeted, In-School-Meals or Take-Home-Rations, and they can be caloric or
micronutrient focused. All SFPs are conditional to the extent that one has to be in school to
eat in school, however THRs allow for additional conditions to be set, such as attendance rate
and can even be targeted for girls. In this way, THRs are thought to be able to help areas with
clear differences in the treatment in boys and girls. More recently, SFPs have begun to focus
not only on caloric intake, but also the micro-nutrients and source of those nutrients. In this
way a SFP can be used to target specific deficiencies within the target population to have the
greatest impact on health and educational outcomes.
Empirical Studies
Previous Studies
Levinger (1986) summarizes the results of 22 SFPs up until the 1980’s splitting them
into three groups consisting of those focused on increasing attendance rates, those focused on
improved nutrition, and those focused on cognitive outcomes. Overall, though many studies
show an increase in attendance, just as many are inconclusive. Similarly, many of the SFPs
have ambiguous effects on cognitive development and nutritional outcomes. Levinger (1986)
argues that most of these ambiguous and inconclusive results are the result of poor study
designs where often times they were retrospective, did not collect a baseline, relied on the
perceptions of the teacher, and did not utilize standard testing methods to evaluate outcomes.
Overall, Levinger (1986) concludes that the assessment of many SFPs are inappropriately
carried out, and also that many SFPs do not take into consideration regional specific factors
and are thus not able to correctly target the deficiencies of the population, limiting the positive
outcomes of the SFP.
RCT in Burkina Faso: THR vs. ISM
Research suggests that the period of gestation and then first 2 years of life are the most
important in terms of proper nutrition and better health and cognitive development outcomes
(Bhutta et al., 2008). For this reason, Kazianga et al. (2009) wanted to compare the effects of
School Feeding Programs: RCTs as effective Evaluation Tools and their Implications
	
	 3	
different SFPs on the at-home siblings of enrolled school children to determine if ISMs show
any effect on the redistribution of intra-household calories. The addition of the conditional
and targeted THRs sought to determine the spillover effects of the at-home siblings of
enrolled children, and to reduce the gap between boys and girls.
RCT Design
Kazianga et al. (2009) conducted their RCT in rural Burkina Faso to compare the
effects of ISMs and THRs side-by-side. Moreover the study differentiates between boys and
girls by making THRs conditional on the enrollment and 90% attendance of a girl. ISMs
where given at lunch time while the THR consisted of a 10kg bag of flour each month. A
visual of the program structure can be seen in Figure 2. The study uses a prospective
randomized design to ensure external validity and randomizes at the village level when
assigning to the treatment in order to achieve internal validity. By conducting both a baseline
and post intervention survey, they are able to construct a counterfactual and calculate the
effect of the treatment. (Kazianga et al., 2009)
Kazianga et al. (2009) argue that other studies have underestimated the effects from
positive spillovers in the household because these effects require testing of children outside of
the home. To capture this effect they also conduct their baseline survey on the at-home
siblings of children enrolled in the program and likewise conduct a post intervention survey
on those siblings. The baseline survey consisted of collecting household backgrounds, wealth,
school enrollment of all children, and anthropometric data. Additionally, they asked children
to solve simple math problems. They concluded that in general children in the program were
experiencing a severe food shortage. The weight for age and height for age anthropometric
measures were on average two standard deviations below the reference population, half of the
children were underweight and stunted and one-third wasted. Finally, children were able to
answer less than 50% of the basic math problems correctly. They then compared their results
between those in the control village and the treatment villages and found that five of the
variables showed spastically significant differences, but they argue that given that four of
these were only significant at the 10% level, they have small coefficients, and that there are a
wide variety of variables which show no significant difference, thus concluding that the
randomization process was a success. (Kazianga et al., 2009)
In the follow up survey, anthropometric measurements were conducted with basic
math problems again. However, additional methods to test for educational outcomes were
used including the WISC and Raven Matrices tests in order to compare the cognitive abilities
between the treatment and control groups. While the Digit Span test was used to test retention
and recall memory as a proxy for short term working memory. Kazianga et al. (2009) explain
School Feeding Programs: RCTs as effective Evaluation Tools and their Implications
	
	 4	
that the idea behind these different testing methods was to enable them to quantify
improvements in long-term hunger with short-term hunger. Furthermore, though these tests
were not administered during the baseline survey, the results could be compared with those
children in the control group. Higher scores in the treatment group over the control group
could therefore be interpreted as a causal effect from one of the treatments. (Kazianga et al.,
2009)
Results
Using a difference in difference method, Kazianga et al. (2009) find that enrollment
increased under both treatments and that boys were not crowded out. While these programs
did not show any significant effect on the absenteeism of boys, they did show that girls in
either the THRs program or the ISMs program missed on average 1 day more school per
month. Kazianga et al. (2009) explain that they were able to determine from their data that
absenteeism in girls was determined by the child labor supply in the household, where
families with four or more children saw absenteeism for girls decrease, and those with less,
increasing with fewer children. The ISMs showed positive effects on short-term memory
function as older girls were able to answer simple math problems faster. However, boys
showed a statistically significant small reduction in the number of correct answers they
provided, possibly a result of decreased learning time. On the other hand, THRs showed no
significant effects on learning outcomes. There were no significant effects for any group on
cognitive development. Anthropometric measures from the ISMs group showed a marginal
weight gain in children between the ages of six and ten, but no impact on their at-home
siblings. However, the THRs program showed gains in weight for age and weight for height
in the at-home sample. (Kazianga et al., 2009)
Kayianga et al. (2009) conclude, that THR have a positive impact on the at-home
siblings of enrolled children, confirming positive spillover effects, but little impact on the
short-run hunger of enrolled children and thus no improvement on educational outcomes.
Whereas the ISMs program showed no evidence of intra-household rationing away from the
enrolled child and improved the short-run nutrition and educational outcomes of enrolled
students, with no impact on their at-home siblings.
Criticisms
While the study by Kaziange et al. (2009) does follow the guideline of an RCT, only
some of the results can be interpreted with causality, specifically, we cannot not compare the
magnitude of the effect between the treatment groups, because we do not know the content of
the ISMs treatment, or what was consumed in the household. Therefore, when applying these
results to a different population, we cannot say whether THRs will have a greater or lesser
School Feeding Programs: RCTs as effective Evaluation Tools and their Implications
	
	 5	
effect compared to ISMs, because those differences maybe attributed to the content of those
programs and would vary based on regional specific factors. It is important to understand that
the idea of providing flour to the household is to free up money for additional food items, but
we do not know how these households will spend their money.
Kazinagne et al. (2009) explain that in effect, a THR is a conditional-in-kind transfer,
citing Gahvari & Mattos (2007) who argue that a THR can be considered in this instance a
conditional-cash-transfer. However, many studies in development including an RCT on non
conditional cash transfer program in Kenya, have concluded that non-conditional transfers are
often just as effective with lower costs (Haushofer & Shapiro, 2013). More importantly, given
that we want to have the largest possible positive impact on as many children as possible,
with particular emphasis on children under the age of two, having a conditional THR may be
inhibiting the program from reaching its full potential.
“In environments with low enrollment rates, perfectly targeting children who
are enrolled might not be the best policy objective if one wants to reach most of the
malnourished children […] nutritionists generally believe that the window of
opportunity for nutrition is during gestation and the first two years of life
[…therefore…] it may be desirable to use THR to reach children who are not enrolled
since they are likely to be the ones who need the intervention the most.” (Kazianga et
al., 2009, p.7)
Unfortunately, because the RCT did not include a group with a THR that was not conditional,
we cannot determine if the 90% attendance condition had a significantly different impact
compared with not having the condition. Furthermore, because there was also no THR group
for boys, we cannot be certain of the impact from targeting the THR at girls.
In this RCT the Hawthorne effect could be exhibited by the treatment group when they
change their behavior and make their children attend school more regularly because they are
being observed. On the other hand, the John Henry effect would be captured in the control
group when they may also increase their enrollment and attendance rates as a result of being
observed. If this were the case, then the treatment effect would not capture these effects and
our results would not be biased (Duflo et al., 2007). On the other hand, if the magnitudes of
these effects differ or their directions, then the treatment effect would be under or over
estimated. Duflo et al. (2007) point out that in most cases, simply conducting the experiment
over a long period of time would help eliminate these effects from our results, as evidence
suggests that the Hawthorne and John Henry effects diminish over time.
School Feeding Programs: RCTs as effective Evaluation Tools and their Implications
	
