1 WFP’S USDA McGovern -Dole International Food for Education and Child Nutrition Program’s Support in Kenya from 2016 to 2020 Baseline Report – Final November 2017 WFP Kenya Country Office Baseline Manager : Beatrice Mwongela Agreement: FFE-615-2016/014-00 Prepared by the independent baseline consultant team: Muriel Visser, Team Leader; Gregory Naulikha, Deputy team leader, Education specialist; Moses Mwangi, Lead Statistician; Ernest Midega, Deputy Statistician, data collection specialist. Decentralized evaluation for evidence-based decision making 2 Acknowledgements The baseline design team would like to thank the various people who provided suggestions and inputs into this work. A particular thank you is extended to WFP baseline committee for their advice and guidance on the baseline process and to the Reference Group for their review and comments on this report. The team also thanks the numerous education officials, teachers, parents and pupils who made themselves available for this study. Disclaimer The opinions expressed in this report are those of the Baseline Team, and do not necessarily reflect those of the World Food Programme. Responsibility for the opinions expressed in this report rests solely with the authors. Publication of this document does not imply endorsement by WFP of the opinions expressed. The designation employed and the presentation of material in the maps do not imply the expression of any opinion whatsoever on the part of WFP concerning the legal or constitutional status of any country, territory or sea area, or concerning the delimitation of frontiers. The authors declare they have no conflict of interest in the present assignment. DISCLAIMER: This publication was produced at the request of the United States Department of Agriculture. It was prepared by an independent third-party evaluation firm. The author’s views expressed in this publication do not necessarily reflect the views of the United States Department of Agriculture or the United States Government. Accessibility Note: An accessible version of this document can be made available by contacting fas.monitoring.evaluation@usda.gov iii Table of Contents Executive Summary......................................................................................viii 1. Introduction.................................................................................................1 1.1. Introduction to the Baseline Study ............................................................... 1 1.2. ObjectivesoftheBaselineSurvey ................................................................ 3 2. Study Methodology ...................................................................................... 4 3. Survey findings...........................................................................................12 3.1. Introduction ....................................................................................... 12 3.2. Characteristicsoftherespondents.............................................................. 12 3.3. LearningOutcomes ............................................................................... 16 3.4. ShortCtermHunger ............................................................................... 32 3.5. School meals and expected outcomes .......................................................... 40 3.6. IncreasedCapacity................................................................................ 49 3.7. Foodutilization and food safety................................................................. 55 4. Associated Factors..................................................................................... 66 4.1. FactorsassociatedwiththehighestlevelofEnglishliteracy (story)foraclass2workamong school going children in class 3 to 8...................................................................... 66 4.2. Factors associated with the highestlevel of Kiswahili literacy (story)for a class 2 work among school going children in class 3 to 8 ............................................................. 67 4.3. Factors associated with the highestlevel of numeracy (division)for a class 2 work among school going children in class 3 to 8...................................................................... 68 5. Discussion and implications ...................................................................... 70 5.1. LearningOutcomes ............................................................................... 70 5.2. Adjusting for other ongoing interventions ..................................................... 72 5.3. Short Term Hunger................................................................................ 73 5.4. School Meals and Expected Outcomes.......................................................... 74 5.5. Pupilandparentalperceptions relatedtohygiene,nutritionandeducation ............... 75 5.6. Progress towards sustainability ................................................................. 75 5.7. NationalCapacity ................................................................................. 76 5.8. Selectedimplications forthemidCandendClinephasesandfor schoolfeedinginKenya . 76 Annex 1 – MGD PerformanceMonitoring Plan ................................................80 Annex 2 – Detailed Baseline Methodology ...........................................................97 Annex 3 – Overview of coordination mechanisms for school feeding in Kenya .............................................................................................................107 iv Annex 5 – Types of school feeding in Kenya and overview of key coordination structures and stakeholders .......................................................................... 109 Annex 6: Summary of BaselineIndicator Values ............................................ 116 Annex 7: Analysis of Factors Associated with Specific Learning Outcomes ..... 124 Annex 8: Comparing distribution of specific variables between study arms stratified by gender of child ............................................................................ 131 Annex 9 – Computation of the Propensity Score ............................................ 148 Annex 10 - UWEZO 2016 results by county ........................................................ 149 Acronyms ................................................................................................... 150 List of figures in this report Figure 1 - Comparison of the quasi-experimental initial design and final situation .................. 6 Figure 2 - Selection of Control and WFPSMP schools using PSM ............................................... 8 Figure 3 - Selection of WFPSMP and HGSMP schoolsusing PSM.............................................8 Figure 4 - Data collection sites for the baseline.........................................................................10 Figure 5 – Percentage of boys and girls across English literacy categories of achievement for all types of schools using the UWEZO methodology (n=5130) .......................................... 18 Figure 6 - Percentage of children who attained the highest level English score in the UWEZO test (story), regardless of type of school, by class and by gender ......................................18 Figure 7 – English literacy scores by category of achievement, comparing control schools with WFPSMP schools, by gender ......................................................................................19 Figure 8 - English literacy scores by category of achievement, comparing WFPSMP schools with HGSMP schools, by gender ........................................................................................20 Figure 9 – Percentage of boys and girls across Kiswahili literacy categories of achievement for all types of schools using the UWEZO methodology (n=5130) .......................................... 21 Figure 10 - Percentage of children who attained the highest level Kiswahili score in the UWEZO test (story), regardless of type of school, by class and by gender ........................ 22 Figure 11 – Kiswahili results by category of achievement, comparing control schools with WFPSMP schools, by gender ................................................................................................. 22 Figure 12 - Kiswahili results by category of achievement, comparing WFPSMP schools with HGSMP schools, by gender .................................................................................................... 23 Figure 13 – Percentage of boys and girls across categories of numeracy achievement for all types of schools (n=5130) ....................................................................................................... 24 Figure 14 - Percentage of children who attained the highest level Numeracy score in the UWEZO test (division), regardless of type of school, by class and by gender ................... 25 v Figure 15 - Numeracy results by category of achievement, comparing control schools with WFPSMP schools, by gender ................................................................................................. 25 Figure 16 - Numeracy results by category of achievement, comparing WFPSMP schools with HGSMP School, by gender ..................................................................................................... 26 Figure 17 - Percentage of boys and girls for all types of schools who report “sometimes” finding it difficult to concentrate in class (n=5130) ............................................................. 28 Figure 18 - Percentage of children who sometimes find it difficult to concentrate in class comparing control and WFPSMP schools, by gender .......................................................... 29 Figure 19 – Percentage of children who sometimes find it difficult to concentrate in class comparing WFPSMP schools with HGSMP School, by gender .......................................... 29 Figure 20 - Reasons why children "sometimes" find it difficult to concentrate in class .......... 31 Figure 21 - Percentage of parents/guardians who reported their children ate daily before going to school, comparing control and WFPSMP schools, by gender ........................................ 33 Figure 22 - Percentage of parents/guardians who reported their children ate daily after going to school, comparing control and WFPSMP schools, by gender ........................................ 33 Figure 23 - Percentage of parents/guardians who reported their children ate daily before going to school, comparing WFPSMP schools and HGSMP schools, by gender ............... 34 Figure 24 - Percentage of parents/guardians who reported their children ate daily after going to school, comparing WFPSMP schools and HGSMP schools, by gender ......................... 34 Figure 25 – Percentage of parent/guardians with acceptable food consumption score (FCS) by gender ...................................................................................................................................... 35 Figure 26 - Percentage of parents/guardians with acceptable FCS by level of education of parent/guardian ...................................................................................................................... 36 Figure 27 - Proportion of children residing in households with acceptable FCS, comparing control and WFPSMP schools ................................................................................................ 36 Figure 28 - Proportion of children residing in households with acceptable FCS, comparing WFPSMP and HGSMP schools .............................................................................................. 37 Figure 29 – Reported coping strategies on days when the family did not have enough food, or money to buy food .................................................................................................................. 38 Figure 30 - CSI comparing WFPSMP and control group schools, by gender...........................38 Figure 31 - CSI comparing HGSMP and WFPSMP schools, by gender ..................................... 39 Figure 32 - Percentage of parents/guardians indicating that their child had received school meals in the current school year (2017), comparing Control and WFPSMP schools, by gender ...................................................................................................................................... 41 Figure 33 - Percentage of parents/guardians indicating that their child had received school meals in the current school year (2017), comparing WFPSMP and HGSMP schools, by gender ...................................................................................................................................... 41 Figure 34 - Percentage of parents/guardians indicating that their child had received school meals in the week of the survey, comparing Control and WFPSMP schools, by gender .42 vi Figure 35 - Percentage of parents/guardians indicating that their child had received school meals in the week of the survey, comparing WFPSMP and HGSMP schools, by gender 43 Figure 36 - Comparison of the average number of students attending school 80% of the time in Control and WFPSMP schools,by gender .....................................................................44 Figure 37 - Comparison of the average number of students attending school 80% of the time in WFPSMP and HGSMP schools, by gender ....................................................................45 Figure 38 - Average number of boys, girls, and overall students enrolled in control and WFPSMP schools ................................................................................................................46 Figure 39 - Average number of boys, girls, and overall students enrolled in WFPSMP and HGSMP schools...................................................................................................................46 Figure 40 - Distribution of parental/guardian responses on the benefits of education (n=5130) .................................................................................................................................. 47 Figure 41 – Parents/guardians in target communities who could name at least three benefits of primary education, by gender ............................................................................................ 48 Figure 42 - Percentage of parents/guardians in control and WFPSMP schools who can name at least three benefits of primary education ......................................................................... 48 Figure 43 - Percentage of parents/guardians in WFPSMP and HGSMP schools who can name at least three benefits of primary education.............................................................49 Figure 44 - Comparison between the percentage of control and WFPSMP schools that store food off the ground..............................................................................................................56 Figure 45 - Comparison between the percentage of WFPSMP schools and HGSMP school that store food off the ground.............................................................................................57 Figure 46 - Percent of food preparers at control and WFPSMP schools who achieve a passing score on a test of safe food preparation and storage..........................................................58 Figure 47 - Percent of food preparers at WFPSMP and HGSMP schools who achieve a passing score on a test of safe food preparation and storage ............................................59 Figure 48 – Frequency and importance of different hygiene habits mentioned by children in response to the survey.........................................................................................................60 Figure 49 - Proportion of children who responded to the survey who mentioned three most important hygiene methods, by research arm (Control and WFPSMP)...........................60 Figure 50 - Proportion of children who responded to the survey who mentioned three most important hygiene methods by research arm (WFPSMP and HGSMP)...........................61 Figure 51 - Frequency by which children mentioned different types of nutritional habits ...... 62 Figure 52 – Proportion of children from Control and WFPSP schools who mentioned three most important nutrition efforts ........................................................................................... 63 Figure 53 - Proportion of children from WFPSP and HGSMP schools who mentioned three most important nutrition efforts ........................................................................................63 vii List of tables Table 1 - Performance of children in class 3 to 8 on class 2 tests across WFPSMP and Control schools ........................................................................................................................................ x Table 2 - Performance of children in class 3 to 8 on class 2 tests across WFPSMP and HGSMP schools ........................................................................................................................................ x Table 3 - Study population in the three arm target counties ....................................................13 Table 4 - Study pupils' characteristics ........................................................................................... 13 Table 5 - Characteristics of the schools covered by the Study...................................................15 Table 6 - Highest English Literacy Level ofthe Child ...............................................................20 Table 7 - Highest Kiswahili Literacy Level (story) ofthe Child.................................................23 Table 8 - Highest Numeracy Level (division) ofthe Child........................................................27 Table 9 - Child sometimes finds it difficult to concentrate in class ..........................................30 Table 10 - Government Funding of HGSMP since inception and the coverage in terms of Number of Pupils and Number of days the children were fed in a year ............................ 54 Table 11 - WFP SMP Funding and Actual Beneficiaries reached (2009-2016) ......................... 54 Table12CFoodpreparationconditionsinthethreearmsofthestudy.Error!Bookmarknotdefined. Table 13 C Storage conditions in the three arms ofthe study ............ Error! Bookmark not defined. Table 14 - Selected county ranks - class 3 who can do class 2 Work (across all competencies) ...................................................................................................................................................... 70 List of boxes Box 1 – PTA, BoM and Student’s Government responses across the three target arms of the study on the impact of hunger/absence of school meals on attention and willingness to go to school ................................................................................................................................... 31 Box 2 – PTA, BoM and Student’s Government responses across the three target arms of the study on food security situation in families .......................................................................... 39 Box 3 - PTA, BoM and pupil responses across the three target arms of the study on the importance of school meals for families and communities...............................................43 Box 4 - PTA, BoM and pupil responses across the target arms of the study on how parents come to know about the benefits of education ..................................................................49 Box 5 - Responses through the qualitative interviews as to the perceived importance of hygiene and suggestions on how to strengthen this component of the intervention .......61 Box 6 - PTA, BoM and pupil responses across the three target arms of the study on the importance of nutrition and how it could be improved going forward ............................64 viii Executive Summary Background This report summarizes the baseline findings for Kenya’s World Food Programme (WFP) implemented School Meals Programme (SMP). The WFPSMP is funded by the United States Department of Agriculture’s (USDA) – Mc Govern Dole (MGD) International Food for Education and Child Nutrition Programme.It consists of a USD 28 million grant for a period of five years (2016-2020) which covers the bulk of annual requirements 1 , although other donors also support school feeding in Kenya. The WFPSMP seeks to contribute to improved enrollment, retention and attentiveness at school level. These outcomes are expected ultimately to contribute to improved literacy and numeracy in primary schools in the intervention areas. Purpose The purpose of the baseline is to establish a clear benchmark for WFP and her partners with information against project indicators. The baseline thus: • Records the situation at the start of the intervention phase in terms of output and performance indicators. • Provides a situational analysis of the conditions for implementation of the SMP. • Forms the foundation for planned midterm and final evaluations. The baseline takes place at a stage when WFP has since 2009 been handing over the management of school feeding in other areas of the country to the Government of Kenya (GoK) run Home Grown School Meals Programme (HGSMP). HGSMP schools receive funding from the GoK to procure food locally. By the end of the MGD funded WFPSMP programme in 2020 all WFP schools will have been handed over to the GoK and integrated into the HGSMP. Methodology An inception report for this study outlined the proposed methodology and was approved by the Internal Committee and USDA. The inception phase concluded with the finding that a quasi-experimental design was feasible for this study given that it was possible to get a match between the intervention and control groups. The inception stage also resulted in the agreement to use a three-arm quasi￾experimental design which involves doing two sets of comparison, namely between WFPSMP schools and a group of WFPSMP control schools, and a second comparison between WFPSMP schools with HGSMP schools. The first comparison (WFPSMP and control schools) provides the means for examining what differences the SMP makes to key education and nutrition indicators. The HGSMP versus WFPSMP arm of the study provides a means to assess progress on sustainability, given that HGSMP schools have been handed over to the GoK. The comparison is therefore meant to inform the transitioning of WFPSMP to HGSMP. The baseline was undertaken using various primary data collection tools at school level. Secondary data was collected from Government and WFP records as well as through interviews. Data for the baseline were collected in March and April of 2017. Selected control and 1 In 2016 and 2017 USDA’s contribution covered 68 percent ofthe contributions.Other main donors (in terms of volume of funding) were Canada, Germany, Japan, Australia, ix HGSMP schools were matched against WFPSMP schools using propensity score matching. Four main data collection tools were used which included a tool to measure literacy and numeracy. Data collection covered a sample of 5130 pupils and an equal number of parents in 90 schools. Sampling took place using a two-step sampling process, across the three arms of the study. Data was collected in five out of the six targeted Arid and Semi-Arid Lands (ASAL) counties (Garissa, Turkana, Mandera, West Pokot, and Wajir). Overview of findings Characteristics of the schools A comparison of characteristics of the schools established differences in terms of the following conditions in the three sets of schools: • A significantly higher proportion of WFPSMP schools had a storage facility (82.6%) compared to control schools (43.5%). • A significantly higher proportion of HGSMP schools had a large enough kitchen for preparing food for pupils (82.6%) compared to WFPSMP schools (43.5%). • A significantly higher proportion of WFPSMP schools indicated that most pupils wash their hands (81.8%) compared to control schools (33.3%). Similarly, a significantly higher proportion of WFPSMP schools indicated that most pupils wash their hands (83.3%) compared to HGSMP schools (40.0%)2. • A significantly higher proportion of WFPSMP schools indicated that their cook is trained in food storage and handling (47.8%) compared to control schools (17.4%). A significantly lower proportion of WFPSMP schools indicated that their cook is trained in food storage and handling (17.4%) compared to HGSMP schools (60.9%). There were no significant differences in other entry characteristics of the three groups of schools.3 The next section examines the situation at baseline against each of the main MGD strategic objectives and high level indicators. In line with the objectives of the baseline, each set of results reports first on the comparison between the WFPSMP schools and the control group, followed by the comparison between HGSMP schools and the WFPSMP schools. MGD SO 1: Improved literacy of school age children This indicator compared literacy scores in English and Kiswahili and numeracy scores at baseline for children across the three arms of the study using the UWEZO literacy and numeracy tool. The test involved doing computations (for mathematics) and reading and comprehension of a text (in English and Kiswhahili) at grade 2 level, by children in grades 3 to 8. The results show that children in the control arm outperformed the children in the 2 It is important to note that there are two sets of WFP SMP school, one set matched with control schools and the other set matched with HGSMP schools and thus different percentages for WFP SMP schools 3 Other characteristics included comparing data on: teacher attendance; pupil attendance; the proportion of pupils completing the last grade of primary; the status of the kitchen; the availability of fuel saving stoves; the water supply; sanitation conditions, and the presence of library facilities. x WFPSMP schools in English literacy level, in Kiswahili literacy, and in numeracy on the highest category of these tests.4 Table 1 - Performance of children in class 3 to 8 on class 2 tests across WFPSMP and Control schools English Literacy Kiswahili Numeracy Control WFPSP Control WFPSP Control WFPSP 55.6% 40.6% 66.0% 51.2% 73.5% 60.9% The baseline also shows that children in the HGSMP outperformed the children in WFPSMP schools in English literacy level, in Kiswahili literacy, and in numeracy, in the highest category of these tests. Table 2 - Performance of children in class 3 to 8 on class 2 tests across WFPSMP and HGSMP schools English Literacy Kiswahili Numeracy HGMSP WFPSP HGMSP WFPSP HGMSP WFPSP 64.6% 45.0% 74.9% 53.5% 77.7% 60.1% MGD 1.2: Improved Attentiveness Regardless of type of school, somewhat less than half of the children in the sample (43.0%) indicated that they sometimes find it difficult to concentrate in class. The proportion of children who sometimes find it difficult to concentrate in class was significantly higher in the control arm (46.4%) than in the WFPSMP schools (41.1%)5. This percentage was also significantly higher in the HGSMP arm (43.5%) than in the WFPSMP schools (37.4%). Stratification by gender revealed consistent results in both cases. Across all groups, “I am hungry”, followed by “I am feeling sick” ranked as the two most prevalent explanations for why children at times find it difficult to concentrate in class. MGD 1.2.1 Reduced Short-Term Hunger Just over one-third of the parents/guardians (38.7%) across all groups of schools indicated their children ate food daily (in the last week) before going to school. The proportion of parents/guardians who indicated their children ate food daily (in the last week) before going to school was significantly higher in control schools 4 The UWEZO literacy test categorizes capacity according to the whether the pupil can read ‘nothing’, only a ‘word’, only a ‘sentence’, or a ‘story’. The highest category therefore corresponds to being able to read the story that is part of the test. Similarly, in numeracy the ‘highest’ UWEZO category skill is division (after addition, subtraction and multiplication). 5 Please note that as the denominators are different for some variables a small percentage difference in one part of the analysis may be significant, while it may not be significant in other analyses where the denominator is much lower. xi (38.0%) than WFPSMP schools (33.0%). This proportion was also higher among children in the HGSMP schools (43.2%) when compared with WFPSMP (38.7%) schools. Looking at Food Consumption Scores (FCS) 6, over a third of the children (39.5%) across all groups resided in households with acceptable FCS, another third (32.2%) lived in households with borderline consumption and approximately three out of 10 children (28.3%) were living in households with poor consumption scores. Additional analysis established that the proportion of households with acceptable FCS was significantly higher among male parents/guardians (42.9%) than female parents/guardians (38.2%). It was also higher among parents/guardians with higher levels of education (college/university and technical), and lower among those without education or who had not completed primary level. No significant difference in FCS was found between children in the control and WFPSMP group and between those in the HGSMP and the WFPSMP. There was no significant difference in coping strategies between the three arms of the study (33.1% for the control schools versus 32.0% for the WFPSMP schools on the first comparison, and 35.2% and 35.6% respectively in the HGSMP versus WFPSMP comparison). In order of importance, the reported coping strategies included: purchase food on credit (80.8%); reliance on less preferred and less expensive foods (80.6%); reduce number of meals eaten in a day (67.4%); limit portion size at mealtimes (66.2%); borrow food, or rely on help from a friend or relative (64.9%); restrict consumption by adults in order for small children to eat (50.6%); and skip entire days without eating (42.3%). Coincidentally, withdrawing children from school did not feature in the coping mechanisms cited. MGD 1.2.1.1/1.3.1.1.Increased Access to Food (School Feeding) Approximately half of the parents/guardians reported that their children have been receiving school meals at school in the current school year (2017). The proportions were consistent among boys and girls. The proportion of parents/guardians who reported that their child had been receiving school meals (at school) in the current school year (2017), was significantly higher in WFPSMP (59.3%) schools compared to control schools (20.0%). The proportion was also significantly higher in HGSMP (80.4%) than in WFPSMP (55.6%). 7 Two out of five of the parents/guardians reported that the school where their child was learning was serving food during the survey week. WFPSMP (51.7%) schools were 6 The “Food consumption score” is a “score calculated using the frequency of consumption of different food groups consumed by a household during the 7 days before the survey” (WFP (2008). Food Consumption Analysis. WFP/VAM, p.8). Details of the methodology can be found at: http://documents.wfp.org/stellent/groups/public/documents/manual_guide_proced/wfP197216.pdf (accessed 06 August 2017). 7 It is important to note that the Baseline Survey was undertaken at a time when the drought was severe in the target counties. Further, there was a pipeline break in the WFPSMP in term 1 because no funding was available for SMP. While there was no direct school feeding from WFP during the survey period therefore, the feeding in some WFSMP schools during the survey period (while not expected) was due to the government and other actors intervening in these areas in response to the drought. In addition, a small number of WFP schools were providing school feeding with carryovers from the previous phase of the SMP. xii more likely to be serving food in the survey week than control schools (16.3%) and HGSMP schools were more likely (51.5%) than WFPSMP schools (43.9%) to be serving food. MGD 1.3 Improved Student Attendance At baseline 85.0% of students in WFP SMP schools, 83.4% of students in control schools and 84.7% of students in HGSMP schools were regularly attending school. There was no significant difference between the average number of students regularly attending WFPSMP schools (232 total, of which 128 boys and 104 girls), compared to control schools (184 total of which 102 boys and 83 girls). However, the average number of students regularly attending (327 total, of which 185 boys and 142 girls) was significantly higher in HGSMP schools than WFPSMP schools (191 total, of which 107 boys and 83 girls). MGD 1.3.4 Increased Student Enrolment A total of 14,2848 students were enrolled in WFP SMP schools as compared to 8133 students in control schools and 9883 students in HGSMP schools. There was no statistically significant difference in average enrolment in the comparison between the schools in the three arms of the study. Average enrolment in control schools (375 total, of which 207 boys and 168 girls) was not significantly higher than WFPSMP schools (280 total, of which 155 boys and 125 girls) when compared to control schools. Average enrolment in HGSMP schools (430 total, of which 243 boys and 186 girls) was also not significantly higher than WFPSMP schools (290 total of which 163 boys and 127 girls). The totals and averages in enrolments will be computed both at midline and end line. MGD 1.3.5 Increased Community Understanding of the Benefits of Education Two out of five parents/guardians in target communities could name at least three benefits of primary education, with a significantly higher proportion of male parents/guardians (47.2%) able to list three benefits than female (39.8%). Parents/ guardians in WFPSMP schools were generally more able to name benefits of primary education when compared to control and HGSMP schools. MGD 1.4.1 Increased Capacity of Government Institutions At the national level, the baseline established that there is room for improving the participation by other ministries in school feeding efforts. A Technical Committee that brings together various stakeholders exists and meets on an ad hoc basis to provide technical support on implementation to the Ministry of Education. A National Inter￾Ministerial Steering Committee does not yet exist and intersectoral county committees that are foreseen remain to be established. 8 This total comprises of all the 46 WFP SMP schools visited compared to 23 schools for control and 23 schools for HGSMP. For WFP SMP school (min=28, aver=317,max=1524 and SD=241), HGSMP schools (min=125, aver=429,max=1113, SD=257) control schools (min=146, aver=364, max=931, SD=187) xiii Progress has been made in strengthening the policy framework. However, the National School Health, Nutrition and Meals Programme Strategy remained to be formally approved at the time of the baseline. Government funding for school feeding has increased in nominal terms but remains insufficient to cover school feeding needs. In 2016, government funding allowed for 77 days of school feeding out of 190. It is worth noting that at the baseline, the GoK led HGSMP was in 19 Counties while WFPSMP was in only 7 Counties. There are no other partners implementing the SMP other than WFP and Feed the Children who focus efforts in the informal settlements in Nairobi. MGD SO 2 Increased Use of Health and Dietary Practices WFPSMP schools are more likely to store food off the ground compared to control schools. The proportion of WFPSMP schools that store food off the ground (56.5%) was significantly higher than control schools (17.4%). However, there was no significant difference between food storage off the ground in HGSMP schools compared to WFPSMP schools (52.2% for HGSMP and 47.8% for WFPSMP). MGD 2.2 Increased Knowledge of Safe Food Prep and Storage Practices Food preparers at WFPSMP schools and at control schools had comparable scores on the test for safe food preparation (43.5% versus 39.1%). However, a significantly higher percentage of food preparers at HGSMP schools achieved a passing score on a test of safe food preparation and storage (73.9%) compared to food preparers in WFPSMP schools (43.5%). MGD 2.3 Increased Knowledge of Nutrition The proportion of children who mentioned at least three hygiene habits was significantly higher in WFPSMP schools (51.0%) compared to the control group (19.8%). Similarly, the proportion of children who mentioned at least three hygiene habit was significantly higher in WFPSMP schools (50.4%) than in HGSMP schools (20.7%). The most important nutrition habits mentioned by children include; balanced diet (42.7%) and food type (39.8%). Associations between variables The baseline assessment also examined associations between different variables. The objective was to determine association/relationship. The approach did not test or prove causal relationship and the results should thus be interpreted with some caution. Using this methodology, factors associated with the highest level of English and Kiswahili literacy as well as highest numeracy for a class 2 Work among school going children in class 3 to 8 were assessed. The analysis revealed that key factors associated with the highest level of English and Kiswahili literacy as well as highest numeracy for a class 2 WORK among school going children in class 3 to 8 include: • Class of the child. • Mode of travel to school. xiv • Number of times child normally eat per day. • Child had a meal today before going to school. • Child thought it is important to go to school. • Child having brothers and sisters who currently study in this school. • Child having brothers and sisters who are old enough to go to school but are NOT currently attending school. • Education level of the parent/guardian. • Number of important nutrition habits mentioned by the parent/guardian. • Number of hygiene habits mentioned by the parent/guardian. • Household Coping Strategy Index. The associations between these variables should enable WFP and its partners to further design and/or improve already existing intervention strategies. Brief reflection on the findings The process of transitioning WFPSMP to HGSMP in Kenya was initiated in 2009. The initial focus was on the counties in the semi-arid areas that were easier to transition given their agro-pastoral economy and the fact that they were better watered, better serviced and had a more developed school system. Consequently, the WFPSMP schools that transitioned to the HGSMP in that first phase are in socio￾economic conditions that are somewhat better than the schools which transitioned in 2015, which are in more arid areas. The schools in the arid locations suffer constrained capacities, considerable enrollment and attendance disparities, and high levels of food insecurity and malnutrition. This is the context in which the schools were selected for the three-arm study – namely WFPSMP, control schools and HGSMP based on vulnerability, food security and education indicators. The baseline compared literacy and numeracy scores for the WFPSMP schools with the control group, and with the HGSMP schools which is the group of schools that have transitioned to the government programme. An important point to note is that the literacy and numeracy scores that were obtained in this baseline are comparable (i.e. in the same range) to those of the 2013 and 2016 UWEZO assessments. Both assessments consistently find low scores for the ASAL areas. This confirms the reliability of the instruments used for this study. A key finding from the study is that in both comparisons (WFPSMP versus control, and HGSMP versus WFPSMP) the WFPSMP schools score lower on literacy and numeracy and on other education indicators such as attentiveness. Enrolment was the only indicator for which no difference was found in both sets of comparison. Differences between the WFPSMP and HGSMP schools are in part likely to be the reflection of the fact that WFPSMP schools/target counties are in the most marginalized, arid and excluded zones of Kenya which as noted above have consistently performed poorly in the UWEZO tests. These arid zones have suffered long drawn and extreme educational marginalization from colonial times through to post independence, because of which they record low rates on virtually all education xv parameters. It is therefore likely that some of the differences can be explained by the fact that the HGSMP schools are in areas of the country that are less marginalized and better served economically, socially and politically, and as noted the HGSMP schools were purposely selected for earlier transitioning as they were considered the easiest