Evaluation of alternative school feeding models on children’s educational outcomes: preliminary findings from a cluster randomised control trial Dr. Clement Adamba School of Education and Leadership, University of Ghana GHANA EDUCATION EVIDENCE SUMMIT 2017 28 March 2017 1
Other colleagues Felix Asante 1 , Elisabetta Aurino 2 , Irene Ayi 1 , Kwasebena Bosompem 1 , Gloria Folson 1 , Lesley Drake 2 , Aulo Gelli 3 , Meena Fernandes 1 , Edoardo Masset 4 , Isaac Osei-Akoto 1 , Getrude Ananse-Baiden 2 1 University of Ghana, Ghana 2 Imperial College London, UK 3 IFPRI, US 4 3ie, UK 2
School feeding as a policy linking education, health and agriculture Figure 1: Education Nutrition and health School feeding Agriculture SME development • Simple idea but programmes are complex! – Managing complexity and trade-offs across objectives is not straightforward 3 3
What we do know so far about school feeding and education is that …. Overall weighted average effect Number of studies ACCESS TO SCHOOLING 0.14* 7 Enrolment 0.09* 6 Attendance – 0.06* 3 Dropout 0 2 Completion LEARNING OUTCOMES 0.09* 8 Language arts scores 0.10* 10 Math scores 0.14 3 Composite test score Source: Snilstveit and others 2015. 4
The Ghana School Feeding Programme • Launched in 2005 • Hot, cooked meal to children in public schools • 2012: 1 in 3 children enrolled in public schools were beneficiaries 5
First rigorous evaluation of school feeding in Ghana • Theory-based impact evaluation designed around scale-up of GSFP in order to expand the linkages between the programme and local agriculture in high food insecurity areas • 116 schools across all regions of Ghana • Randomisation 58 school feeding schools 29 SF standard GSFP model 29 HGSF+ pilot (GSFP + agriculture/nutrition package) 58 no school feeding (pure controls, get SF after year 3) • 3 year study in partnership with Government of Ghana, University of Ghana, Partnership for Child Development at Imperial College and IFPRI • Baseline in 2013 and endline in 2016 6
Randomisation Figure 2: 58 districts (116 schools) 1 st stage randomisation 29 HGSF+ 29 GSFP districts districts (58 schools) (58 schools) 2 nd stage randomisation 29 HGSF+ 29 control 29 GSFP 29 control schools schools schools schools (725 HH) (580 HH) (725 HH) (580 HH)
Main educational outcomes Indicator Metric Children’s Educational access enrolment, attendance, grade repetition and drop out Attention and cognition Digit span, Raven matrices: 12 questions Learning achievement Scores on literacy and maths tests: 15 questions each 8
Descriptive statistics Table 1: Gross enrolment Arm School level 2013 2016 Control KG 78.2 51.4 Primary 97.6 101.0 JHS 46.5 71.4 GSFP KG 81.0 57.5 Primary 101.7 111.0 JHS 49.4 66.0 GHSF+ KG 81.8 62.1 Primary 98.6 102.4 JHS 50.2 71.1 9
Figure 4: Net Enrolment at the Primary School level 90.0 80.0 70.0 60.0 50.0 40.0 30.0 20.0 10.0 - Control GSFP GHSF+ Baseline Endline % Change 10
Figure 5: Net Enrolment at the Junior High School level 70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0 Control GSF HGSF Baseline Endline Change 11
Figure 4: Percent that missed a school day 35.0 30.0 25.0 20.0 15.0 10.0 5.0 0.0 2013 2016 2013 2016 Female Boys Control GSFP 12
Impact evaluation analytical approach Two main approaches: 1. Difference-in-Differences (DD) estimator Design group Baseline Follow-up Difference Treatment A B B - A Control C D D - C Difference A - D B - C DD = (B - A) - (D - C) 2. Mixed effects models • Mixed-effects models (multi-level regression models) account for the hierarchical nature of the data. 13
Impact on School enrolment • The school feeding increases the likelihood of enrolment by about 2%**. • Net enrolment at Kindergarten level increased by nearly 13% and 11% in regular GSF & HGSF schools respectively 14
Table 2: Impact results of GSF and HGSF on Schooling outcomes Estimate Repeated grade Missed a day Dropped out GSF/HGSF (Control) -0.183 0.137 -0.005 (0.126) (0.216) (0.453) HGSF (GSF) 0.017 -0.781** -0.101 (0.190) (0.304) (0.652) Reduce the chance of missing a school day by 6.87% 15
Impact on learning and cognition Table 3: Impact results of GSF and HGSF on learning outcomes Estimate Literacy Maths Raven test Digit span GSF+HGSF (Control) 0.040 0.051 0.039 -0.030 (0.039) (0.041) (0.032) (0.029) HGSF (GSF) -0.010 0.054 0.076 0.035 (0.111) (0.103) (0.087) (0.068) 16
Table 4: Impact results of GSF on learning outcomes, by sex Estimate Literacy Maths Raven test Digit span HGSF (GSF) Boys 0.013 0.064 0.113 0.066 (0.136) (0.105) (0.096) (0.083) Girls -0.037 0.042 0.033 0.001 (0.119) (0.127) (0.096) (0.070) 17
• Girls (6-12 years) in Primary school that received the GSF/HGSF performed significantly higher in Literacy and Maths test. Table 5: Impact results of GSF and Estimate Literacy Maths Raven test Digit span HGSF on learning outcomes by sex of child GSF+HGSF (Control) Boys -0.031 -0.004 0.008 -0.069* (0.055) (0.056) (0.044) (0.040) 0.127** 0.120** 0.079* 0.017 Girls (0.057) (0.059) (0.046) (0.043) 18
This translates to: • Language art -13.5% • Mathematics - 12.7% • Raven test - 8.2% • Conversely, SF appears to have a negative impact on boys’ reasoning ability. Reduced boys reasoning ability by about 7%. • This requires further investigation to unravel the cause of this unintended negative outcome. 19
• One of the pathways to the improvement in learning outcomes, we think, is the changing time use by children in school feeding schools Table 6: Children time use Time-Use SF Control Diff Hours spent on care or chores 1.66 1.60 0.06 Hours spent at work in farm or 0.62 0.69 -0.07* for pay Hours spent in leisurely activities 2.10 2.11 -0.01 Hours spent at school or studying 6.39 5.96 0.43*** Observations 5124 20
• This points to a somewhat protective role of school feeding for children. • Reduced time spent at work and 25.8 additional minutes per day spent at school or studying. • This was strong for girls in school feeding areas, who spent 31.8 additional minutes per day at school or in study as compared to boys (19.8 additional minutes). 21
Concluding remarks • In-school meals have positive impact on enrolment and on girls learning and reasoning outcomes. • Reduces or eliminates delayed enrolment and encourages enrolment at the appropriate age. • There is need to take measures; enhance the nutritional value of the meals provided sustain the gains made in increased enrolment by paying attention to retention and completion. 22
Thank you 23
Recommend
More recommend