	 6	
Kazianga et al. (2009) explain that they tried to minimize non-compliance by
randomizing at the village level. In this way it would be more difficult for households to
switch to the treatment groups or to be impacted by way of spillover effects from neighboring
households. Nonetheless, they explain that noncompliance could still occur if households in
the control group have their children attend school in a treatment village or if a household in
the treatment village fosters children from the control group (Kazianga et al., 2009, p. 11-12).
However, given the geographical barriers, length of time of the intervention, and the
commitment to attend school nearly every day, chances of noncompliance are minimal, but
should be kept in mind.
RCT in Kenya: Nutritional Content
Considerations on the nutritional content of SFPs are especially important in the
developing world, given that many populations in Sub-Saharan Africa and Southeast Asia
have shown to have micronutrient deficiencies and macro nutrient deficiencies stemming
from a lack of protein and low animal consumption (Diaz et al., 2003). Based on these
findings, Neumann et al. (2007) argue that it is essential that studies establish causal evidence
on the effects of different nutritional content in SFPs.
RCT Design
The RCT was conducted in rural Kenya in an area of 60 km2
that was prone to drought
and experienced severe food shortages. The target population consisted of mostly small
holder farmers, which consumed mostly plant-based diets with sources of protein from eggs
and rarely a goat. The program treatment was randomized at the school level between 4
groups. One group was the control where no SFP was given, while the other 3 groups tested
three different treatments consisting of Githeri with meat, with milk, or with oil. The
treatment was administered as a mid-morning snack on school days and the study lasted in
total over a period of 2.25 years. There was an initial baseline survey, followed by
assessments during the program to conclude with a final assessment. Assessments consisted
of Anthropometric measures, cognitive and school performance measures and the levels of
physical activity and behavior were also documented. By taking a baseline survey in every
group, the researches were able to construct a counterfactual to isolate the effect of the
different treatments from other factors. (Neumann et al., 2007)
Initial baseline findings revealed a representative sample of the target population and
in the randomization of assignment to treatment groups with no significant differences
between groups. Through blood work, the program did determine that the meat group had the
highest vitamin B-12 deficiency, which aids in the absorption of iron and zinc in plant-based
diets (Fairbanks, 1999). Overall, they found a high presence of anemia, malaria, stunting,
School Feeding Programs: RCTs as effective Evaluation Tools and their Implications
	
	 7	
underweight, and low muscle mass all of which are associated with not only undernutrition
but also a lack of animal sourced protein. (Neumann et al., 2007)
Results
Unlike the first RCT, this study found evidence of cognitive development, particularly
in the meat group as shown in in Figures 3 & 4. The meat group also showed the highest
percentage of time in leadership, activity during free playtime, and in initiative behavior as
shown in Figure 5. Interestingly, although the treatment with milk was better than the control
who received no feeding, it produced in general lower positive results than the meat and oil
groups. The milk group also produced the lowest results in the Raven Matrices test. Neumann
et al. (2007) explain that this may be a result of the high casein and calcium content in milk
that blocks the absorption of iron. Furthermore, in a study by Whaley et al. (2003) they find
that iron plays a vital role in cognitive function. In essence, introducing SFPs with milk may
impede cognitive growth. In anthropometric measures, all groups who received the treatment
showed greater weight gain than the control group. In terms of height, only the milk group
showed greater results than other groups for children under 6 years of age and stunted
children. All other groups did not show any significant gain in height compared to the control
group. This underscores, that on the other hand, a SFP with milk may be appropriate when the
target population is young and stunted. Finally, the meat group showed the largest increase in
lean muscle mass using the mid-arm muscle area measure. (Neumann et al., 2007)
Criticisms
Even though this RCT is a replicate study, attrition bias can be seen by the different
number of participants between the first and second cohort. In the first cohort, 525 children
were observed while that number dropped to 375 in the second cohort, implying a dropout
rate of over 28%. If there were similar characteristics shared by those who dropped out, the
sample may have no longer be representative and biased the results. Neumann et al. (2007)
explain that they used fixed and mixed effects and used a replicate study design where they
followed the same students overtime and that statistical methods confirmed the validity of
their model. Nonetheless, given that a severe drought and teachers strike occurred during the
study, we cannot be sure if those who moved away exhibited more positive characteristics
because their families could afford to move, if they exhibited more negative characteristics
because their families moved out of desperation, or if they exhibited traits that were on
average no different than the initial sample.
The RCT may have also captured the Hawthorne and John Henry effects, especially
during behavioral observation. From Figure 5 we can see that all groups displayed higher
levels of time engaged in physical activity, leadership, and in initiative behavior at the
School Feeding Programs: RCTs as effective Evaluation Tools and their Implications
	
	 8	
beginning of the study that then declines over time. We cannot be sure if this decline is
because the students changed their behavior because they were being observed initially or if it
is because they are getting older. Duflo et al. (2007) explain that the Hawthorne and John
Henry effects generally decrease overtime as the observed unit becomes accustomed to being
observed. However, because we have a control group who is also aware that it is being
observed, the Hawthorne effect would most likely exhibit the same effect as the John Henry
effect in the treatment group, thus any differences can be interpreted as a result of the
treatment (Duflo et al, 2007). Also, from our results, we can clearly see differences between
groups in their behavior that must go beyond the effect of changed behavior due to being
observed.
However, as we concluded in the first RCT, we also cannot conclude from this RCT
that these results will be similar in other regions, as they may have different nutritional
deficiencies, which may be better targeted through different measures. A study on Iodine
capsules in Tanzania concluded that if mothers in Central and Southern Africa were to take
Iodine capsules while pregnant, there would be a 7.5% increase in educational attainment
(Fields et al., 2009). Another study in Indonesia found that when anemic men took iron
supplements, they saw their productivity and wages increase (Thomas et al., 2004). In
general, it is important to consider not only the calories in a SFP, but also the source of those
calories and the micronutrients they contain, and to design a SFP to meet the needs of the
target population.
Adverse Effects of SFPs
In addition, there are general criticisms on the adverse effects of SFPs, often
concerning decreasing educational performance. Kazianga et al. (2009) summarizes these
theories, first explaining that SFPs often require teachers to spend some of their time
organizing and administering meals to students and therefore have less time in the classroom
to teach or to prepare. Classrooms may also become overcrowded and other resources may
become constrained as a result of increased enrollment if these resources are also not
increased. Also the students who did enroll as a result of the SFP may be less motivated
because they may perceive that they have less to gain from an education, given that they did
not attend school before. Finally, those students who are newly enrolled may have lower
learning abilities and/or have large knowledge gaps, thus decreasing the average scores of the
school. SFPs could therefore be wrongly interpreted as decreasing education outcomes.
(Kaziagna et al., 2009)
In the RCT from Kazianga et al. (2009) we saw that older boys scores on answers to
math problems decreased by a statistically significant, though small amount. This may be
School Feeding Programs: RCTs as effective Evaluation Tools and their Implications
	
	 9	
evidence of decreased learning time as it only occurred in the ISMs group and not the THRs
group, thus discounting the effect of constrained resources. Kazianga et al. (2009) cite
another study in Jamaica where school test scores decreased after introducing an ISMs
program. Grantham-McGregor et al. (1998) explain that this can be attributed to a lack of
organization. Other case studies including one in Bangladesh, have found SFPs to result in
lower test scores (Ahmed & Ninno, 2002). However once the results were compared non-
eligible students in schools with SFPs to students in schools without SFPs, there was no
statistically significant difference in the test scores (Adelman et al., 2008). In general when
assessing SFPs we have to carefully consider the method in which we assess the students,
especially because these programs may attract lower performing students, thereby lowering
the average test results of the school. Furthermore, the amount of learning in the school may
decrease as time and other resources become more constrained. It is therefore essential that
one considers the regional specific factors when implementing a SFP in order to be able to
address possible issues before they become problem (Levinger, 1986).
Other Factors to Consider
Other factors such as the timing of meals and how they interact with other programs
should be considered to ensure maximum effectiveness. School Breakfast Programs in the
United States have clearly shown that by administering meals at the beginning of school helps
to alleviate short-term hunger in children, which allows them to better focus on the learning
environment (Bhattacharya et al., 2004). Yet given this information, many SFPs focus on
either snacks or lunches. While this timing will equally improve long-term measurement
outcomes in child health, these children may be suffering from short-term hunger until snack
or lunch time. As a result, these children will have difficulty concentrating in class and
applying themselves for a significant portion of their school day, thus limiting the positive
outcomes of the SFP. In order for children to reach their full potential, the nature of their
hunger must be carefully considered, and in cases where children suffer from short-term
hunger starting from the beginning of the school day, they may benefit from the addition of a
SBP.
In addition, these RCTs have not considered the effects of how other programs can
increase the effectiveness of SFPs. For example, Dialo (2014) assesses the impact of both
SFPs and deworming programs in Senegal. He finds that both result in positive outcomes on
enrollment and educational outcomes, though deworming is the faster and cheaper of the two
options. More importantly, he finds that when combining both programs, the positive
outcomes are larger than each program independently. In sum, depending on the
School Feeding Programs: RCTs as effective Evaluation Tools and their Implications
	