to transfer. Other differences between the HGSMP schools and the WFPSMP schools that emerge from the baseline may also reflect the relatively better-off status of HGSMP schools/counties. For example, the study baseline also finds that more children in HGSMP schools eat before going to school compared to WFPSMP schools. In terms of the difference between WFPSMP schools and control schools on education indicators it should be noted that while every effort was made to select schools in similar zones to those where the WFPSMP schools were located for the purpose of having a control group, this proved to be very challenging in practice given that WFP targets all schools in each county. The baseline therefore had to select schools in neighboring counties which were identified as being as similar as possible against identified indicators for comparison. However, it is likely that these counties were not similar enough given that WFP-supported schools are exclusively located in Kenya’s northern arid counties which have consistently ranked at the bottom quarter of UWEZOs list (Garissa, Turkana, Mandera and Wajir counties have figured as the bottom four counties for the past two assessments). As none of the ‘control schools’ were drawn from the bottom ranked quarter of the list this would clearly affect comparability, making it rather difficult to effectively compare WFP-supported schools with the control group or the HGSMP. It might therefore also be important to go beyond the comparison between these different schools and focus on progression of the WFP schools over the course of the project in relation to the baseline point which has been measured through the present study. The baseline shows that WFPSMP schools have better scores than control schools on selected indicators related to pupil and parental perceptions and practices in the areas of hygiene, nutrition and education. This could suggest that attention to hygiene, nutrition and the importance of education has been stronger in the ASAL areas, given that these geographical areas have been prioritized by many other actors (UNICEF, and various NGOs) and that the higher awareness may be the result of interventions from other organizations in these areas. Further, previous McGovern-Dole projects (between 2004 and 2013) could have had an impact on indicators in WFP schools. While the baseline was not able to unequivocally establish that this is the case, the mid - and end-line measurements will establish to what extent these indicators will evolve further and how this compares to any change in the comparison groups (the control and the HGSMP). It will be important in the next phases of data collection to further examine some of these differences, and to ensure that the data collection tools (and the qualitative part of the study) focusses on the reasons for these differences. Meanwhile it is important to highlight that despite these differences the baseline has achieved the objective of making it possible to record the values for each of the schools in each of the study arms. This provides the basis for the mid and end-line phases to compare how WFPSMP schools have evolved in terms of these indicators compared to any changes in control and HGSMP schools. xvi The intention of the comparison between the WFPSMP schools and the HGSMP schools was to identify progress towards sustainability. While the significant differences between the locations of the schools make comparison of the educational indicators challenging there are some differences that are notable. HGSMP schools perform well on food preparation scores, but they do not perform well on hygiene and education awareness. As is the case for the control schools this may reflect the lack of exposure to activities that target education and hygiene awareness. It may also suggest that in the transitioning process attention to hygiene has been lost, and that this is reflected in the poorer scores. The higher scores on food preparation for HGSMP schools, would however, suggest that training that has been provided in this area has been relatively successful (WFPSMP cooks are still to be trained, as the baseline was started before activities started). It should also be remembered in this context that pupils are a ‘moving target’ in an intervention of this kind (as they move on to other schools or leave the education system and are replaced by new pupils) whereas cooks are likely to stay in the same position for multiple years, consolidating what they have learnt. Further in terms of sustainability the baseline highlights that there are still important conditions to be met for the transitioning process to be adequately supported, in particularly in terms of coordination, policy framework, and timely budget allocation. These will clearly need attention in the coming period, given that with the transitioning of the WFPSMP schools to the GoK the number of schools that are part of the HGSMP will increase rapidly. Implications for the mid- and end-line phases and for school feeding in Kenya more broadly The baseline survey has demonstrated that a quasi-experimental design is feasible. Going forward through the mid line and end line evaluations, it would be important to ensure that: • The same schools visited during the baseline are visited during the midline and end line. • The changes in school meals programmes in the schools are documented and considered at both the midline and end line • The same sampling strategy is maintained at midline and end line. • Other cofounding factors that might influence the outlined hypothesis are documented and reported at both midline and end line. The experience of the baseline exercise would suggest that for the mid-line and end￾line exercises it would be sensible to do the qualitative data collection after the quantitative analysis. This will make it possible to have a more in-depth understanding of differences that are highlighted from the quantitative data and make for an approach to the qualitative questioning that is more directly related to gaps in understanding. The benchmark values for the indicators in the MGD Performance Monitoring Plan, coupled with the overall baseline findings and the analysis of associated factors, point to the need to focus on the following areas in implementing the WFPSMP: • Progress has been made in drafting the National School Health, Nutrition and xvii Meals Programme Strategy, however it remains to be formally approved. WFP should continue to advocate for a speedy adoption and implementation of this strategy. • Adequate and regular funding through the Government budget is a challenge in transitioning schools to the Government led HGSMP (see also below). WFP should work with partners in advocating with the Ministry of Finance and the Treasury for ring-fencing SMP budgets (which would be consistent with the GOK’s social protection commitments). It should also advocate for regular and timely disbursements of GoK funds to schools, and for a progressive increase in government funding to the HGSMP. • Strong participation by all partners, regular meetings, and better coordination are critical to using scarce resource more efficiently and effectively. With the exception of nutrition, coordination among key ministries and programmes at national level remains weak. WFP should support the Ministry of Education in establishing the National Multi-sectoral Steering Committee, and support the GoK in seeking stronger participation of key ministries and programmes such as social protection and agriculture in this forum. • The baseline highlights poor literacy and numeracy scores of the schools in the WFPSMP areas. School feeding can offer only part of the solution and WFP should therefore actively coordinate its efforts with that of other partners in the same counties to ensure that the factors that affect school participation and achievement are addressed in a holistic manner. • The analysis of associated factors that are presented in this report should inform further research and guide programming work by WFP and its partners to further design and/or improve already existing intervention strategies. WFP could use these findings as input into a meeting with partners to discuss how to strengthen support to education in the targeted areas. • To support the transition WFPSMP schools there should be a strong focus on mentorship and capacity building of school leaders to be able to properly manage school feeding. • Given the importance of involving various sectors and actors in school feeding to cover all dimensions of the programme (nutrition, literacy, local production, etc.), WFP should actively support counties where it is operating in setting up County Level Multi-SectoralSteering Committees (bringing in health, agriculture, academic institutions, the private sector, and other partners as relevant) and support these groups with capacity development if necessary. • WFP should develop a clear and convincing case for decentralizing the management of school feeding to the county level, as is already the case for the management of ECD. This will promote a better quality programme where the key partners respond to/are answerable to their constituents and are more in tune with needs and requirements of the county, and will allow for better allocation and utilization or resources and monitoring programme implementation. A decentralized programme would also allow for the establishment of mechanisms by which counties might advance funds to purchase goods at the best possible time, and make the school feeding interventions more cost effective. • Most of the schools that participated in the baseline were having pipeline xviii breaks. WFP should identify the key factors that are contributing to these pipeline issues and ensure regular delivery to the schools to minimize the number of days without school feeding. Finally, the results of the baseline which compared the HGSMP and WFPSMP schools highlight challenges in the transitioning process. Recommendations in this matter fall outside of the strict scope of the WFPSMP but are still captured here as they are important for the broader group of stakeholders, and if they do not receive attention may in the future also affect the WFPSMP schools that are transitioning. In the view of this evaluation a successful transition from WFPSMP to sustainable HGSMP will need to take into account the following considerations: • Adequate budget allocations should be set aside (and ring fenced) by the GoK and disbursed to schools on time to ensure timely and cost effective purchase of the requisite food. • Capacity should be developed at all levels – national, county, and school to ensure effective implementation of the intervention. • The transition process should be allowed adequate time to ensure contextualization of best models and practices. • Enhanced coordination among all the stakeholders at various levels – national, county and school will be key to the success of the HGSMP. • Strong linkages with local smallholder farmers and traders and enhancement of their capacity to tap into the school markets effectively will be critical to the success of the HGSMP • The baseline finds that HGSMP schools perform well on food preparation scores, but they do not perform well on hygiene and education awareness. This suggests that in the transitioning process attention to hygiene has been lost and that this may need attention. • While it is the Kenya Government’s responsibility to provide food to school going regions in Arid and Semi-Arid areas, particularly with the anticipated transition to HGSMP, donors and other supporters will need to walk with the Ministry of Education over the transitional period to ensure success of the move from WFPSMP to HGSMP. To do this effectively, it would be important for support towards the transition efforts to be more coordinated to reduce uncertainty and breaks in the resources that are necessary for an effective transition. It is also critical that adequate time be given to the transition process as well as predictable technical and other associated support through WFP. Kenya still has a lot to be done on these elements to ensure a sustainable Home Grown School Meals Programme. In so doing, the country will still need the technical and financial support of the various partners that have brought the SMP this far. 1 1. Introduction 1.1.Introduction to the Baseline Study 1. The United States Department of Agriculture (USDA) – Mc Govern Dole (MGD) International Food for Education and Child Nutrition Programme have granted the World Food Programme (WFP) Kenya US$ 28 million to support its programme in Kenya that will run from 2016 -2020. The MGD program supports education, child development and food security in low-income, food-deficit countries around the globe. Support includes United States (US) produced agricultural commodities and financial assistance as well as support to capacity development and to monitoring and reporting. Sustainability of interventions is a critical consideration for USDA. It is worth noting that MGD is not the only funding agency to the programme. It is a multi￾donor supported intervention to which MGD provided 70 percent of the financial contribution in 2016 and 2017.9 2. Since the inception of the School Meals Programme in Kenya in the 1980s, WFP works closely with Kenyan Government Ministries (Education, Agriculture, and Health), with counties in Kenya, and with other partners to provide school meals to vulnerable children in arid counties and in the unplanned settlements of Nairobi. It also works to improve the management and implementation of the national school feeding programme and to strengthen the capacities of national, county and school level actors to ensure reliable and cost-efficient and-effective implementation of the intervention. 3. The current MGD programme is the last of four phases of support, and will result ina full hand-over of the school feeding programme to the Government of Kenya (GoK) by 2019. Previous phases of USDA support included three single year awards in 2004, 2005, and 2006, and three multi-year phases awarded in 2007 (2007-2009), 2010 (2010-2012), and 2013 (2013-2016), respectively. These phases were followed by the current multi-year phase awarded in 2016 (2016-2020). The total funds awarded between 2004 and 2015 amount to approximately 93 million USD. 4. A process of transitioning WFPSMP schools to the Government started in Kenya in 2009, and involves what is known as the Home Grown School Meals Programme (HGSMP). The first phase of transitioning focussed on the semi-arid counties that were relatively easier to transition and which are characterized by a relatively favourable agro-pastoral economy, good rainfall, better services and a more developed school system. The programme includes strengthening linkages with smallholder farmers to enhance agricultural production and promote local purchasing of food as key to the sustainability of HGSMP. A second transitioning process focuses on the arid counties under the current MGD programme. These counties represent a completely different context. They are arid, vast, and poorly populated, food insecure and have suffered marginalization for a long time. They have poor infrastructure in general and schools which are far apart. Consequently, to transition these counties effectively a completely different model from that used the semi-arid counties has been conceived. This model is based on transitional cash transfers. The model has been developed, 9 Annex 5 provides an overview of funding by donors to school to MGD school feeding between 2014 and 2017. 2 piloted and the initial counties where it was operationalized by WFP have been handed over to government. 5. In the spirit of transition, the MGD 2016-2020 programme is divided into two phases. For the first period of three years (2016-2018), the program will provide daily school lunches to a total of 358,000 primary school children in targeted arid and food insecure counties of Kenya. At the end of the first three years the responsibility for the school feeding will have been handed over to the HGSMP and the government will be responsible for managing the programme. Support from USDA will then continue for a further two years (2019-2020) in the form of WFP’s continued technical assistance to further strengthen institutional structures and to ensure that the capacity is in place for the management of the HGSMP in Kenya. 6. Over the five years the programme will be implemented in eight counties: Baringo, Garissa, Mandera, Turkana, Wajir and West Pokot, Marsabit and Tana River. The latter two counties will not receive food but will benefit from complementary activities. The complementary activities focus on: strengthening governance and multi-sectoral coordination and collaboration for the school meals programme; advocacy and dialogue to ensure adequate and regular budget allocations and to maintain political commitment to the programme; strengthening oversight and managementfunctions; and empowering communities to manage school feeding activities through training and capacity building of school managers, teachers, and parents in order to ensure a solid level of awareness about school feeding implementation principles. 7. At the school-level, the MGD School Meals Programme (SMP) includes WFP support to train education officials to monitor school feeding and train trainers among local education, health and agriculture officers, equipping them to facilitate school feeding management trainings at the sub-county level. WFP shares the responsibility for the commodity delivery with the Ministry of Education (MOE), with WFP managing the pipeline and ensuring delivery to central warehouses and the MOE transporting commodities to the sub county level and to schools. The hot lunch with food from MGD funds will be served for 120 out of the 190 school days, comprising 150 grams of bulgur wheat, 40 grams of green split peas, 5 grams of vegetable oil (fortified with vitamin A and D), and 2 grams ofiodized salt–to be procured separately byWFP from funds from other sources. 8. The SMP seeks to contribute to improved enrollment, retention, and attentiveness at school level and ultimately contribute to improved literacy and numeracy in primary schools in the intervention areas – together with actions promoted by other partners like the USAID Funded Tusome Programme. The Tusome (“Let’s Read’’ in Kiswahili) Early Grade Reading Activity is a collaboration between the MOE, USAID and UKAID to improve learning outcomes in English and Kiswahili in Class 1 and 2. The Tusome Programme was conceptualized and developed as a national literacy programme and implemented in all public primary schools in the country (including those in the WFPSMP target counties). It targets approximately 60,000 teachers and 22,600 schools for improvement in literacy instruction and outcomes. It is envisaged that 5.4 million class 1 and 2 pupils will be twice as likely to meet MOE benchmarks for literacy. The programme is being implemented in all public primary schools and 1000 alternative basic education institutions serving low cost urban settlements 3 countrywide. 9. In parallel, there are interventions in place that seek to improve critical gaps in nutrition and hygiene awareness as well as strengthen literacy and numeracy. The aforementioned Tusome programme targets pupil literacy. Within the education sector, the United Nations Children’s Fund (UNICEF) is working with the GoK to update the current national curriculum, an essential step to improve the quality of teaching and pupils’ learning experience. WFP is supporting this process and providing inputs to the review of the national curriculum. UNICEF also aims to increase enrolment, through awareness campaigns - sensitizing communities about the importance of education. UNICEF is also active in the Water, Sanitation and Health (WaSH) sector, providing toilets and running water at school level. These activities complement the SMP. The MOE is implementing an initiative aimed at improving numeracy under the Global Partnership for Education (GPE) grant. In terms of scope, the literacy and numeracy interventions noted are country-wide interventions and overlap with the target counties and schools in the baseline study. The WASH and nutrition interventions are implemented by different actors with a concentration in the counties targeted in the baseline because these are areas where the situation of WASH and nutrition is challenging/dire and warrants attention. 1.2.Objectives of the Baseline Survey 10. The purpose of the Baseline Survey - for which the Terms of Reference (ToR) can be found in the Inception Report (Visser et al, 2017) - is to establish a clear benchmark for WFP and her partners with information against project indicators at the start of the intervention (see Annex 1) and a set of values against which to verify the targets. 11. The baseline thus: • Records the situation at the start of the intervention phase in terms of output and performance indicators for the lower level results in the logical framework. These baseline values will be used to regularly monitor progress. • Provides a situational analysis – based on a desk review of documentation and a small number of interviews – of the conditions for implementation of the SMP at the baseline. • Forms the foundation for planned midterm and final evaluations which will measure performance indicators for MGD strategic objectives as well as the indicators of highest level results that feed into the strategic objectives. 12. The present report is divided into the following chapters: • The next chapter (Chapter 2) outlines the methodology for establishing the baseline for which more information can be found in Annex 2, and discusses study limitations. • Chapter 3 forms the bulk of the findings of the study and covers the findings of the survey for each of the MGD Strategic Objectives (SO) and key indicators. • Chapter 4 examines key associations between variables. • Chapter 5 presents a further discussion of the baseline findings and considers 4 selected implications. 2. Study Methodology 13. The present baseline has been prepared using a combination of primary data collection and secondary data available from Government and WFP records. A detailed methodology for the baseline was drawn up during the inception phase (Visser et al, 2017). An important aspect of the inception phase was to establish whether the envisioned quasi-experimental design for the study was feasible. As the team’s assessment showed that this was feasible - given that it was possible to get a match between the intervention and control groups - the study was designed in line with these parameters. 14. A three-arm quasi-experimental design was employed for the study. The baseline involves two sets of comparison between these three ‘types’ of schools, namely schools where the MGD funded WFPSMP operates, WFPSMP control schools, and schools that are part of the HGSMP (i.e. schools that have been taken over by the government). The two comparisons are as follows: • A comparison of WFPSMP schools with the WFPSMP control schools. • A comparison of HGSMP schools with WFPSMP schools. 15. The HGSMP arm of the study was included in the baseline purely to assess the progress of sustainability through government-led and owned interventions. It is the reason why the choice of sample schools was based on those that have been in the programme for a while and not those that had just been handed over in the recent past (2014-15 in the arid counties). The inclusion was a response to a request that was made during the initial/inception briefings with the WFP Kenya Country Office and USDA/MGD teams and it was included in the inception report which was approved by the baseline reference group. This comparison will provide an opportunity for reflection on how best to transition the WFPSMP arid counties to HGSMP at the end of this phase of the intervention, including what should be done to ensure sustainability given the context of aridity, food insecurity and other factors like long drawn marginalization in the target areas. 16. The research questions and testable hypotheses that underpin the quasi-experimental design focus on examining whether the baseline, mid-term and end-line primary education outcomes (literacy and numeracy levels) and other educational indicators (enrolment, attendance, etc.) in the arid and semi-arid lands (ASAL) areas of Kenya are the same in schools included in WFP/USDA-MGD school meals programme (2016 -2020) as those not included (controls and those transitioning to HGSMP). Four different hypotheses were formulated and proposed for testing at mid-term and end￾line for each indicator. These hypotheses are further explained in Annex 2, and are as follows: Indicator 1: • H0: Enrolment in schools included in WFP/USDA-MGD SMP ≠ Enrolment in schools not included in WFP/USDA-MGD SMP 5 • H1: Enrolment in schools included in WFP/USDA-MGD SMP= Enrolment in schools not included in WFP/USDA-MGD SMP Indicator 2: • H0: Attendance rate in schools included in WFP/USDA-MGD SMP≠ Attendance rate in schools not included in WFP/USDA-MGD SMP • H1: Attendance rate in schools included in WFP/USDA-MGD SMP = Attendance rate in schools not included in WFP/USDA-MGD SMP Indicator 3: • H0: Primary school completion rate in schools included in WFP/USDA-MGD SMP ≠ Primary school completion rate in schools not included in WFP/USDA-MGD SMP • H1: Primary school completion rate in schools included in WFP/USDA-MGD SMP = Primary school completion rate in schools not included in WFP/USDA-MGD SMP Indicator 4: • H0: Literacy/numeracy rate in schools included in WFP/USDA-MGD SMP ≠ Literacy/numeracy rate in schools not included in WFP/USDA-MGD SMP • H1: Literacy/numeracy rate in schools included in WFP/USDA-MGD SMP = Literacy/numeracy rate in schools not included in WFP/USDA-MGD SMP 17. The inception phase identified key parameters for the study including: procedures for sampling and required sample size; data collection approach and tools; and, procedures for data analysis. 18. The conceptual framework for the MGD intervention envisages realization of two results as follows: 1. Results framework #1: MGD Strategic Objective (SO) 1 Improved Literacy of School-Age Children. 2. Results framework #2: MGD SO2 Increased Use of Health and DietaryPractices. 19. Since MGD SO2 is a function of MGD SO1, the sample size was calculated based on MGD SO1. The baseline estimate aligned to MGD SO1 Was interpreted to be the proportion of children ages 7-13 that have attained literacy and numeracy at Standard 2 level. 20.UWEZO10Kenya’s Sixth Learning Assessment Report December 2016, suggested that the learning outcome by selected counties on Class 3 who can do Class 2/Standard 2 level work showed a substantial degree of variance.11 Due to variation in baseline estimate across selected counties and with potential variation in other measurement indicators, this study design used a 50% conservative estimate as the proportion of 10 Uwezo is a five-year initiative that aims to improve competencies in literacy and numeracy among children aged 6-16 years old in Kenya, Tanzania and Uganda, by using an innovative approach to social change that is citizen driven and accountable to the public. 11 The proportions in the proposed intervention areas ranged as follows; Wajir – 9.9%, Mandera – 10.1%, Turkana – 11.4%, Garissa – 12.9%, West Pokot – 15.4%, and Baringo – 16.6%. 6 children aged 7-13 that have attained literacy and numeracy of a Standard 2 level￾Standard 2 competencies in literacy and numeracy. The proportion optimized the sample size to allow for estimation of all indicators devoid of the risk of low sample size calculation. The study presumed a 20% effect size on the primary indicator. 21. The minimum sample size was calculated using the Fleiss et al formula. This resulted in a sample size calculation per study arm (without replacement) of 689. To address gender mainstreaming and women’s empowerment as per WFP’s evaluation principle of gender equality, the overall sample size in both interventions (WFPSMP and HGSMP) and control arms was tripled to 4,134 (2067 boys (689 HGSMP, 689 WFPSMP, 689 controls); 2,067 girls (689 HGSMP, 689 WFPSMP, and 689 control). As each pupil questionnaire also included questions for a corresponding parent (see Annex 4), an equal number of parental responses was sought (i.e. 4,134 parents). Actual participants surpassed the targeted number and added up to 5130 with approximately equal number of boys (2558) and girls (2572). An equal number of parents were reached (5130), of which 1446 were male, and 61 percent (3684) were female. 22.The original design in the inception report for the study anticipated a matching of 30*30*30 for the three groups of schools where these schools would all overlap. In reality, the data collected allowed for the matching of 23 schools from each set. In this manner, 23 WFPSMP schools were matched with 23 control schools, and 23 HGSMP schools were matched to 23 WFPSMP schools.12 Figure 1 - Comparison of the quasi-experimental initial design and final situation 23.While the matching and number of schools is different from the design it had no implications for the study as such as the comparison between WFPSMP and HGSMP 12 A total of 92 schools would thus have been covered by the study. However, data were not collected at two of the selected schools because of challenges of accessibility of the schools during the data collection phase. The final count of schools covered by the study was therefore 90 across the three arms of the study. 7 was not part of the initial design. 24.Primary data collection was undertaken in five (Garissa, Turkana, Mandera, West Pokot, and Wajir)13 out of the six14 targeted ASAL counties. Control schools were selected from the neighboring areas (either within the same county or in a neighboring county in a manner that matched as closely as possible the socio-economic activities and livelihood zones to ensure similarity in terms of vulnerability and food insecurity).15 HGSMP schools were also selected from the neighboring areas with comparable socio-economic activities.16 25.Control and HGSMP schools were matched against WFPSMP schools using propensity score matching (PSM). Selected school characteristics derived from the MOE Education Management Information System (EMIS) tool assisted in facilitating matching of schools using PSM. Characteristics (covariates) that were used in matching included: boy to girl ratio; average pupils/class; pupils to teacher ratio; and residence type (rural/urban). These characteristics are generally known to influence academic performance in schools and thus were identified and/or computed to carry out the PSM. 26.Schools in the first group with a propensity score lower than the lowest observed value in the second group were discarded. Similarly, schools in the second group with a propensity score higher than the highest observed value in the first group were also discarded. The same approach was used for the control group. The remaining schools were in the ‘region of common support’ from which participating schools were selected. This process resulted in the identification of three groups of schools that were as similar as possible in terms of characteristics that influence academic performance. 27. Figure 2 and 3 demonstrate comparison of schools before and after matching. 13 Isiolo, Nairobi, Samburu, and Tana River were excluded from the HGSMP group for the following reasons: Nairobi was excluded because the majority of the counties of focus are in the arid, rural areas, consequently, there were hardly any common contextual similarities that will match the urban context of the capital; the other three have been beneficiaries of the Transitional Cash Transfers to Schools Model developed and implemented by WFP and the Ministry of Education before being handed over to HGSMP – consequently their evolution modality and short history of the same does not approximate to a pure HGSMP modality of government that has been going on in some of the selected counties since 2009. 14 Baringo was initially included as a target county for data collection. However due to security concerns it was not possible to undertake data collection in this county. 15 The control schools were located in Elgeyo Marakwet, Kajiado, Kitui, Laikipia, Machakos, Makueni, Nyeri and Taita Taveta. 16 This covered Elgeyo Marakwet, Embu, Kajiado, Kitui, Laikipia, Machakos, Makueni and Nyeri. 8 Figure 2 - Selection of Control and WFPSMP schools using PSM 100% Distribution of schools by PSM Before Matching 80% 60% 40% 20% 0% 0.000 C 0.100 C 0.200 C 0.300 C 0.400 C 0.500 C 0.600 C 0.700 C 0.800 C 0.900 C 0.099 0.199 0.299 0.399 0.499 0.599 0.699 0.799 0.899 1.000 CONTROL(N=305) WFPSMP (N=899) 100% Distribution of schools by PSM After Matching 50% 0% 0.100 C 0.199 0.400 C 0.499 0.600 C 0.699 0.700 C 0.799 0.800 C0.899 0.900 C 1.000 CONTROL(n=23) WFPSMP (n=23) Figure 3 - Selection of WFPSMP and HGSMP schools using PSM 100% Distribution of schools by PSM Before Matching 80% 60% 40% 20% 0% 0.000 C 0.100 C 0.200 C 0.300 C 0.400 C 0.500 C 0.600 C 0.700 C 0.800 C 0.900 C 0.099 0.199 0.299 0.399 0.499 0.599 0.699 0.799 0.899 1.000 WFPSMP (N=899) HGSMP (N=241) 100% Distribution of schools by PSM After Matching 80% 60% 40% 20% 0% 0.000 C 0.099 0.100 C 0.199 0.200 C 0.299 0.300 C 0.399 0.400 C 0.499 0.500 C 0.599 0.600 C 0.699 WFPSMP (n=23) HGSMP (n=23) 9 28.A A two-stage sampling procedure was employed at the WFPSMP sites as follows. • First stage sampling: involved the selection of primary sampling units (PSUs) - i.e. schools - across the five selected counties. Using probability proportionate to size (PPS) method, the PSUs were distributed across the five counties. Selection of schools within counties was done using simple random sampling, with application of a random number generator. • Second stage sampling: involved the selection of secondary sampling units (SSUs) which were children ages 7-13 years in class 3 to 8, across the selected schools. Distribution of school specific sample size allocation was done across gender and school grade using PPS, where gender specific samples across school grade were drawn. Selection of children within gender and across school grade was done using simple random sampling, with application of a random number generator. 29.Data collection for the baseline took place in March/April 2017. Data collection was preceded by a five-day training of a team of 88 enumerators and supervisors on the process. Data were collected from a total of 90 schools 17 using real time digital data collection and supplemented by manual data registration and audio recording for the focus group discussion (FGD) in schools. A Global Positioning System (GPS) picking capability was integrated into the mobile/electronic version of the data collection script to ensure that data corresponded to the correct schools. 30.A total of 5130 pupils and their parents/guardians were covered by the study. The parent-pupil data collection tool for grades 3 to 8 was the main data collection tool. It was developed as one continuous tool which was responded to first by the parent of the child and then by the child (without the parent present). The parent-pupil tool examined parents’ awareness of the value of education, and views on the barriers to enrolment, participation and learning, situation at home in terms of asset ownership (productive and non-productive), agricultural land holding and land tenure system, issues of food security, nutrition, siblings and whether these go to school, and hygiene. From the pupil’s perspective, the tool examined issues affecting enrolment, attendance, attentiveness, the importance of education, knowledge of nutrition and hygiene, and importantly also included the UWEZO numeracy and literacy test. 18 17 Data collection was planned for a total of 92 schools. However, issues related to the access meant that two schools could not be reached at the time of data collection. 18 Uwezo is a five-year initiative that aims to improve competencies in literacy and numeracy among children aged 6-16 years old in Kenya, Tanzania and Uganda, by using an innovative approach to social change that is citizen driven and accountable to the public. 10 Figure 4 - Data collection sites for the baseline 31. Additional data collection covered head teachers in all schools, selected class teachers, cooks, representatives of the Parent Teacher Associations (PTA) and of the School Board of Management (BOM). FGD complemented survey data collection for all informants and served to gain in-depth insight into the perception of teachers, parents, PTA members, and pupils of the issues behind poor enrolment, attendance and retention. It also explored the role of school feeding and other measures which may impact on performance of pupils. All data collection tools can be found in the inception report (Visser et al, 2017). 32.Ethical considerations were taken on board in the study in the following manner: • Enumerators training included a substantial training on the ethical considerations for conducting surveys in schools, in particular with pupils. • A courtesy call was made to the county district education official before starting the activity. • The head teacher consented to the study before any activity was undertaken inthe school. 11 • The teachers introduced the enumerators to the class to explain the purpose of the exercise. • Participation was voluntary and all participants were told that they could opt not to participate. Participants who consented to being part of the study were informed at the start of the interview that they could discontinue the interview at any time without any repercussions. All participants were thanked at the end of the data collection. • Consent was sought from teachers, pupils and parents. Parents were interviewed prior to the interviews of their respective children so that consent could be sought for the interviews with the children. • All responses were coded and the individual performance of students was not traceable to the student or shared with the participants. All data collected has been kept confidential and none of the information in the report can be traced to specific informants. 33.Data analysis was done using IBM SPSS version 24.0. MS-Excel was used to generate graphical presentation of specific findings. 34.The next Chapter presents the findings of the baseline across the three arms of the design. 12 3. Survey findings 3.1.Introduction 35.The survey findings present the results across the three arms of the study with respect to the USDA MGD indicators. The chapter is divided to cover the main objectives and indicators as follows: • Learning outcomes – this section discusses the findings with respect to indicators of literacy and numeracy for the school children aged 7-13 years, as well as indicators on attentiveness and student attendance. • Short term hunger - this section covers the situation with respect to food consumption by children during the day and week. • School meals and expected outcomes – this section presents the situation with respect to access to food and to school meals during the year of the study (2017) and in the week of the survey. It also reports on the situation with respect to community understanding. • National capacity - examines the situation with respect to capacity, government support, policy and regulatory framework at the time of the baseline • Food utilization and food safety – covers issues related to hygiene and nutrition and provides the baseline with respect to the situation in the schools in terms of food preparation and storage and the knowledge of nutrition. For each of these headings, quantitative findings from the survey instruments are presented first. Where appropriate, qualitative findings are also presented to provide additional insights and further understanding. 3.2. Characteristics of the respondents 36.In this part of the report, an overall picture of population and school characteristics for the three-arm target population baseline survey is presented. The survey was completed in 90 schools in 14 counties, covering 5130 pupils (2558 boys, 2572 girls) 5130 parents (1446 male, 3684 female) 34 head teachers (25 male, 9 female) and 90 PTA and BoMs (Table 3). 