	 10	
characteristics and regional specific factors in the target population, other programs can
increase the effectiveness of SFPs.
Closing Remarks
By using RCTs to control for regional specific effects we can isolate the effects of
SFPs and have found that SFPs can have a measurable impact on the educational and health
outcomes of children, but ensuring that those interventions reach children as early as possible
may have larger effects. In addition, how those outcomes are measured and controlled for can
determine their results. In particular, we have found that the nutritional content of these
programs can be just as important as the caloric value. We were also able to conclude that
THRs have a greater impact on younger siblings at-home, whereas ISMs only seems to
impact those receiving the food directly with no spillover effects. However, we were unable
to draw any causal inferences on the effects of gender based targeting or conditional transfers.
To conclude, RCTs are useful tools to ascertain the impact of SFPs because they can
establish causality and thus their results can be applied in other settings. However, one must
be careful in how they interpret the results of an RCT, because while they can establish the
effect of a specific factor given the conditions of the subjects and their environments, the
magnitude of these effects will most likely differ in different contexts.
Furthermore, there remains much research to be done on the subject of SFPs
especially regarding the interaction with other programs, the effects of timing, and more on
the nutritional makeup of such programs. However, to obtain the effects of these programs,
requires a complex undertaking whereby each variable of interest must be administered to a
separate treatment group, which can be both time consuming and costly. As such, though case
studies lack external validity, they can provide us with in-depth knowledge of a specific
population and allow us to design a more effective SFP.
In general, there is no one size fits all SFP. Each case must be tailored to regional
specific factors to achieve targeted outcomes. If successful, SFPs can have measurable
positive outcomes on the lives of undernourished children, but more emphasis should be
placed on the nutritional content of these programs, their organization and timing, while also
coordinating with other efforts to ensure that the resources are available for students to reap
the full benefits that SFPs seek to provide.
School Feeding Programs: RCTs as effective Evaluation Tools and their Implications
	
	 11	
References
Adelman, S., Gilligan, D. O., & Lehrer, K. (2007) How Effective are Food For Education
Programs? A Critical Assessment of the Evidence from Developing Countries. The
United Nations University. Washington D.C.: International Food Policy Research
Institute.
Adelman, S., Gilligan, D. O., & Lehrer, K. (2008) The Impact of Alternative Food for
Education Programs on Learning achievement and Cognitive Development in Northern
Uganda. The United Nations University. Washington D.C.: International Food Policy
Research Institute.
Ahmed, A. and del Ninno, C. (2002) The Food for Education Program in Bangladesh: An
early Evaluation of its impact on Education Attainment and Food Security. The United
Nations University. Washington D.C.: International Food Policy Research Institute.
Ahmed, A. (2004). Impact of feeding children in school: Evidence from Bangladesh. The
United Nations University. Washington D.C.: International Food Policy Research
Institute.
Baird, S., Hicks, H. J., Kremer, M., & Miguel, E. (2010) Worm at Work: Long-Run Impacts
of Child Health Gains. University of California at Berkeley. Berkley.
Bhattacharya, J., Currie, J., & Haider, S. J. (2004). Evaluating the Impact of School Nutrition
Programs. Washington D.C.: USDA Final Report.
Bhutta, Z., Ahmad, R., Black, S., Cousens, K., Dewey, E., Giugliani, B., Haider, B.,
Kirkwood, S.. Morris, H., Sachdev & Shekar, M. (2008) What works? Interventions for
Maternal and Child Undernutrition and Survival. Lancet, 371(3): 417-440.
Diallo, F. L. (2014). Evalution of Meal and Deworming Programs for PRimary Schools in
Rural Senegal. Masstricht: University of Maastricht.
Diaz, J.R., de las Cagigas, A, Rogriquez, R. (2003) Micronutrient deficiencies in developing
and affluent countries. American Journal of Clinical Nutrition, 57(1): S70-72
Duflo, E., Glennerster, R., & Kremer, M. (2007). Using randomization in development
economics research: A toolkit. Handbook of development economics, 4: 3895-3962.
Dwyer, J. (1995) The School Nutrition Dietary Assessment Surveys. American Journal of
Clinical Nutrition, 61: 173S-177S.
Field, E., Robles, O., & Torero, M. (2009) Iodine Deficiency and School Attainment in
Tanzania. American Economic Journal: Applied Economics, 1(4): 140-169.
Fairbanks, V. F. (1999) Iron in medicine and nutrition. Modern nutrition in health and
disease, 9: 223-239
School Feeding Programs: RCTs as effective Evaluation Tools and their Implications
	
	 12	
Food and Agricultural Organization-FAO. (2005) The state of food insecurity in the world:
Eradicating world hunger-key to achieving the Millennium Development Goal. Italy:
FAO
Gahvari, F., and Mattos, E. (2007) Conditional Cash Transfers, Public Provision of Private
Goods, and Income Redistribution. American Economic Review, 97(1): 491-502
Grantham-McGregor, S. M., Chang, S., & Walker, S. P. (1998) Evaluation of School Feeding
Programs: Some Jamaican Examples. American Journal of Clinical Nutrition, 67S:
785S-789S.
Haushofer, J., & Shapiro, J. (2013). Household response to income changes: Evidence from
an unconditional cash transfer program in Kenya. Massachusetts Institute of
Technology.
Kazianga, H., De Walque, D., & Alderman, H. (2009). Educational and health impact of two
school feeding schemes: evidence from a randomized trial in rural Burkina Faso.
Washington, DC: World Bank.
Levinger, B. (1986) School Feeding Programs in Developing Countries: An Analysis of
Actual and Potential Impact. Aid Evaluation Special Study 30. Washington D.C.: U.S.
Agency for International Development.
Neumann, C. G., Murphy, S. P., Gewa, C., Grillenberger, M., & Bwibo, N. O. (2007). Meat
supplementation improves growth, cognitive, and behavioral outcomes in Kenyan
children. Journal of Nutrition, 137(4), 1119-1123.
Thomas, D., Frankenberg, E., Friedman, J., Habicht, J. P., Jones, N., McKelvey, C., ... &
Suriastini, W. (2004). Causal effect of health on labor market outcomes: Evidence from
a random assignment iron supplementation intervention. California Center for
Population Research.
UNESCO (2013) A growing number of children and adolescents are out of school as aid fails
to meet the mark. (Policy Paper 22). Paris: UNESCO
Victora, C. G., Adair, L., Fall, C., Hallal, P. C., Martorell, R., Richter, L., … for the Maternal
and Child Undernutrition Study Group. (2008). Maternal and child undernutrition:
consequences for adult health and human capital. Lancet, 371(9609), 340–357.
Whaley, S. E., Sigman, M., Neumann, C., Bwibo, N., Guthrie, D., Weiss, R. E., ... & Murphy,
S. P. (2003). The impact of dietary intervention on the cognitive development of
Kenyan school children. The Journal of nutrition, 133(11), 3965S-39
World Bank. (2004). School feeding programs: why they should be scaled up now.
Washington, D.C.: World Food Program.
School Feeding Programs: RCTs as effective Evaluation Tools and their Implications
	
	 13	
World Bank. (2012). Scaling up school feeding: Keeping children in school while improving
their learning and health. Washington, D.C.: World Bank.
Yendaw, E., & Dayour, F. (2015) Effect of the National School Feeding Programme on
Pupils’ Enrolment, Attendance and Retention: A Case Study of Nyoglo of the
Savelugu-Nantong Municipality, Ghana. British Journal of Education, Society &
Behavioural Science, 5(3): 341-353.
School Feeding Programs: RCTs as effective Evaluation Tools and their Implications
	
	 14	
Appendix
Figure 2. RCT Design. (Kazianga et al., 2009, p. 23).
Figure 1. Conceptual Framework of SFPs. (Yendaw & Dayour, 2015, p. 345).
Figure 3. SFP Impact on Cognitive Outcomes. (Neumann, 2007, p. 1121).
School Feeding Programs: RCTs as effective Evaluation Tools and their Implications
	
	 15	
Figure 4. SFP Impact on Education Outcomes. (Neumann, 2007, p. 1121).
Figure 5. SFP Impact on Behavioral Outcomes. (Neumann, 2007, p. 1121).

More Related Content

What's hot

Volume 5 Issue 3 Paper 7
Volume 5 Issue 3 Paper 7Volume 5 Issue 3 Paper 7
Volume 5 Issue 3 Paper 7AnnMarie Kuzel
 
Number of meals consumed by the pre school age going children
Number of meals consumed by the pre school age going childrenNumber of meals consumed by the pre school age going children
Number of meals consumed by the pre school age going childrenAlexander Decker
 
Kessler Research Issue Brief
Kessler Research Issue BriefKessler Research Issue Brief
Kessler Research Issue BriefJordan Kessler
 
child obesity - final draft
child obesity - final draftchild obesity - final draft
child obesity - final draftClaudia Willis
 
Chappell_OT674_Poster
Chappell_OT674_PosterChappell_OT674_Poster
Chappell_OT674_PosterAnya Chappell
 
Childhood obesity prevention literature review
Childhood obesity prevention literature reviewChildhood obesity prevention literature review
Childhood obesity prevention literature reviewAmber Breidel
 
Segal (2012) theory! the missing link
Segal (2012) theory! the missing linkSegal (2012) theory! the missing link
Segal (2012) theory! the missing linkLucieCluver
 
Correlation between Eating Habits and Nutritional Status
Correlation between Eating Habits and Nutritional StatusCorrelation between Eating Habits and Nutritional Status
Correlation between Eating Habits and Nutritional StatusJanrey Tiña
 
Steve Kelder on Child Obesity
Steve Kelder on Child ObesitySteve Kelder on Child Obesity
Steve Kelder on Child Obesityshkelder
 
6 Week Sport-Based Healthy Eating Intervention for Year 6 Students in Barnsta...
6 Week Sport-Based Healthy Eating Intervention for Year 6 Students in Barnsta...6 Week Sport-Based Healthy Eating Intervention for Year 6 Students in Barnsta...
6 Week Sport-Based Healthy Eating Intervention for Year 6 Students in Barnsta...Tom Lake
 
A project proposal for East Timor on improving health and nutrition for women...
A project proposal for East Timor on improving health and nutrition for women...A project proposal for East Timor on improving health and nutrition for women...
A project proposal for East Timor on improving health and nutrition for women...Kazuko Yoshizawa
 
Cooley project poster 7_31_2013_final submission
Cooley project poster 7_31_2013_final submissionCooley project poster 7_31_2013_final submission
Cooley project poster 7_31_2013_final submissionShannon Stephanie Cooley
 
What influences choice 3
What influences choice 3What influences choice 3
What influences choice 3guest9916186
 

What's hot (19)

JOE_v52_6tt6
JOE_v52_6tt6JOE_v52_6tt6
JOE_v52_6tt6
 
Volume 5 Issue 3 Paper 7
Volume 5 Issue 3 Paper 7Volume 5 Issue 3 Paper 7
Volume 5 Issue 3 Paper 7
 
Number of meals consumed by the pre school age going children
Number of meals consumed by the pre school age going childrenNumber of meals consumed by the pre school age going children
Number of meals consumed by the pre school age going children
 
Kessler Research Issue Brief
Kessler Research Issue BriefKessler Research Issue Brief
Kessler Research Issue Brief
 
U01821129153
U01821129153U01821129153
U01821129153
 
10.1186%2 f1471 2334-13-590
10.1186%2 f1471 2334-13-59010.1186%2 f1471 2334-13-590
10.1186%2 f1471 2334-13-590
 
child obesity - final draft
child obesity - final draftchild obesity - final draft
child obesity - final draft
 
Chappell_OT674_Poster
Chappell_OT674_PosterChappell_OT674_Poster
Chappell_OT674_Poster
 
Childhood obesity prevention literature review
Childhood obesity prevention literature reviewChildhood obesity prevention literature review
Childhood obesity prevention literature review
 
Segal (2012) theory! the missing link
Segal (2012) theory! the missing linkSegal (2012) theory! the missing link
Segal (2012) theory! the missing link
 
Correlation between Eating Habits and Nutritional Status
Correlation between Eating Habits and Nutritional StatusCorrelation between Eating Habits and Nutritional Status
Correlation between Eating Habits and Nutritional Status
 
Steve Kelder on Child Obesity
Steve Kelder on Child ObesitySteve Kelder on Child Obesity
Steve Kelder on Child Obesity
 
6 Week Sport-Based Healthy Eating Intervention for Year 6 Students in Barnsta...
6 Week Sport-Based Healthy Eating Intervention for Year 6 Students in Barnsta...6 Week Sport-Based Healthy Eating Intervention for Year 6 Students in Barnsta...
6 Week Sport-Based Healthy Eating Intervention for Year 6 Students in Barnsta...
 
Awareness of Infant Feeding: Nursing Program Student Survey
Awareness of Infant Feeding: Nursing Program Student Survey Awareness of Infant Feeding: Nursing Program Student Survey
Awareness of Infant Feeding: Nursing Program Student Survey
 
A project proposal for East Timor on improving health and nutrition for women...
A project proposal for East Timor on improving health and nutrition for women...A project proposal for East Timor on improving health and nutrition for women...
A project proposal for East Timor on improving health and nutrition for women...
 
Cooley project poster 7_31_2013_final submission
Cooley project poster 7_31_2013_final submissionCooley project poster 7_31_2013_final submission
Cooley project poster 7_31_2013_final submission
 
What influences choice 3
What influences choice 3What influences choice 3
What influences choice 3
 
V42n5a10
V42n5a10V42n5a10
V42n5a10
 
Problem statements
Problem statementsProblem statements
Problem statements
 

Viewers also liked

Еженедельный геологический обзор № 4
Еженедельный геологический обзор № 4Еженедельный геологический обзор № 4
Еженедельный геологический обзор № 4geologykz
 
Multi – Objective Two Stage Fuzzy Transportation Problem with Hexagonal Fuzzy...
Multi – Objective Two Stage Fuzzy Transportation Problem with Hexagonal Fuzzy...Multi – Objective Two Stage Fuzzy Transportation Problem with Hexagonal Fuzzy...
Multi – Objective Two Stage Fuzzy Transportation Problem with Hexagonal Fuzzy...IJERA Editor
 
Comunicacion
ComunicacionComunicacion
Comunicacionceciharo
 
Haiti | Jan-1 | Initiative Smart Villages
Haiti | Jan-1 | Initiative Smart Villages Haiti | Jan-1 | Initiative Smart Villages
Haiti | Jan-1 | Initiative Smart Villages Smart Villages
 
Programa de coaching Shine professional coach
Programa de coaching Shine professional coachPrograma de coaching Shine professional coach
Programa de coaching Shine professional coachMAX GALARZA HERNANDEZ
 
Massive bone allografts
Massive bone allograftsMassive bone allografts
Massive bone allograftsRaunak Milton
 
Pedagogy notes ef 26012017104008
Pedagogy notes ef  26012017104008Pedagogy notes ef  26012017104008
Pedagogy notes ef 26012017104008EugenioFouz
 
3Com 3CR008-90
3Com 3CR008-903Com 3CR008-90
3Com 3CR008-90savomir
 

Viewers also liked (16)

Еженедельный геологический обзор № 4
Еженедельный геологический обзор № 4Еженедельный геологический обзор № 4
Еженедельный геологический обзор № 4
 
Multi – Objective Two Stage Fuzzy Transportation Problem with Hexagonal Fuzzy...
Multi – Objective Two Stage Fuzzy Transportation Problem with Hexagonal Fuzzy...Multi – Objective Two Stage Fuzzy Transportation Problem with Hexagonal Fuzzy...
Multi – Objective Two Stage Fuzzy Transportation Problem with Hexagonal Fuzzy...
 
Scaling up blended finance
Scaling up blended financeScaling up blended finance
Scaling up blended finance
 
Wifi y bluetooth
Wifi y bluetoothWifi y bluetooth
Wifi y bluetooth
 
Presentacion elmer hidalgoo
Presentacion elmer hidalgooPresentacion elmer hidalgoo
Presentacion elmer hidalgoo
 
Comunicacion
ComunicacionComunicacion
Comunicacion
 
Haiti | Jan-1 | Initiative Smart Villages
Haiti | Jan-1 | Initiative Smart Villages Haiti | Jan-1 | Initiative Smart Villages
Haiti | Jan-1 | Initiative Smart Villages
 
Fenómenos de trasporte cap1
Fenómenos de trasporte cap1Fenómenos de trasporte cap1
Fenómenos de trasporte cap1
 
Programa de coaching Shine professional coach
Programa de coaching Shine professional coachPrograma de coaching Shine professional coach
Programa de coaching Shine professional coach
 
MSC_rad_summary
MSC_rad_summaryMSC_rad_summary
MSC_rad_summary
 
article1714
article1714article1714
article1714
 
Massive bone allografts
Massive bone allograftsMassive bone allografts
Massive bone allografts
 
cv_of pale moeketsi
cv_of pale moeketsicv_of pale moeketsi
cv_of pale moeketsi
 
Pedagogy notes ef 26012017104008
Pedagogy notes ef  26012017104008Pedagogy notes ef  26012017104008
Pedagogy notes ef 26012017104008
 
Partner For IT - IT solutions
Partner For IT - IT solutionsPartner For IT - IT solutions
Partner For IT - IT solutions
 
3Com 3CR008-90
3Com 3CR008-903Com 3CR008-90
3Com 3CR008-90
 

Similar to Dingus_SchoolFeedingPrograms

Running head CHILDHOOD OBESITY 1CHILDHOOD OBESITY 7.docx
Running head CHILDHOOD OBESITY 1CHILDHOOD OBESITY 7.docxRunning head CHILDHOOD OBESITY 1CHILDHOOD OBESITY 7.docx
Running head CHILDHOOD OBESITY 1CHILDHOOD OBESITY 7.docxsusanschei
 
Research Critique Guidelines Essay Example Paper.docx
Research Critique Guidelines Essay Example Paper.docxResearch Critique Guidelines Essay Example Paper.docx
Research Critique Guidelines Essay Example Paper.docxwrite22
 
Childhood Obesity Scholarly Paper.docx
Childhood Obesity Scholarly Paper.docxChildhood Obesity Scholarly Paper.docx
Childhood Obesity Scholarly Paper.docx4934bk
 