13 Table 3 - Study population in the three arm target counties Characteristic Frequency Number of Counties 14 Number of Schools 90 WFSMP schools 44 Control Schools 23 HGSMP Schools 23 Number of Pupils sampled for the survey 5130 Boys 2558 Girls 2572 Head Teachers Interviewed 34 Male 25 Female 9 Teachers Interviewed 56 Male 34 Female 22 Parents Interviewed 5130 Male 1446 Female 3684 PTA and BoMs reached 90 37. Table 4 presents the distribution of study pupils by study arm and grade. Enrolment per study arm (Control (1396), WFPSMP (2221), and HGSMP (1513)) was approximately in the ratio of 1:2:1, while that of gender (boys (2558), girls (2572)) was approximately 1:1. Enrolment by grade was almost equal across class 3 to 7, with class 8 slightly lower. Table 4 - Study pupils' characteristics Variable Boys (n=2558) Girls (n=2572) Total (n=5130) N % n % n % Study arm Control 675 26.4% 721 28.0% 1396 27.2% WFPSMP 1146 44.8% 1075 41.8% 2221 43.3% HGSMP 737 28.8% 776 30.2% 1513 29.5% Grade Class 3 438 17.1% 459 17.8% 897 17.5% Class4 443 17.3% 464 18.0% 907 17.7% Class 5 472 18.5% 428 16.6% 900 17.5% Class 6 455 17.8% 439 17.1% 894 17.4% Class 7 442 17.3% 438 17.0% 880 17.2% Class 8 308 12.0% 344 13.4% 652 12.7% 14 38.Table 5 presents the characteristics of the schools that were reached in the study. It captures aspects related to teacher and pupil attendance rates, completion rates,food storage and cooking facilities (kitchens), sources of water supply, sanitation facilities and associated hygiene practices in schools and in relation to the preparation of school meals. This information shows that: • A significantly high proportion of WFPSMP schools had a storage facility (82.6%) compared to control schools (43.5%); (p=0.006). • A significantly high proportion of HGSMP schools had a large enough kitchen/facility for preparing food for pupils (82.6%) compared to WFPSMP schools (43.5%); (p=0.004). • A significantly high proportion of WFPSMP schools indicated that most pupils wash their hands (81.8%) compared to control schools (33.3%); (p=0.046). Similarly, a significantly high proportion of WFPSMP schools indicated that most pupils wash their hands (83.3%) compared to HGSMP schools (40.0%); (p=0.030). • A significantly low proportion of WFPSMP schools indicated that their cook is trained in food storage and handling (47.8%) compared to control schools (17.4%); (p=0.028). Similarly, a significantly low proportion of WFPSMP schools indicated that their cook is trained in food storage and handling (17.4%) compared to HGSMP schools (60.9%); (p=0.003). 39.Other characteristics were not significantly different between the three study arms. 15 Table 5 - Characteristics of the schools covered by the Study Control WFPSMP p HGSMP WFPSMP p Characteristics (n=23) (n=23) value (n=23) (n=23) value Teacher Attendance Rates 92.0% 92.9% 0.656 94.1% 94.0% 0.955 Pupil Attendance Rates 83.8% 81.3% 0.534 84.7% 86.6% 0.488 Proportion of Pupils who completed the last grade of school 77.7% 81.9% 0.513 78.1% 77.4% 0.919 Schools with a storage facility 43.5% 82.6% 0.006 65.2% 82.6% 0.179 Sufficient kitchen for preparing pupils food Yes, and large enough to prepare food 52.9% 43.5% 0.385 82.4% 30.4% 0.004 Yes, but not large enough for food preparation 23.5% 43.5% 17.6% 56.5% No Status of the Kitchen 23.5% 13.0% 0.0% 13.0% Can’t repair 7.7% 10.0% 0.993 29.4% 15.0% 0.052 Good Condition 23.1% 25.0% 35.3% 20.0% Slight Repair 38.5% 35.0% 35.3% 30.0% Serious repair Has enough fuel-efficient stoves 30.8% 30.0% 0.0% 35.0% Yes, and sufficient quantity 46.2% 40.0% 0.059 70.6% 40.0% 0.156 Yes, but not sufficient quantity 15.4% 50.0% 23.5% 40.0% No School main water supply 38.5% 10.0% 5.9% 20.0% No water source 0.0% 4.3% 0.0% 8.7% Water tank 35.3% 8.7% 11.8% 8.7% Water piped into school 11.8% 13.0% 23.5% 8.7% Water brought by pupils 0.0% 4.3% 0.0% 8.7% Public taps/stand pipes 0.0% 4.3% 11.8% 4.3% Tube well/borehole 23.5% 8.7% 23.5% 4.3% Rainwater collection 5.9% 4.3% 5.9% 0.0% Unprotected spring 0.0% 8.7% 11.8% 4.3% Cart with small tank 0.0% 4.3% 0.0% 8.7% Tanker truck 0.0% 21.7% 0.0% 8.7% Surface water 11.8% 4.3% 0.0% 13.0% Children carry water to school 11.8% 8.7% 11.8% 4.3% Other 0.0% 4.3% 0.0% 17.4% School Sanitation Facilities Has Latrines 100.0% 100.0% 100.0% 100.0% Has separate facilities for boys and girls 94.1% 87.0% 0.624 100.0% 87.0% 0.248 Has hand washing facilities that are used by pupils 35.3% 47.8% 0.428 58.8% 52.2% 0.676 Proportion of pupils who wash their hands Most pupils wash their hands 33.3% 81.8% 0.046 40.0% 83.3% 0.030 Only some pupils wash their hands 66.7% 18.2% 60.0% 8.3% There is no water to wash hands close to the latrine 0.0% 0.0% 0.0% 8.3% The School has a library School Cook 5.9% 26.1% 0.096 30.0% 47.1% 0.283 The school has a cook 100.0% 100.0% 100.0% 100.0% The cook is trained in food preparation 43.5% 26.1% 0.216 34.8% 21.7% 0.326 The cook is trained in food storage and handling 47.8% 17.4% 0.028 60.9% 17.4% 0.003 The Cook has a health certificate 30.4% 47.8% 0.227 60.9% 43.5% 0.238 Control WFPSMP p HGSMP WFPSMP p Characteristics (n=23) (n=23) value (n=23) (n=23) value Teacher Attendance Rates 92.0% 92.9% 0.656 94.1% 94.0% 0.955 Pupil Attendance Rates 83.8% 81.3% 0.534 84.7% 86.6% 0.488 Proportion of Pupils who completed the last grade of school 77.7% 81.9% 0.513 78.1% 77.4% 0.919 16 Control WFPSMP p HGSMP WFPSMP p Characteristics (n=23) (n=23) value (n=23) (n=23) value Schools with a storage facility 43.5% 82.6% 0.006 65.2% 82.6% 0.179 Sufficient kitchen for preparing pupils food Yes, and large enough to prepare food 52.9% 43.5% 0.385 82.4% 30.4% 0.004 Yes, but not large enough for food preparation 23.5% 43.5% 17.6% 56.5% No Status of the Kitchen 23.5% 13.0% 0.0% 13.0% Can’t repair 7.7% 10.0% 0.993 29.4% 15.0% 0.052 Good Condition 23.1% 25.0% 35.3% 20.0% Slight Repair 38.5% 35.0% 35.3% 30.0% Serious repair Has enough fuel-efficient stoves 30.8% 30.0% 0.0% 35.0% Yes, and sufficient quantity 46.2% 40.0% 0.059 70.6% 40.0% 0.156 Yes, but not sufficient quantity 15.4% 50.0% 23.5% 40.0% No 38.5% 10.0% 5.9% 20.0% School main water supply No water source 0.0% 4.3% 0.0% 8.7% Water tank 35.3% 8.7% 11.8% 8.7% Water piped into school 11.8% 13.0% 23.5% 8.7% Water brought by pupils 0.0% 4.3% 0.0% 8.7% Public taps/stand pipes 0.0% 4.3% 11.8% 4.3% Tube well/borehole 23.5% 8.7% 23.5% 4.3% Rainwater collection 5.9% 4.3% 5.9% 0.0% Unprotected spring 0.0% 8.7% 11.8% 4.3% Cart with small tank 0.0% 4.3% 0.0% 8.7% Tanker truck 0.0% 21.7% 0.0% 8.7% Surface water 11.8% 4.3% 0.0% 13.0% Children carry water to school 11.8% 8.7% 11.8% 4.3% Other 0.0% 4.3% 0.0% 17.4% School Sanitation Facilities Has Latrines 100.0% 100.0% 100.0% 100.0% Has separate facilities for boys and girls 94.1% 87.0% 0.624 100.0% 87.0% 0.248 Has hand washing facilities that are used by pupils 35.3% 47.8% 0.428 58.8% 52.2% 0.676 Proportion of pupils who wash their hands Most pupils wash their hands 33.3% 81.8% 0.046 40.0% 83.3% 0.030 Only some pupils wash their hands 66.7% 18.2% 60.0% 8.3% There is no water to wash hands close to the latrine 0.0% 0.0% 0.0% 8.3% The School has a library School Cook 5.9% 26.1% 0.096 30.0% 47.1% 0.283 The school has a cook 100.0% 100.0% 100.0% 100.0% The cook is trained in food preparation 43.5% 26.1% 0.216 34.8% 21.7% 0.326 The cook is trained in food storage and handling 47.8% 17.4% 0.028 60.9% 17.4% 0.003 The Cook has a health certificate 30.4% 47.8% 0.227 60.9% 43.5% 0.238 40.A detailed comparison of background and other characteristics of parents and children is presented in Annex 8. 3.3. Learning Outcomes 41. This section discusses the findings with respect to indicators of literacy and numeracy of school age children (7-13 years): improved attentiveness; and improved student attendance. It presents the findings organized under specific objectives and outcomes 17 in the MGD Performance Monitoring Plan (PMP). MGD SO 1: Improved literacy of school age children 42.Three specific performance indicators were agreed on for the monitoring of these learning outcomes. Indicator 1: Proportion of 7-13 year olds that can solve Class 2 numeracy and literacy problems 43.As noted in the methodology, the data analyzed in this section was collected through the incorporation and use of the UWEZO test booklet in the survey tool for pupils. As part of the survey 5130 Pupils (2572 girls, 2558 boys) from class 3-8 were tested for literacy in Kiswahili and English and for numeracy using this tool. 44.The results followed the UWEZO parameters for numeracy acquisition which breaks down the level of language acquisition into five categories distinguishing between: children who are not able to read anything (marked as ‘nothing’), those who can distinguish letters (‘letter’); those who can read words (‘word’); those who can read sentences (‘sentence’) and those who can read a story (‘story’). All these capabilities are measured at a level that corresponds to Class 2. This means learning outcomes are assessed among children aged six to 16 years through tests set at what one would expect to have achieved in terms of literacy (and numeracy) at Standard (Grade/Class) 2 level in Kenya. The assumption behind this is that children need to acquire the basic skills in literacy and numeracy by the end of grade 2 to be able to acquire higherskills in the later grades. This is the standard at which many of learning assessments, including those in Kenya, peg their learning outcomes interventions. Literacy (English) 45.The results show that across the full set of children, and regardless of the types of school, half of the children (52.5%) were able to handle the highest English literacy Summary of main findings • Children in the control arm of the study (55.6%) performed significantly better on the highest category of the English literacy level (story) compared to the WFPSMP arm (40.6%). • Children in the HGSMP arm of the study (64.6%) performed significantly better on the highest category of the English literacy level (story) compared to the WFPSMP arm (45.0%). • The proportion of children with highest Kiswahili literacy level (story) was significantly higher in the control arm (66.0%) than WFPSMP (51.2%). • The proportion of children with highest Kiswahili literacy level (story) was significantly higher in HGSMP arm (74.9%) than WFPSMP (53.5%). • The proportion of children with highest numeracy level (division) was significantly higher in control arm (73.5%) than WFPSMP (60.9%). • The proportion of children with highest numeracy level (division) was significantly higher in HGSMP arm (77.7%) than WFPSMP (60.1%). 18 80% 70% 60% 50% 40% 30% 20% 10% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% level (story) of a class 2 child. The proportion was not significantly different between boys (51.2%) and girls (53.7%); (p=0.073) 19 (Figure 5). Figure 5 – Percentage of boys and girls across English literacy categories of achievement for all types of schools using the UWEZO methodology (n=5130) 0% Boys (n=2558) Girls (n=2572) Total (n=5130) Nothing 3,6% 4,4% 4,0% Letter 10,9% 10,8% 10,8% Word 16,9% 15,4% 16,1% Paragraph 17,5% 15,7% 16,6% Story 51,2% 53,7% 52,5% 46.The percentage of children who could read at Class 2 level increased with the class of the child. The trend was consistent among boys and girls. However, while at class 3, the girls have a slightly higher level than boys, the latter catch up with girls at class 4 and 5. At class 6 and 7 girls perform better than boys, but the boys catch up with them again at class 8 (Figure 6). Figure 6 - Percentage of children who attained the highest level English score in the UWEZO test (story), regardless of type of school, by class and by gender Class 3 Class 4 Class 5 Class 6 Class 7 Class 8 Boys (n=2558) 15,1% 30,9% 46,0% 60,9% 77,1% 88,3% Girls (n=2572) 20,7% 31,5% 44,6% 65,8% 82,9% 86,6% Total (n=5130) 17,9% 31,2% 45,3% 63,3% 80,0% 87,4% 19 P is a probability value that helps to determine the significance of the results. A small p value (typically < 0.05) indicates strong evidence of difference (significant difference) between the compared parameter estimates. A large p value (typically > 0.05) indicates weak evidence of difference (no significant difference) between parameter estimates; p value very close to the cut off (0.05), is considered to infer marginal difference. 19 80% 70% 60% 50% 40% 30% 20% 10% 0% 47.A further analysis examined the difference across target arms. Comparing pupils from the control group of school with the WFPSMP schools, Figure 7 and Table 6 show that, the proportion of children with highest English literacy level (story) was significantly higher in the control arm (55.6%) than in the WFPSMP arm (40.6%), (p=0.003). 48.Further analysis by gender revealed varying results. The proportion of children with highest English literacy level(story) among boys was significantly higher in the control arm (54.0%) than in the WFPSMP arm (42.8%), (p=0.019). However, among girls, the difference was not statistically significant (p=0.327). Figure 7 – English literacy scores by category of achievement, comparing control schools with WFPSMP schools, by gender CONTROL WFPSMP CONTROL WFPSMP CONTROL WFPSMP (n=675) (n=579) (n=721) (n=565) (n=1396) (n=1144) Boys(n=1254) Girls(n=1286) Total (n=2540) Nothing 3,4% 5,7% 3,1% 8,3% 3,2% 7,0% Letter 11% 12,1% 9,3% 17,3% 10,1% 14,7% Word 15% 20,4% 15,0% 18,6% 15,0% 19,5% Paragraph 17% 19,0% 15,5% 17,5% 16,0% 18,3% Story 54% 42,8% 57,1% 38,2% 55,6% 40,6% 49.A further comparison was done between the HGSMP schools and the WFPSMP schools. This analysis showed that a significantly higher proportion of pupils from the HGSMP arm obtained the highest English literacy level (story) (64.6%), compared to the WFPSMP pupils (45.0%), (p=0.001) (Figure 8 and Table 6). 20 80% 70% 60% 50% 40% 30% 20% 10% 0% HGSMP Figure 8 - English literacy scores by category of achievement, comparing WFPSMP schools with HGSMP schools, by gender Nothing 5,4% 2,4% 8,5% 1,9% 6,9% 2,1% Letter 10,3% 9,5% 14,4% 6,2% 12,3% 7,8% Word 18,7% 13,9% 14,4% 11,2% 16,7% 12,5% Paragraph 19,1% 13,2% 19,4% 12,8% 19,2% 13,0% Story 46,5% 61,0% 43,3% 68,0% 45,0% 64,6% 50.The proportion of girls with highest English literacy level (story) was significantly higher in HGSMP (68.0%) than in WFPSMP (43.3%), (p<0.001). However, the comparison between boys in both arms was not statistically significant (p=0.131). Table 6 - Highest English Literacy Level of the Child Variables Boys (n=1254) Girls (n=1286) Total (N=2540) AOR 95% CI Lower Upper p value AOR 95% CI Lower Upper p value AOR 95% CI Lower Upper p valu Group WFPSMP CONTROL 0.60 1.00 0.39 0.92 0.019 0.80 1.00 0.52 1.24 0.327 0.63 1.00 0.47 0.86 0.003 Propensity score quintiles First 0.80 0.46 1.39 0.428 1.90 1.08 3.32 0.026 1.11 0.75 1.64 0.602 Second Third Fourth Fifth 1.03 0.91 0.95 1.00 0.60 0.58 0.67 1.76 1.41 1.36 0.927 0.661 0.794 1.89 2.12 0.94 1.00 1.08 1.33 0.66 3.29 3.39 1.36 0.025 0.002 0.751 1.38 1.18 1.04 1.00 0.94 0.85 0.81 2.03 1.62 1.34 0.099 0.32 0.765 Variables Boys (n=1306) Girls (n=1284) Total (N=2590) AOR 95% CI p Lower Upper value AOR 95% CI p Lower Upper value AOR 95% CI p Lower Upper value Group HGSMP WFPSMP 1.38 1.00 0.91 2.09 0.131 2.44 1.00 1.59 3.75 <0.001 1.64 1.00 1.23 2.19 0.001 Propensity score quintiles First 0.62 0.36 1.07 0.084 0.82 0.47 1.43 0.474 0.60 0.41 0.88 0.009 Second 0.63 0.38 1.07 0.085 0.66 0.39 1.11 0.113 0.54 0.38 0.77 0.001 Third 0.73 0.50 1.07 0.105 0.63 0.43 0.93 0.021 0.65 0.50 0.86 0.002 Fourth 0.79 0.56 1.13 0.200 0.88 0.60 1.28 0.489 0.77 0.59 0.99 0.042 Fifth 1.00 1.00 1.00 AOR – Adjusted Odds Ratio 21 80% 70% 60% 50% 40% 30% 20% 10% Literacy (Kiswahili) 51. The results on literacy in Kiswahili indicate that 62.2% of children in class 3-8 were able to handle the highest Kiswahili literacy level (story) of a Class 2 child (Figure 9) showing therefore an acquisition of Kiswahili that was superior to that of the English literacy (see Section above). The proportion was significantly higher in girls (63.6%) than boys (60.9%); (p=0.046). 52.As was the case for the English results (reported above) the proportion of children by class achieving the highest Kiswahili literacy level (story) increased as the class level of the child went up. The trend was consistent in boys and girls. Figure 9 – Percentage of boys and girls across Kiswahili literacy categories of achievement for all types of schools using the UWEZO methodology (n=5130) 0% Boys (n=2558) Girls (n=2572) Total (n=5130) Nothing 4,9% 5,2% 5,1% Letter 6,2% 6,3% 6,3% Word 10,9% 10,8% 10,9% Paragraph 17,1% 14,0% 15,5% Story 60,9% 63,6% 62,2% 53. However, comparing across boys and girls, at class 3, the girls start at a slightly higher level than boys. Boys later catch up with girls at class 4 and 5. At class 6 and 7 girls perform better than boys, but the boys catch up with them again at class 8 (Figure 10). 22 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Figure 10 - Percentage of children who attained the highest level Kiswahili score in the UWEZO test (story), regardless of type of school, by class and by gender Class 3 Class 4 Class 5 Class 6 Class 7 Class 8 Boys (n=2558) 22,1% 45,8% 60,2% 72,3% 82,6% 90,6% Girls (n=2572) 29,4% 43,5% 60,3% 78,6% 88,6% 89,5% Total (n=5130) 25,9% 44,7% 60,2% 75,4% 85,6% 90,0% 54.As was the case for the English results, further analysis, comparing the differentarms was done for the Kiswahili results. Adjusting for propensity score, the proportion of children with highest Kiswahili literacy level (story) was significantly higher in control arm (66.0%) than WFPSMP (51.2%); (p=0.034), and significantly higher in HGSMP arm (74.9%) than WFPSMP (53.5%); (p<0.001) (Figure 11 and Table 7). Like the English results, the stratification by gender again revealed contrary results. Disaggregated by gender the proportion of children with highest Kiswahili literacy level (story) was not significantly different between control and WFPSMP arms, in both boys (p=0.183) and girls (p=0.236). Figure 11 – Kiswahili results by category of achievement, comparing control schools with WFPSMP schools, by gender 80% 70% 60% 50% 40% 30% 20% 10% 0% Nothing Letter Word Story Boys (n=1254) 5,5% 4,7% 5,5% 4,3% 8,7% 5,5% 4,5% 9,4% 6,4% 8,8% 23 80% 70% 60% 50% 40% 30% 20% 10% 0% 55. However, comparing the HGSMP arm, with the WFPSMP arm, it was found that the proportion of children with the highest Kiswahili literacy level (story) was significantly higher in HGSMP arm (78.1%) than WFPSMP (53.2%); (p<0.001). This was similar among the boys, where the proportion of children with highest Kiswahili literacy level (story) was also significantly higher in HGSMP arm (71.5%) than WFPSMP (53.8%); (p=0.055) (Figure 12). Figure 12 - Kiswahili results by category of achievement, comparing WFPSMP schools with HGSMP schools, by gender WFPSMP HGSMP WFPSMP HGSMP WFPSMP HGSMP (n=593) (n=713) (n=541) (n=743) (n=1134) (n=1456) Boys(n=1306) Girls(n=1284) Total (n=2590) Nothing 5,4% 4,6% 8,5% 2,8% 6,9% 3,7% Letter 7,6% 4,2% 7,9% 3,1% 7,8% 3,6% Word 12,3% 8,0% 12,9% 6,1% 12,6% 7,0% Paragraph 20,9% 11,6% 17,4% 10,0% 19,2% 10,8% Story 53,8% 71,5% 53,2% 78,1% 53,5% 74,9% Table 7 - Highest Kiswahili Literacy Level (story) of the Child Variables Boys (n=1254) Girls (n=1286) Total (N=2540) AOR 95% CI Lower Upper p value AOR 95% CI Lower Upper p value AOR 95% CI Lower Upper p value Group WFPSMP 0.75 0.49 1.15 0.183 0.76 0.49 1.19 0.236 0.72 0.53 0.98 0.034 CONTROL 1.00 1.00 1.00 Propensity score quintiles First 1.09 0.63 1.89 0.772 1.75 0.99 3.10 0.055 1.36 0.92 2.02 0.127 Second 1.46 0.84 2.52 0.177 1.74 0.99 3.05 0.056 1.51 1.03 2.23 0.037 Third 1.10 0.70 1.71 0.684 1.87 1.17 2.99 0.009 1.32 0.96 1.82 0.088 Fourth 0.94 0.66 1.33 0.716 0.97 0.68 1.38 0.865 0.99 0.78 1.27 0.958 Fifth 1.00 1.00 1.00 Variables Boys (n=1306) Girls (n=1284) Total (N=2590) 95% CI p AOR Lower Upper value 95% CI p AOR Lower Upper value 95% CI p AOR Lower Upper value Group HGSMP 1.53 0.99 2.35 0.055 2.77 1.76 4.37 <0.001 1.74 1.28 2.35 <0.001 WFPSMP 1.00 1.00 1.00 Propensity score quintiles 24 80% First 0.49 0.28 0.86 0.013 0.65 0.36 1.19 0.164 0.45 0.30 0.68 <0.001 Second 0.58 0.33 1.00 0.049 0.87 0.49 1.55 0.640 0.57 0.39 0.83 0.004 Third 0.72 0.48 1.09 0.117 0.71 0.46 1.10 0.123 0.66 0.49 0.89 0.007 Fourth 0.68 0.46 1.00 0.050 0.81 0.53 1.24 0.322 0.74 0.56 0.99 0.042 Fifth 1.00 1.00 1.00 AOR – Adjusted Odds Ratio Numeracy 56. Similar analyses were done on the numeracy portion of the UWEZO test. The numeracy test includes eight levels of acquisition which are ordered from ‘nothing’ to ‘division’ with the latter reflecting the highest level of acquisition. 57. The results show that over two-thirds of children in Class 3-8 (69.7%) can solve the highest numeracy level tasks (division) of a Class 2 child. The proportion was consistent between boys (69.9%) and girls (69.4%) (Figure 13). Figure 13 – Percentage of boys and girls across categories of numeracy achievement for all types of schools (n=5130) 70% 60% 50% 40% 30% 20% 10% 0% Boys (n=2558) Girls (n=2572) Total (n=5130) Nothing 2,0% 1,7% 1,9% Counting and matching 1,9% 2,8% 2,4% Numerical recognition 10U99 1,7% 2,0% 1,9% Which one is greater 0,2% 0,3% 0,3% Addition 5,8% 4,5% 5,1% Subtraction 8,8% 9,8% 9,3% Multiplication 9,7% 9,4% 9,6% Division 69,9% 69,4% 69,7% 58.There was an increasing trend in highest numeracy level (division) of a Class 2 child with increase in class of the child. The trend was consistent among boys and girls. However, at class 3, the boys start at a slightly higher level than girls, the latter catch up with boys at class 4 and above (Figure 14). 25 80% 70% 60% 50% 40% 30% 20% 10% Figure 14 - Percentage of children who attained the highest level Numeracy score in the UWEZO test (division), regardless of type of school, by class and by gender Class 3 Class 4 Class 5 Class 6 Class 7 Class 8 Boys (n=2558) 38,8% 55,1% 72,2% 80,7% 86,4% 92,5% Girls (n=2572) 34,0% 55,6% 72,4% 81,3% 87,0% 93,9% Total (n=5130) 36,3% 55,3% 72,3% 81,0% 86,7% 93,3% 59.The results after adjusting for differences across the three arms of study show that, the proportion of children with highest numeracy level (division) was significantly higher in control arm (73.5%)than WFPSMP (60.9%); (p=0.034)(Figure 15 and Table 8). Figure 15 - Numeracy results by category of achievement, comparing control schools with WFPSMP schools, by gender 0% CONTROL (n=675) WFPS Boys (n=1254) MP (n=579) CONTROL (n=721) WFPS Girls (n=1286) MP (n=565) CONTROL (n=1396) WFPS Total (n=2540) MP(n=1 Nothing 1,3% 2,8% 0,8% 3,4% 1,1% 3,1% Counting and matching 1,8% 2,4% 1,8% 5,3% 1,8% 3,8% Numerical recognition 10U99 0,9% 3,5% 1,5% 2,5% 1,2% 3,0% Which one is greater 0,0% 0,3% 0,3% 0,2% 0,1% 0,3% Addition 5,2% 7,9% 3,7% 7,6% 4,4% 7,8% Subtraction 8,6% 10,9% 8,9% 10,8% 8,7% 10,8% Multiplication 9,6% 10,4% 8,6% 10,3% 9,1% 10,3% Division 72,6% 61,8% 74,3% 60,0% 73,5% 60,9% 60.Similarly, the proportion of children with the highest numeracy level (division) was significantly different between the HGSMP (77.7%) and WFPSMP (60.1%); (p=0.001), 26 (n (n n= 3 (Figure 13 and Table 8). Figure 16 - Numeracy results by category of achievement, comparing WFPSMP schools with HGSMP School, by gender 0% WFPSMP =59 Bo 3) HGS ys (n=1306) MP (n= 1 ) WFPSMP =54 Gi 1) HGS rls(n=1284) MP (n= 4 ) WFPSMP ( 11 To 4) HGS tal(n=2590) MP(n=145 Nothing 4,0% 1,4% 3,1% 0,9% 3,6% 1,2% Counting and matching 3,2% 1,5% 6,8% 1,2% 4,9% 1,4% Numerical recognition 10U99 1,3% 1,3% 2,8% 1,3% 2,0% 1,3% Which one is greater 0,2% 0,3% 0,6% 0,0% 0,4% 0,1% Addition 7,4% 3,6% 7,0% 2,7% 7,2% 3,2% Subtraction 9,9% 6,5% 11,3% 8,9% 10,6% 7,7% Multiplication 11,3% 6,6% 10,9% 8,3% 11,1% 7,5% Division 62,6% 78,8% 57,5% 76,6% 60,1% 77,7% 61. Stratification by gender revealed contrary results. The proportion of children with highest numeracy level (division) was not significantly different between control and WFPSMP arms, in both boys (p=0.103) and girls (p=0.205). However, this was not the case for HGSMP and WFPSMP where stratification by gender revealed consistent results. Among the girls, the proportion of children with highest numeracy level (division) was significantly higher in HGSMP arm (76.6%) than WFPSMP (57.5%), (p=0.007). Among the boys, the proportion of children with highest numeracy level (division) was not significantly higher in HGSMP arm (78.8%) than WFPSMP (68.8%) (p=0.073). 80% 70% 60% 50% 40% 30% 20% 10% 27 Table 8 - Highest Numeracy Level (division) of the Child Variable s Boys (n=1254) Girls (n=1286) Total (N=2540) AO R 95% CI Lowe Uppe r r p value AO R 95% CI Lowe Upp r er p valu e AO R 95% CI Lowe Uppe r r p value Group WFPSMP 0.69 0.44 1.08 0.103 0.74 0.47 1.18 0.205 0.71 0.51 0.97 0.034 CONTRO L 1.00 1.00 1.00 Propensity score quintiles First 0.97 0.54 1.73 0.906 1.56 0.86 2.82 0.144 1.24 0.82 1.88 0.309 Second 1.23 0.69 2.18 0.491 1.57 0.87 2.83 0.131 1.33 0.89 2.00 0.17 Third 1.07 0.68 1.70 0.763 1.32 0.82 2.13 0.259 1.19 0.86 1.66 0.301 Fourth 0.87 0.61 1.25 0.449 0.95 0.67 1.36 0.789 0.91 0.71 1.17 0.454 Fifth 1.00 1.00 1.00 Variable s Boys (n=1306) Girls (n=1284) Total (N=2590) AO R 95% CI Lowe Uppe r r p value AO R 95% CI Lowe Upp r er p valu e AO R 95% CI Lowe Uppe r r p value Group 0.00 HGSMP 1.52 0.96 2.40 0.073 1.85 1.18 2.89 7 1.72 1.26 2.35 0.001 WFPSMP L 1.00 1.00 1.00 Propensity score quintiles First 0.55 0.30 1.01 0.052 0.65 0.36 1.16 0.143 0.65 0.43 0.98 0.038 Second 0.61 0.34 1.09 0.096 0.73 0.42 1.27 0.27 0.68 0.46 1.01 0.053 Third 0.72 0.46 1.12 0.142 0.77 0.51 1.18 0.226 0.82 0.60 1.11 0.198 Fourth 0.95 0.62 1.46 0.804 0.99 0.65 1.50 0.957 0.96 0.71 1.29 0.764 Fifth 1.00 1.00 1.00 AOR – Adjusted Odds Ratio Indicator 2: Number of individuals benefiting directly from USDA-funded interventions 62.No activities, USDA - WFP interventions, had been planned prior to the time of the baseline. The value for this indicator is therefore zero at baseline. Indicator 3: Number of individuals benefiting indirectly from USDA-funded interventions 63.No activities, USDA - WFP interventions, had been planned prior to the time of the baseline. The value for this indicator is therefore zero at baseline. 64.The direct beneficiaries of the USDA interventions are primarily the girls and boys 28 receiving school meals and attending schools supported by WFP. 65.The indirect beneficiaries are primarily the household members who do not receive school meals themselves (parents, other adults, and out-of-school siblings), but who benefit, as the food security of the entire household is relieved if the school-going children receive healthy school meals. 66.Data for this indicator would be realized through interviews with parents to determine the average number of children per household going to schools supported by WFP. Since the average household size in target areas is known. Indirect beneficiaries = Number of Households (HH) * (HH size- average number of children per HH going to school). MGD 1.2: Improved Attentiveness Indicator 4: Percent of students in classrooms identified as inattentive by their teachers 67. The data on this indicator was collected using the parent-child tool. A relatively high proportion of the children (43.0%) indicated that they sometimes find it difficult to concentrate in class. The findings indicate that the proportion of children who sometimes find it difficult to concentrate in class was comparable between boys (43.8%) and girls (42.2%) (Figure 17). Figure 17 - Percentage of boys and girls for all types of schools who report “sometimes” finding it difficult to concentrate in class (n=5130) 68.Adjusting for propensity score, the proportion of children who indicated that sometimes they find it difficult to concentrate in class was significantly higher in 50% 45% 40% 35% 30% 25% 20% 15% 10% 5% 0% 29 50% 45% 40% 35% 30% 25% 20% 15% 10% 5% 0% 50% 45% 40% 35% 30% 25% 20% 15% 10% 5% 0% control arm (46.4%) than WFPSMP (41.1%) (p=0.016) (Figure 18 and Table 9). Figure 18 - Percentage of children who sometimes find it difficult to concentrate in class comparing control and WFPSMP schools, by gender CONTROL WFPSMP (n=675) (n=579) Boys (n=1254) CONTROL WFPSMP (n=721) (n=565) Girls (n=1286) CONTROL WFPSMP (n=1396) (n=1144) Total (n=2540) 69.On the other hand, adjusting for difference between HGSMP and WFPSMP shows the proportion to be significantly higher in HGSMP arm (43.5%) than WFPSMP (37.4%), (p=0.028) (Figure 19 and Table 9). Stratification by gender reveals consistent results in both cases. Figure 19 – Percentage of children who sometimes find it difficult to concentrate in class comparing WFPSMP schools with HGSMP School, by gender WFPSMP HGSMP (n=593) (n=713) Boys (n=1306) WFPSMP HGSMP (n=541) (n=743) Girls (n=1284) WFPSMP HGSMP (n=1134) (n=1456) Total (n=2590) 30 Table 9 - Child sometimes finds it difficult to concentrate in class Variabl es Boys (n=1254) Girls (n=1286) Total (N=2540) AO R 95% CI Lowe Uppe r r p value AO R 95% CI Lowe Uppe r r p value AO R 95% CI Lowe Uppe r r p value Group WFPSM P 0.70 0.46 1.08 0.108 0.67 0.43 1.03 0.068 0.69 0.51 0.93 0.016 CONTR OL 1.00 1.00 1.00 Propensity score quintiles First 0.71 0.41 1.24 0.227 0.98 0.56 1.73 0.954 0.79 0.53 1.16 0.228 Second 0.79 0.46 1.35 0.384 1.01 0.58 1.77 0.961 0.89 0.61 1.31 0.568 Third 0.79 0.50 1.23 0.292 1.14 0.71 1.83 0.581 0.91 0.66 1.26 0.581 Fourth 0.93 0.66 1.33 0.699 1.43 1.00 2.05 0.049 1.06 0.82 1.36 0.673 Fifth 1.00 1.00 1.00 Variabl es Boys (n=1306) Girls (n=1284) Total (N=2590) AO R 95% CI Lowe Uppe r r p value AO R 95% CI Lowe Uppe r r p value AO R 95% CI Lowe Uppe r r p value Group HGSMP 1.57 1.03 2.39 0.038 1.19 0.78 1.82 0.420 1.39 1.04 1.86 0.028 WFPSM PL 1.00 1.00 1.00 Propensity score quintiles First 1.10 0.64 1.90 0.722 1.04 0.60 1.81 0.881 1.05 0.72 1.54 0.798 Second 1.25 0.74 2.11 0.402 1.32 0.80 2.20 0.278 1.33 0.93 1.89 0.117 Third 1.33 0.92 1.92 0.136 1.35 0.94 1.96 0.108 1.41 1.08 1.83 0.01 Fourth 0.93 0.65 1.31 0.662 1.32 0.93 1.88 0.124 1.05 0.82 1.34 0.716 Fifth 1.00 1.00 1.00 AOR – Adjusted Odds Ratio 70.The findings of the key reasons why children at times find it difficult to concentrate in class (across all types of school) are shown in the table below. This shows that “I am hungry” ranked as the most prevalent explanation (62 percent), followed by “feeling sick” (36 percent) (Figure 20). 31 Figure 20 - Reasons why children "sometimes" find it difficult to concentrate in class 71. Results from the focus group discussions in the survey schools with parents and pupils provided additional insights on how hunger and the absence of school meals impacts on the capacity for concentration and the willingness of children to stay in school (Box 1). Box 1 – PTA, BoM and Student’s Government responses across the three target arms of the study on the impact of hunger/absence of school meals on attention and willingness to go to school 0% Impact of hunger/absence of school meals on attention and willingness to go to school: • Absenteeism rises as many pupils stay at home. • Pupils are released from school earlier than usual. • Poor concentration in class- pupils are inattentive in class. • Pupils transfer or drop out of school. • Pupils perform poorly in school – they fail examinations. • There is an increase in indiscipline among pupils including sneaking out of school to look for food. • Pupils fall sick or fake sickness to leave school. • Pupils go to the nearest market to beg or resort to stealing. • Pupils spend most of their time discussing about food and do not cooperate when asked to do school work. 32 3.4. Short-term Hunger 72. This section covers the situation with respect to food consumption by children during the day and week. It also looks at results for the Food Consumption Scores of households covered by the survey and associated coping mechanisms. The data was collected through the parent/child tool with the parents as respondents. MGD 1.2.1 Reduced Short-Term Hunger Indicator 5: Number of daily school meals (breakfast, snack, lunch) provided to school￾age children because of USDA assistance 73. No activities, USDA-WFP interventions, had been planned prior to the time of the baseline. This indicator will therefore be measured at mid- and end-line and is zero at baseline. Indicator 6: Number of school-aged children receiving daily school meals (breakfast, snack, lunch) because of USDA assistance 74. No activities, USDA-WFP interventions, had been planned prior to the time of the baseline. This indicator will therefore be measured at mid- and end-line and is zero at baseline. Indicator 7: Percent of students in target schools who regularly consume a meal before the school day 75. Data for this indicator was collected through the parent/pupil tool. A total of 5130 parents and 5130 pupils were interviewed. 76. More than one-third of the parents/guardians (38.7%) indicated their children ate food daily (in the last 1 Week) before going to school. More than one-half of the parents/guardians (59.2%) indicated their children ate food daily (in the last 1 Week) after coming from school. 77. Adjusted for propensity score, the results shown in Figure 21 indicate that the proportion of parents/guardians who indicated their children ate food daily (in the last 1 Week) before going to school was significantly higher among those in control Summary of main findings • Just over one-third of the parents/guardians (38.7%) indicated their children ate food daily (in the last 1 Week) before going to school. • The proportion of parents/guardians who indicated their children ate food daily (in the last 1 Week) before going to school was significantly higher among those in control schools (38.0%) than WFPSMP schools (33.0%). • The proportion of parents/guardians who indicated their children ate food daily (in the last 1 Week) before going to school was slightly higher among those in HGSMP (43.2%) than WFPSMP (38.7%). 33 80% 70% 60% 50% 40% 30% 20% 10% 0% 80% 70% 60% 50% 40% 30% 20% 10% 0% schools (38.0%) than WFPSMP schools (33.0%); (p=0.009). Figure 21 - Percentage of parents/guardians who reported their children ate daily before going to school, comparing control and WFPSMP schools, by gender CONTROL WFPSMP (n=675) (n=579) Boys (n=1254) CONTROL WFPSMP (n=721) (n=565) Girls (n=1286) CONTROL WFPSMP (n=1396) (n=1144) Total (n=2540) 78.The percentage of parents/guardians whose children ate food daily (in the last 1 Week) after coming from school (Figure 22) was significantly higher among those in control (70.6%) than WFPSMP (45.5%); (p<0.001). The results in both cases were consistent by gender. Figure 22 - Percentage of parents/guardians who reported their children ate daily after going to school, comparing control and WFPSMP schools, by gender CONTROL WFPSMP (n=675) (n=579) Boys (n=1254) CONTROL WFPSMP (n=721) (n=565) Girls (n=1286) CONTROL WFPSMP (n=1396) (n=1144) Total (n=2540) 79. An analysis of the difference between WFPSMP and HGSMP schools (Figure 23) showed a similar trend. The proportion of parents/guardians whose children ate food daily (in the last 1 Week) before going to school was slightly higher among children in 34 0% 0% the HGSMP (43.2%) than in the WFPSMP (38.7%); (p=0.021). Figure 23 - Percentage of parents/guardians who reported their children ate daily before going to school, comparing WFPSMP schools and HGSMP schools, by gender WFPSMP HGSMP (n=593) (n=713) Boys (n=1306) WFPSMP HGSMP (n=541) (n=743) Girls (n=1284) WFPSMP HGSMP (n=1134) (n=1456) Total (n=2590) 80.Similarly, the proportion of parents/guardians whose children ate food daily (in the last 1 Week) after coming from school(Figure 24) was significantly higher among those in HGSMP (71.2%) than WFPSMP (46.9%); (p<0.001). The results were consistent by gender. Figure 24 - Percentage of parents/guardians who reported their children ate daily after going to school, comparing WFPSMP schools and HGSMP schools, by gender WFPSMP HGSMP (n=593) (n=713) Boys (n=1306) WFPSMP HGSMP (n=541) (n=743) Girls (n=1284) WFPSMP HGSMP (n=1134) (n=1456) Total (n=2590) 35 81. Food Consumption Scores:20 To further anchor the preceding results in the context, an analysis of the household Food Consumption Score was undertaken. The results show that over a third of the children (39.5%) reside in households with acceptable food consumption score, 32.2% in households with borderline and 28.3% in households with poor consumption score. The proportion of households with acceptable food consumption score was significantly higher among male parents/guardians (42.9%) than female (38.2%); (p=0.002) (Figure 25). Figure 25 – Percentage of parent/guardians with acceptable food consumption score (FCS) by gender 82.Looking at these results in more detail against level of education of the parents (Figure 26), the proportion of households with acceptable food consumption scores was significantly higher among parents/guardians who attained madrasa/adult learning level of education (62.7%) and those with technical college/university level (58.9%), and lower among parents/guardians who have never attended school (36.7%) and those who did not complete primary education (34.1%); (p<0.001). 20 The Food consumption score was calculated using WFP’s guidelines as set out in: WFP VAM Unit (2008). Food consumption analysis - Calculation and use of the food consumption score in food security analysis. World Food Programme, Vulnerability Analysis and Mapping. 50% 45% 40% 35% 30% 25% 20% 15% 10% 5% 0% 36 5% 0% Figure 26 - Percentage of parents/guardians with acceptable FCS by level of education of parent/guardian 83.The results for the analysis of differences across the three study arms in the survey (Figure 27) show that the proportion of children residing in households with acceptable FCS was not significantly different among those in control (35.4%) and WFPSMP (35.2%) (p=0.916). Figure 27 - Proportion of children residing in households with acceptable FCS, comparing control and WFPSMP schools CONTROL WFPSMP (n=675) (n=579) Boys (n=1254) CONTROL WFPSMP (n=721) (n=565) Girls (n=1286) CONTROL WFPSMP (n=1396) (n=1144) Total (n=2540) 84.Similarly, the results for children residing in households with acceptable FCS was not significantly different among those in HGSMP (41.2%) and WFPSMP (42.7%); 0% Never attended school (n=2211) Madrasa/Adult learning centre (n=102) Not complete primary (n=1001) Completed primary (n=1066) Not compete secondary (238) Completed secondary (n=322) Technical college/university (n=190) Total(n=5130) 37 (p=0.443) (Figure 28). The results were consistent by gender. Figure 28 - Proportion of children residing in households with acceptable FCS, comparing WFPSMP and HGSMP schools 85.Coping Mechanisms: The survey sought to establish what the most common coping strategies were by parents/guardians on days when the family did not have enough food or money to buy food. 86.In order of importance the reported coping strategies include: purchase food on credit (80.8%), reliance on less preferred and less expensive foods (80.6%), reduce number of meals eaten in a day (67.4%), limit portion size at mealtimes (66.2%), borrow food, or rely on help from a friend or relative (64.9%), restrict consumption by adults in order for small children to eat (50.6%) or skip entire days without eating (42.3%) (Figure 29). Coincidentally, withdrawing children from school did not feature in the coping mechanisms cited. 5% 0% 42,5% WFPSMP HGSMP WFPSMP HGSMP WFPSMP HGSMP 38 Figure 29 – Reported coping strategies on days when the family did not have enough food, or money to buy food 87.The computed mean coping strategy index (CSI)21 (Figure 30) was comparable between control (33.1) and WFPSMP (32.0). Figure 30 - CSI comparing WFPSMP and control group schools, by gender 21 The Coping Strategies Index was calculated using the methodology proposed in: Maxwell, D. & R. Caldwell (2008). A tool for rapid measurement of household food security and the impact of food aid programs in humanitarian emergencies. Field Methods Manual, Second Edition. members 12,1% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% WFPSMP CONTROL WFPSMP CONTROL WFPSMP CONTROL 39 In response to a question on alternatives to school meals where the latter are not consistently available the respondents indicated the following that shed light on the food security situation in the families: 88.The CSI was also comparable between HGSMP (35.2) and WFPSMP (35.6). The results were consistent by gender. Figure 31 - CSI comparing HGSMP and WFPSMP schools, by gender HGSMP WFPSMP (n=713) (n=593) Boys (n=1306) HGSMP WFPSMP (n=743) (n=541) Girls (n=1284) HGSMP WFPSMP (n=1456) (n=1134) Total (n=2590) 89.Results from the focus group discussions in the survey schools with parents and pupils provided additional insights on the food security situation of families. Box 2 – PTA, BoM and Student’s Government responses across the three target arms of the study on food security situation in families Responses from WFPSMP schools Responses from Control schools Responses from HGSMP schools • Parents are required to pay or their children stay without lunch. • Parents are asked to bring food to school • Parents and teachers contribute money for buying food • children stay without lunch • Parents give their children cash to buy food from nearby markets • The children survive without lunch • The learners go home for lunch • Some learners carry food from home. • Learners go home for lunch • Learners carry lunch from home • Parents give children cash to buy food from nearby markets • Pupils share with others what they carried or some borrow food from nearby homesteads. • Pupils drink water 40 3.5. School meals and expected outcomes 90.This section presents the situation with respect to access to food and to school meals during the year of the study (2017) and in the week of the survey. It also reports on the situation with respect to key expected outcomes of school feeding, namely attendance, enrollment and community understanding. 91. It is important to note that the Baseline Survey was undertaken at a time when the drought was severe in the target counties. At the same time, WFP had a complete pipeline break in term one of 2017. No funding was availed for any school meals in the arid counties. While there was no direct school feeding from WFP during the survey period there was school feeding in some of the schools where this was not expected, mainly because there were interventions from Government and other actors to mitigate the effects of the drought. In addition, a small number of WFP schools were providing school feeding with carryovers from the previous phase of the SMP. MGD 1.2.1.1/1.3.1.1.Increased Access to Food (School Feeding) Indicator 8: Percent of students in target schools who regularly consume a meal 92.This section of the survey examined regularity of school meal consumption during the school year, and during the week of the survey. School meals situation in the current school year (2017) 93.Approximately half of the parents/guardians (53.7%) responded that their children have been receiving school meals at school in the current school year (2017). As might be expected, the proportions were consistent among boys (53.4%) and girls (54.0%). 94.The analysis of differences across the three arms show that the proportion of parents/guardians who indicated that the child has been receiving school meals (at Summary of main findings • Approximately half of the parents/guardians reported that their children had been receiving school meals at school in the current school year (2017). The proportions were consistent among boys and girls. • The proportion of parents/guardians who reported that the child has been receiving school meals (at school) in the current school year (2017), was significantly higher in WFPSMP (59.3%) than in Control (20.0%). The proportion was also significantly higher in HGSMP (80.4%) than in WFPSMP (55.6%). The results were consistent for boys and girls. • Two out of five of the parents/guardians reported that the school where the child was learning was serving food during the survey week. WFPSMP (51.7%) schools were more likely to be serving food in the survey week than control schools (16.3%), and HGSMP schools were more likely (51.5% than WFPSMP schools (43.9%) to be serving food, as reported by parents/guardians. The results were consistent for boys and girls. 41 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% school) in the current school year (2017), was significantly higher in WFPSMP (59.3%) than in the Control schools (20.0%) (p<0.001) (Figure 32). Figure 32 - Percentage of parents/guardians indicating that their child had received school meals in the current school year (2017), comparing Control and WFPSMP schools, by gender CONTROL WFPSMP (n=675) (n=579) Boys (n=1254) CONTROL WFPSMP (n=721) (n=565) Girls (n=1286) CONTROL WFPSMP (n=1396) (n=1144) Total (n=2540) 95.The proportion was also significantly higher in HGSMP (80.4%) than in WFPSMP (55.6%); (p<0.001) (Figure 33). The patterns were consistent among boys and girls. Figure 33 - Percentage of parents/guardians indicating that their child had received school meals in the current school year (2017), comparing WFPSMP and HGSMP schools, by gender WFPSMP HGSMP (n=593) (n=713) Boys (n=1306) WFPSMP HGSMP (n=541) (n=743) Girls (n=1284) WFPSMP HGSMP (n=1134) (n=1456) Total (n=2590) 42 60% 50% 40% 30% 20% 10% 0% School meals situation in the current week 96.More than one-third of the parents/guardians (40.6%) reported that the school where the child was learning was serving food during the survey week. The proportions were found to be consistent for boys (39.8%) and girls (41.5%) (figure 34). Figure 34 - Percentage of parents/guardians indicating that their child had received school meals in the week of the survey, comparing Control and WFPSMP schools, by gender CONTROL WFPSMP (n=675) (n=579) Boys (n=1254) CONTROL WFPSMP (n=721) (n=565) Girls (n=1286) CONTROL WFPSMP (n=1396) (n=1144) Total (n=2540) 97. The proportion of parents/guardians that reported that the school where the child was learning was serving food during the survey week, was significantly higher in WFPSMP (51.7%) than in Control (16.3%); (p<0.001). Comparing the HGSMP with WFPSMP it was found that the proportion of children having food during the week of the survey was significantly higher in HGSMP (51.5%) than in WFPSMP (43.9%) (Figure 35); (p<0.001). The patterns were consistent among boys and girls. 43 0% Figure 35 - Percentage of parents/guardians indicating that their child had received school meals in the week of the survey, comparing WFPSMP and HGSMP schools, by gender WFPSMP HGSMP (n=593) (n=713) Boys (n=1306) WFPSMP HGSMP (n=541) (n=743) Girls (n=1284) WFPSMP HGSMP (n=1134) (n=1456) Total (n=2590) 98.Qualitative data provided further insights into the importance of school meals from the perspective of the family and community. These are highlighted in Box 3 below. Box 3 - PTA, BoM and pupil responses across the three target arms of the study on the importance of school meals for families and communities Indicator 9: Total amount of commodities that have been provided as a part of USDA￾funded intervention. 99.No activities, USDA-WFP interventions, had been planned prior to the time of the baseline. This indicator is therefore zero at baseline. School meals are important in: • Supplementing food at home • Reducing the burden of carrying food from home • Curbing hunger because some pupils do not get food at home and so depend on only one meal they get in school • Relieving the burden of providing three meals on parents 44 MGD 1.3 Improved Student Attendance Indicator 10: Number of students regularly (80%) attending USDA supported classrooms/schools 100. The average number of students regularly attending22 Was not significantly higher in control schools (232 total, of which 128 boys and 104 girls) than WFPSMP schools (184 total of which 102 boys and 83 girls); (p=0.157) (Figure 36). Figure 36 - Comparison of the average number of students attending school 80% of the time in Control and WFPSMP schools, by gender Control WFPSMP (n=23) (n=23) Boys Control WFPSMP (n=23) (n=23) Girls Control WFPSMP (n=23) (n=23) Overall 101. The average number of students regularly attending (327 total, of which 185 boys and 142 girls) was significantly higher in HGSMP schools than WFPSMP schools (191 total, of which 107 boys and 83 girls); (p=0.024) (Figure 37). 22 Regular attendance was defined as attending school >80% of the time. Summary of main findings • There was no significant difference between the average number of students regularly attending WFPSMP schools compared to control schools • However, there was a significant difference between the average numbers of students regularly attending HGSMP schools compared to WFPSMP schools, with this number being higher in the HGSMP schools. 45 Figure 37 - Comparison of the average number of students attending school 80% of the time in WFPSMP and HGSMP schools, by gender WFPSMP HGSMP (n=23) (n=23) Boys WFPSMP HGSMP (n=23) (n=23) Girls WFPSMP HGSMP (n=23) (n=23) Overall MGD 1.3.4 Increased Student Enrolment Indicator 11: Number of students enrolled in schools receiving USDA assistance 102. Average enrolment in control schools (375 total, of which 207 boys and 168 girls) was not significantly higher than WFPSMP schools (280 total, of which 155 boys and 125 girls); (p=0.076) (Figure 38). Summary of main findings • There was no statistically significant difference in average enrolment in the comparison between control schools and WFPSMP schools. • Similarly, there was no statistically significant difference in average enrolment between WFPSMP and HGSMP schools. 46 Figure 38 - Average number of boys, girls, and overall students enrolled in control and WFPSMP schools Control WFPSMP (n=23) (n=23) Boys Control WFPSMP (n=23) (n=23) Girls Control WFPSMP (n=23) (n=23) Overall 103. Average enrolment in HGSMP schools (430 total, of which 243 boys and 186 girls) was not significantly higher than WFPSMP schools (290 total of which 163 boys and 127 girls); (p=0.095) (Figure 39). Figure 39 - Average number of boys, girls, and overall students enrolled in WFPSMP and HGSMP schools WFPSMP HGSMP (n=23) (n=23) Boys WFPSMP HGSMP (n=23) (n=23) Girls WFPSMP HGSMP (n=23) (n=23) Overall 47 MGD 1.3.5 Increased Community Understanding of the Benefits of Education Indicator 12: Percent of parents in target communities who can name at least three benefits of primary education 104. The parent tool was responded to by 28.2% male and 71.8% female parents. This section examines the findings on the benefits of primary education. 105. The most commonly mentioned benefits of education by the parents/guardians related to the benefit of education in improving literacy (67.5%); increasing the chance of the pupil’s future economic self-reliance (64.6%); and that education helps break the cycle of poverty (47.5%) (Figure 40). Figure 40 - Distribution of parental/guardian responses on the benefits of education (n=5130) Education helps girls remain more in school and early marriages are delayed Through girls' education, improves the general wellbeing of households (nutrition, health etc) Education improves cohesion in the community Education increases ability to learn new skills (adoption of technology) Education develops social skills 9,6% 13,0% 17,3% 21,4% 23,5% Education helps break the cycle of poverty Education increases the chances of the pupils' future economic selfUreliance Education improves literacy 47,5% 64,6% 67,5% 0% 10% 20% 30% 40% 50% 60% 70% 80% Summary of main findings • Two out of five parents/guardians in target communities could name at least three benefits of primary education, with a significantly higher proportion of male parents/guardians (47.2%) able to list three benefits than female (39.8%). • Parents/guardians in WFPSMP schools were generally more able to name benefits of primary education when compared to control and HGSMP schools. • Specifically, the proportion of parents/guardians in target communities who could name at least three benefits of primary education, was significantly higher in WFPSMP arm (57.2%) than in control (26.1%). Similarly, the proportion of parents/guardians in target communities who could name at least three benefits of primary education, was significantly higher in WFPSMP (57.3%) than in HGSMP (30.0%). 48 0% 106. The results show that 41.9% of parents/guardians could name at least three benefits of primary education, with a significantly higher proportion of male parents/guardians (47.2%) than female (39.8%); (p<0.001) (Figure 41). Figure 41 – Parents/guardians in target communities who could name at least three benefits of primary education, by gender 107. The results of the analysis from adjusting for differences across the three arms show that the proportion of parents/guardians in target communities who could name at least three benefits of primary education, was significantly higher in WFPSMP (57.2%) than in control (26.1%); (p<0.001). Figure 42 - Percentage of parents/guardians in control and WFPSMP schools who can name at least three benefits of primary education CONTROL WFPSMP (n=675) (n=579) Boys (n=1254) CONTROL WFPSMP (n=721) (n=565) Girls (n=1286) CONTROL WFPSMP (n=1396) (n=1144) Total (n=2540) 108. A similar pattern was evident between WFPSMP and HGSMP where the proportion of parents/guardians in target communities who could name at least three benefits of primary education, was significantly higher in WFPSMP (57.3%) than in 5% 0% 49 HGSMP (30.0%); (p<0.001) (Figure 42 and 43). The pattern was in both cases consistent among boys and girls. Figure 43 - Percentage of parents/guardians in WFPSMP and HGSMP schools who can name at least three benefits of primary education 70% Figure 39) Parents/guardians in target communities who can name at least three benefits of primary education 57,7% 56,9% 57,3% 30,0% 30,0% 30,0% 60% 50% 40% 30% 20% 10% 0% WFPSMP HGSMP (n=593) (n=713) Boys (n=1306) WFPSMP HGSMP (n=541) (n=743) Girls (n=1284) WFPSMP HGSMP (n=1134) (n=1456) Total (n=2590) 109. The following responses from FGDs indicate how parents came to know about the benefits of education across the three arms: Box 4 - PTA, BoM and pupil responses across the target arms of the study on how parents come to know about the benefits of education 3.6. Increased Capacity Ways in which parents are informed about the benefits of education: • Through organized public “barazas”/meetings • Through government sensitization of community members • Through having parents’ meeting at school • Through role modelling and motivational initiatives • Through guidance and counselling both parents and pupils Summary of main findings • The overall view from key informant interviews was that the policy and institutional environment has improved in the period preceding this baseline. • In terms of the baseline: o The absence of a national steering committee or equivalent that brings together the key stakeholders reduces the effectiveness of coordination at national level. In general, there is insufficient participation by other 50 110. WFP, with the support of USDA and others, will through this intervention seek to strengthen national capacity to improve the performance of the HGSFP and the WFPSMP and to ensure adequate transitioning of the schools that are still managed by WFP to the Government within the agreed time-frame. 111. This section examines the situation with respect to capacity, government support and regulatory framework at the time of the baseline, as drawn from documentary review, key informant interviews and school and county visits during the baseline data collection phase. MGD 1.4.1 Increased Capacity of Government Institutions Indicator 13: Number of county-level inter-ministerial committees for HGSMP established 112. Effective implementation of the school meals programmes at decentralized levels requires strong inter-ministerial coordination at the county level. 113. At baseline, there were no county level inter-ministerial committees in place for the control and the WFPSMP schools. 114. The key informant interviews, as well as the school visits, established several key limitations: • At present the departments of health and agriculture are not fully involved in all the county level coordination, with some exceptions. This is essential to ensure complementary activities and approaches in nutrition and key areas such as agricultural production and marketing. • Challenges arise from the different levels at which school feeding is managed. The pre-primary level (ECDE) has been decentralized to the counties while the primary ministries at national level in the school feeding efforts (with the exception of nutrition). The establishment of the national committee is a target for the MGD WFPSMP. o County level committees are not in place but are foreseen. Departments of health and agriculture are not fully involved in all the county level coordination, with some exceptions. o Coordination and management of school feeding takes place at decentralized level for ECDE and at national level for primary. Efficiency challenges are reported to arise from the different levels at which school feeding is managed. o A key gap is that the National School Health, Nutrition and Meals Programme Strategy was yet to be formally approved at time of the baseline. o Government funding for school feeding has increased in nominal terms but remains insufficient to cover the needs with school feeding in 2016, only covering 77 days out of 190. 51 education level remains the responsibility of MOE. Coordination between these two levels presents challenges, with each claiming and focusing on what is their jurisdiction without due attention to the bigger picture. • Most key informant interviews expressed the view that decentralizing school feeding for primary to county level would improve school feeding management across the different levels of education. In the absence of such decentralization it was felt that improved co-ordination between the MOE national level school feeding structures (MOE and national level committees) and the county level would help streamline management and increase the level of ownership of school feeding. This was felt to be insufficient in evidence at the baseline. Indicator 14: Number of national-level inter-ministerial coordination committees for HGSMP established 115. Effective implementation of the schools meals programmes also presumes strong and broad-based multi-sectoral coordination. 116. A schematic overview of the national coordination mechanisms for school feeding in the country is provided in Annex 3. The annex also outlines the responsibilities for the implementation of the school meals initiatives across ministries and non– governmental stakeholders. At present, there is no national level inter-ministerial coordination committee. The baseline for this indicator is therefore zero (the targetis one). 117. Stakeholder analysis informed by documentary research and key informant interviews (see Annex 4 for list of persons interviewed) identified and confirmed the critical roles played by government ministries, development partners, other government entities and departments and civil society organizations in the implementation of school meals programmes in Kenya. In particular, the MOE, the Ministry of Agriculture and the Ministry of Health stand out in their respective roles and responsibilities in designing and implementing the school meals programmes. 118. However, qualitative data from KIIs indicated that the participation of other ministries in school feeding coordination continues to be ad-hoc and participation and commitment is not always ensured. The process continues to be mostly driven by the education sector and the envisioned multi-sectoral ownership has been lacking, although there have been some improvements. A challenge has been frequent changes in leadership and senior positions in the MOE. The MOE acknowledges, and confirms the important roles played by the development partners and civil society organizations, with roles that range from resource/funding provision (USDA/MGD, GAC etc.) to implementing partners (WFP, Deworm the World, PCD etc.). However, inadequate multi-sectoral commitment to school feeding is evident, with the programme being mostly run by the MOE, although nutrition has been participating well. 119. At baseline, there continues to be a need for much stronger multi-sectoral approach, as evidenced by: • The absence of a national steering committee or equivalent that brings together 52 the key stakeholders and provides a stronger anchoring and coordination role at national level. • The need for increased integration of the program with the national social protection programming. • The need for increased integration of SMP with agricultural production and marketing programs. MGD 1.4.2/2.7.2 Improved Policy and Regulatory Framework Indicator 15: Number of educational policies, regulations, and/or administrative procedures in each of the following stages of development because of USDA assistance (Stage 5) Indicator 16: Number of child health and nutrition policies, regulations, and/or administrative procedures in each of the following stages of development because of USDA assistance (Stage 5) 120. As indicators 15 and 16 are closely linked, the baseline findings are discussed together. 121. At the time of the baseline the following overall policy documents were in place. • The Vision 2030 which highlights the importance of agriculture and the need for improved market access for small-scale farmers. This document is in place and is a key guideline for government and other stakeholder interventions. • The Agriculture Sector Development Strategy (2010) – focusing (among other priorities) on food and nutrition security for all Kenyans and increased employment and incomes in rural areas. • The National Social Protection Policy (2011) – in which school meals are one of the approaches to ensuring social protection. • The National School Health Policy (2009). • National School Health Guidelines (2009) – which emphasize that school meal programmes should have three components: balanced meals for children in all schools; encourage children in day schools to carry nutritious snacks and lunch; and supplementary feeding for children from the most underserved, food insecure regions, etc. • Food and Nutrition Security Policy (2011) – stresses that school meal programmes decrease short-term hunger, help pupils to concentrate and learn, encourage parents to enrol their children; and can provide iodine and iron and other micronutrients. • Home Grown School Meals Programme -Technical Development Plan (2012). • National School Health, Nutrition and Meals Programme Strategy (Draft Version 2016) which outlines the specific strategy for implementing school feeding in Kenya and which has received support from WFP. The Strategy 53 focusses on bringing together the broader policy into a more concrete strategy of government vision. It is based on a multi-sectoral commitment, coordination and ownership. 122. It should be noted that these documents were not developed as a result of the USDA assistance under this project, and that therefore strictly speaking the baseline for indicator 15 is zero. 123. The predominant view expressed in national level interviews was that there has been a lot of work done to improve the policy environment. There was also much appreciation of the GOK commitment to implementing: “Kenya has done a great job on the HGSMP, and it has taken up responsibilities” (source: interview). 124. A gap at the time of the baseline is that the key National School Health, Nutrition and Meals Programme Strategy remained to be formally approved. Key informant interviews specifically highlighted the importance of this document (which has been under preparation for a while), being approved. The expectation is that this would happened after the national election process in August 2017. MGD 1.4.3/2.7.3 Increased Government Support Indicator 17: Value of new public and private sector investments leveraged as a result of USDA assistance 125. At baseline the value of new public and private sector investments because of USDA assistance is zero given that work in this phase was yet to start. 126. However, to provide a benchmark against which progress can be measured the baseline examined the current public sector commitment (by the Government of Kenya) to school meals. This is by far the most substantial financial commitment to this initiative (see tables 10 and 11). Table 10 shows the trend in funding and coverage by the government led HGSMP, while table 11 shows the same by WFP. It is important to note the reducing budget commitment and coverage by WFP which reflects the transition from WFPSMP to HGSMP that has been going on since 2009. In addition, Annex 5 provides an overview of other school meals initiatives in the country including private initiatives, although it was beyond the focus of this study to quantify the financial investments in these interventions. 54 Table 10 - Government Funding of HGSMP since inception and the coverage in terms of Number of Pupils and Number of days the children were fed in a year Year Budget/Funding allocated (kshs) Number of Pupils reached Number of days children were fed in the year 2009 400 million 550,000 In most cases schools received 50% of the budget required and therefore were only able to feed the increasing number of children for two terms or even less. 2010 600 million 620,000 2011 650 million 770,000 2012 800 million 810,000 2013 850 million 810,000 2014 850 million 920,000 During this time, the MOE was only able to feed children for 77 (40% of the days) out of 190 days 2015 850 million 998,000 2016 850 million 1.2 million 2017 2.5 Billion 1.2 million 2017 -2018 likely to feed children for the three terms Source: Information provided by MOE to the Baseline Team, July 2017. Table 11 - WFP SMP Funding and Actual Beneficiaries reached (2009-2016) Number of Pupils reached Year Budget/Funding allocated (USD) Boys Girls Total 2009 27,816,522 493,518 368,730 862,248 2010 19,029,232 457,524 346,145 803,669 2011 18,290,159 483,631 380,470 864,101 2012 17,614,011 469,598 376,623 846,221 2013 9,515,965 425,435 341,673 767,108 2014 10,451,862 417,865 378,251 796,116 2015 9,782,097 451,871 333,582 785,453 2016 7,274,713 297,559 233,908 531,467 Source: WFP records 127. It should be noted that in 2016 and 2017, the GoK allocated additional funds to school feeding from the Drought Response Fund. Funds have also been provisionally marked in the GoK budget for 2018 for this priority. 128. Evidence reviewed at the baseline stage acknowledges the efforts made by the GoK in funding school feeding. However, concerns were voiced about: • Limited resourcing by GoK - actual implementation is limited by resources and that there has not been a commensurate scale up of financial resources to match the growing number of pupils. • Impact on the number of days of school feeding in schools that have transitioned to government ownership - thus, in 2016 school feeding was only provided 77 days 55 out of 190, compromising the quality of programme. 23 • Challenges in budgeting and timeliness of the budget - allocations per year are not sufficient. There is as yet insufficient appreciation from treasury of the commitments the government has made in policy documents and of the corresponding budget that is needed. Thus, while the GoK has taken over the HGSMP and the number of children has been increasing per year, the funding amounts have remained the same, with exception (as noted above) of the funds that were received from the Drought Response Fund. Indicator 18: Number of public-private partnerships formed as a result of USDA assistance 129. No USDA-WFP interventions had been implemented in this respect at the time of the baseline. This indicator will therefore be measured at mid- and end-line and is zero at baseline. Indicator 19: Number of Parent-Teacher Associations (PTAs) or similar “school” governance structures supported as a result of USDA assistance 130. No USDA-WFP interventions had been implemented in this respect at the time of the baseline. This indicator will therefore be measured at mid- and end-line and is zero at baseline. 3.7. Food utilization and food safety 131. This final section of the survey reports on issues related to hygiene and nutrition and provides the baseline with respect to the situation in the schools in terms of food preparation and storage. MGD SO 2 Increased Use of Health and Dietary Practices 23 These figures would suggest that the situation has become worse. The previous HGSMP review of 2012 had found that during the programme’s first three years (2009-2012), school meals had actually only been provided on about 54 percent of school days. Summary of main findings • The proportion of schools that store food off the ground in WFPSMP schools (56.5%) was significantly higher than control schools (17.4%). • The proportion of schools that store food off the ground in HGSMP schools (52.2%) was not significantly high than WFPSMP schools (47.8%); (p=0.271). 56 Indicator 20: Percent of schools in target counties that store food off the ground 24 132. The proportion of WFPSMP schools with a storage facility (82.6%) was significantly higher than of control schools (43.5%); (p=0.006). Contrary to what was expected some control school had a storage facility meaning that they prepare meals at the school. 133. The proportion of WFPSMP schools that store food off the ground (56.5%) was significantly higher than control schools (17.4%); (p=0.006) (Figure 44). Figure 44 - Comparison between the percentage of control and WFPSMP schools that store food off the ground 134. The proportion of WFPSMP schools with a storage facility (82.6%) was not significant higher than HGSMP schools (65.2%); (p=0.179). 135. The proportion of HGSMP schools that store food off the ground (52.2%) was not significantly higher than WFPSMP schools (47.8%); (p=0.768) (Figure 45). 24 The denominator is 23 schools (with or without food store) per arm. Please note that as the denominators are different for some variables a small percentage difference in one part of the analysis may be significant, while it may not be significant in other analyses where the denominator is much lower. 56,5% 17,4% WFPSMP(n=23) 57 Figure 45 - Comparison between the percentage of WFPSMP schools and HGSMP school that store food off the ground MGD 2.2 Increased Knowledge of Safe Food Prep and Storage Practices Indicator 21: Percent of food preparers at target schools who achieve a passing score on a test of safe food preparation and storage 136. The percentage of food preparers at schools who achieve a passing score25 on a test of safe food preparation and storage in WFPSMP schools (43.5%) was not significantly higher than in the control schools (39.1%); (p=0.765) (Figure 46). 25 The passing score was computed using the following variable; Q1 - Have you been trained in safe food preparation? Q2 - Have you been trained in food storage and handling? Q3 - Do you have a valid health certificate? Q4 - Do you have a uniform or apron to use in the kitchen? Q5 - When do you clean the kitchen? Q6 - When do you usually wash your hands for food preparation? Q7 - How do you ensure the food is clean before cooking? 73,9% 43,5% WFPSMP(n=23) HGSMP(n=23) Summary of main findings • There was no significant difference in the percentage of food preparers at WFPSMP schools who achieve a passing score on a test of safe food preparation in (43.5%), when compared to control schools (39.1%). • There was significantly higher percentage of food preparers at HGSMP schools who achieve a passing score on a test of safe food preparation and storage (73.9%) than those in WFPSMP schools (43.5%). 58 Figure 46 - Percent of food preparers at control and WFPSMP schools who achieve a passing score on a test of safe food preparation and storage 137. The percentage of food preparers at schools who achieved a passing score on a test of safe food preparation and storage in HGSMP schools (73.9%) was significantly higher than in the WFPSMP schools (43.5%); (p=0.036) (Figure 47). Q8 - How do you verify that food is in good condition before cooking? Q9 - How do you store food prior to serving it? It was computed as follows: Q1 (Yes=1) + Q2 (Yes=1) + Q3 (Yes=1) + Q4 (Yes=1) + Q5 (Every morning before food preparation=1) + Q6 (Before handling food=1) + Q7 (Use clean containers to collect food from store, remove foreign matters and then wash with clean water thoroughly before cooking=1) + Q8 (Look at Expiry date =1)+ Q9 (Store cooked food in covered cooking pots in a clean, safe place before serving the pupils=1). A % score was calculated using the attained score divided by 9 [attained score/9]. Those who score 50% and above are considered to achieve a passing score. 30% 20% 10% 0% 59 Figure 47 - Percent of food preparers at WFPSMP and HGSMP schools who achieve a passing score on a test of safe food preparation and storage MGD 2.3 Increased Knowledge of Nutrition Indicator 22: Number of schools benefitting from nutrition and hygiene education Hygiene 138. The most important hygiene habits mentioned by children included: general body hygiene/cleanliness (72.4%), hand washing (70.4%), safe drinking water (27.7%) and sanitation (25.0%) (Figure 48). 50% 40% 30% 20% 10% 0% Summary of main findings • The proportion of children who mentioned at least three hygiene habits was significantly higher in WFPSMP schools (51.0%) when compared to the control group (19.8%). • Similarly, the proportion of children who mentioned at least three hygiene habits was significantly higher in WFPSMP schools (50.4%) than in HGSMP schools (20.7%). 60 0% Figure 48 – Frequency and importance of different hygiene habits mentioned by children in response to the survey 139. The number of most important hygiene habits mentioned by the children varied significantly between study arms. However, the proportion of those that did not mention any habit was not significantly different between control (9.2%) and WFPSMP (8.2%); (p=0.375) (Figure 49). Figure 49 - Proportion of children who responded to the survey who mentioned three most important hygiene methods, by research arm (Control and WFPSMP) CONTROL WFPSMP (n=675) (n=579) Boys (n=1254) CONTROL WFPSMP (n=721) (n=565) Girls (n=1286) CONTROL WFPSMP (n=1396) (n=1144) Total (n=2540) 140. Differences were also not significant between the HGSMP (7.4%) and WFPSMP (6.9%); (p=0.625) (Figure 50). The proportions were consistent across the genders. 0% 61 mostimportant hygiene habits 0% Figure 50 - Proportion of children who responded to the survey who mentioned three most important hygiene methods by research arm (WFPSMP and HGSMP) WFPSMP HGSMP (n=593) (n=713) Boys (n=1306) WFPSMP HGSMP (n=541) (n=743) Girls (n=1284) WFPSMP HGSMP (n=1134) (n=1456) Total (n=2590) 141. Data from the qualitative study on hygiene practices in schools provided a series of responses that underlined a strong rationale for hygiene in schools and resulted in suggestions as to how it could be improved further as shown below. Box 5 - Responses through the qualitative interviews as to the perceived importance of hygiene and suggestions on how to strengthen this component of the intervention Encouraging handwashing before eating and after visiting the toilet, observation of general body cleanliness, maintaining cleanliness in the school compound and teaching pupils the importance of hygiene in science lessons and while on parade were considered as some of the key ways of enhancing good hygiene practices in schools (from FGD). Respondents suggested that such hygiene practices are important to promote good health among learners, prevent diseases, build the body; enhance concentration in class which improves performance; prevent absenteeism thus increasing school attendance; and enhance pupils’ confidence. Suggestions for improving hygiene practices in schools included: • Strengthening the provision of water for drinking and handwashing • Conducting education and capacity building through the creation of hygiene clubs • Having hygiene campaigns in schools and communities • Training teachers on hygiene camp • Ensuring regular deworming of learners and encouraging proper sanitation practices. 62 Nutrition 142. Results from this study show that the most important nutrition habits mentioned by children include; balanced diet (42.7%) and food type (39.8%) (Figure 51). However, the responses varied between study groups. Figure 51 - Frequency by which children mentioned different types of nutritional habits Over two-thirds of the respondents in WFPSMP group (72.3%) mentioned at least one important nutritional habit, compared to 61.7% of their counterparts in control (p<0.001) (data not shown). On the other hand, close to two-thirds of the respondents in HGSMP group (63.3%) mentioned at least one important nutritional habit/practice compared to 78.6% of their counterparts in WFPSMP (p<0.001) (Figure 51). The pattern was consistent among the genders in both cases (data now shown). Figure 52 and 53 below shows the percentage of children in different study arms who were able to mention three or more important nutritional habits. 7,1% 7,2% 39,8% 0% 5% 10% 15% 20% 25% 30% 35% 40% 63 30% 25% 20% 15% 10% 5% 0% 30% 25% 20% 15% 10% 5% 0% Figure 52 – Proportion of children from Control and WFPSP schools who mentioned three most important nutrition efforts CONTROL WFPSMP (n=675) (n=579) Boys (n=1254) CONTROL WFPSMP (n=721) (n=565) Girls (n=1286) CONTROL WFPSMP (n=1396) (n=1144) Total (n=2540) Figure 53 - Proportion of children from WFPSP and HGSMP schools who mentioned three most important nutrition efforts WFPSMP HGSMP (n=593) (n=713) Boys (n=1306) WFPSMP HGSMP (n=541) (n=743) Girls (n=1284) WFPSMP HGSMP (n=1134) (n=1456) Total (n=2590) 64 143. Qualitative data responses from parents and pupils indicated the importance of nutrition and how this could be improved. These responses are shown below. Box 6 - PTA, BoM and pupil responses across the three target arms of the study on the importance of nutrition and how it could be improved going forward Indicator 23: Number of individuals trained in child health and nutrition as a result of USDA assistance 144. No activities, USDA-WFP interventions, had been planned prior to the time of the baseline. This indicator will therefore be measured at mid- and end-line and is zero at baseline. MGD 2.6 Increased Access to Requisite Food Prep and Storage Tools Indicator 24: Number of target schools with increased access to improved food preparation and storage equipment (kitchens, storerooms, stoves, kitchen utensils) 145. No activities, USDA-WFP interventions, had been planned prior to the time of the baseline. This indicator will therefore be measured at mid- and end-line and is On importance of nutrition, both sets of respondents indicated that good nutrition: • promotes good learning processes • boost self-esteem and confidence • reduces sickness • reduces absenteeism • promotes concentration • promotes good health • motivates learning • keeps learner’s active • Enables retention of learners in school • Ensures pupils perform better On how nutritional habits practices could be improved, the following were noted as key: • Ensuring the child has three meals (including those in school and at home) • Providing enough food • Providing variety of foods and building proper kitchens and food stores. Summary of main findings • The baseline for schools with improved food preparation and storage facilities and for the number of meals provided under the USDA programme is zero. 65 zero at baseline. However, 34.1% (WFP SMP schools), 26.1% (control schools) and 52.3% (HGSPM) schools had access to fuel efficient stoves as shown in Table 12. Table 12 U Food preparation conditions in the three arms of the study Food preparation Control (n=23) p value WFP SMP (n=44) HGSMP (n=23) p value n % n % n % Sufficient kitchen for preparing pupils food 9 39.1% 0.428 13 29.5% 14 60.9% 0.013 Kitchen have fuel efficient stoves in sufficient quantity? 6 26.1% 0.502 15 34.1% 12 52.2% 0.152 Enough utensils 0 0.0% 20 45.5% 7 30.4% 0.234 146. Table 13 shows the distribution of storage properties among the different study arms. There was a significantly high number of WFP schools with better storage conditions compared to control schools. However, there was no significant difference in number of schools with improved better storage conditions between WFPSMP and HGSMP schools. Table 13 U Storage conditions in the three arms of the study Storage properties Control (n=23) p value WFP SMP (n=44) HGSMP (n=23) n % n % n % p value Storage locked 7 30.4% <0.001 37 84.1% 15 65.2% 0.078 Storage ventilated 6 26.1% 0.001 30 68.2% 15 65.2% 0.806 Humidity free storage 7 30.4% <0.001 33 75.0% 13 56.5% 0.123 Store have pallets 4 17.4% 0.003 24 54.5% 12 52.2% 0.853 Store have weighing scale 1 4.3% 0.016 13 29.5% 7 30.4% 0.940 Indicator 25: Number of school-aged children receiving daily school meals (breakfast, snack, lunch) as a result of USDA assistance 147. No activities, USDA-WFP interventions, had been planned prior to the time of the baseline. This indicator will therefore be measured at mid- and end-line and is zero at baseline. 66 4. Associated Factors 148. Factors associated with the highest level of English and Kiswahili literacy (story) as well as highest numeracy for a class 2 Work among school going children in class 3 to 8 were established. 149. The analysis of associated factors forms part of the key findings that could inform programming work and should enable WFP and its partners to further design and/or improve already existing intervention strategies. 4.1.Factors associated with the highest level of English literacy (story) for a class 2 Work among school going children in class 3 to 8. 