CHILDHOOD OBESITY 4 Research HW.docx
CHILDHOOD OBESITY 4 Research HW.docxCHILDHOOD OBESITY 4 Research HW.docx
CHILDHOOD OBESITY 4 Research HW.docxbkbk37
 
RESEARCH Open AccessThe impact of an innovative web-based.docx
RESEARCH Open AccessThe impact of an innovative web-based.docxRESEARCH Open AccessThe impact of an innovative web-based.docx
RESEARCH Open AccessThe impact of an innovative web-based.docxronak56
 
Nutritional Status of School Age Children in Private Elementary Schools: Basi...
Nutritional Status of School Age Children in Private Elementary Schools: Basi...Nutritional Status of School Age Children in Private Elementary Schools: Basi...
Nutritional Status of School Age Children in Private Elementary Schools: Basi...IJAEMSJORNAL
 
Aene project a medium city public students obesity study
Aene project   a medium city public students obesity studyAene project   a medium city public students obesity study
Aene project a medium city public students obesity studyCIRINEU COSTA
 
The Effect of Breastfeeding on  Children’s CognitionAlgham.docx
The Effect of Breastfeeding on   Children’s CognitionAlgham.docxThe Effect of Breastfeeding on   Children’s CognitionAlgham.docx
The Effect of Breastfeeding on  Children’s CognitionAlgham.docxcherry686017
 
PROGRAM ANALYSIS2 PROGRAM ANALYSIS8 Program A.docx
PROGRAM ANALYSIS2 PROGRAM ANALYSIS8 Program A.docxPROGRAM ANALYSIS2 PROGRAM ANALYSIS8 Program A.docx
PROGRAM ANALYSIS2 PROGRAM ANALYSIS8 Program A.docxgertrudebellgrove
 
PROGRAM ANALYSIS2 PROGRAM ANALYSIS8 Program A.docx
 PROGRAM ANALYSIS2 PROGRAM ANALYSIS8 Program A.docx PROGRAM ANALYSIS2 PROGRAM ANALYSIS8 Program A.docx
PROGRAM ANALYSIS2 PROGRAM ANALYSIS8 Program A.docxgertrudebellgrove
 
Community trial design
Community trial designCommunity trial design
Community trial designSruthi Bhat
 
Running head LITERATURE REVIEW 1LITERATURE REVIEW 5.docx
Running head LITERATURE REVIEW 1LITERATURE REVIEW 5.docxRunning head LITERATURE REVIEW 1LITERATURE REVIEW 5.docx
Running head LITERATURE REVIEW 1LITERATURE REVIEW 5.docxcowinhelen
 
Nutritional status of boarding and non boarding children in selected schools ...
Nutritional status of boarding and non boarding children in selected schools ...Nutritional status of boarding and non boarding children in selected schools ...
Nutritional status of boarding and non boarding children in selected schools ...Alexander Decker
 
Statistic in Health Care Management Assignment Week 3Case Study.docx
Statistic in Health Care Management Assignment Week 3Case Study.docxStatistic in Health Care Management Assignment Week 3Case Study.docx
Statistic in Health Care Management Assignment Week 3Case Study.docxrafaelaj1
 
In this students will pull together the change proposal project.pdf
In this students will pull together the change proposal project.pdfIn this students will pull together the change proposal project.pdf
In this students will pull together the change proposal project.pdfsdfghj21
 

Similar to Dingus_SchoolFeedingPrograms (20)

Childhood Obesity EBR
Childhood Obesity EBRChildhood Obesity EBR
Childhood Obesity EBR
 
Running head CHILDHOOD OBESITY 1CHILDHOOD OBESITY 7.docx
Running head CHILDHOOD OBESITY 1CHILDHOOD OBESITY 7.docxRunning head CHILDHOOD OBESITY 1CHILDHOOD OBESITY 7.docx
Running head CHILDHOOD OBESITY 1CHILDHOOD OBESITY 7.docx
 
Research Critique Guidelines Essay Example Paper.docx
Research Critique Guidelines Essay Example Paper.docxResearch Critique Guidelines Essay Example Paper.docx
Research Critique Guidelines Essay Example Paper.docx
 
Childhood Obesity Scholarly Paper.docx
Childhood Obesity Scholarly Paper.docxChildhood Obesity Scholarly Paper.docx
Childhood Obesity Scholarly Paper.docx
 
CHILDHOOD OBESITY 4 Research HW.docx
CHILDHOOD OBESITY 4 Research HW.docxCHILDHOOD OBESITY 4 Research HW.docx
CHILDHOOD OBESITY 4 Research HW.docx
 
The Impact of a Food for Education Program on Schooling in Cambodia
The Impact of a Food for Education Program on Schooling in CambodiaThe Impact of a Food for Education Program on Schooling in Cambodia
The Impact of a Food for Education Program on Schooling in Cambodia
 
RESEARCH Open AccessThe impact of an innovative web-based.docx
RESEARCH Open AccessThe impact of an innovative web-based.docxRESEARCH Open AccessThe impact of an innovative web-based.docx
RESEARCH Open AccessThe impact of an innovative web-based.docx
 
Nutritional Status of School Age Children in Private Elementary Schools: Basi...
Nutritional Status of School Age Children in Private Elementary Schools: Basi...Nutritional Status of School Age Children in Private Elementary Schools: Basi...
Nutritional Status of School Age Children in Private Elementary Schools: Basi...
 
Aene project a medium city public students obesity study
Aene project   a medium city public students obesity studyAene project   a medium city public students obesity study
Aene project a medium city public students obesity study
 
The Effect of Breastfeeding on  Children’s CognitionAlgham.docx
The Effect of Breastfeeding on   Children’s CognitionAlgham.docxThe Effect of Breastfeeding on   Children’s CognitionAlgham.docx
The Effect of Breastfeeding on  Children’s CognitionAlgham.docx
 
PROGRAM ANALYSIS2 PROGRAM ANALYSIS8 Program A.docx
PROGRAM ANALYSIS2 PROGRAM ANALYSIS8 Program A.docxPROGRAM ANALYSIS2 PROGRAM ANALYSIS8 Program A.docx
PROGRAM ANALYSIS2 PROGRAM ANALYSIS8 Program A.docx
 
PROGRAM ANALYSIS2 PROGRAM ANALYSIS8 Program A.docx
 PROGRAM ANALYSIS2 PROGRAM ANALYSIS8 Program A.docx PROGRAM ANALYSIS2 PROGRAM ANALYSIS8 Program A.docx
PROGRAM ANALYSIS2 PROGRAM ANALYSIS8 Program A.docx
 
Community trial design
Community trial designCommunity trial design
Community trial design
 
Ramification of gulayan sa paaralan progran in tingloy
Ramification of gulayan sa paaralan progran in tingloyRamification of gulayan sa paaralan progran in tingloy
Ramification of gulayan sa paaralan progran in tingloy
 
Running head LITERATURE REVIEW 1LITERATURE REVIEW 5.docx
Running head LITERATURE REVIEW 1LITERATURE REVIEW 5.docxRunning head LITERATURE REVIEW 1LITERATURE REVIEW 5.docx
Running head LITERATURE REVIEW 1LITERATURE REVIEW 5.docx
 
Nutritional status of boarding and non boarding children in selected schools ...
Nutritional status of boarding and non boarding children in selected schools ...Nutritional status of boarding and non boarding children in selected schools ...
Nutritional status of boarding and non boarding children in selected schools ...
 
Statistic in Health Care Management Assignment Week 3Case Study.docx
Statistic in Health Care Management Assignment Week 3Case Study.docxStatistic in Health Care Management Assignment Week 3Case Study.docx
Statistic in Health Care Management Assignment Week 3Case Study.docx
 
In this students will pull together the change proposal project.pdf
In this students will pull together the change proposal project.pdfIn this students will pull together the change proposal project.pdf
In this students will pull together the change proposal project.pdf
 