150. Analysis of factors associated with the highest level of English literacy (story) fora class 2 Work among school going children in class 3 to 8 was done as presented in Annex 2. The analysis is based on binary logistic model, with a single dependent variable (dichotomous) against multiple independent variable. 151. The objective is to determine association/ relationship. The approach does not test or prove causal relationship and the results should thus be interpreted with some Summary of main findings Key factors associated with the highest level of English and Kiswahili literacy (story); as well as highest numeracy for a class 2 Work among school going children in class 3 to 8 include: • Class of the child • Mode of travel to school • Number of times child normally eat per day • Child had a meal today before going to school • Child thought it is important to go to school • Child having brothers and sisters who currently study in this school • Child having brothers and sisters who are old enough to go to school but are NOT currently attending school • Education level of the parent/guardian • Number of important nutrition habits mentioned by the parent/guardian • Number of hygiene habits mentioned by the parent/guardian • Household Coping Strategy Index 67 caution. In summary, this analysis finds that: • The highest level of English literacy was significantly associated with higher grades (ranging from 31.2% in class 4 to 87.4% in class 8) compared to class 3 (17.9%). • Highest level of English literacy was significantly higher in children provided with transport (76.4%) compared to those walking to school (52.1%). • Providing a child with a meal three times or more per day was significantly associated with highest English literacy performance (60.5%) compared to two times or less (47.5%). • Providing a child with a meal before going to school was significantly associated with improved highest English literacy performance (55.0%) compared to lack of meal before going to school (47.0%). • Understanding that it was important to go to school was significantly associated with improved highest English literacy performance (53.0%) compared to not appreciating (26.5%). • Not having brothers and/or sisters in the same school was significantly associated with improved highest English literacy performance (57.0%) compared to having them (50.6%). • Similarly, not having brothers and sisters who are old enough to go to school but are not currently attending school was significantly associated with improved highest English literacy performance (54.2%) compared to having them (41.0%). • Improved performance on highest English literacy was significantly associated with parent/guardian higher levels of education (ranging from 51.5% among those who did not complete primary school to 70.5% among those who reached Technical college/ university) compared to those with no formal education (43.1%). • A significantly high proportion of children whose parents/guardians mentioned at least one important nutrition habits, achieved highest level of English literacy (58.1%) compared to those whose parents could not mention any (40.7%). • Similarly, a significantly high proportion of children whose parents mentioned 1 to 2, or 3 and above important hygiene habits achieved highest level of English literacy (52.2% and 57.3%) compared to those whose parents could not mention any (34.8%). 4.2.Factors associated with the highest level of Kiswahili literacy (story) for a class 2 Work among school going children in class 3 to 8 152. Analysis of factors associated with the highest level of Kiswahili literacy (story)for a class 2 Work among school going children in class 3 to 8 was done as presented in Annex 2. In summary, this analysis shows that: • The highest level of Kiswahili literacy was significantly associated with higher grades (ranging from 44.7% in class 4 to 90.0% in class 8) compared to class 3 (25.9%). 68 • Children provided with transport (79.2%) had highest level of Kiswahili literacy compared to those walking to school (62.0%). • Providing a child with a meal three time or more per day was significantly associated with improved highest Kiswahili literacy performance (69.2%) compared to two times or less (57.9%). • Providing a child with a meal before going to school was significantly associated with improved highest Kiswahili literacy performance (63.6%) compared to lack of meal before going to school (59.9%). • Not having brothers and sisters who are old enough to go to school but are not currently attending school was significantly associated with improved highest Kiswahili literacy performance (64.0%) compared to having them (50.7%). • Improved performance on highest Kiswahili literacy was significantly associated with parent/guardian higher levels of education (ranging from 64.7% among those who did not complete primary school to 78.9% among those who reached technical college/university) compared to those with no formal education (52.4%). • A significantly high proportion of children whose parents mentioned 1 to 2, or 3 and above important hygiene habits achieved highest level of Kiswahili literacy (63.1% and 64.4%) compared to those whose parents could not mention any (48.2%). • The highest level of Kiswahili literacy was significantly associated with lower Coping Strategy Index (CSI) quintiles (ranging from 57.0% in third quintile to 61.3% in first quintile) compared to the fifth quintile (46.1%). 4.3.Factors associated with the highest level of numeracy (division) for a class 2 Work among school going children in class 3 to 8. 153. Analysis of factors associated with the highest level of numeracy (division) for a class 2 Work among school going children in class 3 to 8 was done as presented in Annex 2. Key associations were as follows: • The highest level of numeracy was significantly associated with high grades (ranging from 55.3% in class 4 to 93.3% in class 8) compared to class 3 (36.3%). • Highest level of numeracy was significantly higher in children provided with transport (86.1%) compared to those walking to school (69.4%). • Providing a child with a meal three time or more per day was significantly associated with improved highest numeracy performance (74.4%) compared to two times or less (66.7%). • Understanding that it was important to go to school was significantly associated with improved highest numeracy performance (70.2%) compared to not appreciating (44.1%). • Improved performance on highest numeracy was significantly associated with parent/guardian higher levels of education (ranging from 68.6% among those who 69 did not complete primary school to 82.4% among those who reached Technical college/ university) compared to those with no formal education (63.5%). • A significantly high proportion of children whose parents/guardians mentioned three or more important benefits of education achieved highest level of numeracy (71.1%) compared to those whose parents could not mention two or less (67.7%). • A significantly high proportion of children whose parents/guardians mentioned at least one important nutrition habits, achieved highest level of numeracy (74.2%) compared to those whose parents could not mention any (60.2%). • Similarly, a significantly high proportion of children whose parents mentioned 1 to 2, or 3 and above important hygiene habits achieved highest level of numeracy (70.6% and 72.2%) compared to those whose parents could not mention any (52.8%). 70 5. Discussion and implications 154. This chapter reflects on the results of the baseline and brings out implications for improving existing intervention strategies or designing new ones. It also seeks to inform the mid-line and end line evaluations of the intervention. The discussion is structured as follows: i) learning outcomes; ii) adjusting for other on-going interventions; iii) short term hunger; iv) school meals and expected outcomes; v) Pupil and parental perceptions related to hygiene, nutrition and education; vi) progress towards sustainability; and vii) national capacity. The chapter ends with a discussion of implications of the baseline for the implementation of the WFPSMP and the broader school feeding efforts in Kenya. 5.1. Learning Outcomes 155. Overall, the results in the preceding section are in agreement with the ranking of counties by UWEZO Kenya in their last Learning Assessment Survey (2016) (Table14 below).In this survey, the target WFSMP counties were among the poorest performers (see annex for the genesis of this situation in the WFPSMP Counties). This was also the case in the 2013 survey, which found large difference between urban and agricultural districts compared to arid and other less developed districts. The UWEZO Learning Assessment reports have consistently shown that Class 3 pupils in Turkana, Wajir, Mander, and Garissa cannot competently do standard 2 numeracy and literacy work. Table 14 - Selected county ranks - class 3 who can do class 2 Work (across all competencies) County Rank County Name OUTCOMES (Percentage) Class 3 who can do class 2 Work (%) Teacher presence (%) Pupil Presence (%) 1 Nyeri 51.8 88.3 88.9 5 Kajiado 42.3 83.9 88.1 8 Laikipia 39.2 90.4 86.0 12 Taita Taveta 35.1 82.8 87.3 17 Elgeyo Marakwet 31.0 91.0 84.0 20 Embu 29.5 86.7 86.6 23 Machakos 28.5 87.1 91.5 26 Kitui 26.1 91.3 81.4 30 Makueni 24.1 91.7 89.0 42 West Pokot 15.4 88.5 79.3 44 Garissa 15.3 86.1 83.9 45 Turkana 12.9 83.9 76.7 46 Mandera 10.1 89.0 77.7 47 Wajir 9.9 89.2 83.6 71 156. The current study used the UWEZO methodology to assess literacy (English/Kiswahili) and numeracy of class 3 to 8 pupils for class 2 Work. Thebaseline compared literacy and numeracy schools for the WFPSMP schools with the control group, and with the HGSMP schools which is the group of schools that have transitioned under the government programme. 157. A first important point to note is that the literacy and numeracy scores that were obtained in this baseline are comparable (i.e. in the same range) to those of the 2013 and 2016 UWEZO assessments. Both those assessments consistently find low scores for the ASAL areas. This confirms the reliability of the instruments used for this study. 158. A key finding from these comparisons is that in both comparisons (WFPSMP versus control, and HGSMP versus WFPSMP) the WFPSMP schools score lower on literacy and numeracy and in other education indicators such as attentiveness. Enrolment was the only indicator for which no difference was found in both sets of comparison. 159. A likely explanation for this is the location of the WFPSMP schools. Differences between the WFPSMP and HGSMP schools are possibly partly a reflection of the fact that WFPSMP schools/target counties are in the most marginalized and excluded zones of Kenya which have consistently in the UWEZO tests performed poorly, as already noted above. These arid zones have suffered long drawn and extreme educational marginalization from colonial times through to post independence, as a consequence of which they record low rates on virtually all education parameters. It is therefore likely that some of the difference can be explained by the fact that the HGSMP schools are in areas of the country that are less marginalized and better served economically, socially and politically. HGSMP schools in the semi-arid counties and were purposely selected for earlier transitioning as they were considered the easiest to transfer. 160. Other differences between the HGSMP schools and the WFPSMP schools that emerge from the baseline may also reflect the relatively better-off status of HGSMP schools/counties. For example, the study baseline finds that more children in HGSMP schools eat before going to school compared to WFPSMP schools. 161. In terms of the difference between WFPSMP schools and control schools on education indicators it should be noted that while every effort was made to select schools in similar zones to those where the WFPSMP schools were located for the purpose of having a control group, this proved to be very challenging in practice given that WFP targets all schools in each county. The baseline therefore had to select schools in neighboring counties which were identified as being as similar as possible against identified indicators for comparison, but probably not similar enough. 162. WFP-supported schools are exclusively located in Kenya’s northern arid counties. UWEZO’s published ranking of counties by learning outcome consistently highlights that the arid counties are largely in the bottom quarter of that list, with Garissa, Turkana, Mandera and Wajir as the bottom four for the past two assessments. As ‘control schools’ are not drawn from the bottom ranked quarter of the list it is very difficult to see comparability. This caveat is significant. It makes it rather difficult to effectively compare WFP-supported schools with the control group or the HGSMP. 72 For this reason, it would be worth looking at the progression of WFP over the course of the project in relation to the baseline point, in addition to the comparison with the other groups of schools as was done at the baseline phase. 163. This baseline study also identified key factors associated with the highest level of English and Kiswahili literacy (story); as well as highest numeracy for a class 2 Work among school going children in class 3 to 8. The high level of consistency betweenthe factors across different tests suggests strong reliability of the survey. It also presents a consistent picture of the various determinants of performance of pupils. It is worth noting that the UWEZO Kenya Learning assessment reports equally underscore that the determinants of performance by pupils are numerous and varied. And it is important to also remember that the same factors that impinge on the learners’ performance in literacy and numeracy may also affect other outcome level indicators (attendance, retention, etc.). These findings and those on other indicators, as well as on the association between variables should enable WFP and its partners to further design and/or improve already existing intervention strategies 5.2. Adjusting for other ongoing interventions 164. The Tusome (“Let’s Read’’ in Kiswahili) Early Grade Reading Activity is a collaboration between the Ministry of Education, USAID and UKAID to improve learning outcomes in English and Kiswahili in class 1 and 2 and is operational in the whole country – it covers public primary schools in the whole country including those in the three arms in this study. A midline evaluation of Tusome had just been done (2016) at the time of this study (and following a Tusome baseline in 2015). The results of the Tusome midline indicate that: • The Tusome approach is having a strong, positive influence on reading outcomes, with the data showing a strong relationship between project interventions and reading outcomes. • Reading outcomes for class 1 and 2 pupils greatly improved during the one-year period between the baseline and midline evaluations. While impressive gains have been made, continuing with the Tusome approach will be critical to sustaining or improving on those gains. 165. The end-line for Tusome project will take place in 2018. Learners in the intervention sites for the MGD WFPSMP will be in class 3 by that time and will therefore be captured in the mid-line evaluation of this project. 166. Further, another key project that is likely to influence learning outcomes going forward is the GPE funded Kenya Primary Education Development (PRIEDE) whose objective is to improve competencies in early grade Mathematics and strengthen education management and accountability both at the national and school level. The component on improving early grade Mathematics covers all public primary schools in the country including those across the three arms in this study. By the time the WFPSMP/USDA -MGD project will undertake its end line, the effects of this second project would have taken root in the target schools and therefore will need to be 73 considered in evaluating the numeracy component. 167. Since the two ongoing interventions are rolled out in all the target counties, their effect (if an effect exists) is assumed to be present in equal measure in all the design arms (Control, WFPSMP, and HGSMP). The control arm will assist in removing any such effects and will therefore enable measurement of any effect attributable to the WFPSMP and the HGSFP at midline and end line evaluations. 5.3. Short Term Hunger 168. The findings on short term hunger, are a clear pointer to the fact that a large majority of the counties targeted in the survey are arid and semi-arid and faced by an increasing frequency of droughts which impacts on local food production and food security in many homes, hence the prevalent hunger levels. Further, qualitative data from this baseline on the number of meals eaten per day indicate that food is not consistently available and underscore the important role that a school feeding programme will play in this context. High food prices are prevalent among a population that is largely poor, with over 60% of the population living below the poverty line in many of these areas, and even higher levels in counties like Turkana, Wajir and Mandera (between 74% and 97% of people live below the absolute poverty line in these areas). Arid lands, poor soil quality and unreliable rain are common and the trend is towards further deterioration given environmental degradation and climate change related natural disasters. Thus, food consumption at the household level is pragmatic. Individual families eat what is available and a substantial portion of family time is dedicated to finding food in detriment of other occupations such as education. 169. This baseline survey found that two out of five children (39.5%) reside in households with acceptable food consumption score with a mean coping index of 34. This also means that three out of five children live in families that do not have an acceptable food consumption score, underscoring the chronic food insecurity challenge in the targeted counties. Most districts were facing a serious drought and its associated effects at the time of the survey. The Famine Early Warning Systems Network (FEWS NET) (February 2017) reported Crisis (IPC Phase 3) food security outcomes in parts of the pastoral areas of Turkana, Marsabit, West Pokot, Baringo, Wajir, Mandera, Tana River and Garissa, and parts of the coastal marginal agricultural areas of Kilifi and Lamu. In all these pastoral areas, food security was projected to continue worsening through to July 2017 (i.e. beyond the time that the survey was carried out in April 2017). Households in the IPC Phase 3 category are only marginally able to meet their minimum food needs and only by depleting their assets more rapidly and thus undermining their food consumption. Further, households in the south￾eastern and coastal marginal agricultural areas, as well as some pastoral and agro￾pastoral areas of Narok, Kajiado, Laikipia, Kieni, Baringo and West Pokot counties were reported as being Stressed (IPC Phase 2), meaning they could afford minimally adequate food consumption but were unable to afford essential non-food expenditures. This situation is further compounded by inconsistent access to income or employment, or both, which affect the population’s ability to combat hunger. The resultant food insecurity is particularly detrimental for children. The latter are more vulnerable to the harmful effects of food insecurity and the long-term consequences 74 can be more severe. Poor nutrition and episodes of hunger subject children to increased health risks and impaired cognitive development. Families in these area are quite pragmatic about food consumption; they only eat what is available but this means there are hardly any considerations for nutrient rich foods whose prices are likely to be prohibitive. In effect, coping strategies at the household level are an expression of negotiated decisions to minimize the impact of food insecurity. Food sharing practices, may be a sustainable mechanism for coping with hunger. Such practices tend to be rooted in cultural and social customs. Hence, understanding these food insecurity coping strategies could be a good starting point to develop and formulate community based contextually sensitive interventions to improve household food security. 5.4. School Meals and Expected Outcomes 170. The findings indicate that approximately half of the parents/guardians reported that their children have been receiving school meals at school in the current school year (2017) with the proportions higher in the WFPSMP than control schools on one hand; and also higher in the HGSMP than in WFPSMP schools. 171. Two out of five of the parents/guardians reported that the school where the child was learning was serving food during the survey week, with WFPSMP schools more likely to be serving food in the survey week, than control schools, and HGSMP schools more likely to be serving food than WFPSMP schools. Yet the school attendance data of this survey show that, there was no significant difference between the average numbers of students regularly attending WFPSMP schools compared to control schools; while there is a significant difference between the average numbers of students regularly attending HGSMP schools compared to WFPSMP schools (with the former having higher attendance rates). Further, there was no statistically significant difference in average enrolment in the comparison between control schools and WFPSMP schools on one hand; and between WFPSMP and HGSMP schools on the other. 172. Could these outcomes be explained by the fact that the survey was undertaken at a time when the country was experiencing extreme drought conditions in many places, which coincided with the pipeline break in the WFPSMP in term one of 2017 because no funding was available for SMP? In effect, while it was expected that control schools would have no school meals programme, three control schools had a school meals programme in place at the time of the survey. Further, the fact that 12 WFPSMP schools did not have a school meals programme since the third term of 2016, although most WFPSMP schools (32) had the school meals programme likely played a role. Finally, most of the HGSMP schools (15) reported having the programme except three schools that indicated the programme existed but it was not running and strangely three schools which are classified as HGSMP schools by the MOE but where the programme does not exist according the persons responsible for these schools. 173. The dire drought situation prevalent in the target counties coupled with government’s budgetary constraints to meet the demands of all schools under the HGSMP might offer explanations for these situations across the three arms of the survey. Increasing frequency of droughts and associated famine has led the 75 government to put in place school feeding in areas that were not previously covered (i.e. in agro- ecological zones that were previously not considered as subject to severe food insecurity) and has also led some communities towards seeking local solutions. On the other hand, government budgets have not been sufficient to meet the needs of schools that have transitioned to the HGSMP and this may explain why some ofthese schools did not have school feeding at the time of the survey. 5.5.Pupil and parental perceptions related to hygiene, nutrition and education 174. The baseline shows that WFPSMP schools have better scores than control schools on selected indicators related to pupil and parental perceptions and practices related to hygiene, nutrition and education. This could suggest that attention to hygiene, nutrition and the importance to education has been stronger in the ASAL areas, given that these geographical areas have been prioritized by many other actors (UNICEF, NGOs) and that the higher awareness may be the result of interventions from other organizations in these areas. Further, previous McGovern￾Dole projects (FY 2004, 2005, 2006, 2007-2009, 2010-2013) could have had an impact among the WFP Schools on indicators related to pupil and parental perceptions and practices related to hygiene, nutrition, and education. While the baseline was not able to unequivocally establish that this is the case, the mid- and end￾line measurements will establish to what extent these indicators will evolve further and how this compares to any change in the comparison groups (the control and the HGSMP). 5.6. Progress towards sustainability 175. The intention of the comparison between the WFPSMP schools and the HGSMP schools was to identify progress towards sustainability. It is worth noting that the HGSMP schools as identified in the study are those that were handed over in the first phase of transitioning WFPSMP to HGSMP (2009 -2013) and do not include those transitioned in phase two that were in the more arid counties and were transitioned on a completely different modality. Consequently, while the significant differences between the locations of the schools make a comparison of the educational indicators challenging, there are some differences that are notable. HGSMP schools perform well on food preparation scores, but they do not perform well on hygiene and education awareness. As is the case for the control schools this may reflect the lack of exposure to activities that target education and hygiene awareness. It may also suggest that in the transitioning process attention to hygiene has been lost, and that this is reflected in the poorer scores. The higher scores on food preparation for HGSMP schools, would however, suggest that training that has been provided in this area has been relatively successful (WFPSMP cooks are still to be trained, as the baseline was started before activities started). It should also be remembered in this context that pupils are a ‘moving target’ in an 76 intervention of this kind, as they move on, whereas cooks are likely to stay in the same position for multiple years, consolidating what they have learnt. 5.7. National Capacity 176. The baseline establishes that Kenya has made progress in enacting policy and legal frameworks and in having a government owned and led school meals programme. It also underscores key areas that still need attention. Gaps include ensuring a strong national multi-sectoral coordination mechanism, and higher levels of more predictable government funding of school meals. Approval of the National School, Health Nutrition and Meals Strategy (2016) is pending, yet it is key in implementing school feeding. Further, the need for Kenya to have a stable funding source independent of external support for its school meals programme as a prerequisite for sustainability, cannot be emphasized enough. Consequently, the degree to which school feeding is included in county and national level planning and will determine to what extent school feeding can be sustained and produce the anticipated outcome and impact. 5.8.Selected implications for the mid- and end-line phases and for school feeding in Kenya 177. It will be important in the next phases of data collection to further examine some of the differences that have been identified between arms, and to ensure that the data collection tools (especially the qualitative part of the study) focus on the reasons for these differences. 178. Meanwhile it is important to highlight that despite these differences the baseline has achieved the objective of making it possible to record the values for each of the schools in each of the groups. This provides the basis for the mid and end-line phases to compare how WFPSMP schools have evolved in terms of these indicators compared to any changes control and HGSMP schools. 179. The baseline survey has demonstrated that a quasi-experimental design is feasible. Going forward through the mid line and end line evaluations, it would be important to ensure that: • The same schools visited during the baseline are visited during the midline and end line. • The changes in school meals programmes in the schools are documented and considered at both the midline and end line • The same sampling strategy is maintained at midline and end line. • Other cofounding factors that might influence the outlined hypothesis are documented and reported at both midline and end line. 180. The experience of the baseline exercise would suggest that for the mid-line and end-line exercises it would be sensible to do the qualitative data collection after the quantitative analysis. This will make it possible to have a more in-depth understanding of differences that are highlighted from the quantitative data and make 77 for an approach to the qualitative questioning that is more directly related to gaps in understanding. 181. The benchmark values for the indicators in the MGD Performance Monitoring Plan, coupled with the overall baseline findings and the analysis of associated factors, point to the need to focus on the following areas in implementing the WFPSMP: • Progress has been made in drafting the National School Health, Nutrition and Meals Programme Strategy, however it remains to be formally approved. WFP should continue to advocate for a speedy adoption and implementation of this strategy. • Adequate and regular funding through the Government budget is a challenge in transitioning schools to the Government led HGSMP (see also below). WFP should work with partners in advocating with the Ministry of Finance and the Treasury for ring-fencing SMP budgets (which would be consistent with the GOK’s social protection commitments). It should also advocate for regular and timely disbursements of GoK funds to schools, and for a progressive increase in government funding to the HGSMP. • Strong participation by all partners, regular meetings, and better coordination are critical to using scarce resource more efficiently and effectively. With the exception of nutrition, coordination among key ministries and programmes at national level remains weak. WFP should support the Ministry of Education in establishing the National Multi-sectoral Steering Committee, and support the GoK in seeking stronger participation of key ministries and programmes such as social protection and agriculture in this forum. • The baseline highlights poor literacy and numeracy scores of the schools in the WFPSMP areas. School feeding can offer only part of the solution and WFP should therefore actively coordinate its efforts with that of other partners in the same counties to ensure that the factors that affect school participation and achievement are addressed in a holistic manner. • The analysis of associated factors that are presented in this report should inform further research and guide programming work by WFP and its partners to further design and/or improve already existing intervention strategies. WFP could use these findings as input into a meeting with partners to discuss how to strengthen support to education in the targeted areas. • To support the transition WFPSMP schools there should be a strong focus on mentorship and capacity building of school leaders to be able to properly manage school feeding. • Given the importance of involving various sectors and actors in school feeding to cover all dimensions of the programme (nutrition, literacy, local production, etc.), WFP should actively support counties where it is operating in setting up County Level Multi-SectoralSteering Committees (bringing in health, agriculture, academic institutions, the private sector, and other partners as relevant) and 78 support these groups with capacity development if necessary. • WFP should develop a clear and convincing case for decentralizing the management of school feeding to the county level, as is already the case for the management of ECD. This will promote a better quality programme where thekey partners respond to/are answerable to their constituents and are more in tune with needs and requirements of the county, and will allow for better allocation and utilization or resources and monitoring programme implementation. A decentralized programme would also allow for the establishment of mechanisms by which counties might advance funds to purchase goods at the best possible time, and make the school feeding interventions more cost effective. • Most of the schools that participated in the baseline were having pipeline breaks. WFP should identify the key factors that are contributing to these pipeline issues and ensure regular delivery to the schools to minimize the number of days without school feeding. 182. Finally, the results of the baseline which compared the HGSMP and WFPSMP schools highlight challenges in the transitioning process. Recommendations in this matter fall outside of the strict scope of the WFPSMP but are still captured here as they are important for the broader group of stakeholders, and if they do not receive attention may in the future also affect the WFPSMP schools that are transitioning. In the view of this evaluation a successful transition from WFPSMP to sustainable HGSMP will need to consider the following: • Adequate budget allocations should be set aside (and ring fenced) by the GoK and disbursed to schools on time to ensure timely and cost effective purchase of the requisite food. • Capacity should be developed at all levels – national, county, and school to ensure effective implementation of the intervention. • The transition process should be allowed adequate time to ensure contextualization of best models and practices. • Enhanced coordination among all the stakeholders at various levels – national, county and school will be key to the success of the HGSMP. • Strong linkages with local smallholder farmers and traders and enhancement of their capacity to tap into the school markets effectively will be critical to the success of the HGSMP • The baseline finds that HGSMP schools perform well on food preparation scores, but they do not perform well on hygiene and education awareness. This suggests that in the transitioning process attention to hygiene has been lost and that this may need attention. • While it is the Kenya Government’s responsibility to provide food to school going regions in Arid and Semi-Arid areas, particularly with the anticipated transition to HGSMP, donors and other supporters will need to walk with the Ministry of Education over the transitional period to ensure success of the move from WFPSMP to HGSMP. To do this effectively, it would be important for support 79 towards the transition efforts to be more coordinated to reduce uncertainty and breaks in the resources that are necessary for an effective transition. It is also critical that adequate time be given to the transition process as well as predictable technical and other associated support through WFP. 183. Kenya still has a lot to be done on these elements to ensure a sustainable Home Grown School Meals Programme. In so doing, the country will still need the technical and financial support of the various partners that have brought the SMP this far. 80 Annex 1 – MGD Performance Monitoring Plan DRAFT Performance Monitoring Plan (PMP) Kenya FY 16 Award *NOTE: The first section includes results and performance indicators. The second section includes activities and activity output indicators. There is some overlap between the two sections where output indicators are also result indicators. Performance Indicator and Activity output indicator Indicator Definition and Unit of Measurement Data Source Method/ Approach of Data Collection or Calculation Data Collection Analysis, Use and Reporting When Who Why Who Result: MGD SO1 Improved Literacy of SchoolKAge Children Proportion of 7913 years olds that can solve Class 2 This indicator measures the proportion of children ages 7913 that UWEZO annual reports Review of UWEZO data Baselin e, Midter External evaluator s Indicates whether children’s’ WFP, MOE, Donors, numeracy and literacy problems (Outcome Indicator: Custom; Responsible Organization: UWEZO, USAID, Tusome Project Participants) have attained literacy and numeracy at a Standard 2 level Unit of measure: Percentage Disaggregation: TBD m, and final evaluat ion literacy and numeracy learning outcomes are being achieved through the USAID9funded Tusome project. This project developm ent and NGO partners, other Governm ent of Kenya institution s overlaps with USDA McGovern9 Dole9targeted counties and the schools are being co9 located for the achievement of MGD SO1 81 Number of individuals benefiting directly from USDA￾funded interventions This indicator measures the number of individuals directly benefitting from USDA9 WFP standard Project reports, Review and analysis of project records and reports Annuall y and quarter ly WFP and MOE Indicates the breadth and scale of the project's WFP, MOE, Donors, developm (Output Indicator: Standard; Responsible Organization: WFP and MOE) funded interventions. These individuals must come into direct contact with project interventions (i.e. goods or services). Direct beneficiaries include: children, teachers, school administrators, parents, cooks, storekeepers, farmers, and government staff. School termly reports impact in the target districts To inform annual review meetings with education stakeholders To inform annual reporting to USDA and WFP HQ ent and NGO partners, other Governm ent of Kenya institution s Unit of measure: individuals Data will be disaggregated by gender, new and continuing. Number of individuals benefiting indirectly from USDA-funded interventions This indicator measures the number of individuals indirectly benefitting from USDA9 funded interventions. Survey: Household /parent interviews Interviews with parents to determine the average number of children per household going to school. The average household size in target areas is known. Indirect beneficiaries=N umber of HH * (HH size￾average number of children per HH going to school) Baselin e, midter m, and final Independ ent consultan ts Indicates the breadth and scale of the project's impact. WFP, MOE, Donors, developm ent and (Output Indicator: Standard; Responsible Organization: WFP and MOE) These individuals will not come into direct contact with project interventions but will benefit tangentially. Indirect beneficiaries assumed for this project are siblings of children receiving school meals and parents of children who are not direct evaluat ion To inform annual review meetings with education stakeholders To inform annual reporting to USDA and WFP HQ NGO partners, other Governm ent of Kenya institution s beneficiaries through PTA training Unit of measure: individuals Data will be disaggregated by gender 82 Result: MGD 1.2 Improved Attentiveness Percentofstudentsin classrooms identified as inattentive by their teachers (Outcome Indicator: Custom; Responsible Organization: WFP, MOE) This indicator measures the percentage of students in any given classroom that is identified as inattentive by the teacher. Unit of measure: percent Survey: Teachers interviews Primary data collection by asking teachers of the sampled schools their perception of the share of students that appeared inattentive in classes Baselin e, midter m, and final evaluat ion Independ ent consultan ts To determine whether the interventions have influenced students’ ability to be attentive. WFP, MoE, Donors, developm ent and NGO partners, other Governm ent of Kenya institution s Result: MGD 1.2.1 Reduced ShortKTerm Hunger Number of daily school meals (breakfast, snack, lunch) provided to school9age children as a result of USDA assistance This indicator measures the total number of school meals provided to students in MGD￾supported schools, as reported by school managers and cooperating partners. WFP and MOE project records, School Termly Reports Review and analysis of project records and reports Bi annual and Annual, monthl y reports by School Administr ators, WFP To measure the number of school meals given to students. WFP, MOE, Donors, developm ent and NGO partners , other (Output Indicator: Standard; Responsible Organization: WFP, MOE) Unit of measure: no. of meals MOE, daily school records Governm ent of Kenya institution s Number of school9 aged children receiving daily school meals (breakfast, snack,lunch) as a This indicator measures the total number of students receiving a daily cooked meal per year over the life of the WFP and MOE project records, School Review and analysis of project records and reports Bi annual and Annual, monthl School Administr ators, WFP To measure the percentage of students reached with a WFP, MOE Donors, developm ent and result of USDA assistance (Output Indicator: Standard; Responsible Organization: WFP,MOE) project, as reported by school managers and CPs Unit of measure: individuals Data will be disaggregated by gender, new and records y reports by MOE, daily school records daily school meal NGO partners , other Governm ent of Kenya institution s continuing 83 Percent of students in target schools who regularly consume a meal before the school day (Outcome Indicator: Custom; Responsible Organization: WFP) This indicator measures what percentage of childrenreceiveameal at home prior to the school meal at lunch time. Unit of measure: percent Survey: Parent interviews Primary data collection by asking parents from sampled schools if their children eat before going to school and if yes, how often i.e. always, sometimes or never. Baselin e, midter m, and final evaluat ion Independ ent consultan ts To measure the percentage of children who may experience short9term hunger resulting in lack of concentration WFP, MOE Donors , developm ent and NGO partners , other Governm ent of Kenya as a result of not takinga meal before institution s going to school Percent of students in target schools who regularly consume a meal during the school day This indicator measures what percentage of childrenreceiveameal during the school day. Unit of measure: percent WFP and MOE project records, School records Review and analysis of project records and reports complemented by monitoring reports Bi annual and Annual, monthl y reports School Administr ators To measure percentage of students regularly reached with a daily school meal WFP, MOE, Donors, developm ent and NGO partners , (Outcome Indicator: Custom; Responsible Organization: WFP) by MOE daily collecti on by school other Governm ent of Kenya institution s Result: MGD 1.2.1.1/1.3.1.1.Increased Access to Food (School Feeding) Number of social assistance beneficiaries participating in productive safety nets as a result of USDA assistance (Output Indicator: Standard; Organization: WFP) This indicator measures thenumberofstudents who consume a daily meal at school Unit of measure: individuals Data will be disaggregated by new, continuing and gender. WFP and MOE project records, School records Review and analysis of project records and reports Bi annual and Annual, monthl y reports by MOE, daily collecti on by school School Administr ators, WFP To measure the number of students reached with a daily school meal WFP, MOE, Donors, developm ent and NGO partners , other Governm ent of Kenya institution s Total quantity of This indicator measures WFP WFP analysis of Bi9 WFP To measure WFP, commodities provided tostudentsasaresult the total amount of commodities that have Logistics Data reports annual report; the quantity of commodities MOE, Donors, of USDA assistance. been provided as a part that have been imported and developm ent and 84 (Output Indicator: Custom; Organization: WFP) of this USDA9funded intervention. Unit of measure: MT quarter ly are to be distributed. NGO partners , other Governm ent of Kenya institution s Result: MGD 1.3 Improved Student Attendance Number of students regularly (80%) This indicator measures thenumberofstudents School records Collection and analysis of Baselin e, Independ ent To track progress WFP, MOE, attending USDA supported classrooms/schools in MGD9supported schools who attend classes at least 80 students attendance data from midter m, and final consultan ts towards improved student Donors, developm ent and (Performance Indicator: Standard; Organization: WFP) percent of the time that school is in session, as reported by school directors Unit of measure: individuals school attendance records for a sample of students in sampled schools evaluat ion attendance NGO partners , other Governm ent of Kenya institution Data will be disaggregated by gender. Result: MGD 1.3.4 Increased Student Enrolment Number of students This indicator measures School Collection and Baselin Independ To track WFP, enrolled in schools receiving USDA the number of students officially registered in records analysis of school records e, midter ent consultan progress towards MOE, Donors, assistance MGD9supported primary schools in a given school year. on enrolment m, and final evaluat ts, WFP, MOE increasing student enrolment developm ent and NGO (Output Indicator: Standard; Responsible Organization: WFP) Unit of measure: individuals ion. Termly by schools, termly partners , other Governm ent of Kenya Data will be disaggregated by gender. by WFP throug h mVAM institution Result: MGD 1.3.5 Increased Community Understanding of Benefits of Education 85 Percent of parents in target communities who can name at least three benefits of primary education (Performance Indicator: Custom; Organization: WFP) This indicator measures the percentage of parents who can name at least three benefits of primary education Unit of measure: percent Survey: Parent interviews Primary data collection by asking parents from sampled schools to name at least three benefits of primary education Baselin e, midter m, and final evaluat ion Independ ent consultan ts To track communities understanding of engagement with their communities education system and services. WFP, MOE, Donors, developm ent and NGO partners , other Governm ent of Kenya institution s Result: MGD 1.4.1 Increased Capacity of Government Institutions Number of county9 level inter9ministerial committees for This indicator will measuretheNumberof county9levelinter9 Committe e meetings minutes Review of committee minutes midter m, and final Independ ent consultan To track progress of strengthening governance and multi￾sectoral coordination and collaboration for the school meals programme at county level WFP, MOE, Donors, HGSMP established ministerial committees forHGSMPestablished evaluat ion ts developm ent and (Output Indicator: Custom; Organization: WFP) at county level Unit of measure: Number of committees NGO partners , other Governm ent of Kenya institution s Number of national9 This indicator will Committe Review of midter Independ To track WFP, level inter9ministerial coordination measure the Number of county9level inter9 e meetings minutes committee minutes m, and final ent consultan progress of strengthening MOE Donors , committees for HGSMP established ministerial committees for HGSMP established at national level evaluat ion ts governance and multi9 sectoral developm ent and NGO (Output Indicator: Custom; Organization: WFP) Unit of measure: Number of committees coordination and collaboration for the school meals partners , other Governm ent of Kenya programme at national institution s level Result: MGD 1.4.2/2.7.2 Improved Policy and Regulatory Framework Number of educational policies, regulations, and/or administrative This indicator measures the number of policies/regulations/adm inistrative procedures in Governme nt of Kenya policy Review and analysis of GOK Annual, Baselin e, Independ ent consultan To track progress made following advocacy and WFP, MOE, Donors, developm 86 procedures in each of the various stages of related policy related Midter ts, WFP; dialogue ent and the following stages of development as a result of USDA progress towards an enhanced enabling environment for reports documents m and final evaluat MOE related activities to ensure NGO partners , other assistance (Stage 5) education. Specifically, this ions adequate and regular budget Governm ent of (Performance Indicator: Standard; Organization: WFP, MOE) includes: 1. School Nutrition and Meals Strategy 2. Revised HGSMP Guidelines allocations and maintain political commitment to the programme Kenya institution s Unit of measure: no. of policies in process and relevant stage Number of child This indicator measures Governme Review and Annual, Independ To track WFP, health and nutrition policies, regulations, and/or administrative procedures in each of the following stages of development as a result of USDA assistance (Stage 5) the number of policies/regulations/adm inistrative procedures in the various stages of progress towards an enhanced enabling environment for education. Specifically, this includes: nt of Kenya policy related reports analysis of GOK policy related documents Baselin e, Midter m and final evaluat ions ent consultan ts, WFP; MOE progress made following advocacy and dialogue related activities to ensure adequate and regular budget allocations and MOE, Donors, developm ent and NGO partners , other Governm ent of Kenya (Performance Indicator: Standard; Organization: WFP, MOE) 1. School Health Policy (revised) Unit of measure: no. of maintain political commitment to the programme institution s policies in process and relevant stage Result: MGD 1.4.3/2.7.3 Increased Government Support 87 Value of new public and private sector investments leveraged as a result of USDA assistance (Performance Indicator: Standard; Organization: WFP, MOE) This indicator measures the value of public sector resources intended to complement USDA-funded activities – specifically the increased government investment in the HGSMP. Unit of measure: US Dollar Data will be disaggregated by type of investment WFP and GOK project reports Review and analysis of project reports Baselin e, Midter m and final evaluat ions, Annual Independ ent consultan ts, WFP To measure level of complementar y support of the project outside of USDA funding. WFP, MOE Donors, developm ent and NGO partners , other Governm ent of Kenya institution s Number of public￾private partnerships formed as a result of USDA assistance (Performance Indicator: Standard; Organization: WFP, MOE) This indicator measures the number of private partnerships generated in CTS counties during the transition year. Unit of measure: no of partnerships (suppliers/small traders, farmer organizations) WFP reports; school tender data Review and analysis of project records and reports Annual WFP To measure level of complementar y support of the project outside of USDA funding. WFP, MOE Donors, developm ent partners, county governme nts; communit ies. Result: MGD 1.4.4/2.7.4 Increased Engagement of Local Organizations and Community Groups Number of Parent￾Teacher Associations (PTAs) or similar “school” governance structures supported as a result of USDA assistance (Performance Indicator: Standard; Organization: WFP) This indicator measures the number of schools that benefit from the establishment and training of PTAs Unit of measure: No. of school governance structures School and project records Review and analysis of project reports Bi9 annual WFP and MOE To measure the effects of the project on promoting the capacity of organizations at school level WFP, MOE, Donors, developm ent and NGO partners , other Governm ent of Kenya institution s Result: SO 2 Increased Use of Health and Dietary Practices Percentof schools in targetcountiesthat store food off the ground This indicator will measurethenumberof schools where food is stored off the ground Survey reports, Monitoring reports School stores will be observed to check if food has been stored off the ground. Baselin e, Midter m and final evaluati ons, monthl Independe nt Consultan ts, WFP and MOE To measure the effects of promoting good hygiene and health practices, WFP, MOE, Donors, developm ent and NGO partners , other 88 (Performance Indicator: Custom; Responsible Organization: WFP) Unit of measure: No. of school y through monthl y monitor ing visits at school level Governm ent of Kenya institution s Result: MGD 2.2 Increased Knowledge of Safe Food Prep and Storage Practices Percent of food preparers at target schools who achieve a passing score on a test of safe food preparation and storage (Outcome indicator: Custom; Responsible Organization: WFP) This indicator will measure the percentage of food preparers (cooks) at school who achieve a passing score on a test of safe food preparation and storage Unit of measure: individuals Data will be Survey report: Results of tests administer ed to cooks Primary data collection by administering a test on safe food preparation and storage to cooks in representative sampled schools Baselin e, midter m, and final evaluat ion Independ ent consultan ts To measure effects of promoting safe food preparation and storage practices WFP, MOE, Donors, developm ent and NGO partners , other Governm ent of Kenya institution s disaggregated by gender. Result: MGD 2.3 Increased Knowledge of Nutrition Number of schools benefitting from nutrition and hygiene education (Output indicator: Custom; Responsible Organization: WFP) This indicator will measure the number of schools benefitting from nutrition and hygiene education Unit of measure: No. of school project reports Review and analysis of project reports Quarter ly, Bi9 annual WFP and MOE To measure number of schools that have received nutrition and hygiene related education WFP, MOE, Donors, developm ent and NGO partners , other Governm ent of Kenya institution s Number of individuals trained in child health and nutrition as a result of USDA assistance Total number of individuals trained in health and nutrition in MGD9supported schools and communities, Project reports Review and analysis of project training reports Termly Bi9 annual WFP and MOE Enables to know the number of people in communities’ WFP, MOE, Donors, developm ent and (Output Indicator: Standard; including Canteen Management Staff and School Management Committee members. target who have knowledge in health and nutrition. NGO partners , other Governm ent of Sentinel Kenya 89 Responsible Organization: WFP) Unit of Measure: Individuals Data will be disaggregated by gender indicator for project theory of change: people trained shared nutrition and health information through communities institution s Result: MGD 2.6 Increased Access to Requisite Food Prep and Storage Tools Number of target schools with increased access to improved food prep and storage This indicator measures the number of schools fully supplied with new or rehabilitated kitchens, storerooms, fuel￾Project reports Review and analysis of project reports Quarter ly, Bi9 annual WFP and MOE To track s progress towards improving access to food WFP, MOE, Donors, , developm ent and NGO partners , other Governm ent of Kenya institution s equipment (kitchens, storerooms, stoves, kitchen utensils) efficient stoves and kitchen utensils prep and storage equipment Unit of measure: no. of schools (Output indicator: Custom; Organization: WFP) Activity 1: Provide School Meals Number of school9 This indicator measures Project Review and Monthl WFP and To measure WFP, aged children receiving daily school the total number of students receiving a reports analysis of project reports y, quarter MOE the success of school meals MOE Donors, meals (breakfast, snack, lunch) as a result of USDA daily cooked meal per year over the life of the project, as reported by ly Bi9 annual at reducing short term hunger developm ent and NGO assistance school managers and partners , (Output Indicator: Standard; Organization: WFP, MOE) CPs Unit of measure: individuals other Governm ent of Kenya institution Data will be s disaggregated by gender. Activity 2: Build the Capacity of National and CountyUlevel Actors to Manage School Feeding Programs 90 Number of parents This indicator measures Project Review and Bi9 WFP and To track progress in building capacity of school –level actors (BoM members) to manage school feeding programs WFP, trained or certified as a result of USDA assistance the number of parents that have been trained as a result of USDA reports analysis of project training reports annual MOE MOE, Donors, developm assistance ent and NGO (Output Indicator: Custom; Organization: WFP) Unit of measure: individuals Data will be disaggregated by partners , other Governm ent of Kenya gender. institution s Number of school administrators and This will measure the number of school head Project reports Review and analysis of Bi9 annual WFP and MOE To track progress in building capacity of school head teachers to manage school feeding programs WFP, MOE, Donors, developm ent and NGO partners , other Governm ent of Kenya institution s officials in target schools trained or certified as a result of teachers trained on school meals programme project training reports USDA assistance management (Output Indicator: Standard; Responsible Organization: WFP) Unit of measure: individuals Data will be disaggregated by gender. Number of county9 level officials trained or certified as a result of USDAassistance (Output Indicator: Standard; Responsible Organization: WFP) This will measure the number of education officials trained on school meals programme management Unit of measure: individuals Data will be disaggregated by gender. Project reports Review and analysis of project training reports Bi9 annual WFP and MOE To track progress in building capacity of school head teachers to manage school feeding programs WFP, MOE, Donors, developm ent and NGO partners , other Governm ent of Kenya institution s Number of school administrators and officials in target schools who demonstrate use of new techniques or toolsas a resultof This will measure the number of school head teachers trained on school meals programme management Project reports Review and analysis of project training reports Bi9 annual WFP and MOE To track progress in building capacity of school head teachers to manage school feeding programs WFP, MOE, Donors, developm ent and NGO partners , other Governm ent of Kenya USDA assistance Unit of measure: individuals Data will be (Output Indicator: disaggregated by Standard; gender. 91 Responsible Organization: WFP) institution s Number of county9 level officials in target schools who demonstrate use of new techniques or tools as a result of USDA assistance (Output Indicator: Standard; Responsible Organization: WFP) This will measure the number of education officials trained on school meals programme management Unit of measure: individuals Data will be disaggregated by gender. Project reports Review and analysis of project training reports Bi9 annual WFP and MOE To track progress in building capacity of school head teachers to manage school feeding programs WFP, MOE, Donors, developm ent and NGO partners , other Governm ent of Kenya institution s Activity 3: Raise Awareness on the importance of Education Number of radio spots held (Output Indicator: This indicator will measure the number of radio spots held to pass messages on benefits of education. These will Project reports Review and analysis of project reports Monthl y, Quarter ly, Bi9 annual WFP and MOE To track the number of radio spots held WFP, MOE, Donors, developm ent and NGO partners , other Governm ent of Kenya institution s Custom; Organization: WFP) target communities where the programme is implemented Unit of measure: number of radio spots Number of community members benefiting from radio spots (Output Indicator: Custom; Organization: WFP) This indicator will measure the number of community members in targeted counties (Baringo, Garissa, Mandera, Turkana, Wajir and West Pokot) reached through radio spots with messages on benefits of education. Project reports Review and analysis of project reports Monthl y, Quarter ly, Bi9 annual WFP and MOE To track the number of community members reached through the radio spots WFP, MOE, Donors, developm ent and NGO partners , other Governm ent of Kenya institution s 92 Number of posters, fliers, leaflets distributed (Output Indicator: Custom; Organization: WFP) This indicator will measure the number of posters, fliers, leaflets distributed Unit of measure: number of posters, fliers, leaflets project reports Review and analysis of project reports Termly Bi9 annual WFP and MOE To track number of posters, fliers, leaflets distributed WFP, MOE, Donors, developm ent and NGO partners , other Governm ent of Kenya institution s Activity 4: Build/Rehabilitate: Kitchens, Cook Areas and Other School Grounds or Buildings Number of educational facilities (i.e. school buildings, classrooms, and latrines) This indicator will measure the number of kitchensand/orstorage facilities constructed as a result of USDA assistance Unit of measure: number of kitchens project reports compleme nted by monitoring reports Review and analysis of project reports Bi9 annual, monthl y monitor ing reports WFP and MOE To track number of kitchens constructed WFP, MOE, Donors, developm ent and NGO partners , other Governm ent of Kenya institution s rehabilitated/cons tructed as a result of USDA assistance (Output Indicator: standard; Organization: WFP) Activity 5: Provide EnergyUSaving Stoves to Schools Number of energy saving jikos installed in schools as a result of USDAassistance (Output indicator: Custom; Responsible Organization: WFP) This indicator will measure the Number of energy saving jikos installed in schools as a result of USDA assistance Unit of measure: number of energy saving jikos project reports compleme nted by monitoring reports Review and analysis of project reports Bi9 annual, monthl y monitor ing reports WFP and MOE To track number of energy saving jikos installed at school level WFP, MOE, Donors, developm ent and NGO partners , other Governm ent of Kenya 93 institution s Activity 6: Conduct Awareness Campaigns and Trainings on Nutrition and Hygiene Number schools benefitting from nutrition education and hygiene (Output Indicator: Custom; Responsible Organization: WFP) This indicator measures the number of schools benefitting from nutrition and hygiene education Unit of measure: number of schools project reports compleme nted by monitoring reports Review and analysis of project reports Bi9 annual, monthl y monitor ing reports WFP and MOE To track the number of schools benefitting from nutrition education and hygiene WFP, MOE, Donors, developm ent and NGO partners , other Governm ent of Kenya institution s Number of children This indicator measures project Review and Bi9 WFP and To track the WFP, MOE, Donors, developm ent and NGO partners , other Governm ent of Kenya institution s benefitting from nutrition education the number of children benefitting from reports compleme analysis of project reports annual, monthl MOE number of children and hygiene (Output Indicator: Custom; Responsible Organization: WFP) nutrition and hygiene education Unit of measure: individuals nted by monitoring reports y monitor ing reports benefitting from nutrition education and hygiene Data will be disaggregated by gender Activity 7: Empower the Community to Manage School Feeding Programs Number of counties where beneficiary feedback has been has been incorporated into community training and awareness activities (Output Indicator: Custom; Organization: WFP) This indicator will measure the number of counties where beneficiary feedback has been rolled out Follow up to increase awareness on the helpline will include radio spots, public meetings and distribution of posters and leaflets project reports compleme nted by monitoring reports Review and analysis of project reports Quarter ly, Bi9 annual, monthl y monitor ing reports WFP and MOE To track the number of counties with beneficiary feedback mechanism in place WFP, MOE Donors , developm ent and NGO partners , other Governm ent of Kenya institution s 94 Unit of measure: Number of counties Activity 8: Promote Food Safety and Quality in the HGSMP 95 Number of officials trained on food quality in HGSMP supply chain (Output Indicator: Custom; Organization: WFP, MOE) This indicator measures thenumberofofficials (County Public Health Officers, County School Meals Programme Officers, School Meals Procurement Committee and traders )trained on food quality in HGSMP supply chain Unit of measure: individuals Data will be disaggregated by gender project reports Review and analysis of project training reports Bi9 annual, WFP and MOE To track to the number of officials trained on food quality in HGSMP supply chain. WFP, MOE, Donors, developm ent and NGO partners , other Governm ent of Kenya institution s Number of farmer organizations trained on food quality (Output Indicator: Custom; Organization: WFP) This indicator measures the number of farmer organizations trained on food quality Unitofmeasure:farmer organizations project reports Review and analysis of project training reports Bi9 annual, WFP and MOE To track to the number of farmer organizations trained on food quality WFP, MOE, MOALF, Donors, developm ent and NGO partners , other Governm ent of Kenya institution s Number of traders trained on food quality (Output Indicator: Custom; Organization: WFP) This indicator measures the number of traders trained on food quality Unit of measure: individuals Data will be disaggregated by gender project reports Review and analysis of project training reports Bi9 annual, WFP and MOE To track to the number of traders trained on food quality WFP, MOE, MOH, Donors , developm ent and NGO partners , other Governm ent of Kenya institution s 96 Number of individuals who demonstrate use of new safe food preparation and storage practices as a result of USDA assistance (Outcome Indicator: Standard ; Organization: WFP) This indicator measures the number of farmer organization, officials and traders applying improved food quality practises after undergoing training on food quality. Unit of measure: Number of farmer organizations , officials and traders Data will be disaggregated by farmer organizations, officials and traders Survey reports compleme nted by project reports Primary data collection through observation and interviewing traders and farmer organization representatives on what improved food quality practices they are applying that they did not before the training Baselin e, midter m, and final evaluat ion Independ ent consultan ts To measure effectiveness of the training WFP, MoE, Donors, developm ent and NGO partners, other Governm ent of Kenya institution s Number of testing kits (Blue Boxes) distributed to public health officials (Output Indicator: Custom; Organization: WFP) This indicator will measure the number of testing kits (Blue Boxes) distributed to public health officials Unit of measure: Number of blue boxes project reports Review and analysis of project reports and blue boxes distribution reports Bi9 annual, annual WFP and MOH To track to the number of testing kits (Blue Boxes) distributed to public health officials WFP, MOE,MO H, MOALF, Donors, developm ent and NGO partners, other Governm ent of Kenya institution s 97 Annex 2 – Detailed Baseline Methodology Overview A detailed methodology for the baseline was drawn up during the inception phase and presented in an Inception Report (Visser et al, 2017). An important aspect of the Inception phase was to establish whether the envisioned quasi-experimental design for the study was feasible. As the team’s assessment showed that this was feasible the study was designed in line with these parameters. The inception phase also identified key parameters for the study including the required sample size, data collection approach and tools, and the approach to data analysis. Feasibility of the proposed quasi-experimental design The Inception phase confirmed that a quasi-experimental design could be employed in this study. The assessment was based on the fact that a quasi – experimental design is feasible when one can get a match between the intervention and control. This was deemed feasible in this case because the study team was able to: i) Generate variables ‘good enough’ for the PSM. ii) Other data sets (livelihoods and food security data) were found to be available and sufficiently suitable for identification of locations iii) Successfully carry out the PSM. iv) Successfully identify matching: WFPSMP-Controls and WFSMP - HGSMP Schools. Overall evaluation design A pretest posttest quasi-experimental design was set up to measure both the difference before and after the intervention in the treatment groups, and also the difference between control and treatment. The study quasi-experimental design thus compares three groups: • WFPSMP: Selected schools located in counties where WFPSMP under the USDA – MGD funding is to be implemented (the intervention schools). • HGSMP: Selected schools located in counties where WFPSMP was being implemented but now transitioning to HGSMP. • Control: Selected schools located in counties where neither WFPSMP nor HGSMP is to be implemented. 98 Research question and hypotheses The Research question and testable hypotheses that underpin the quasi –experimental design will allow WFP, USDA and its partners to establish examine whether the baseline, mid-term and end-term primary education outcomes (literacy and numeracy levels) and other educational indicators (enrolment, attendance, completion, parental involvement, etc.) in the arid and semi-arid lands (ASAL) areas of Kenya are the same in schools included in WFP/USDA-MGD school meals programme (2016 -2020) as those not included (controls and those transitioning to HGSMP).Four different hypotheses were formulated and proposed for testing at Mid-term and End term evaluation for each indicator: Indicator 1: • H0: Enrolment in schools included in WFP/USDA-MGD SMP ≠ Enrolment in schools not included in WFP/USDA-MGD SMP • H1: Enrolment in schools included in WFP/USDA-MGD SMP= Enrolment in schools not included in WFP/USDA-MGD SMP Indicator 2: • H0: Attendance rate in schools included in WFP/USDA-MGD SMP≠ Attendance rate in schools not included in WFP/USDA-MGD SMP • H1: Attendance rate in schools included in WFP/USDA-MGD SMP = Attendance rate in schools not included in WFP/USDA-MGD SMP Indicator 3: • H0: Primary school completion rate in schools included in WFP/USDA-MGD SMP ≠ Primary school completion rate in schools not included in WFP/USDA-MGD SMP • H1: Primary school completion rate in schools included in WFP/USDA-MGD SMP = Primary school completion rate in schools not included in WFP/USDA-MGD SMP Indicator 4: • H0: Literacy/numeracy rate in schools included in WFP/USDA-MGD SMP ≠ Literacy/numeracy rate in schools not included in WFP/USDA-MGD SMP • H1: Literacy/numeracy rate in schools included in WFP/USDA-MGD SMP = Literacy/numeracy rate in schools not included in WFP/USDA-MGD SMP Sampling Since the WFPSMP will run in all schools located within six selected ASAL counties (Baringo, Garissa, Turkana, Mandera, West Pokot, and Wajir)26, control schools were 26 Isiolo, Nairobi, Samburu, and Tana River which were targeted under the previous phases of the USDA support will not be included. These counties were excluded from the HGSMP group for the following reasons. Nairobi was excluded because of urban context issues. The majority of the counties of focus are in the Arid, rural areas, consequently, there were hardly any common contextual similarities that will match Nairobi with them. The other three have been beneficiaries of the Cash Transfers to schools Model developed and implemented by WFP before being handed over to HGSMP – consequently their evolution modality and short history of the same does not approximate to a pure HGSMP modality of government that has been going on in some of the counties selected since 2009. 99 selected from the neighboring areas (either within the same county or in a neighboring county (in a manner that matched as closely as possible the socio-economic activities - livelihood zones - to ensure similarity in terms of vulnerability and food insecurity). Similarly, the HGSMP schools were selected from the neighboring areas with comparable socio-economic activities. Selected control and HGSMP schools were matched against WFPSMP schools. Group comparison based on schools: Prior to data collection propensity score matching (PSM) was used to compare and match schools using selected school characteristics derived from Education Management Information System (EMIS) tool. Selection of matching characteristics was based on theoretical background knowledge 27 of confounders of the measurement indicator(s). The matching characteristics were selected to be unrelated (unaffected) by the proposed intervention (WFPSMP or HGSMP). Propensity scores were constructed using the ‘participation equation’, derived from a logit regression28 with programme participation as the dependent variable coded as follows: • WFPSMP school = 1, versus Control school = 0, and • HGSMP school = 1, versus WFPSMP school = 0. Each school belonging to one of the intervention groups was matched to one school of the control group by matching each to their ‘nearest neighbor’ using propensity score. Characteristics that were used in matching included: boy: girl ratio, averagepupils/class, pupils: teacher ratio, residence type (rural/urban). This data was taken from the Ministry of Education EMIS data set. Schools in the first group with a propensity score lower than the lowest observed value in the second group were discarded. Similarly, schools in the second group with a propensity score higher than the highest observed value in the first group were also discarded. The same approach was used for the control group. The remaining schools were in the region of common support from which participating schools were selected. This process resulted in the identification of three groups of schools that were as similar as possible from the perspective of livelihoods and socio-economic characteristics. The original design in the IR anticipated a matching of 30*30*30 for the three groups of schools where these schools would all overlap. The data collected allowed for the 27 Theoretical background knowledge refers to knowledge about factors that are plausible or known to confound the relationship between the outcome(s) and the intervention. They are potential or are confirmed to be independently related to the outcome(s). 28A Logistic regression is a statistical method for analyzing a dataset in which there are one or more independent variables that determine an outcome. The outcome is measured with a dichotomous variable (in which there are only two possible outcomes). 100 matching of 23 schools from each set where 23 WFPSMP schools were matched with 23 control schools, and 23 HGSMP schools were matched to 23 WFPSMP schools. In this manner, the study obtained: 23 WFPSMP matched with 23 control schools and 23 HGSMP matched with 23 WFPSMP schools. While this is different from the design it had no implications for the study as such as the comparison between WFPSMP and HGSMP was not part of the initial design. Group comparison based on children: This process took place after data collection where propensity score matching was done to ensure comparability of pupils (between the groups) using selected characteristics captured during data collection, therefore reducing selection bias (the possibility that those enrolled in a particular group are systematically different from those enrolled in another group). The matching characteristics were those that are unaffected by the intervention (WFPSMP or HGSMP). Like in school comparison, each member of a specific group was matched to one member of the comparison group by matching each to their ‘nearest neighbor’ using propensity score. Baseline data was used for calculating propensity scores. The propensity score constructed using children characteristics was used as a weighting factor to balance the groups during analysis. The same technique will apply at mid-term and final evaluation using the same characteristics. Sample size The results conceptual framework for the MGD intervention envisages realization of two results as follows: 3. Results framework #1: MGD Strategic Objective (SO)1 Improved Literacy of School-Age Children. 4. Results framework #2: MGD SO2 Increased Use of Health and DietaryPractices. Since MGD SO2 is a function of MGD SO1, the sample size was calculated based on MGD SO1. The baseline estimate aligned to MGD SO1 Was interpreted to be the proportion of children ages 7-13 that have attained literacy and numeracy at Standard 2 level. 101 1 UWEZO29 Kenya’s Sixth Learning Assessment Report December 2016, suggested that the learning outcome by selected counties on Class 3 who can do Class 2/Standard 2 level work showed a substantial degree of variance. 30 Due to variation in baseline estimate across selected counties and with potential variation in other measurement indicators, this study design decided to use a 50% conservative estimate as the proportion of children ages 7-13 that have attained literacy and numeracy of a Standard 2 level- Standard 2 competencies in literacy and numeracy. The proportion optimized the sample size to allow for estimation of all indicators devoid of the risk of low sample size calculation. The study presumed a 20% effect size on the primary indicator. The minimum sample size was calculated using Fleiss, et al (15) formula as follows: (Z + Z ) 2 *(P (1P)+ P (1P )) n = D* 1 / 2 1 1 1 2 2 Where; (P2 P ) 2 Performance indicators presented as percentages (P1, P2) P1 (estimated value of indicators at baseline) 50% P2 (estimated value of indicators at final evaluation) 70% P2-P1 (estimated change over time) 20% α (Type 1 error) 0.05 β (Type 2 error) 0.10 Zα (Z score at desired statistical significance) 0.975 1.96 Zβ (Z score at desired statistical power) 0.90 1.28 D (design effect = 1 + δ (m – 1); where m is the average enrolment per school (200) and δ is the estimated intra-class correlation coefficient, referenced from literature (0.02)) 5.0 620 29 Uwezo is a five-year initiative that aims to improve competencies in literacy and numeracy among children aged 6-16 years old in Kenya, Tanzania and Uganda, by using an innovative approach to social change that is citizen driven and accountable to the public. 30 The proportions in the proposed intervention areas ranged as follows; Wajir – 9.9%, Mandera – 10.1%, 102 Turkana – 11.4%, Garissa – 12.9%, West Pokot – 15.4%, and Baringo – 16.6%. 103 The sample size (n) of measurement unit - number of sampled children ages 7-13 in Standard 3 to 8 Allowing for 10% non-response, the sample size is adjusted upwards (n/ (1-L) where L is the provision of 10% non-response). Adjusted sample size = 620/ (1-0.1) = 688.88889, rounded upwards to 689 children. Therefore; number of sampled children per study arm (without replacement) 689 Overall sample size in both intervention and control arms 2,067 In order to address gender mainstreaming and women’s empowerment as per WFP’s evaluation principle of gender equality, the evaluation will be conducted with a view to elucidating the effect of the intervention (WFPSMP or HGSMP) among boys and girls. To the greatest extent possible, the consultants will ensure both men and women are targeted as respondents. Therefore, the overall sample size in both interventions (WFPSMP and HGSMP) and control arms will triple to 4,134 (2067 boys (689 HGSMP, 689 WFPSMP, 689 Controls); 2,067 girls (689 HGSMP, 689 WFPSMP, and 689 Control). As each pupil questionnaire also includes questions for a corresponding parent (see Annex 4), there will be an equal number of parental responses. Care will be taken to have at least 40 percent female parents participating in the study. In order to address gender mainstreaming and women’s empowerment as per WFP’s evaluation principle of gender equality, the overall sample size in both interventions (WFPSMP and HGSMP) and control arms was tripled to 4,134 (2067 boys (689 HGSMP, 689 WFPSMP, 689 Controls); 2,067 girls (689 HGSMP, 689 WFPSMP, and 689 Control). As each pupil questionnaire also included questions for a corresponding parent (see Annex 4), there were also an equal number of parental responses. The baseline targeted having at least 40 percent female parents participating in the study. In practice this target was largely surpassed. Sample procedure A two-stage sampling procedure was employed at the WFPSMP sites and was set up as follows. First stage: involved selection of 30 primary sampling units (PSUs) i.e. schools, across 104 the six selected counties (Baringo, Garissa, Turkana, Mandera, West Pokot, and Wajir).31 Using probability proportionate to size (PPS) method, the 30 PSUs were distributed across the six counties. Selection of schools within counties was done using simple random sampling, with application of a random number generator. Second stage: involved the selection of secondary sampling units (SSUs) which were children ages 7-13 years in class 3 to 8, across the thirty selected schools. Distributionof school specific sample size allocation was done across gender and school grade using PPS, where gender specific samples across school grade were drawn. Selection of children within gender and across school grade was done using simple random sampling, with application of a random number generator. Data collection a) Desk research The desk research consisted of two sets of work: a documentation review, supplemented by key informant interviews. Key informant (KI) interviews used semi-structured guidelines to collect information on the key roles of the various stakeholders in the intervention, their views on the policy, institutional and operational context, and their views regarding how it could be improved further, lessons learned and the potential for sustainability of the school feeding programme going forward. The respondentsincluded a selection of WFP staff, implementing partners, donors, and education officials. The key informant interviews were done after the data collection in the schools. The second part of the desk research used secondary data sets from WFP and the Ministry of Education to establish the baseline for key indicators in the monitoring framework for which primary data was not collected. b) Tool development, and School Level Data Collection The tools that were developed and used in the English Language. The team used real time digital data collection for four of the instruments. This was supplemented by manual data registration and audio recording for the focus group discussion in schools. A Global Positioning System (GPS) picking capability was integrated into the mobile/electronic version of the data collection script. This allowed for the tracking of interviewers to ensure that data collection was indeed carried out at the sampled sites. Teams of enumerators were gender balanced to ensure that interviews with girl pupils could be done by female enumerators to the extent possible. Each team of enumerators was headed by a supervisor. In addition to overseeing the data collection process and quality assurance the supervisors also provided technical guidance to the teams and did any trouble shooting on technology. Selection of 31 Isiolo, Nairobi, Samburu, and Tana River counties were excluded from the HGSMP group for the following reasons. Nairobi was excluded because of urban context issues. The majority of the counties of focus are in the arid, rural areas, consequently, there were hardly any common contextual similarities that will match Nairobi with them. The other three have been beneficiaries of the Cash Transfers to schools Model developed and implemented by WFP before being handed over to HGSMP – consequently their evolution modality and short history of the same does not approximate to a pure HGSMP modality of government that has been going on in some of the counties selected since 2009. 