THESIS 2.pptx
THESIS 2.pptxTHESIS 2.pptx
THESIS 2.pptx
 
feasibility study
feasibility studyfeasibility study
feasibility study
 

Dingus_SchoolFeedingPrograms

  • 1. 0 “Poor Economics WiSe 2015” School Feeding Programs: RCTs as effective Evaluation Tools and their Implications David Dingus Göttingen, 11403109 October 26, 2015 Supervisors: Dr. Sebastian Vollmer & Ms. Jana Kuhnt
  • 2. 1 Introduction “Undernourished children are more likely to become short adults, to have lower educational achievement, and to give birth to smaller infants. Under nutrition is also associated with lower economic status in Adulthood.” (Victoria et al., 2008) Education is widely considered a key component on the pathway to development, yet approximately 59 million primary aged children are not in school, with preferential treatment given to boys over girls in many regions (UNESCO, 2013, World Bank, 2012). In total there are an estimated 300 million chronically hungry children in the world and of those who do attend school, an estimated 66 million children go to school hungry (World Bank, 2004, 2012). This hunger prevents many children from reaching their full potential in school, with life-long consequences. In a case study in Kenya, deworming was used as a proxy for undernourishment and found that children who were given a deworming pill for two years, stayed in school longer and earned up to 20% more than children who received the pill for only 1 year (Baird et al., 2010). Clearly improved nutrition can have long-term effects on the lives of children. School feeding programs began as early as 1790 in Munich as a way to feed hungry children, spreading to France in 1867 (FAO, 2005). Around the same time SFP were first seen in the United States (Dwyer, 1995). Today, SFPs can been found in almost every country, with many different organizational structures and goals. In the developing world SFPs are used as a method to increase student enrollment, attendance, improve child health outcomes, and often target girls. According to estimates from the World Bank (2012), SFPs have a global turnover of approximately $100 billion USD, with the WFP alone reaching 22 million children. Many SFPs have come under criticism for having unclear, ambiguous effects. In this paper we will focus on two Randomized-Controlled-Trials, to show that RCTs can be used to more accurately isolate the effects of SFPs on children and establish that SFPs have clear benefits, but that there are many key factors to be considered with an emphasis on nutrition content. We will begin by discussing the Conceptual framework of SFPs, then discuss the empirical evidence and other factors of consideration, before we conclude with closing remarks. Background Conceptual Framework of SFP SFPs seek to improve both the educational and nutritional outcomes of children and thereby improve their lives in the long-run (Grantham-McGregor et al., 1998). The dynamics of these effects can be seen in Figure 1. SFPs have positive effects by first attracting
  • 3. School Feeding Programs: RCTs as effective Evaluation Tools and their Implications 2 households to enroll their children in school, then by keeping them enrolled longer, and by increasing their attendance because they essentially subsidize the costs of school, by reducing food costs at home (Adelman et al., 2007). In addition, evidence has shown that long-term and short-term child hunger has several adverse effects on children including: reduced cognitive ability and the ability to partake in the learning environment, increased probability of suffering from illness and a higher reate of absence, and result in lower academic performance (Adelman et al., 2007, 2008, Ahmed, 2004). By reducing short and long term hunger, SFPs seek to mitigate these effects. Not all SFPs are the same, there can be many differences in their design such as conditional, targeted, In-School-Meals or Take-Home-Rations, and they can be caloric or micronutrient focused. All SFPs are conditional to the extent that one has to be in school to eat in school, however THRs allow for additional conditions to be set, such as attendance rate and can even be targeted for girls. In this way, THRs are thought to be able to help areas with clear differences in the treatment in boys and girls. More recently, SFPs have begun to focus not only on caloric intake, but also the micro-nutrients and source of those nutrients. In this way a SFP can be used to target specific deficiencies within the target population to have the greatest impact on health and educational outcomes. Empirical Studies Previous Studies Levinger (1986) summarizes the results of 22 SFPs up until the 1980’s splitting them into three groups consisting of those focused on increasing attendance rates, those focused on improved nutrition, and those focused on cognitive outcomes. Overall, though many studies show an increase in attendance, just as many are inconclusive. Similarly, many of the SFPs have ambiguous effects on cognitive development and nutritional outcomes. Levinger (1986) argues that most of these ambiguous and inconclusive results are the result of poor study designs where often times they were retrospective, did not collect a baseline, relied on the perceptions of the teacher, and did not utilize standard testing methods to evaluate outcomes. Overall, Levinger (1986) concludes that the assessment of many SFPs are inappropriately carried out, and also that many SFPs do not take into consideration regional specific factors and are thus not able to correctly target the deficiencies of the population, limiting the positive outcomes of the SFP. RCT in Burkina Faso: THR vs. ISM Research suggests that the period of gestation and then first 2 years of life are the most important in terms of proper nutrition and better health and cognitive development outcomes (Bhutta et al., 2008). For this reason, Kazianga et al. (2009) wanted to compare the effects of
  • 4. School Feeding Programs: RCTs as effective Evaluation Tools and their Implications 3 different SFPs on the at-home siblings of enrolled school children to determine if ISMs show any effect on the redistribution of intra-household calories. The addition of the conditional and targeted THRs sought to determine the spillover effects of the at-home siblings of enrolled children, and to reduce the gap between boys and girls. RCT Design Kazianga et al. (2009) conducted their RCT in rural Burkina Faso to compare the effects of ISMs and THRs side-by-side. Moreover the study differentiates between boys and girls by making THRs conditional on the enrollment and 90% attendance of a girl. ISMs where given at lunch time while the THR consisted of a 10kg bag of flour each month. A visual of the program structure can be seen in Figure 2. The study uses a prospective randomized design to ensure external validity and randomizes at the village level when assigning to the treatment in order to achieve internal validity. By conducting both a baseline and post intervention survey, they are able to construct a counterfactual and calculate the effect of the treatment. (Kazianga et al., 2009) Kazianga et al. (2009) argue that other studies have underestimated the effects from positive spillovers in the household because these effects require testing of children outside of the home. To capture this effect they also conduct their baseline survey on the at-home siblings of children enrolled in the program and likewise conduct a post intervention survey on those siblings. The baseline survey consisted of collecting household backgrounds, wealth, school enrollment of all children, and anthropometric data. Additionally, they asked children to solve simple math problems. They concluded that in general children in the program were experiencing a severe food shortage. The weight for age and height for age anthropometric measures were on average two standard deviations below the reference population, half of the children were underweight and stunted and one-third wasted. Finally, children were able to answer less than 50% of the basic math problems correctly. They then compared their results between those in the control village and the treatment villages and found that five of the variables showed spastically significant differences, but they argue that given that four of these were only significant at the 10% level, they have small coefficients, and that there are a wide variety of variables which show no significant difference, thus concluding that the randomization process was a success. (Kazianga et al., 2009) In the follow up survey, anthropometric measurements were conducted with basic math problems again. However, additional methods to test for educational outcomes were used including the WISC and Raven Matrices tests in order to compare the cognitive abilities between the treatment and control groups. While the Digit Span test was used to test retention and recall memory as a proxy for short term working memory. Kazianga et al. (2009) explain
  • 5. School Feeding Programs: RCTs as effective Evaluation Tools and their Implications 4 that the idea behind these different testing methods was to enable them to quantify improvements in long-term hunger with short-term hunger. Furthermore, though these tests were not administered during the baseline survey, the results could be compared with those children in the control group. Higher scores in the treatment group over the control group could therefore be interpreted as a causal effect from one of the treatments. (Kazianga et al., 2009) Results Using a difference in difference method, Kazianga et al. (2009) find that enrollment increased under both treatments and that boys were not crowded out. While these programs did not show any significant effect on the absenteeism of boys, they did show that girls in either the THRs program or the ISMs program missed on average 1 day more school per month. Kazianga et al. (2009) explain that they were able to determine from their data that absenteeism in girls was determined by the child labor supply in the household, where families with four or more children saw absenteeism for girls decrease, and those with less, increasing with fewer children. The ISMs showed positive effects on short-term memory function as older girls were able to answer simple math problems faster. However, boys showed a statistically significant small reduction in the number of correct answers they provided, possibly a result of decreased learning time. On the other hand, THRs showed no significant effects on learning outcomes. There were no significant effects for any group on cognitive development. Anthropometric measures from the ISMs group showed a marginal weight gain in children between the ages of six and ten, but no impact on their at-home siblings. However, the THRs program showed gains in weight for age and weight for height in the at-home sample. (Kazianga et al., 2009) Kayianga et al. (2009) conclude, that THR have a positive impact on the at-home siblings of enrolled children, confirming positive spillover effects, but little impact on the short-run hunger of enrolled children and thus no improvement on educational outcomes. Whereas the ISMs program showed no evidence of intra-household rationing away from the enrolled child and improved the short-run nutrition and educational outcomes of enrolled students, with no impact on their at-home siblings. Criticisms While the study by Kaziange et al. (2009) does follow the guideline of an RCT, only some of the results can be interpreted with causality, specifically, we cannot not compare the magnitude of the effect between the treatment groups, because we do not know the content of the ISMs treatment, or what was consumed in the household. Therefore, when applying these results to a different population, we cannot say whether THRs will have a greater or lesser
  • 6. School Feeding Programs: RCTs as effective Evaluation Tools and their Implications 5 effect compared to ISMs, because those differences maybe attributed to the content of those programs and would vary based on regional specific factors. It is important to understand that the idea of providing flour to the household is to free up money for additional food items, but we do not know how these households will spend their money. Kazinagne et al. (2009) explain that in effect, a THR is a conditional-in-kind transfer, citing Gahvari & Mattos (2007) who argue that a THR can be considered in this instance a conditional-cash-transfer. However, many studies in development including an RCT on non conditional cash transfer program in Kenya, have concluded that non-conditional transfers are often just as effective with lower costs (Haushofer & Shapiro, 2013). More importantly, given that we want to have the largest possible positive impact on as many children as possible, with particular emphasis on children under the age of two, having a conditional THR may be inhibiting the program from reaching its full potential. “In environments with low enrollment rates, perfectly targeting children who are enrolled might not be the best policy objective if one wants to reach most of the malnourished children […] nutritionists generally believe that the window of opportunity for nutrition is during gestation and the first two years of life […therefore…] it may be desirable to use THR to reach children who are not enrolled since they are likely to be the ones who need the intervention the most.” (Kazianga et al., 2009, p.7) Unfortunately, because the RCT did not include a group with a THR that was not conditional, we cannot determine if the 90% attendance condition had a significantly different impact compared with not having the condition. Furthermore, because there was also no THR group for boys, we cannot be certain of the impact from targeting the THR at girls. In this RCT the Hawthorne effect could be exhibited by the treatment group when they change their behavior and make their children attend school more regularly because they are being observed. On the other hand, the John Henry effect would be captured in the control group when they may also increase their enrollment and attendance rates as a result of being observed. If this were the case, then the treatment effect would not capture these effects and our results would not be biased (Duflo et al., 2007). On the other hand, if the magnitudes of these effects differ or their directions, then the treatment effect would be under or over estimated. Duflo et al. (2007) point out that in most cases, simply conducting the experiment over a long period of time would help eliminate these effects from our results, as evidence suggests that the Hawthorne and John Henry effects diminish over time.
  • 7. School Feeding Programs: RCTs as effective Evaluation Tools and their Implications 6 Kazianga et al. (2009) explain that they tried to minimize non-compliance by randomizing at the village level. In this way it would be more difficult for households to switch to the treatment groups or to be impacted by way of spillover effects from neighboring households. Nonetheless, they explain that noncompliance could still occur if households in the control group have their children attend school in a treatment village or if a household in the treatment village fosters children from the control group (Kazianga et al., 2009, p. 11-12). However, given the geographical barriers, length of time of the intervention, and the commitment to attend school nearly every day, chances of noncompliance are minimal, but should be kept in mind. RCT in Kenya: Nutritional Content Considerations on the nutritional content of SFPs are especially important in the developing world, given that many populations in Sub-Saharan Africa and Southeast Asia have shown to have micronutrient deficiencies and macro nutrient deficiencies stemming from a lack of protein and low animal consumption (Diaz et al., 2003). Based on these findings, Neumann et al. (2007) argue that it is essential that studies establish causal evidence on the effects of different nutritional content in SFPs. RCT Design The RCT was conducted in rural Kenya in an area of 60 km2 that was prone to drought and experienced severe food shortages. The target population consisted of mostly small holder farmers, which consumed mostly plant-based diets with sources of protein from eggs and rarely a goat. The program treatment was randomized at the school level between 4 groups. One group was the control where no SFP was given, while the other 3 groups tested three different treatments consisting of Githeri with meat, with milk, or with oil. The treatment was administered as a mid-morning snack on school days and the study lasted in total over a period of 2.25 years. There was an initial baseline survey, followed by assessments during the program to conclude with a final assessment. Assessments consisted of Anthropometric measures, cognitive and school performance measures and the levels of physical activity and behavior were also documented. By taking a baseline survey in every group, the researches were able to construct a counterfactual to isolate the effect of the different treatments from other factors. (Neumann et al., 2007) Initial baseline findings revealed a representative sample of the target population and in the randomization of assignment to treatment groups with no significant differences between groups. Through blood work, the program did determine that the meat group had the highest vitamin B-12 deficiency, which aids in the absorption of iron and zinc in plant-based diets (Fairbanks, 1999). Overall, they found a high presence of anemia, malaria, stunting,
  • 8. School Feeding Programs: RCTs as effective Evaluation Tools and their Implications 7 underweight, and low muscle mass all of which are associated with not only undernutrition but also a lack of animal sourced protein. (Neumann et al., 2007) Results Unlike the first RCT, this study found evidence of cognitive development, particularly in the meat group as shown in in Figures 3 & 4. The meat group also showed the highest percentage of time in leadership, activity during free playtime, and in initiative behavior as shown in Figure 5. Interestingly, although the treatment with milk was better than the control who received no feeding, it produced in general lower positive results than the meat and oil groups. The milk group also produced the lowest results in the Raven Matrices test. Neumann et al. (2007) explain that this may be a result of the high casein and calcium content in milk that blocks the absorption of iron. Furthermore, in a study by Whaley et al. (2003) they find that iron plays a vital role in cognitive function. In essence, introducing SFPs with milk may impede cognitive growth. In anthropometric measures, all groups who received the treatment showed greater weight gain than the control group. In terms of height, only the milk group showed greater results than other groups for children under 6 years of age and stunted children. All other groups did not show any significant gain in height compared to the control group. This underscores, that on the other hand, a SFP with milk may be appropriate when the target population is young and stunted. Finally, the meat group showed the largest increase in lean muscle mass using the mid-arm muscle area measure. (Neumann et al., 2007) Criticisms Even though this RCT is a replicate study, attrition bias can be seen by the different number of participants between the first and second cohort. In the first cohort, 525 children were observed while that number dropped to 375 in the second cohort, implying a dropout rate of over 28%. If there were similar characteristics shared by those who dropped out, the sample may have no longer be representative and biased the results. Neumann et al. (2007) explain that they used fixed and mixed effects and used a replicate study design where they followed the same students overtime and that statistical methods confirmed the validity of their model. Nonetheless, given that a severe drought and teachers strike occurred during the study, we cannot be sure if those who moved away exhibited more positive characteristics because their families could afford to move, if they exhibited more negative characteristics because their families moved out of desperation, or if they exhibited traits that were on average no different than the initial sample. The RCT may have also captured the Hawthorne and John Henry effects, especially during behavioral observation. From Figure 5 we can see that all groups displayed higher levels of time engaged in physical activity, leadership, and in initiative behavior at the
  • 9. School Feeding Programs: RCTs as effective Evaluation Tools and their Implications 8 beginning of the study that then declines over time. We cannot be sure if this decline is because the students changed their behavior because they were being observed initially or if it is because they are getting older. Duflo et al. (2007) explain that the Hawthorne and John Henry effects generally decrease overtime as the observed unit becomes accustomed to being observed. However, because we have a control group who is also aware that it is being observed, the Hawthorne effect would most likely exhibit the same effect as the John Henry effect in the treatment group, thus any differences can be interpreted as a result of the treatment (Duflo et al, 2007). Also, from our results, we can clearly see differences between groups in their behavior that must go beyond the effect of changed behavior due to being observed. However, as we concluded in the first RCT, we also cannot conclude from this RCT that these results will be similar in other regions, as they may have different nutritional deficiencies, which may be better targeted through different measures. A study on Iodine capsules in Tanzania concluded that if mothers in Central and Southern Africa were to take Iodine capsules while pregnant, there would be a 7.5% increase in educational attainment (Fields et al., 2009). Another study in Indonesia found that when anemic men took iron supplements, they saw their productivity and wages increase (Thomas et al., 2004). In general, it is important to consider not only the calories in a SFP, but also the source of those calories and the micronutrients they contain, and to design a SFP to meet the needs of the target population. Adverse Effects of SFPs In addition, there are general criticisms on the adverse effects of SFPs, often concerning decreasing educational performance. Kazianga et al. (2009) summarizes these theories, first explaining that SFPs often require teachers to spend some of their time organizing and administering meals to students and therefore have less time in the classroom to teach or to prepare. Classrooms may also become overcrowded and other resources may become constrained as a result of increased enrollment if these resources are also not increased. Also the students who did enroll as a result of the SFP may be less motivated because they may perceive that they have less to gain from an education, given that they did not attend school before. Finally, those students who are newly enrolled may have lower learning abilities and/or have large knowledge gaps, thus decreasing the average scores of the school. SFPs could therefore be wrongly interpreted as decreasing education outcomes. (Kaziagna et al., 2009) In the RCT from Kazianga et al. (2009) we saw that older boys scores on answers to math problems decreased by a statistically significant, though small amount. This may be