105 Data collection was done by a total of 88 enumerators. Enumerator training was done by the evaluation team to ensure independence and took place over a period of five days. Training included rigorous pre-testing of tools in the field, allowing for the tools to be revised prior to use. Enumerators were selected using detailed criteria established at the inception phase (see IR), were from the regions covered by the study and had the capacity to translate each item into Kiswahili and the local language. A debriefing took place after each day of field data collection. In addition, the consultant team was mobilized and carried out data collection spot-checks in all school during the two-week data collection process. The key respondents at the school level were the head teacher, selected class teachers, learners (grade 3-8) and their parents, cooks, and representatives of the Parent Teacher Associations (PTA) and the School Board of Management (BOM). These were selected as follows: • The head teacher was automatic selection • A school committee members were identified based on the lists of members at the schools and was preferably the chairperson and a PTA representative available in the school. • Pupils were selected from each class. The number of –girls and boys was pegged on attendance on that day. • A sample of parents per school – Equal numbers of male and female parents were selected for each school to correspond to the selected pupils. There was one parent for each child. • A cook and a store keeper was selected automatically in the schools where they are available. Both male and female cooks were covered. The following tools were used for primary data collection: a. A School Audit tool - Focused on establishing a baseline of the conditions in the school with respect to facilities including kitchens, water supply, latrines and school gardens. b. A parent-pupil data collection tool for grades 3 to 8 – was one continuous tool responded to first by the parent of the child and then by the child itself (without the parent present. The tool examined parents’ awareness of the value of education, and views on the barriers to enrolment, participation and learning, situation at home in terms of asset ownership (productive and non-productive), agricultural land holding and land tenure system, issues of food security, nutrition, siblings and whether these go to school, and hygiene. From the pupil’s perspective, the tool examined issues affecting enrolment, attendance, attentiveness, the importance of education, knowledge of nutrition and hygiene, and importantly also included the UWEZO a numeracy and literacy test. focusing on literacy and health and nutrition. c. A head teacher data collection tool - covered the head teacher perspectives on enrolment, attendance, retention and learning achievement., challenges and barriers in school access, to materials, and supplies; priority materials for teaching and learning to improve literacy and numeracy. The items included assessing gaps in skills and knowledge of school administration; as well as 106 support to the school feeding programme. d. A data collection tool for class teachers focusing on teachers in grades 3 to 8 - focused on issues affecting enrolment, attendance and educational achievement. It covered issues of teacher attendance, and hygiene and nutrition. The tool also served to identify the percentage of learners that are inattentive on a given day (using a spectrum from attentive – to inattentive) and to probe the reasons for this. e. A focus group discussion (FGD) guideline for a focus group with the PTA, including parents, and teachers – This served to gain in-depth insight into the perception of teachers, parents and PTA members of the issues behind poor enrolment, attendance and retention. It also explored the role of school feeding and other measures which may impact on performance of pupils. f. A FGD guideline for a focus group with pupils – served to gain insights into learner perspectives on enrolment, attendance and retention and explore views on the role of school feeding and other measures which may impact on performance of pupils. Ethical considerations in the study • Enumerator training included a substantial training on the ethical considerations for conducting surveys in schools, in particular with the pupils. • A courtesy call was made to the county district education official before starting the activity • The head teacher consented to the study before any activity was undertaken inthe school • The teachers introduced the enumerators to the class to explain the purpose of the exercise. • Participation was voluntary and all participants were told that they could opt not to participate and could discontinue the interview at any time without any repercussions. All participants were thanked at the end of the data collection. • Consent was sought from teachers, pupils and parents. Parents were interviewed prior to the interviews of their respective children so that consent could be sought for the interviews with the children. • All responses were coded and the individual performance of students was not traceable to the student or shared with the participants. Data analysis Data analysis was done using IBM SPSS version 24.0. MS-Excel was used to generate graphical presentation of specific findings. Univariate analysis: Descriptive statistics such as measures of central tendency (mean, standard deviations) were used for analysis of continuous variables, while frequencies and percentages were used for categorical variables. Bivariate analysis: Pearson’s Chi-square or Fisher Exact test was used to compare the 107 distribution of indicator variables and other observable characteristics between interventions and control groups. T-test were used to compare mean difference between interventions and control groups. Where normality assumptions were violated, appropriate non-parametric methods were used. Multiple regression analysis: Binary logistic regression was used to estimate the difference in the proportion of children ages 7-13 that have attained literacy and numeracy for a Standard 2 level adjusting for baseline characteristics, identified to be significantly different between interventions and control groups at bivariate analysis. Threshold for statistical significance was set at p<0.05. Estimation of programme effects: The programme effect will be measured at midterm and final evaluations. Difference-in-differences (DID), also known as the ‘double difference’ method, will be used to compare changes in outcome (effect size) over time between specific intervention (HGSMP and WFPSMP) and control group. Applying the DID method removed the difference in the outcome between both interventions (HGSMP and WFPSMP) and control group at baseline. Effect of WFPSMP: To identify the effects of WFPSMP at midterm and final evaluation, the difference in the measurement indicator between WFPSMP and control groups will first be calculated at baseline, midterm and final evaluation. The calculated baseline difference will then be differenced from the midterm and final evaluation differences to ascertain the accurate difference attributable to the WFPSMP at midterm and final evaluation. Evaluating sustainability of SMP: To determine whether transitioning schools from WFPSMP to HGSMP sustains school performance, the comparison of HGSMP and WFPSMP was done. The indicators were measured and compared at baseline, and this will also be done at midterm and final evaluation. Owing to its rigorous programme implementation, the bench mark will be WFPSMP. Propensity score matching was used in adjusting for differences in distribution of characteristics at baseline. A similar approach will be used during midterm and final evaluation. Strengths and limitations of propensity score match in the study The PSM was able to balance between the treatments (WFPSMP and HGSMP) and control on several identified covariates without losing observations however, none observed factors that affected assignment to either treatment or control could not be accounted for. Annex 3 – Overview of coordination mechanisms for school feeding in Kenya Source: Republic of Kenya (MOE, MoH, MoAL &F) School Nutrition and Meals Strategy for Kenya (Draft 2016) pg. 40 107 109 Annex 5 – Types of school feeding in Kenya and overview of key coordination structures and stakeholders 1. This annex provides additional background information to supplement the baseline of the institutional context for school feeding. Different school meals initiatives in Kenya - overview 2. There are several types of school feeding programmes operational in Kenya born out of innovations that seek to localize and contextualize the school meals programmes to enhance their effectiveness and sustainability. These include: The WFPSMP Regular School Meals Programme (SMP) 3. WFP and the Ministry of Education in Kenya have since the 1980s been jointly implementing schools meals programme targeting the mostly food insecure counties. The target counties are characterized by low school enrolment and completion rates and high gender disparities. Consequently, the Regular programme involves the physical distribution of food commodities to schools; the bulk of the food is imported though there are certain cases when food is procured locally especially cereals. It is implemented with support of World Food Programme (WFP). This final phase will entail providing hot lunch to 358,000 learners in 6 arid counties in the country; and, supporting the hand over process to government through training, joint missions and exchange of staff to build national capacity in procurement, data collection, reporting, monitoring, evaluation and programme management. After over three decades of collaboration and programming for school meals programme between WFP and the Ministry of Education, transition to the government run school meals programme is due for completion in 2019. Home Grown School Meals Programme: 4. To ensure long term sustainability of the School Meals Programme there was need to move from donor supported school meals programmes to a nationally supported programme. Therefore, the idea of a Home Grown School Meals (HGSMP) was conceived in 2009 as a government led programme. It was subsequently agreed that 50,000 pupils will be offloaded annually from the Regular SMP to the Home Grown SMP as part of the transition arrangement. It entails cash transfers by government to schools. The amount of money disbursed depends on the school enrolment and funds available. Disbursement is done as per the number of learning days in a term. The programme started off with a beneficiary level of 538,000 children in 1,777 schools in 66 semi-arid districts. By 2011 it had reached a beneficiary level of 592,638 children in approximately 1,800 schools in 72 semi-arid districts. By 2016, 950,000 children had been transitioned to this programme. Going forward, the programme has been working at strengthening links with smallholder farmers to enhance local agricultural production so more food for the meals can be purchased locally. This is key to the scale up and sustainability of the initiative. The initiative represents multi sectoral understanding and support that brings together various government ministries 110 (Education, health, water and irrigation etc.) development partners and other key stakeholders. The Government of Kenya is demonstrating leadership in this initiative regionally. However, a key concern that was raised from the qualitative data is that, that actual implementation of HGSMP is limited by resources and that there has not been a commensurate scale up of financial resources to match the growing number of pupils. As a result, in 2016 school feeding was only provided on 70 days out of 190(37% of the school days) compromising the quality of programme and a worsening situation since, the HGSMP review of 2012 had found that during the programme’s first three years (2009-2012), school meals had only been provided on about 54 percent of school days. “Njaa Marufuku Kenya” (Eradication of Hunger in Kenya): 5. This was an initiative of the Ministry of Agriculture that was started in 2005 targeting areas of high poverty that have high and medium potential to grow food yet have high levels of school dropouts, poor primary performance and high levels of malnutrition. It is one of the HGSMP models which attempts to achieve the dual objectives of increasing national food production and ensuring children go to school. By 2012 it was targeting 44, 229 children in 66 schools across the country. Cash Transfers to schools: 6. This was an innovation in the long-established WFP school meals programme in Kenya. The objective of the cash transfer programme was to improve the educational attainment of school children in the most disadvantaged arid areas of the country while paving the way for more appropriate and accountable government owned school meals model for these northern counties. The process of transitioning the schools in the Arid Counties from WFP support is challenging. This is due to poor infrastructure, remoteness, vast terrain and lower agricultural potential that translates into volatile and elevated food prices. The transition strategy for the arid counties from WFP support to HGSMP therefore: o Calls for the need to draw lessons from the HGSMP in the Semi-arid counties and identifying strategies for enhancing and adopting the programme design and implementation processes in the arid areas. o Should tap into the WFP/GoK developed strategy that seeks to inform the sustainable expansion of the HGSMP into the arid lands. 7. In actualizing the latter, the transitional cash to schools model was piloted in Isiolo County, expanded into Samburu, Tana River and Marsabit as a means towards ensuring suppliers and the community are prepared for their roles in the programme. The pilot and expansion process entailed building the capacity of farmers and traders in the HGSMP so they could increasingly sell food to schools and other institutional markets. The approach defined a clear exit strategy that could enhance the sustainability of the HGSMP through cash transfers to schools. However, the qualitative data from KIIs indicate that the programme in the Arid counties that have since been transitioned to government are beleaguered by budgetary constraints (insufficient funds) on one hand and late disbursements of even the little that is available. Are there lessons to be learnt about transitioning arid lands to HGSMPthat this phase of WFP/USDA/MGD could draw on as it seeks to transition 6 arid counties 111 at ago? Has a clear exit strategy that will ensure sustainability been clearly thought through and laid out? Community supported initiatives 8. There are other school meals initiatives undertaken by communities and school authorities without the support of county or national government. These initiatives are not regular. They only take place when there is a good harvest. The salient characteristics of these ad hoc school meals initiatives include the fact that; parents contribute money for school meals as part of the school fees per 3-month school term or with cash or food donations in kind; the menu is in most cases boiled maize and beans; and, School Meals Committees buy the food and make arrangements for cooking and serving to students. Some support is received from private and non-profit sectors. Complementary Initiatives: 9. a) School Deworming Initiatives: Over 5 million school-age children in Kenya are at risk of intestinal parasitic worms, including soil-transmitted helminthes (STH) and schistosomiases. It is in view of the negative impact such worms would have on the children’s health and education that the Government of Kenya launched the National School Based Deworming Programme(NSBDP) in 2009, upon which 3.6 million children were dewormed. The programme is implemented by the Ministry of Education Science and Technology (MOE)in collaboration with the Ministry of Health (MoH) with technical assistance from the Deworm the World Initiative (DtWI) at Evidence Action. The goal of the programme is to eradicate parasitic worms as a public health problem in Kenya. It seeks to treat at least 5 million Kenyan children each year for at least five years (2012-2016) in all primary schools in areas endemic for parasitic worms according to WHO criteria. The programme is complementary to the school meals in that it seeks to improve the health and education status of children and secure Kenya’s future. It has since been proven that regularly providing deworming tablets to children through schools is a cost-effective treatment strategy that is readily available and sustained educational infrastructure. In view of this, WHO has certified the safety of the administration of deworming tablets by teachers with support from the local health personnel. The key elements of the programme’s success include facilitating teacher trainings, distributing deworming tablets to schools, managing community sensitization activities and monitoring deworming activities. So far, the initiative has proven that deworming programmes reduce school absenteeism and it is cheaper than other ways of increasing school participation. 10. b) Nutrition interventions: Malnutrition is a significant health problem in Kenya. Micronutrient deficiencies are widespread and are exacerbated by low consumption of vitamin A- and iron-rich foods. Today in Kenya, an estimated 2.1 million children are stunted which is a serious national development concern as these children will never reach their full physical and mental potential. Regional disparities in nutrition indicators in Kenya are significant with North Eastern province having the highest proportion of children exhibiting severe wasting (8%) while Eastern province has highest level of stunted children (44%). Consequently, Kenya made a commitment to accelerate reduction of malnutrition by signing up to the global SUN movement in November 2012 as the 30th country member. Scaling Up Nutrition (SUN) is a unique 112 movement founded on the principle that all people have a right to food and good nutrition.It unites people from governments, civil society, the United Nations, donors, businesses & researchers in a collective effort to improve nutrition. As a consequent of research evidence and such initiatives, the statistics on the nutrition situation in Kenya caught the attention of local manufacturing companies who started drafting strategies towards contributing to the health sector through food fortification. Mumias Sugar Company, Unga limited, Tetrapak, the GAIN initiative among others led the way towards fortifying their products – sugar, flour and milk with requisite micronutrients. To further enhance commitment in this area, on 27 April 2017, the Government of Kenya unveiled a new 6-year partnership between the Ministry of Health and Jomo Kenyatta University of Agriculture and Technology (JKUAT) as well as private sector partners to strengthen and address gaps in food fortification. The project, funded by the European Union, focuses on improving the capacity of manufacturers to fortify maize flour and other staples consumed by poor households where the levels of malnutrition are higher. b) The School Meals Program Governance and Operational Framework. 11. Effective implementation of the school meals programmes including achievement of goals and objectives requires broad-based multi-sectoral coordination. There is need for strong governance and institutional arrangements anchored in clearly share responsibilities and accountability protocols. The Kenyan institutional framework for such coordination and management is shown in the chart below. 12. Table 5 outlines the shared responsibilities for the implementation of the school meals initiatives across ministries and non–governmental stakeholders. The need for an effective coordination mechanism across all actors involved in the programme cannot be emphasized enough. Located at different levels in the structure they are critical in the effective implementation of the SNM programmes under different modalities, monitoring and of improvement in strategy. The qualitative data, confirmed the need for a national steering committee that brings together the key stakeholders and provides a stronger anchoring and coordination role at national level. Governance and Coordination Institutions and structures Structure Role Governance Structures National Committee • Development of mechanisms to coordinate with the rest of government fora on school meals. School Nutrition and meals coordination unit at MOE • Convening and promoting Inter￾Ministerial and Inter agency dialogue including information sharing. • Sectoral planning and budgeting • Hosting on-going capacity building and leadership nurturing activities for 113 effective implementation and monitoring of school meals strategy. School Nutrition and Meals committee • Coordinating discussions of the SNMP implementation issues that bring together various ministries – MOE, MoH, Treasury etc. County Level Committees • Defining County levelimplementation strategy including roles to be played by school level committees • Budgeting at County level￾actualization of decentralization/devolution School Meals programme Committee • A structure that is so critical for the implementation of the programme at the school level – it should be established at every school constituting of a chair, procurement supervisor, food quality supervisor and reporting supervisor (could be expanded based on need. Roles and Responsibilities of Institutions involved in programme Governance National Government ( MOE) • Coordination and oversight of school meals interventions country wide • Integration and building of linkages with other ministries • Policy guidance and guidelines for implementation • capacity building opportunities • Promotion and enabling of participation by counties and other stakeholders in the development of policies and procedures. Agricultural Sector (MoA) • Developing capacities for increased production and access to markets by smallholder farmers Health Sector ( MoH) • Contribute nutritionists’ expertise to guide menu preparation that informs purchases • Ensure food quality at delivery and monitoring during storage • Training implementers in the SNMP food chain. NGO community • Implementing Initiatives at community level- school gardens etc. 114 • Providing voice for the un-empowered on rights and interests of the disadvantaged (children, women and smallholder farmers). • Credit provision • Building skills among smallholder farmers • Establishment of cooperatives and other farmer based organizations. Private Sector • Creating income earning opportunities in agricultural and nonagricultural sectors. • Encouraging agricultural commercialization and agro- business • Promoting new farming practices through research. Academic and Research Institutions • Providing evidence for policy and programme improvement. Development Partners • Support – funding and technical to SNMP initiatives through working closely with governments, civil society and communities. County Governments • Extending participation in SNMP beyond ECD. Sub- County level • Implementation of SNMP interventions with oversight by the county level. Local Committees • Their roles are dependent on the prevailing contexts which maybe different county to county. 115 Funding to the WFP SMP between 2014 and 2017 (Source: WFP data provided to the Evaluation Team) 116 Annex 6: Summary of Baseline Indicator Values Table 6a - Summary of baseline values for WFPSMP Result Indicator WFPSMP Schools Improved Literacy of School-Age Children Proportion of 7-13 year olds that can solve Class 2 numeracy and literacy problems English: 42.8% Kiswahili: 52.35% Numeric: 60.5% Number of individuals benefiting directly from USDA￾funded interventions 0 Number of interventions individuals benefiting indirectly from USDA-funded 0 Number of social assistance beneficiaries participating in productive safety nets as a result of USDA assistance 0 Increased access to food (school feeding) Number of daily school meals (breakfast, snack, lunch) provided to school￾age children as a result of USDA assistance 0 Number of school-aged children receiving daily school meals (breakfast, snack, lunch) as a result of USDA assistance 0 Percent of students in target schools who regularly consume a meal before the school day • Daily before going to school (last 1 Week): 35.8% • Had a meal on interview day before going to school: 59.75 Percent of students in target schools who regularly consume a meal during the school day • Has been receiving school meals in the current school year 117 Result Indicator WFPSMP Schools (2017): 57.45% • The school in which the child was learning at currently (same week) serving food: 47.8% Improved student attendance Number of students classrooms/schools regularly (80%) attending USD supported 252,90632 Improved Attentiveness Percent of students in classrooms identified as inattentive by their teachers 39.25% Increased Community Understanding of Benefits of Education Percent of parents in target communities who can name at least three benefits of primary education (Disaggregated Male and Female) 57.3% Increased Capacity of Government Institutions Number of county-level inter-ministerial committees for HGSMP established 0 Number of national-level inter-ministerial coordination committees for HGSMP established 0 Increased Government Support Number of Parent-Teacher Associations (PTAs) or similar “school” governance structures supported as a result of USDA assistance 0 Number of public-private partnerships formed as a result of USDA assistance 0 32 This is based on the actual number of students enrolled in the school receiving USDA assistance and an 85% regular attendance rate from the WFP SMP schools sampled during the study. 118 Result Indicator WFPSMP Schools Number of school administrators and officials in target schools who demonstrate use of new techniques or tools as a results of USDA assistance 0 Number of school administrators and officials trained or certified as a result of USDA assistance 0 Value of new public and private sector investments leveraged as a result of USDA assistance 0 Improved Policy and Regulatory Framework Number of child health and nutrition policies, regulations, and/or administrative procedures in each of the following stages of development as a result of USDA assistance (Stage 5) 0 Number of educational policies, regulations, and/or administrative procedures in each of the following stages of development as a result of USDA assistance (Stage 5) 0 Increased Use of Health and Dietary Practices Percent of schools in target counties that store food off the ground 50.0% Increased Knowledge of Safe Food Prep and Storage Practices Percent of food preparers at target schools who achieve a passing score on a test of safe food preparation and storage 33.7% Increased knowledge of nutrition Number of individuals trained in child health and nutrition as a result of USDA assistance 0 Improved school infrastructure Number of educational facilities (e.g. school buildings, classrooms and latrines) rehabilitated and constructed as a result of USDA assistance 0 119 Result Indicator WFPSMP Schools Increased Student Enrolment Number of students enrolled in schools receiving USDA assistance 297,53633 Increased Access to Requisite Food Prep and Storage Tools Number of target schools with increased access to improved food prep and storage equipment (kitchens, storerooms, stoves, kitchen utensils) 0 Table 6 b - Summary of baseline values for the Three Arms: WFPSMP, Control and HGSMP Schools Result Indicator Baseline Indicator Value WFPSMP Schools Control Schools HGSMP Schools WFPSMP Schools Improved Literacy of School-Age Children Proportion of 7-13 year olds that can solve Class 2 numeracy and literacy problems English: 40.6% Kiswahili: 51.2% Numeric: 60.9% English: 55.6% Kiswahili: 66.0% Numeric: 73.5% English: 64.6% Kiswahili: 74.9% Numeric: 77.7% English: 45.0% Kiswahili: 53.5% Numeric: 60.1% Number of individuals benefiting directly from USDA-funded interventions 0 0 0 353,000 Number of individuals benefiting indirectly from USDA-funded interventions 0 0 0 0 Number of social assistance beneficiaries participating in productive safety nets as a result of 0 0 0 0 33 This number includes children receiving meals in six arid counties and Marsabit where 44,100 children benefited from nutrition and hygiene related activities. It is important to note that the number of students enrolled in the schools sampled for the survey were 7142 (min=28, max=1524, mean= 317, SD=241) 120 Result Indicator Baseline Indicator Value WFPSMP Schools Control Schools HGSMP Schools WFPSMP Schools USDA assistance Increased access to food (school feeding) Number of daily school meals (breakfast, snack, lunch) provided to school-age children as a result of USDA assistance 0 0 0 0 Number of school-aged children receiving daily school meals (breakfast, snack, lunch) as a result of USDA assistance 0 0 0 0 Percent of students in target schools who regularly consume a meal before the school day • Daily before going to school (last 1 week): 33.0% • Daily before going to school (last 1 week): 38.0% • Daily before going to school (last 1 week): 43.2% • Daily before going to school (last 1 week): 38.7% • Had a meal on interview day before going to school: 54.9% • Had a meal on interview day before going to school: 63.6% • Had a meal on interview day before going to school: 67.6% • Had a meal on interview day before going to school: 64.6% Percent of students in target schools who regularly consume a meal during the school day • Has been receiving school meals in the current school year (2017): 59.3% • Has been receiving school meals in the current school year (2017): 20.0% • Has been receiving school meals in the current school year (2017): 80.4% • Has been receiving school meals in the current school year (2017): 55.6% • The school in which the child was learning at • The school in which the child was learning at • The school in which the child was learning at • The school in which the child was learning at 121 Result Indicator Baseline Indicator Value WFPSMP Schools Control Schools HGSMP Schools WFPSMP Schools currently (same week) serving food: 51.7% currently (same week) serving food: 16.3% currently (same week) serving food: 51.5% currently (same week) serving food: 43.9% Improved student attendance Number of students regularly (80%) attending USDA supported classrooms/schools 252,906 N/A N/A 252,906 Improved Attentiveness Percent of students in classrooms identified as inattentive by their teachers 41.1% 46.4% 43.5% 37.4% Increased Community Understanding of Benefits of Education Percent of parents in target communities who can name atleast three benefits of primary education (Disaggregated Male and Female) 57.3% 26.1% 30.0% 57.3% Increased Capacity of Government Institutions Number of county-level inter￾ministerial committees for HGSMP established 0 0 0 0 Number of national-level inter￾ministerial coordination committees for HGSMP established 0 0 0 0 Increased Government Support Number of Parent-Teacher Associations (PTAs) or similar “school” governance structures supported as a result of USDA assistance 0 0 0 0 122 Result Indicator Baseline Indicator Value WFPSMP Schools Control Schools HGSMP Schools WFPSMP Schools Number of public-private partnerships formed as a result of USDA assistance 0 0 0 0 Number of school administrators and officials in target schools who demonstrate use of new techniques or tools as a results of USDA assistance 0 0 0 0 Number of school administrators and officials trained or certified as a result of USDA assistance 0 0 0 0 Value of new public and private sector investments leveraged as a result of USDA assistance 0 0 0 0 Improved Policy and Regulatory Framework Number of child health and nutrition policies, regulations, and/or administrative procedures in each of the following stages of development as a result of USDA assistance (Stage 5) 0 0 0 0 Number of educational policies, regulations, and/or administrative procedures in each of the following stages of development as a result of USDA assistance (Stage 5) 0 0 0 0 Increased Use of Health and Dietary Practices Percent of schools in target counties that store food off the ground 56.5% 17.4% 73.9% 43.5% 123 Result Indicator Baseline Indicator Value WFPSMP Schools Control Schools HGSMP Schools WFPSMP Schools Increased Knowledge of Safe Food Prep and Storage Practices Percent of food preparers at target schools who achieve a passing score on a test of safe food preparation and storage 33.6% 32.1% 39.7% 33.8% Increased knowledge of nutrition Number of individuals trained in child health and nutrition as a result of USDA assistance 0 0 0 0 Improved school infrastructure Number of educational facilities (e.g. school buildings, classrooms and latrines) rehabilitated and constructed as a result of USDA assistance 0 0 0 0 Increased Student Enrolment Number of students enrolled in schools receiving USDA assistance 297,536 N/A N/A 297,536 Increased Access to Requisite Food Prep and Storage Tools Number of target schools with increased access to improved food prep and storage equipment (kitchens, storerooms, stoves, kitchen utensils) 0 0 0 0 Variables Boys (n=2558) Girls (n=2572) Literate (n=1310) n % Not literate (n=1248) n % OR 95% CI Lowe Uppe r r p value Literate (n=1382) n % Not literate (n=1190) n % OR 95% Lowe r Class of the child 84.9 Third 66 15.1% 372 % 1.00 95 20.7% 364 79.3% 1.00 <0.00 68.5 Fourth 137 30.9% 306 69.1% 2.52 1.81 3.51 1 146 31.5% 318 % 1.76 1.31 54.0 <0.00 Fifth 217 46.0% 255 % 4.80 3.49 6.59 1 191 44.6% 237 55.4% 3.09 2.30 <0.00 34.2 Sixth 277 60.9% 178 39.1% 8.77 6.35 12.11 1 289 65.8% 150 % 7.38 5.47 22.9 <0.00 82.9 Seventh 341 77.1% 101 % 19.03 13.50 26.82 1 363 % 75 17.1% 18.55 13.26 42.5 <0.00 86.6 24.8 Eighth 272 88.3% 36 11.7% 9 27.56 65.80 1 298 % 46 13.4% 2 16.91 Mode of travel to school 128 118 46.6 On foot 0 50.9% 1237 49.1% 1.00 1357 53.4% 4 % 1.00 26.8 80.6 Bicycle/car/motor cycle 30 73.2% 11 % 2.64 1.32 5.28 0.006 25 % 6 19.4% 3.64 1.49 Number of times child normally eat per day 52.4 2 times or less 742 47.4% 825 52.6% 1.00 757 47.6% 833 % 1.00 <0.00 36.4 3 times or more 568 57.3% 423 42.7% 1.49 1.27 1.75 1 625 63.6% 357 % 1.93 1.64 Child had a meal today before going to school <0.00 Yes 897 54% 759 46% 1.40 1.19 1.65 1 915 57% 700 43% 1.37 1.17 No 413 46% 489 54% 467 49% 490 51% Child thought it is important to go to school 48.2 <0.00 136 45.8 Yes 1297 51.8% 1208 % 3.30 1.76 6.21 1 8 54.2% 1155 % 2.96 1.59 28.6 No 13 24.5% 40 75.5% 1.00 14 % 35 71.4% 1.00 The child brothers and sisters who currently study in this school 48.2 Yes 870 49.3% 895 50.7% 1.00 956 51.8% 890 % 1.00 No 440 55.5% 353 44.5% 1.28 1.08 1.52 0.004 426 58.7% 300 41.3% 1.32 1.11 The child had brothers and sisters who are old enough to go to school but are NOT currently attending school 58.9 40.9 Yes 140 41.1% 201 % 1.00 139 % 201 59.1% 1.00 Did not complete primary school 239 48.5% 25 4 5 1 .5% 1 .16 0.94 1 . 4 4 0 . 1 6 5 277 54.5% 23 1 45 .5% 1 . 7 1 1 . 3 8 39.8 <0.00 Completed primary school 30 1 60 . 2 % 19 9 % 1 . 8 7 1 . 5 1 2 .31 1 358 63.3% 20 8 36 .7% 2 . 4 5 1 . 9 9 43.8 33.3 Did not compete secondary 63 56.3% 49 % 1 . 5 9 1 . 0 7 2 .35 0 .02 1 84 66.7% 4 2 % 2 . 8 5 1 . 9 3 35.0 <0.00 Completed secondary school 10 2 65.0% 55 % 2 .29 1 . 6 2 3 .24 1 13 4 8 1 . 2 % 3 1 18 . 8 % 6 . 1 6 4.09 <0.00 Technical college/ university 66 67.3% 3 2 32 .7% 2 .55 1 . 6 4 3.95 1 68 73.9% 2 4 26 . 1 % 4.04 2 . 4 9 Number of important nutrition habits mentioned by the parent/guardian None 311 38.9% 48 9 61 . 1 % 1 . 0 0 359 42 .3% 489 57.7% 1 . 0 0 43 . 2 <0.00 10 2 40.7 One and above 999 56.8% 759 % 2 . 0 7 1 . 7 5 2 . 4 6 1 3 59.3% 70 1 % 1 . 9 9 1 . 6 8 Number of hygiene habits mentioned by the parent/guardian 66 . 2 64.3 None 71 33.8% 13 9 % 1 . 0 0 80 35.7% 14 4 % 1 . 0 0 50.4 <0.00 45 . 2 1 to 2 727 49.6% 739 % 1 . 9 3 1 . 4 2 2 . 6 1 1 808 54.8% 666 % 2.1 8 1 . 6 3 42.0 <0.00 3 and above 512 58.0% 370 % 2 .71 1 . 9 8 3 .72 1 494 56.5% 380 43.5% 2 .34 1 . 