  • 10. School Feeding Programs: RCTs as effective Evaluation Tools and their Implications 9 evidence of decreased learning time as it only occurred in the ISMs group and not the THRs group, thus discounting the effect of constrained resources. Kazianga et al. (2009) cite another study in Jamaica where school test scores decreased after introducing an ISMs program. Grantham-McGregor et al. (1998) explain that this can be attributed to a lack of organization. Other case studies including one in Bangladesh, have found SFPs to result in lower test scores (Ahmed & Ninno, 2002). However once the results were compared non- eligible students in schools with SFPs to students in schools without SFPs, there was no statistically significant difference in the test scores (Adelman et al., 2008). In general when assessing SFPs we have to carefully consider the method in which we assess the students, especially because these programs may attract lower performing students, thereby lowering the average test results of the school. Furthermore, the amount of learning in the school may decrease as time and other resources become more constrained. It is therefore essential that one considers the regional specific factors when implementing a SFP in order to be able to address possible issues before they become problem (Levinger, 1986). Other Factors to Consider Other factors such as the timing of meals and how they interact with other programs should be considered to ensure maximum effectiveness. School Breakfast Programs in the United States have clearly shown that by administering meals at the beginning of school helps to alleviate short-term hunger in children, which allows them to better focus on the learning environment (Bhattacharya et al., 2004). Yet given this information, many SFPs focus on either snacks or lunches. While this timing will equally improve long-term measurement outcomes in child health, these children may be suffering from short-term hunger until snack or lunch time. As a result, these children will have difficulty concentrating in class and applying themselves for a significant portion of their school day, thus limiting the positive outcomes of the SFP. In order for children to reach their full potential, the nature of their hunger must be carefully considered, and in cases where children suffer from short-term hunger starting from the beginning of the school day, they may benefit from the addition of a SBP. In addition, these RCTs have not considered the effects of how other programs can increase the effectiveness of SFPs. For example, Dialo (2014) assesses the impact of both SFPs and deworming programs in Senegal. He finds that both result in positive outcomes on enrollment and educational outcomes, though deworming is the faster and cheaper of the two options. More importantly, he finds that when combining both programs, the positive outcomes are larger than each program independently. In sum, depending on the
  • 11. School Feeding Programs: RCTs as effective Evaluation Tools and their Implications 10 characteristics and regional specific factors in the target population, other programs can increase the effectiveness of SFPs. Closing Remarks By using RCTs to control for regional specific effects we can isolate the effects of SFPs and have found that SFPs can have a measurable impact on the educational and health outcomes of children, but ensuring that those interventions reach children as early as possible may have larger effects. In addition, how those outcomes are measured and controlled for can determine their results. In particular, we have found that the nutritional content of these programs can be just as important as the caloric value. We were also able to conclude that THRs have a greater impact on younger siblings at-home, whereas ISMs only seems to impact those receiving the food directly with no spillover effects. However, we were unable to draw any causal inferences on the effects of gender based targeting or conditional transfers. To conclude, RCTs are useful tools to ascertain the impact of SFPs because they can establish causality and thus their results can be applied in other settings. However, one must be careful in how they interpret the results of an RCT, because while they can establish the effect of a specific factor given the conditions of the subjects and their environments, the magnitude of these effects will most likely differ in different contexts. Furthermore, there remains much research to be done on the subject of SFPs especially regarding the interaction with other programs, the effects of timing, and more on the nutritional makeup of such programs. However, to obtain the effects of these programs, requires a complex undertaking whereby each variable of interest must be administered to a separate treatment group, which can be both time consuming and costly. As such, though case studies lack external validity, they can provide us with in-depth knowledge of a specific population and allow us to design a more effective SFP. In general, there is no one size fits all SFP. Each case must be tailored to regional specific factors to achieve targeted outcomes. If successful, SFPs can have measurable positive outcomes on the lives of undernourished children, but more emphasis should be placed on the nutritional content of these programs, their organization and timing, while also coordinating with other efforts to ensure that the resources are available for students to reap the full benefits that SFPs seek to provide.
  • 12. School Feeding Programs: RCTs as effective Evaluation Tools and their Implications 11 References Adelman, S., Gilligan, D. O., & Lehrer, K. (2007) How Effective are Food For Education Programs? A Critical Assessment of the Evidence from Developing Countries. The United Nations University. Washington D.C.: International Food Policy Research Institute. Adelman, S., Gilligan, D. O., & Lehrer, K. (2008) The Impact of Alternative Food for Education Programs on Learning achievement and Cognitive Development in Northern Uganda. The United Nations University. Washington D.C.: International Food Policy Research Institute. Ahmed, A. and del Ninno, C. (2002) The Food for Education Program in Bangladesh: An early Evaluation of its impact on Education Attainment and Food Security. The United Nations University. Washington D.C.: International Food Policy Research Institute. Ahmed, A. (2004). Impact of feeding children in school: Evidence from Bangladesh. The United Nations University. Washington D.C.: International Food Policy Research Institute. Baird, S., Hicks, H. J., Kremer, M., & Miguel, E. (2010) Worm at Work: Long-Run Impacts of Child Health Gains. University of California at Berkeley. Berkley. Bhattacharya, J., Currie, J., & Haider, S. J. (2004). Evaluating the Impact of School Nutrition Programs. Washington D.C.: USDA Final Report. Bhutta, Z., Ahmad, R., Black, S., Cousens, K., Dewey, E., Giugliani, B., Haider, B., Kirkwood, S.. Morris, H., Sachdev & Shekar, M. (2008) What works? Interventions for Maternal and Child Undernutrition and Survival. Lancet, 371(3): 417-440. Diallo, F. L. (2014). Evalution of Meal and Deworming Programs for PRimary Schools in Rural Senegal. Masstricht: University of Maastricht. Diaz, J.R., de las Cagigas, A, Rogriquez, R. (2003) Micronutrient deficiencies in developing and affluent countries. American Journal of Clinical Nutrition, 57(1): S70-72 Duflo, E., Glennerster, R., & Kremer, M. (2007). Using randomization in development economics research: A toolkit. Handbook of development economics, 4: 3895-3962. Dwyer, J. (1995) The School Nutrition Dietary Assessment Surveys. American Journal of Clinical Nutrition, 61: 173S-177S. Field, E., Robles, O., & Torero, M. (2009) Iodine Deficiency and School Attainment in Tanzania. American Economic Journal: Applied Economics, 1(4): 140-169. Fairbanks, V. F. (1999) Iron in medicine and nutrition. Modern nutrition in health and disease, 9: 223-239
  • 13. School Feeding Programs: RCTs as effective Evaluation Tools and their Implications 12 Food and Agricultural Organization-FAO. (2005) The state of food insecurity in the world: Eradicating world hunger-key to achieving the Millennium Development Goal. Italy: FAO Gahvari, F., and Mattos, E. (2007) Conditional Cash Transfers, Public Provision of Private Goods, and Income Redistribution. American Economic Review, 97(1): 491-502 Grantham-McGregor, S. M., Chang, S., & Walker, S. P. (1998) Evaluation of School Feeding Programs: Some Jamaican Examples. American Journal of Clinical Nutrition, 67S: 785S-789S. Haushofer, J., & Shapiro, J. (2013). Household response to income changes: Evidence from an unconditional cash transfer program in Kenya. Massachusetts Institute of Technology. Kazianga, H., De Walque, D., & Alderman, H. (2009). Educational and health impact of two school feeding schemes: evidence from a randomized trial in rural Burkina Faso. Washington, DC: World Bank. Levinger, B. (1986) School Feeding Programs in Developing Countries: An Analysis of Actual and Potential Impact. Aid Evaluation Special Study 30. Washington D.C.: U.S. Agency for International Development. Neumann, C. G., Murphy, S. P., Gewa, C., Grillenberger, M., & Bwibo, N. O. (2007). Meat supplementation improves growth, cognitive, and behavioral outcomes in Kenyan children. Journal of Nutrition, 137(4), 1119-1123. Thomas, D., Frankenberg, E., Friedman, J., Habicht, J. P., Jones, N., McKelvey, C., ... & Suriastini, W. (2004). Causal effect of health on labor market outcomes: Evidence from a random assignment iron supplementation intervention. California Center for Population Research. UNESCO (2013) A growing number of children and adolescents are out of school as aid fails to meet the mark. (Policy Paper 22). Paris: UNESCO Victora, C. G., Adair, L., Fall, C., Hallal, P. C., Martorell, R., Richter, L., … for the Maternal and Child Undernutrition Study Group. (2008). Maternal and child undernutrition: consequences for adult health and human capital. Lancet, 371(9609), 340–357. Whaley, S. E., Sigman, M., Neumann, C., Bwibo, N., Guthrie, D., Weiss, R. E., ... & Murphy, S. P. (2003). The impact of dietary intervention on the cognitive development of Kenyan school children. The Journal of nutrition, 133(11), 3965S-39 World Bank. (2004). School feeding programs: why they should be scaled up now. Washington, D.C.: World Food Program.
  • 14. School Feeding Programs: RCTs as effective Evaluation Tools and their Implications 13 World Bank. (2012). Scaling up school feeding: Keeping children in school while improving their learning and health. Washington, D.C.: World Bank. Yendaw, E., & Dayour, F. (2015) Effect of the National School Feeding Programme on Pupils’ Enrolment, Attendance and Retention: A Case Study of Nyoglo of the Savelugu-Nantong Municipality, Ghana. British Journal of Education, Society & Behavioural Science, 5(3): 341-353.
  • 15. School Feeding Programs: RCTs as effective Evaluation Tools and their Implications 14 Appendix Figure 2. RCT Design. (Kazianga et al., 2009, p. 23). Figure 1. Conceptual Framework of SFPs. (Yendaw & Dayour, 2015, p. 345). Figure 3. SFP Impact on Cognitive Outcomes. (Neumann, 2007, p. 1121).
  • 16. School Feeding Programs: RCTs as effective Evaluation Tools and their Implications 15 Figure 4. SFP Impact on Education Outcomes. (Neumann, 2007, p. 1121). Figure 5. SFP Impact on Behavioral Outcomes. (Neumann, 2007, p. 1121).