7 3 Variables in the Equation AOR Lower Upper value AOR L Class of the child Class 3 1.00 1.00 Class 4 2.45 1.74 3.44 <0.001 1.71 Class 5 4.62 3.33 6.41 <0.001 2.93 Class 6 8.36 6.00 11.65 <0.001 7.00 Class 7 18.22 12.77 25.98 <0.001 17.31 Class 8 42.72 27.31 66.82 <0.001 24.63 Mode of travel to school On foot 1.00 1.00 Bicycle/car/motor cycle 2.50 1.13 5.54 0.024 3.01 Number of times child normally eat per day 2 times or less 1.00 1.00 3 times or more 1.41 1.15 1.72 0.001 1.68 Child had a meal today before going to school Yes 1.24 1.01 1.52 0.037 1.12 No 1.00 1.00 Child thought it is important to go to school Yes 2.21 1.08 4.53 0.030 1.65 No 1.00 1.00 The child brothers and sisters who currently study in this school Yes 1.00 1.00 No 1.28 1.05 1.57 0.016 1.34 The child had brothers and sisters who are old enough to go to school but are NOT currently attending school Yes 1.00 1.00 No 1.34 1.01 1.76 0.040 1.36 Education level of the respondent Never attended school 1.00 1.00 Madrasa/Adult learning center 1.02 0.54 1.92 0.959 0.91 Did not complete primary school 1.14 0.88 1.47 0.327 1.37 Completed primary school 1.81 1.39 2.35 <0.001 2.20 Did not compete secondary 1.65 1.04 2.62 0.035 2.36 Completed secondary school 2.03 1.35 3.05 0.001 4.64 Technical college/ university 1.96 1.18 3.26 0.010 2.77 Number of important nutrition habits mentioned by the parent/guardian None 1.00 1.00 At least 1 1.20 0.96 1.49 0.108 1.34 Number of hygiene habits mentioned by the parent/guardian None 1.00 1.00 1 to 2 1.43 0.99 2.07 0.058 1.81 3 or more 1.95 1.31 2.91 0.001 1.84 Literate (n=1557) literate (n=1001) 95% CI Literate (n=1636) literate (n=936) Variables n % n % OR Lower Upper p value n % n % OR L Class of the child Class 3 97 22.1% 341 77.9% 1.00 135 29.4% 324 70.6% 1.00 Class 4 203 45.8% 240 54.2% 2.97 2.22 3.99 <0.001 202 43.5% 262 56.5% 1.85 Class 5 284 60.2% 188 39.8% 5.31 3.97 7.11 <0.001 258 60.3% 170 39.7% 3.64 Class 6 329 72.3% 126 27.7% 9.18 6.77 12.45 <0.001 345 78.6% 94 21.4% 8.81 Class 7 365 82.6% 77 17.4% 16.66 11.94 23.26 <0.001 388 88.6% 50 11.4% 18.62 Class 8 Mode of travel to school 279 90.6% 29 9.4% 33.82 21.70 52.72 <0.001 308 89.5% 36 10.5% 20.53 On foot 1524 60.5% 993 39.5% 1.00 1612 63.5% 929 36.6% 1.00 Bicycle/car/motor cycle 33 80.5% 8 19.5% 2.69 1.24 5.84 0.013 24 77.4% 7 22.6% 1.97 Number of times child normally eat per day 2 times or less 889 56.7% 678 43.3% 1.00 938 59.0% 652 41.0% 1.00 3 times or more Child had a meal today before going to 668 67.4% 323 32.6% 1.58 1.34 1.86 <0.001 698 71.1% 284 28.9% 1.71 school Yes 1027 62.0% 629 38.0% 1.15 0.97 1.35 0.107 1052 65.1% 563 34.9% 1.19 No 530 58.8% 372 41.2% 1.00 584 61.0% 373 39.0% 1.00 The child had brothers and sisters who are old enough to go to school but are NOT currently attending school Yes 171 50.1% 170 49.9% 1.00 174 51.2% 166 48.8% 1.00 No Education level of the respondent 1386 62.5% 831 37.5% 1.66 1.32 2.09 0.000 1462 65.5% 770 34.5% 1.81 Never attended school 614 53.8% 528 46.2% 1.00 544 50.9% 525 49.1% 1.00 Madrasa/Adult learning center 27 48.2% 29 51.8% 0.80 0.47 1.37 0.417 29 63.0% 17 37.0% 1.65 Did not complete primary school 306 62.1% 187 37.9% 1.41 1.13 1.75 0.002 342 67.3% 166 32.7% 1.99 Completed primary school 345 69.0% 155 31.0% 1.91 1.53 2.39 <0.001 408 72.1% 158 27.9% 2.49 Did not compete secondary 75 67.0% 37 33.0% 1.74 1.16 2.63 0.008 99 78.6% 27 21.4% 3.54 Completed secondary school 112 71.3% 45 28.7% 2.14 1.49 3.08 <0.001 142 86.1% 23 13.9% 5.96 Technical college/ university 78 79.6% 20 20.4% 3.35 2.02 5.56 <0.001 72 78.3% 20 21.7% 3.47 Number of hygiene habits mentioned by the parent/guardian None 96 45.7% 114 54.3% 1.00 113 50.4% 111 49.6% 1.00 1 to 2 891 60.8% 575 39.2% 1.84 1.38 2.46 <0.001 963 65.3% 511 34.7% 1.85 3 or more Quintiles for Coping Strategy Index (CSI) 570 64.6% 312 35.4% 2.17 1.60 2.94 <0.001 560 64.1% 314 35.9% 1.75 First 605 59.0% 421 41.0% 1.82 1.53 2.17 <0.001 306 56.9% 232 43.1% 1.81 Second 561 54.7% 465 45.3% 1.53 1.28 1.82 <0.001 271 52.7% 243 47.3% 1.53 Third 573 55.8% 453 44.2% 1.60 1.34 1.91 <0.001 269 54.6% 224 45.4% 1.65 Fourth 500 48.7% 526 51.3% 1.20 1.01 1.43 0.038 253 49.4% 259 50.6% 1.34 Variables in the Equation Uppe p AOR Lower r value AOR L Class of the child Class 3 Class 4 Class 5 Class 6 Class 7 Class 8 Mode of travel to school On foot Bicycle/car/motor cycle Number of times child normally eat per day 2 times or less 3 times or more Child had a meal today before going to school Yes No The child had brothers and sisters who are old enough to go to school but are NOT currently attending school Yes No Education level of the respondent Never attended school Madrasa/Adult learning center Did not complete primary school Completed primary school Did not compete secondary Completed secondary school Technical college/ university Number of hygiene habits mentioned by the parent/guardian None 1 to 2 3 or more Quintiles for Coping Strategy Index (CSI) First Second Third Fourth Fifth 1.00 2.86 2.11 3.86 <0.001 5.11 3.79 6.90 <0.001 8.62 6.30 11.80 <0.001 16.23 11.50 22.91 <0.001 34.29 21.79 53.95 <0.001 1.00 2.27 0.96 5.36 0.062 1.00 1.45 1.18 1.77 <0.001 1.63 0.86 3.08 0.135 1.00 1.00 1.30 0.99 1.71 0.058 1.00 0.58 0.31 1.09 0.091 1.28 1.00 1.65 0.053 1.70 1.30 2.21 <0.001 1.72 1.07 2.75 0.025 1.73 1.13 2.63 0.011 2.65 1.50 4.68 0.001 1.00 1.51 1.08 2.12 0.016 1.77 1.24 2.53 0.002 1.51 1.11 2.05 0.008 1.10 0.82 1.48 0.512 1.54 1.14 2.08 0.005 1.10 0.82 1.46 0.523 1.00 1.00 1.82 3.52 8.61 17.23 19.99 1.00 1.49 1.00 1.41 1.66 1.00 1.00 1.34 1.00 1.58 1.55 2.11 3.01 4.42 2.33 1.00 1.67 1.50 1.06 1.11 1.22 0.92 1.00 Variables Numerate (n=1789) n % Numerate (n=769) n % OR 95% CI Lower Upper p value Num (n= n erate 1785) % Numerate (n=787) n % OR Class of the child Class 3 170 38.8% 268 61.2% 1.00 156 34.0% 303 66.0% 1.0 Class 4 244 55.1% 199 44.9% 1.93 1.48 2.53 <0.001 258 55.6% 206 44.4% 2.4 Class 5 341 72.2% 131 27.8% 4.10 3.11 5.42 <0.001 310 72.4% 118 27.6% 5.1 Class 6 367 80.7% 88 19.3% 6.58 4.86 8.89 <0.001 357 81.3% 82 18.7% 8.4 Class 7 382 86.4% 60 13.6% 10.04 7.19 14.01 <0.001 381 87.0% 57 13.0% 12.9 Class 8 285 92.5% 23 7.5% 19.54 12.25 31.14 <0.001 323 93.9% 21 6.1% 29.8 Mode of travel to school On foot 1754 69.7% 763 30.3% 1.00 1758 69.2% 783 30.8% 1.0 Bicycle/car/motor cycle 35 85.4% 6 14.6% 2.54 1.06 6.06 0.036 27 87.1% 4 12.9% 3.0 Number of times child normally eat per day 2 times or less 1059 67.6% 508 32.4% 1.00 1048 65.9% 542 34.1% 3 times or more 730 73.7% 261 26.3% 1.34 1.13 1.60 0.001 737 75.1% 245 24.9% 1.5 Child thought it is important to go to school Yes 1764 70.4% 741 29.6% 2.67 1.54 4.60 <0.001 1765 70.0% 758 30.0% 3.3 No 25 47.2% 28 52.8% 1.00 20 40.8% 29 59.2% 1.0 Education level of the respondent Never attended school 739 64.7% 403 35.3% 1.00 664 62.1% 405 37.9% 1.0 Madrasa/Adult learning center 37 66.1% 19 33.9% 1.06 0.60 1.87 0.835 31 67.4% 15 32.6% 1.2 Did not complete primary school 341 69.2% 152 30.8% 1.22 0.98 1.53 0.081 346 68.1% 162 31.9% 1.3 Completed primary school 375 75.0% 125 25.0% 1.64 1.29 2.07 <0.001 424 74.9% 142 25.1% 1.8 83.0 Did not compete secondary 93 % 19 17.0% 2.67 1.61 4.44 <0.001 103 81.7% 23 18.3% 2.7 Completed secondary school 123 78.3% 34 21.7% 1.97 1.32 2.94 0.001 141 85.5% 24 14.5% 3.5 Technical college/ university 81 82.7% 17 17.3% 2.60 1.52 4.44 <0.001 76 82.6% 16 17.4% 2.9 Number of important benefits of education mentioned by the parent/guardian 2 or less 1048 71.9% 410 28.1% 1.00 1071 70.3% 453 29.7% 1.0 3 or more 741 67.4% 359 32.6% 0.81 0.68 0.96 0.014 714 68.1% 334 31.9% 0.9 Number of important nutrition habits mentioned by the parent/guardian None 471 58.9% 329 41.1% 1.00 521 61.4% 327 38.6% 1.0 At least 1 1318 75.0% 440 25.0% 2.09 1.75 2.50 <0.001 1264 73.3% 460 26.7% 1.7 Number of hygiene habits mentioned by the parent/guardian None 110 52.4% 100 47.6% 1.00 119 53.1% 105 46.9% 1.0 1 to 2 1039 70.9% 427 29.1% 2.21 1.65 2.97 <0.001 1038 70.4% 436 29.6% 2.1 3 or more 640 72.6% 242 27.4% 2.40 1.77 3.27 <0.001 628 71.9% 246 28.1% 2.2 Variables in the Equation AOR Lower Upper value Lower Upper Class of the child Class 3 1.00 1.00 Class 4 1.80 1.36 2.37 <0.001 2.41 1.83 3.17 Class 5 3.86 2.90 5.14 <0.001 5.06 3.76 6.81 Class 6 6.06 4.45 8.26 <0.001 8.15 5.93 11.21 Class 7 9.15 6.50 12.89 <0.001 11.95 8.41 16.97 Class 8 17.69 11.01 28.42 <0.001 29.03 17.75 47.48 Mode of travel to school On foot 1.00 1.00 Bicycle/car/motor cycle 2.03 0.79 5.19 0.139 2.85 0.92 8.86 Number of times child normally eat per day 2 times or less 1.00 1.00 3 times or more 1.21 0.99 1.48 0.061 1.33 1.08 1.64 Child thought it is important to go to school Yes 1.76 0.95 3.25 0.072 2.03 1.05 3.91 No 1.00 1.00 Education level of the respondent Never attended school 1.00 1.00 Madrasa/Adult learning center 0.96 0.51 1.82 0.907 1.09 0.55 2.17 Did not complete primary school 1.09 0.84 1.41 0.539 0.93 0.71 1.21 Completed primary school 1.43 1.09 1.88 0.010 1.46 1.12 1.91 Did not compete secondary 2.87 1.65 4.98 <0.001 2.34 1.38 3.99 Completed secondary school 1.65 1.07 2.57 0.025 2.44 1.48 4.03 Technical college/ university 2.02 1.13 3.62 0.019 1.93 1.04 3.57 Number of important benefits of education mentioned by the parent/guardian 2 or less 1.00 1.00 3 or more 1.35 1.10 1.66 0.004 1.08 0.87 1.33 Number of important nutrition habits mentioned by the parent/guardian None 1.00 1.00 At least 1 1.47 1.18 1.83 0.001 1.12 0.90 1.40 Number of hygiene habits mentioned by the parent/guardian None 1.00 1.00 1 to 2 1.64 1.15 2.32 0.006 1.88 1.33 2.67 3 or more 1.66 1.13 2.45 0.010 1.74 1.18 2.57 Annex 8: Comparing distribution of specific variables between study arms stratified by gender of child Table 1a: Socio-demographic of parents/guardians distributed by CONTROL and WFPSMP stratified by gender of the child Variables Male (n=1254) Female (n=1286) Total (N=2540) CONTROL (n=675) WFPSMP (n=579) p value CON (n= TROL 721) WFPSMP (n=565) p value CON (n=1 TROL 396) WFPSMP (n=1144) p value Age of guardian in years <20 1 0.1% 3 0.5% 0.534 1 0.1% 2 0.4% 0.593 2 0.1% 5 0.4% 0.183 20 - 29 71 10.5% 70 12.1% 90 12.5% 82 14.5% 161 11.5% 152 13.3% 30 - 39 279 41.3% 227 39.2% 290 40.2% 208 36.8% 569 40.8% 435 38.0% 40 - 49 192 28.4% 177 30.6% 200 27.7% 169 29.9% 392 28.1% 346 30.2% 50 - 59 84 12.4% 70 12.1% 87 12.1% 69 12.2% 171 12.2% 139 12.2% 60 and above 48 7.1% 32 5.5% 53 7.4% 35 6.2% 101 7.2% 67 5.9% Gender of the guardian Male 195 28.9% 200 34.5% 0.032 171 23.7% 163 28.8% 0.037 366 26.2% 363 31.7% 0.002 Female 480 71.1% 379 65.5% 550 76.3% 402 71.2% 1030 73.8% 781 68.3% Relationship of guardian to the child Mother/Father 582 86.2% 486 83.9% 0.001 620 86.0% 458 81.1% 0.002 1202 86.1% 944 82.5% <0.001 Brother/Sister 11 1.6% 25 4.3% 19 2.6% 25 4.4% 30 2.1% 50 4.4% Uncle/Aunt 19 2.8% 24 4.1% 19 2.6% 34 6.0% 38 2.7% 58 5.1% Grand parent 49 7.3% 22 3.8% 45 6.2% 25 4.4% 94 6.7% 47 4.1% Guardian 14 2.1% 22 3.8% 18 2.5% 23 4.1% 32 2.3% 45 3.9% Guardian was the household head Yes 395 58.5% 354 61.1% 0.345 447 62.0% 348 61.6% 0.882 842 60.3% 702 61.4% 0.590 No 280 41.5% 225 38.9% 274 38.0% 217 38.4% 554 39.7% 442 38.6% Main occupation of the guardian Too old to work 18 2.7% 22 3.8% <0.001 14 1.9% 20 3.5% <0.001 32 2.3% 42 3.7% <0.001 Student 4 0.6% 2 0.3% 6 0.8% 2 0.4% 10 0.7% 4 0.3% Farmer 274 40.6% 40 6.9% 293 40.6% 49 8.7% 567 40.6% 89 7.8% Pastoralist 31 4.6% 87 15.0% 19 2.6% 87 15.4% 50 3.6% 174 15.2% Salaried employee 19 2.8% 27 4.7% 25 3.5% 12 2.1% 44 3.2% 39 3.4% Casual laborer 154 22.8% 60 10.4% 183 25.4% 57 10.1% 337 24.1% 117 10.2% Self-employed business 51 7.6% 42 7.3% 42 5.8% 56 9.9% 93 6.7% 98 8.6% Not currently working 110 16.3% 234 40.4% 115 16.0% 216 38.2% 225 16.1% 450 39.3% Others 14 2.1% 65 11.2% 24 3.3% 66 11.7% 38 2.7% 131 11.5% Education level of the guardian Never attended school 114 16.9% 447 77.2% <0.001 95 13.2% 454 80.4% <0.001 209 15.0% 901 78.8% <0.001 Madrasa/Adult learning center 0 0.0% 25 4.3% 0 0.0% 18 3.2% 0 0.0% 43 3.8% Did not complete primary school 206 30.5% 41 7.1% 228 31.6% 42 7.4% 434 31.1% 83 7.3% Completed primary school 228 33.8% 20 3.5% 252 35.0% 19 3.4% 480 34.4% 39 3.4% Did not compete secondary 37 5.5% 11 1.9% 58 8.0% 9 1.6% 95 6.8% 20 1.7% Completed secondary school 59 8.7% 17 2.9% 59 8.2% 11 1.9% 118 8.5% 28 2.4% Completed technical college 30 4.4% 15 2.6% 19 2.6% 12 2.1% 49 3.5% 27 2.4% Completed university/graduate school 1 0.1% 3 0.5% 10 1.4% 0 0.0% 11 0.8% 3 0.3% Table 1b: Socio-demographic of parents/guardians distributed by WFPSMP and HGSMP stratified by gender of the child Variables Male (n=1306) Female (n=1284) Total (N=2590) WFPSMP (n=593) HGSMP (n=713) p value WF (n= PSMP 541) HGSMP (n=743) p value WF (n= PSMP 1134) HGSMP (n=1456) p value Age of guardian in years <20 5 0.8% 2 0.3% 0.004 4 0.7% 2 0.3% 0.212 9 0.8% 4 0.3% 0.001 20 - 29 61 10.3% 77 10.8% 63 11.6% 75 10.1% 124 10.9% 152 10.4% 30 - 39 225 37.9% 297 41.7% 211 39.0% 325 43.7% 436 38.4% 622 42.7% 40 - 49 204 34.4% 183 25.7% 164 30.3% 202 27.2% 368 32.5% 385 26.4% 50 - 59 66 11.1% 91 12.8% 69 12.8% 84 11.3% 135 11.9% 175 12.0% 60 and above 32 5.4% 63 8.8% 30 5.5% 55 7.4% 62 5.5% 118 8.1% Gender of the guardian Male 211 35.6% 180 25.2% <0.001 156 28.8% 177 23.8% 0.043 367 32.4% 357 24.5% <0.001 Female 382 64.4% 533 74.8% 385 71.2% 566 76.2% 767 67.6% 1099 75.5% Relationship guardian to the child Mother/Father 487 82.1% 606 85.0% <0.001 434 80.2% 627 84.4% <0.001 921 81.2% 1233 84.7% <0.001 Brother/Sister 23 3.9% 12 1.7% 35 6.5% 14 1.9% 58 5.1% 26 1.8% Uncle/Aunt 29 4.9% 16 2.2% 27 5.0% 19 2.6% 56 4.9% 35 2.4% Grand parent 20 3.4% 55 7.7% 21 3.9% 56 7.5% 41 3.6% 111 7.6% Guardian 34 5.7% 24 3.4% 24 4.4% 27 3.6% 58 5.1% 51 3.5% Guardian was the household head Yes 371 62.6% 431 60.4% 0.435 326 60.3% 463 62.3% 0.455 697 61.5% 894 61.4% 0.974 No 222 37.4% 282 39.6% 215 39.7% 280 37.7% 437 38.5% 562 38.6% Main occupation of the guardian Too old to work 23 3.9% 11 1.5% <0.001 15 2.8% 9 1.2% <0.001 38 3.4% 20 1.4% <0.001 Student 3 0.5% 5 0.7% 5 0.9% 4 0.5% 8 0.7% 9 0.6% Farmer 41 6.9% 279 39.1% 30 5.5% 297 40.0% 71 6.3% 576 39.6% Pastoralist 91 15.3% 29 4.1% 65 12.0% 23 3.1% 156 13.8% 52 3.6% Salaried employee 24 4.0% 34 4.8% 18 3.3% 41 5.5% 42 3.7% 75 5.2% Casual laborer 78 13.2% 166 23.3% 50 9.2% 187 25.2% 128 11.3% 353 24.2% Self-employed business 40 6.7% 89 12.5% 59 10.9% 83 11.2% 99 8.7% 172 11.8% Not currently working 239 40.3% 87 12.2% 247 45.7% 89 12.0% 486 42.9% 176 12.1% Others 54 9.1% 13 1.8% 52 9.6% 10 1.3% 106 9.3% 23 1.6% Education level of the guardian Never attended school 466 78.6% 118 16.5% <0.001 427 78.9% 105 14.1% <0.001 893 78.7% 223 15.3% <0.001 Madrasa/Adult learning center 30 5.1% 1 0.1% 24 4.4% 0 0.0% 54 4.8% 1 0.1% 28.4 Did not complete primary school 31 5.2% 212 29.7% 28 5.2% 202 27.2% 59 5.2% 414 % Completed primary school 22 3.7% 224 31.4% 22 4.1% 264 35.5% 44 3.9% 488 33.5% Did not compete secondary 6 1.0% 58 8.1% 8 1.5% 53 7.1% 14 1.2% 111 7.6% Completed secondary school 26 4.4% 61 8.6% 17 3.1% 80 10.8% 43 3.8% 141 9.7% Completed technical college 10 1.7% 34 4.8% 14 2.6% 37 5.0% 24 2.1% 71 4.9% Completed university/graduate school 2 0.3% 5 0.7% 1 0.2% 2 0.3% 3 0.3% 7 0.5% Table 2A: Number of males and females in the household distributed by CONTROL and WFPSMP stratified by gender of the child Variables Male (n=1254) Female (n=1286) Total (N=2540) CONTROL (n=675) WFPSMP (n=579) p value CON (n= TROL 721) WFPSMP (n=565) p value CON (n= TROL 1396) WFPSMP (n=1144) p value Total males in the household <0.00 None 6 0.9% 1 0.2% <0.001 35 4.9% 21 3.7% 1 41 2.9% 22 1.9% <0.001 1 to 2 187 27.7% 104 18.0% 307 42.6% 150 26.5% 494 35.4% 254 22.2% 3 to 4 316 46.8% 273 47.2% 287 39.8% 269 47.6% 603 43.2% 542 47.4% 5 to 6 136 20.1% 153 26.4% 74 10.3% 106 18.8% 210 15.0% 259 22.6% 7 to 8 30 4.4% 48 8.3% 18 2.5% 19 3.4% 48 3.4% 67 5.9% Total females in the household None 26 3.9% 23 4.0% <0.001 2 0.3% 4 0.7% 0.021 28 2.0% 27 2.4% 0.002 1 to 2 316 46.8% 208 35.9% 200 27.7% 148 26.2% 516 37.0% 356 31.1% 3 to 4 260 38.5% 259 44.7% 346 48.0% 235 41.6% 606 43.4% 494 43.2% 5 to 6 70 10.4% 74 12.8% 135 18.7% 145 25.7% 205 14.7% 219 19.1% 7 to 8 3 0.4% 15 2.6% 38 5.3% 33 5.8% 41 2.9% 48 4.2% Total males between 7-18 years attending school <0.00 None 15 2.2% 13 2.2% 0.028 196 27.2% 100 17.7% 1 211 15.1% 113 9.9% <0.001 1 to 2 477 70.7% 368 63.6% 438 60.7% 351 62.1% 915 65.5% 719 62.8% 3 to 4 164 24.3% 175 30.2% 77 10.7% 101 17.9% 241 17.3% 276 24.1% 5 to 6 12 1.8% 20 3.5% 9 1.2% 11 1.9% 21 1.5% 31 2.7% 7 to 8 7 1.0% 3 0.5% 1 0.1% 2 0.4% 8 0.6% 5 0.4% Total females between 7-18 years attending school None 203 30.1% 162 28.0% 0.065 24 3.3% 31 5.5% 0.204 227 16.3% 193 16.9% 0.516 1 to 2 408 60.4% 337 58.2% 508 70.5% 401 71.0% 916 65.6% 738 64.5% 3 to 4 58 8.6% 68 11.7% 167 23.2% 111 19.6% 225 16.1% 179 15.6% 5 to 6 3 0.4% 10 1.7% 18 2.5% 18 3.2% 21 1.5% 28 2.4% 7 to 8 3 0.4% 2 0.3% 4 0.6% 4 0.7% 7 0.5% 6 0.5% Table 2B: Number of males and females in the household distributed by WFPSMP and HGSMP stratified by gender of the child Variables Male (n=1306) Female (n=1284) Total (N=2590) WFPSMP (n=593) HGSMP (n=713) p value WF (n= PSMP 541) HGSMP (n=743) p value WF (n= PSMP 1134) HGSMP (n=1456) p value Total males in the household None 5 0.8% 3 0.4% <0.001 26 4.8% 47 6.3% <0.001 31 2.7% 50 3.4% <0.001 1 to 2 110 18.5% 218 30.6% 161 29.8% 354 47.6% 271 23.9% 572 39.3% 48.8 3 to 4 265 44.7% 348 % 235 43.4% 259 34.9% 500 44.1% 607 41.7% 5 to 6 161 27.2% 111 15.6% 96 17.7% 65 8.7% 257 22.7% 176 12.1% 7 to 8 52 8.8% 33 4.6% 23 4.3% 18 2.4% 75 6.6% 51 3.5% Total females in the household None 22 3.7% 68 9.5% <0.001 4 0.7% 3 0.4% 0.014 26 2.3% 71 4.9% <0.001 1 to 2 198 33.4% 320 44.9% 143 26.4% 234 31.5% 341 30.1% 554 38.0% 3 to 4 264 44.5% 241 33.8% 234 43.3% 346 46.6% 498 43.9% 587 40.3% 5 to 6 98 16.5% 74 10.4% 124 22.9% 121 16.3% 222 19.6% 195 13.4% 7 to 8 11 1.9% 10 1.4% 36 6.7% 39 5.2% 47 4.1% 49 3.4% Total males between 7-18 years attending school None 13 2.2% 18 2.5% 0.168 97 17.9% 246 33.1% <0.001 110 9.7% 264 18.1% <0.001 1 to 2 388 65.4% 509 71.4% 334 61.7% 410 55.2% 722 63.7% 919 63.1% 3 to 4 163 27.5% 155 21.7% 96 17.7% 76 10.2% 259 22.8% 231 15.9% 5 to 6 22 3.7% 24 3.4% 13 2.4% 9 1.2% 35 3.1% 33 2.3% 7 to 8 7 1.2% 7 1.0% 1 0.2% 2 0.3% 8 0.7% 9 0.6% Total females between 7-18 years attending school None 151 25.5% 267 37.4% <0.001 26 4.8% 24 3.2% 0.298 177 15.6% 291 20.0% 0.029 1 to 2 355 59.9% 369 51.8% 382 70.6% 545 73.4% 737 65.0% 914 62.8% 3 to 4 80 13.5% 66 9.3% 114 21.1% 145 19.5% 194 17.1% 211 14.5% 5 to 6 7 1.2% 8 1.1% 14 2.6% 26 3.5% 21 1.9% 34 2.3% 7 to 8 0 0.0% 3 0.4% 5 0.9% 3 0.4% 5 0.4% 6 0.4% Table 3a: Availability of food at home distributed by CONTROL and WFPSMP stratified by gender of the child Variables Male (n=1254) Female (n=1286) Total (N=2540) CONTROL (n=675) WFPSMP (n=579) p value CON (n= TROL 721) WFPSMP (n=565) p value CON (n= TROL 1396) WFPSMP (n=1144) p value Number of days child ate before going to school None 104 15.4% 158 27.3% <0.001 125 17.3% 153 27.1% <0.001 229 16.4% 311 27.2% <0.001 1 - 2 days 96 14.2% 83 14.3% 107 14.8% 101 17.9% 203 14.5% 184 16.1% 3 - 4 days 203 30.1% 149 25.7% 231 32.0% 122 21.6% 434 31.1% 271 23.7% 5 days 272 40.3% 189 32.6% 258 35.8% 189 33.5% 530 38.0% 378 33.0% Number of days child ate after coming from school None 23 3.4% 36 6.2% <0.001 29 4.0% 35 6.2% <0.001 52 3.7% 71 6.2% <0.001 1 - 2 days 42 6.2% 81 14.0% 48 6.7% 95 16.8% 90 6.4% 176 15.4% 3 - 4 days 129 19.1% 206 35.6% 140 19.4% 171 30.3% 269 19.3% 377 33.0% 5 days 481 71.3% 256 44.2% 504 69.9% 264 46.7% 985 70.6% 520 45.5% Child had a meal on interview day before going to school No 221 32.7% 256 44.2% <0.001 287 39.8% 260 46.0% <0.001 508 36.4% 516 45.1% <0.001 Yes: Not enough 298 44.1% 152 26.3% 279 38.7% 156 27.6% 577 41.3% 308 26.9% Yes: Enough 156 23.1% 171 29.5% 155 21.5% 149 26.4% 311 22.3% 320 28.0% Food consumption score (FCS) Poor 164 24.3% 216 37.3% <0.001 182 25.2% 213 37.7% <0.001 346 24.8% 429 37.5% <0.001 Borderline 268 39.7% 154 26.6% 288 39.9% 158 28.0% 556 39.8% 312 27.3% Acceptable 243 36.0% 209 36.1% 251 34.8% 194 34.3% 494 35.4% 403 35.2% Table 3b: Availability of food at home distributed by WFPSMP and HGSMP stratified by gender of the child Variables Male (n=1306) Female (n=1284) Total (N=2590) WFPSMP (n=593) HGSMP (n=713) p value WF (n= PSMP 541) HGSMP (n=743) p value WF (n= PSMP 1134) HGSMP (n=1456) p value Number of days child ate before going to school None 116 19.6% 90 12.6% 0.007 97 17.9% 80 10.8% <0.001 213 18.8% 170 11.7% <0.001 1 - 2 days 80 13.5% 98 13.7% 99 18.3% 103 13.9% 179 15.8% 201 13.8% 3 - 4 days 165 27.8% 224 31.4% 138 25.5% 232 31.2% 303 26.7% 456 31.3% 5 days 232 39.1% 301 42.2% 207 38.3% 328 44.1% 439 38.7% 629 43.2% Number of days child ate after coming from school None 24 4.0% 41 5.8% <0.001 22 4.1% 42 5.7% <0.001 46 4.1% 83 5.7% <0.001 1 - 2 days 89 15.0% 44 6.2% 94 17.4% 41 5.5% 183 16.1% 85 5.8% 3 - 4 days 204 34.4% 125 17.5% 169 31.2% 126 17.0% 373 32.9% 251 17.2% 5 days 276 46.5% 503 70.5% 256 47.3% 534 71.9% 532 46.9% 1037 71.2% Child had a meal on interview day before going to school No 209 35.2% 231 32.4% <0.001 193 35.7% 241 32.4% <0.001 402 35.4% 472 32.4% <0.001 Meal not enough 162 27.3% 312 43.8% 154 28.5% 332 44.7% 316 27.9% 644 44.2% Enough meal 222 37.4% 170 23.8% 194 35.9% 170 22.9% 416 36.7% 340 23.4% Food consumption score (FCS) Poor 197 33.2% 148 20.8% <0.001 193 35.7% 156 21.0% <0.001 390 34.4% 304 20.9% <0.001 Borderline 135 22.8% 281 39.4% 125 23.1% 271 36.5% 260 22.9% 552 37.9% Acceptable 261 44.0% 284 39.8% 223 41.2% 316 42.5% 484 42.7% 600 41.2% Table 4a: Availability of food at school distributed by CONTROL and WFPSMP stratified by gender of the child Variables Male (n=1254) Female (n=1286) Total (N=2540) CONTROL (n=675) WFPSMP (n=579) p value CON (n= TROL 721) WFPSMP (n=565) p value CON (n= TROL 1396) WFPSMP (n=1144) p value Child has been receiving school meals at school in the current school year (2017) Yes 158 23.4% 322 55.6% <0.001 121 16.8% 356 63.0% <0.001 279 20.0% 678 59.3% <0.001 No 517 76.6% 257 44.4% 600 83.2% 209 37.0% 1117 80.0% 466 40.7% The school in which the child was learning at currently (same week) serving food Yes 114 16.9% 289 49.9% <0.001 113 15.7% 302 53.5% <0.001 227 16.3% 591 51.7% <0.001 No 561 83.1% 290 50.1% 608 84.3% 263 46.5% 1169 83.7% 553 48.3% When school meals are not provided: Child carried food from home No 410 60.7% 573 99.0% <0.001 383 53.1% 562 99.5% <0.001 793 56.8% 1135 99.2% <0.001 Yes 265 39.3% 6 1.0% 338 46.9% 3 0.5% 603 43.2% 9 0.8% When school meals are not provided: Child buys lunch No 664 98.4% 572 98.8% 0.532 711 98.6% 558 98.8% 0.818 1375 98.5% 1130 98.8% 0.546 Yes 11 1.6% 7 1.2% 10 1.4% 7 1.2% 21 1.5% 14 1.2% When school meals are not provided: Child goes home for lunch No 537 79.6% 246 42.5% <0.001 558 77.4% 278 49.2% <0.001 1095 78.4% 524 45.8% <0.001 Yes 138 20.4% 333 57.5% 163 22.6% 287 50.8% 301 21.6% 620 54.2% When school meals are not provided: Child remains at home No 670 99.3% 562 97.1% 0.003 715 99.2% 544 96.3% <0.001 1385 99.2% 1106 96.7% <0.001 Yes 5 0.7% 17 2.9% 6 0.8% 21 3.7% 11 0.8% 38 3.3% When school meals are not provided: Child goes without lunch No 255 37.8% 293 50.6% <0.001 315 43.7% 262 46.4% 0.337 570 40.8% 555 48.5% <0.001 Yes 420 62.2% 286 49.4% 406 56.3% 303 53.6% 826 59.2% 589 51.5% Child missed a complete day of school during the 1st term of the year (2017) Yes 357 52.9% 211 36.4% <0.001 360 49.9% 190 33.6% <0.001 717 51.4% 401 35.1% <0.001 No 318 47.1% 368 63.6% 361 50.1% 375 66.4% 679 48.6% 743 64.9% Table 4b: Availability of food at school distributed by WFPSMP and HGSMP stratified by gender of the child Variables Male (n=1306) Female (n=1284) Total (N=2590) WFPSMP (n=593) HGSMP (n=713) p value WF (n= PSMP 541) HGSMP (n=743) p value WFP (n= SMP 1134) HGSMP (n=1456) p value Child has been receiving school meals at school in the current school year (2017) Yes 318 53.6% 573 80.4% <0.001 313 57.9% 598 80.5% <0.001 631 55.6% 1171 80.4% <0.001 No 275 46.4% 140 19.6% 228 42.1% 145 19.5% 503 44.4% 285 19.6% The school in which the child was learning at currently (same week) serving food Yes 263 44.4% 351 49.2% 0.079 235 43.4% 399 53.7% <0.001 498 43.9% 750 51.5% <0.001 No 330 55.6% 362 50.8% 306 56.6% 344 46.3% 636 56.1% 706 48.5% When school meals are not provided: Child carried food from home No 583 98.3% 410 57.5% <0.001 537 99.3% 387 52.1% <0.001 1120 98.8% 797 54.7% <0.001 Yes 10 1.7% 303 42.5% 4 0.7% 356 47.9% 14 1.2% 659 45.3% When school meals are not provided: Child buys lunch No 581 98.0% 698 97.9% 0.919 532 98.3% 732 98.5% 0.794 1113 98.1% 1430 98.2% 0.900 Yes 12 2.0% 15 2.1% 9 1.7% 11 1.5% 21 1.9% 26 1.8% When school meals are not provided: Child goes home for lunch No 197 33.2% 557 78.1% <0.001 202 37.3% 602 81.0% <0.001 399 35.2% 1159 79.6% <0.001 Yes 396 66.8% 156 21.9% 339 62.7% 141 19.0% 735 64.8% 297 20.4% When school meals are not provided: Child remains at home No 580 97.8% 707 99.2% 0.042 535 98.9% 726 97.7% 0.116 1115 98.3% 1433 98.4% 0.848 Yes 13 2.2% 6 0.8% 6 1.1% 17 2.3% 19 1.7% 23 1.6% When school meals are not provided: Child goes without lunch No 376 63.4% 291 40.8% <0.001 320 59.1% 328 44.1% <0.001 696 61.4% 619 42.5% <0.001 Yes 217 36.6% 422 59.2% 221 40.9% 415 55.9% 438 38.6% 837 57.5% Child missed a complete day of school during the 1st term of the year (2017) Yes 181 30.5% 402 56.4% <0.001 186 34.4% 399 53.7% <0.001 367 32.4% 801 55.0% <0.001 No 412 69.5% 311 43.6% 355 65.6% 344 46.3% 767 67.6% 655 45.0% Table 5a: Coping strategy on days when the family did not have enough food or money to buy food distributed by CONTROL and WFPSMP stratified by gender of child Variables Male (n=1254) Female (n=1286) Total (N=2540) CONTROL (n=675) WFPSMP (n=579) p value CON (n= TROL 721) WFPSMP (n=565) p value CON (n= TROL 1396) WFPSMP (n=1144) p value Quintiles of Coping Strategy Index (CSI) First 108 16.0% 130 22.5% 0.001 119 16.5% 109 19.3% 0.117 227 16.3% 239 20.9% 0.001 Second 126 18.7% 132 22.8% 142 19.7% 104 18.4% 268 19.2% 236 20.6% Third 140 20.7% 93 16.1% 147 20.4% 121 21.4% 287 20.6% 214 18.7% Fourth 161 23.9% 103 17.8% 173 24.0% 105 18.6% 334 23.9% 208 18.2% Fifth 140 20.7% 121 20.9% 140 19.4% 126 22.3% 280 20.1% 247 21.6% Table 5b: Coping strategy on days when the family did not have enough food or money to buy food distributed by WFPSMP and HGSMP stratified by gender of th Variables Male (n=1306) Female (n=1284) Total (N=2590) WFPSMP (n=593) HGSMP (n=713) p value WF (n= PSMP 541) HGSMP (n=743) p value WF (n= PSMP 1134) HGSMP (n=1456) p value Quintiles of Coping Strategy Index (CSI) First 178 30.0% 126 17.7% 0.001 132 24.4% 145 19.5% 0.161 310 27.3% 271 18.6% 0.001 Second 133 22.4% 132 18.5% 110 20.3% 154 20.7% 243 21.4% 286 19.6% Third 92 15.5% 170 23.8% 100 18.5% 167 22.5% 192 16.9% 337 23.1% Fourth 77 13.0% 165 23.1% 96 17.7% 144 19.4% 173 15.3% 309 21.2% Fifth 113 19.1% 120 16.8% 103 19.0% 133 17.9% 216 19.0% 253 17.4% Table 6a: Views on benefits of education, school absenteeism, sources of information on school feeding and hygiene distributed by CONTROL and WFPSMP strat by gender of the child Variables Male (n=1254) Female (n=1286) Total (n=2540) CONTROL (n=675) WFPSMP (n=579) p value CON (n= TROL 721) WFPSMP (n=565) p value CON (n= TROL 1396) WFPSMP (n=1144) p value Number of important benefits of education mentioned by the parent/guardian <2 227 33.6% 78 13.5% <0.001 242 33.6% 80 14.2% <0.001 469 33.6% 158 13.8% <0.00 2 to 3 404 59.9% 310 53.5% 436 60.5% 284 50.3% 840 60.2% 594 51.9% 4 to 5 41 6.1% 104 18.0% 38 5.3% 96 17.0% 79 5.7% 200 17.5% 6 and above 3 0.4% 87 15.0% 5 0.7% 105 18.6% 8 0.6% 192 16.8% Number of sources of information on school feeding in the past year mentioned by the parent/guardian None 469 69.5% 172 29.7% <0.001 511 70.9% 164 29.0% <0.001 980 70.2% 336 29.4% <0.00 One 181 26.8% 304 52.5% 184 25.5% 302 53.5% 365 26.1% 606 53.0% Two 19 2.8% 73 12.6% 24 3.3% 69 12.2% 43 3.1% 142 12.4% Three and above 6 0.9% 30 5.2% 2 0.3% 30 5.3% 8 0.6% 60 5.2% Number of sources of information on hygiene in the past year mentioned by the parent/guardian None 337 49.9% 218 37.7% <0.001 361 50.1% 182 32.2% <0.001 698 50.0% 400 35.0% <0.00 One 257 38.1% 233 40.2% 288 39.9% 239 42.3% 545 39.0% 472 41.3% Two 55 8.1% 83 14.3% 51 7.1% 99 17.5% 106 7.6% 182 15.9% Three and above 26 3.9% 45 7.8% 21 2.9% 45 8.0% 47 3.4% 90 7.9% Number of reasons why the child missed a complete day of school during the 1st term of this year None 318 47.1% 368 63.6% <0.001 361 50.1% 375 66.4% <0.001 679 48.6% 743 64.9% <0.00 One 261 38.7% 143 24.7% 282 39.1% 118 20.9% 543 38.9% 261 22.8% Two 80 11.9% 39 6.7% 61 8.5% 45 8.0% 141 10.1% 84 7.3% Three and above 16 2.4% 29 5.0% 17 2.4% 27 4.8% 33 2.4% 56 4.9% Table 6b: Views on benefits of education, school absenteeism, sources of information on school feeding and hygiene distributed by WFPSMP and HGSMP stratifi gender of the child Variables Male (n=1306) Female (n=1284) Total (N=2590) WFPSMP (n=593) HGSMP (n=713) p value WF (n= PSMP 541) HGSMP (n=743) p value WF (n= PSMP 1134) HGSMP (n=1456) p value Number of important benefits of education mentioned <2 63 10.6% 202 28.3% <0.001 64 11.8% 183 24.6% <0.001 127 11.2% 385 26.4% <0.001 63.8 2 to 3 358 60.4% 455 % 301 55.6% 500 67.3% 659 58.1% 955 65.6% 4 to 5 109 18.4% 55 7.7% 93 17.2% 57 7.7% 202 17.8% 112 7.7% 6 and above 63 10.6% 1 0.1% 83 15.3% 3 0.4% 146 12.9% 4 0.3% Number of sources of information on school feeding in the past year None 182 30.7% 299 41.9% <0.001 169 31.2% 318 42.8% <0.001 351 31.0% 617 42.4% <0.001 One 310 52.3% 357 50.1% 271 50.1% 377 50.7% 581 51.2% 734 50.4% Two 71 12.0% 42 5.9% 66 12.2% 37 5.0% 137 12.1% 79 5.4% Three and above 30 5.1% 15 2.1% 35 6.5% 11 1.5% 65 5.7% 26 1.8% Number of sources of information on hygiene in the past year None 235 39.6% 244 34.2% <0.001 192 35.5% 258 34.7% <0.001 427 37.7% 502 34.5% <0.001 One 217 36.6% 372 52.2% 200 37.0% 368 49.5% 417 36.8% 740 50.8% Two 102 17.2% 76 10.7% 109 20.1% 92 12.4% 211 18.6% 168 11.5% Three and above 39 6.6% 21 2.9% 40 7.4% 25 3.4% 79 7.0% 46 3.2% Number of reasons why the child missed a complete day of school during the 1st term of this year None 412 69.5% 311 43.6% <0.001 355 65.6% 344 46.3% <0.001 767 67.6% 655 45.0% <0.001 One 124 20.9% 304 42.6% 124 22.9% 323 43.5% 248 21.9% 627 43.1% Two 34 5.7% 76 10.7% 42 7.8% 58 7.8% 76 6.7% 134 9.2% Three and above 23 3.9% 22 3.1% 20 3.7% 18 2.4% 43 3.8% 40 2.7% Table 7a: Socio-demographic characteristics of children distributed by CONTROL and WFPSMP stratified by gender of the child Variables Male (n=1254) Female (n=1286) Total (N=2540) CONTROL (n=675) WFPSMP (n=579) p value CON (n= TROL 721) WFPSMP (n=565) p value CON (n=1 TROL 396) WFPSMP (n=1144) va Age of child in years 7 to 8 43 6.4% 32 5.5% 0.078 67 9.3% 33 5.8% 0.002 110 7.9% 65 5.7% <0. 9 to 10 135 20.0% 87 15.0% 158 21.9% 97 17.2% 293 21.0% 184 16.1% 11 to 12 188 27.9% 163 28.2% 202 28.0% 162 28.7% 390 27.9% 325 28.4% 13 to 14 204 30.2% 181 31.3% 210 29.1% 172 30.4% 414 29.7% 353 30.9% >14 105 15.6% 116 20.0% 84 11.7% 101 17.9% 189 13.5% 217 19.0% Class of the child Third 99 14.7% 136 23.5% 0.006 116 16.1% 123 21.8% 0.001 215 15.4% 259 22.6% <0. Fourth 116 17.2% 94 16.2% 109 15.1% 108 19.1% 225 16.1% 202 17.7% Fifth 122 18.1% 96 16.6% 112 15.5% 96 17.0% 234 16.8% 192 16.8% Sixth 122 18.1% 95 16.4% 128 17.8% 95 16.8% 250 17.9% 190 16.6% Seventh 133 19.7% 94 16.2% 135 18.7% 76 13.5% 268 19.2% 170 14.9% Eighth 83 12.3% 64 11.1% 121 16.8% 67 11.9% 204 14.6% 131 11.5% Time taken to get to school Less than 15 minutes 151 22.4% 262 45.3% <0.001 153 21.2% 273 48.3% <0.001 304 21.8% 535 46.8% <0. Between 15 and 30 minutes 250 37.0% 189 32.6% 255 35.4% 158 28.0% 505 36.2% 347 30.3% Between 30 and 60 minutes 219 32.4% 78 13.5% 238 33.0% 85 15.0% 457 32.7% 163 14.2% More than 1 hour 55 8.1% 50 8.6% 75 10.4% 49 8.7% 130 9.3% 99 8.7% Mode of travel to school On foot 668 99.0% 574 99.1% 0.753 719 99.7% 562 99.5% 0.468 1387 99.4% 1136 99.3% 0. Bicycle/ Bus/ Motor cycle 7 1.0% 5 0.9% 2 0.3% 3 0.5% 9 0.6% 8 0.7% Brothers and sisters currently studying in the same school Yes 484 71.7% 382 66.0% 0.029 547 75.9% 387 68.5% 0.003 1031 73.9% 769 67.2% <0. No 191 28.3% 197 34.0% 174 24.1% 178 31.5% 365 26.1% 375 32.8% Having brothers and sisters who are old enough to go to school but are NOT currently attending school Yes 69 10.2% 131 22.6% <0.001 62 8.6% 124 21.9% <0.001 131 9.4% 255 22.3% <0. No 606 89.8% 448 77.4% 659 91.4% 441 78.1% 1265 90.6% 889 77.7% Table 7b: Socio-demographic characteristics of children distributed by WFPSMP and HGSMP stratified by gender of the child Variables Male (n=1306) Female (n=1284) Total (N=2590) WFPSMP (n=593) HGSMP (n=713) p value WF (n= PSMP 541) HGSMP (n=743) p value WFP (n= SMP 1134) HGSMP (n=1456) p value Age of child in years 7 to 8 28 4.7% 27 3.8% 0.001 38 7.0% 54 7.3% 0.005 66 5.8% 81 5.6% <0.001 9 to 10 83 14.0% 163 22.9% 97 17.9% 176 23.7% 180 15.9% 339 23.3% 11 to 12 184 31.0% 191 26.8% 164 30.3% 236 31.8% 348 30.7% 427 29.3% 30.9 13 to 14 206 34.7% 220 % 164 30.3% 213 28.7% 370 32.6% 433 29.7% >14 92 15.5% 112 15.7% 78 14.4% 64 8.6% 170 15.0% 176 12.1% Class of the child Third 125 21.1% 95 13.3% <0.001 123 22.7% 111 14.9% <0.001 248 21.9% 206 14.1% <0.001 Fourth 119 20.1% 119 16.7% 121 22.4% 122 16.4% 240 21.2% 241 16.6% Fifth 120 20.2% 130 18.2% 95 17.6% 126 17.0% 215 19.0% 256 17.6% Sixth 94 15.9% 139 19.5% 98 18.1% 125 16.8% 192 16.9% 264 18.1% Seventh 84 14.2% 123 17.3% 66 12.2% 149 20.1% 150 13.2% 272 18.7% Eighth 51 8.6% 107 15.0% 38 7.0% 110 14.8% 89 7.8% 217 14.9% Time taken to get to school Less than 15 minutes 298 50.3% 187 26.2% <0.001 262 48.4% 190 25.6% <0.001 560 49.4% 377 25.9% <0.001 Between 15 and 30 minutes 194 32.7% 273 38.3% 174 32.2% 306 41.2% 368 32.5% 579 39.8% Between 30 and 60 minutes 68 11.5% 203 28.5% 67 12.4% 191 25.7% 135 11.9% 394 27.1% More than 1 hour 33 5.6% 50 7.0% 38 7.0% 56 7.5% 71 6.3% 106 7.3% Mode of travel to school On foot 592 99.8% 685 96.1% <0.001 539 99.6% 720 96.9% <0.001 1131 99.7% 1405 96.5% <0.001 Bicycle/ Bus/ Motor cycle 1 0.2% 28 3.9% 2 0.4% 23 3.1% 3 0.3% 51 3.5% Brothers and sisters currently studying in the same school 66.8 Yes 378 63.7% 476 % 0.254 355 65.6% 512 68.9% 0.214 733 64.6% 988 67.9% 0.085 No 215 36.3% 237 33.2% 186 34.4% 231 31.1% 401 35.4% 468 32.1% Having brothers and sisters who are old enough to go to school but are NOT currently attending school Yes 116 19.6% 56 7.9% <0.001 104 19.2% 66 8.9% <0.001 220 19.4% 122 8.4% <0.001 No 477 80.4% 657 92.1% 437 80.8% 677 91.1% 914 80.6% 1334 91.6% Table 8a: Children feeding distributed by CONTROL and WFPSMP stratified by gender of the child Variables Male (n=1254) Female (n=1286) Total (N=2540) CONTROL (n=675) WFPSMP (n=579) p value CON (n TROL =721) WFPSMP (n=565) p value CON (n= TROL 1396) WFPSMP (n=1144) p value Had a meal today BEFORE coming to school No 230 34.1% 221 38.2% <0.001 296 41.1% 237 41.9% 0.007 526 37.7% 458 40.0% <0.00 Yes: Not enough 252 37.3% 149 25.7% 238 33.0% 146 25.8% 490 35.1% 295 25.8% Yes: Enough 193 28.6% 209 36.1% 187 25.9% 182 32.2% 380 27.2% 391 34.2% Number of days child ate before going to school None 89 13.2% 135 23.3% <0.001 103 14.3% 124 21.9% <0.001 192 13.8% 259 22.6% <0.00 1 - 2 days 111 16.4% 93 16.1% 113 15.7% 114 20.2% 224 16.0% 207 18.1% 3 - 4 days 219 32.4% 162 28.0% 232 32.2% 159 28.1% 451 32.3% 321 28.1% 5 days 256 37.9% 189 32.6% 273 37.9% 168 29.7% 529 37.9% 357 31.2% Number of times child normally eat per day 1 time 107 15.9% 119 20.6% <0.001 133 18.4% 118 20.9% 0.007 240 17.2% 237 20.7% <0.00 2 times 291 43.1% 292 50.4% 326 45.2% 293 51.9% 617 44.2% 585 51.1% 3 times 266 39.4% 164 28.3% 249 34.5% 147 26.0% 515 36.9% 311 27.2% More than 3 time 11 1.6% 4 0.7% 13 1.8% 7 1.2% 24 1.7% 11 1.0% Number of times child ate yesterday 1 time 148 21.9% 139 24.0% 0.050 196 27.2% 134 23.7% 0.024 344 24.6% 273 23.9% 0.00 2 times 276 40.9% 261 45.1% 290 40.2% 275 48.7% 566 40.5% 536 46.9% 3 times 234 34.7% 173 29.9% 220 30.5% 148 26.2% 454 32.5% 321 28.1% More than 3 time 17 2.5% 6 1.0% 15 2.1% 8 1.4% 32 2.3% 14 1.2% The last time meals were provided for pupils in the school Yesterday 82 12.1% 238 41.1% <0.001 69 9.6% 263 46.5% <0.001 151 10.8% 501 43.8% <0.00 One week ago 94 13.9% 24 4.1% 74 10.3% 18 3.2% 168 12.0% 42 3.7% One month ago 9 1.3% 4 0.7% 7 1.0% 5 0.9% 16 1.1% 9 0.8% One term ago 19 2.8% 201 34.7% 25 3.5% 186 32.9% 44 3.2% 387 33.8% Two terms ago 5 0.7% 65 11.2% 4 0.6% 51 9.0% 9 0.6% 116 10.1% One year ago 62 9.2% 43 7.4% 51 7.1% 39 6.9% 113 8.1% 82 7.2% More than one year ago 404 59.9% 4 0.7% 491 68.1% 3 0.5% 895 64.1% 7 0.6% Table 8b: Children feeding distributed by WFPSMP and HGSMP stratified by gender of the child Variables Male (n=1306) Female (n=1284) Total (N=2590) WFPSMP (n=593) HGSMP (n=713) p value WF (n PSMP =541) HGSMP (n=743) p value WF (n= PSMP 1134) HGSMP (n=1456) p value Had a meal today BEFORE coming to school No 175 29.5% 194 27.2% <0.001 182 33.6% 246 33.1% 0.006 357 31.5% 440 30.2% <0.00 Yes: Not enough 145 24.5% 257 36.0% 137 25.3% 244 32.8% 282 24.9% 501 34.4% Yes: Enough 273 46.0% 262 36.7% 222 41.0% 253 34.1% 495 43.7% 515 35.4% Number of days child ate before going to school None 88 14.8% 65 9.1% <0.001 78 14.4% 83 11.2% <0.001 166 14.6% 148 10.2% <0.00 1 - 2 days 111 18.7% 107 15.0% 100 18.5% 119 16.0% 211 18.6% 226 15.5% 3 - 4 days 169 28.5% 198 27.8% 165 30.5% 182 24.5% 334 29.5% 380 26.1% 5 days 225 37.9% 343 48.1% 198 36.6% 359 48.3% 423 37.3% 702 48.2% Number of times child normally eat per day 1 time 81 13.7% 111 15.6% <0.001 92 17.0% 106 14.3% <0.001 173 15.3% 217 14.9% <0.00 2 times 275 46.4% 242 33.9% 246 45.5% 240 32.3% 521 45.9% 482 33.1% 3 times 227 38.3% 339 47.5% 193 35.7% 371 49.9% 420 37.0% 710 48.8% More than 3 time 10 1.7% 21 2.9% 10 1.8% 26 3.5% 20 1.8% 47 3.2% Number of times child ate yesterday 1 time 110 18.5% 126 17.7% 0.061 114 21.1% 141 19.0% 0.001 224 19.8% 267 18.3% <0.00 2 times 240 40.5% 256 35.9% 226 41.8% 242 32.6% 466 41.1% 498 34.2% 3 times 233 39.3% 305 42.8% 188 34.8% 330 44.4% 421 37.1% 635 43.6% More than 3 time 10 1.7% 26 3.6% 13 2.4% 30 4.0% 23 2.0% 56 3.8% The last time meals were provided for pupils in the school Yesterday 211 35.6% 305 42.8% <0.001 197 36.4% 352 47.4% <0.001 408 36.0% 657 45.1% <0.00 One week ago 32 5.4% 72 10.1% 27 5.0% 67 9.0% 59 5.2% 139 9.5% One month ago 7 1.2% 26 3.6% 10 1.8% 22 3.0% 17 1.5% 48 3.3% One term ago 246 41.5% 200 28.1% 225 41.6% 189 25.4% 471 41.5% 389 26.7% Two terms ago 47 7.9% 2 0.3% 32 5.9% 4 0.5% 79 7.0% 6 0.4% One year ago 47 7.9% 17 2.4% 47 8.7% 17 2.3% 94 8.3% 34 2.3% More than one year ago 3 0.5% 91 12.8% 3 0.6% 92 12.4% 6 0.5% 183 12.6% Table 9a: Hygiene, nutrition, concentration in class, importance of education and school absenteeism distributed by CONTROL and WFPSMP stratified by gende the child Variables Male (n=1254) Female (n=1286) Total (n= CONTROL (n=675) WFPSMP (n=579) p value CON (n TROL =721) WFPSMP (n=565) p value CON (n= TROL 1396) WFP (n=1 In the past month the teacher talked to students about hygiene Yes 582 86.2% 497 85.8% 0.845 628 87.1% 503 89.0% 0.293 1210 86.7% 1000 No 93 13.8% 82 14.2% 93 12.9% 62 11.0% 186 13.3% 144 Number of hygiene habits mentioned None 64 9.5% 49 8.5% <0.001 64 8.9% 45 8.0% <0.001 128 9.2% 94 1 to 2 479 71.0% 249 43.0% 513 71.2% 217 38.4% 992 71.1% 466 3 to 4 127 18.8% 205 35.4% 140 19.4% 227 40.2% 267 19.1% 432 5 and above 5 0.7% 76 13.1% 4 0.6% 76 13.5% 9 0.6% 152 In the past month the teacher talked to students about nutrition Yes 460 68.1% 394 68.0% 0.970 495 68.7% 399 70.6% 0.447 955 68.4% 793 No 215 31.9% 185 32.0% 226 31.3% 166 29.4% 441 31.6% 351 Number of important nutrition habits mentioned None 251 37.2% 163 28.2% <0.001 284 39.4% 154 27.3% <0.001 535 38.3% 317 One 267 39.6% 143 24.7% 261 36.2% 134 23.7% 528 37.8% 277 Two 114 16.9% 126 21.8% 117 16.2% 121 21.4% 231 16.5% 247 Three and above 43 6.4% 147 25.4% 59 8.2% 156 27.6% 102 7.3% 303 Number of reasons why missed school Never missed 343 50.8% 382 66.0% <0.001 366 50.8% 376 66.5% <0.001 709 50.8% 758 One 289 42.8% 164 28.3% 304 42.2% 146 25.8% 593 42.5% 310 Two 41 6.1% 23 4.0% 48 6.7% 27 4.8% 89 6.4% 50 Three or more 2 0.3% 10 1.7% 3 0.4% 16 2.8% 5 0.4% 26 Number of reasons why it was difficult to concentrate in class Never missed 361 53.5% 331 57.2% <0.001 387 53.7% 343 60.7% <0.001 748 53.6% 674 One 234 34.7% 104 18.0% 245 34.0% 98 17.3% 479 34.3% 202 Two 72 10.7% 81 14.0% 74 10.3% 61 10.8% 146 10.5% 142 Three or more 8 1.2% 63 10.9% 15 2.1% 63 11.2% 23 1.6% 126 Number of most important benefits of education mentioned by the child <2 356 52.7% 138 23.8% <0.001 383 53.1% 143 25.3% <0.001 739 52.9% 281 2 to 3 299 44.3% 280 48.4% 306 42.4% 258 45.7% 605 43.3% 538 4 to 5 20 3.0% 93 16.1% 32 4.4% 68 12.0% 52 3.7% 161 6 and above 0 0.0% 68 11.7% 0 0.0% 96 17.0% 0 0.0% 164 Table 9b: Hygiene, nutrition, concentration in class, importance of education and school absenteeism distributed by WFPSMP and HGSMP stratified by gender o child Variables Male (n=1306) Female (n=1284) Total (N=2 WFPSMP (n=593) HGSMP (n=713) p value WF (n PSMP =541) HGSMP (n=743) p value WF (n= PSMP 1134) HGS (n=14 In the past month the teacher talked to students about hygiene Yes 508 85.7% 629 88.2% 0.171 462 85.4% 631 84.9% 0.815 970 85.5% 1260 No 85 14.3% 84 11.8% 79 14.6% 112 15.1% 164 14.5% 196 Number of hygiene habits mentioned by the child None 39 6.6% 44 6.2% <0.001 39 7.2% 64 8.6% <0.001 78 6.9% 108 1 to 2 248 41.8% 520 72.9% 236 43.6% 526 70.8% 484 42.7% 1046 3 to 4 246 41.5% 139 19.5% 207 38.3% 142 19.1% 453 39.9% 281 5 and above 60 10.1% 10 1.4% 59 10.9% 11 1.5% 119 10.5% 21 In the past month the teacher talked to students about nutrition Yes 413 69.6% 479 67.2% 0.341 367 67.8% 500 67.3% 0.838 780 68.8% 979 No 180 30.4% 234 32.8% 174 32.2% 243 32.7% 354 31.2% 477 Number of important nutrition habits mentioned by the child None 125 21.1% 256 35.9% <0.001 118 21.8% 278 37.4% <0.001 243 21.4% 534 One 138 23.3% 267 37.4% 120 22.2% 282 38.0% 258 22.8% 549 Two 164 27.7% 109 15.3% 157 29.0% 105 14.1% 321 28.3% 214 Three and above 166 28.0% 81 11.4% 146 27.0% 78 10.5% 312 27.5% 159 Number of reasons why missed school Never missed 410 69.1% 342 48.0% <0.001 359 66.4% 372 50.1% <0.001 769 67.8% 714 One 159 26.8% 333 46.7% 143 26.4% 321 43.2% 302 26.6% 654 Two 17 2.9% 34 4.8% 26 4.8% 48 6.5% 43 3.8% 82 Three or more 7 1.2% 4 0.6% 13 2.4% 2 0.3% 20 1.8% 6 Number of reasons why it was difficult to concentrate in class Never missed 371 62.6% 390 54.7% <0.001 339 62.7% 432 58.1% <0.001 710 62.6% 822 One 105 17.7% 249 34.9% 101 18.7% 251 33.8% 206 18.2% 500 Two 66 11.1% 65 9.1% 63 11.6% 50 6.7% 129 11.4% 115 Three or more 51 8.6% 9 1.3% 38 7.0% 10 1.3% 89 7.8% 19 Number of most important benefits of education mentioned by the child <2 136 22.9% 342 48.0% <0.001 122 22.6% 346 46.6% <0.001 258 22.8% 688 2 to 3 313 52.8% 337 47.3% 281 51.9% 356 47.9% 594 52.4% 693 4 to 5 87 14.7% 34 4.8% 73 13.5% 38 5.1% 160 14.1% 72 6 and above 57 9.6% 0 0.0% 65 12.0% 3 0.4% 122 10.8% 3 Annex 9 – Computation of the Propensity Score All variables whose distribution was significantly different (p<0.05) between the study arm (CONTROL, WFPSMP and HGSMP) were used to construct t propensity score. The propensity score was constructed using the ‘participation equation’, derived from a logit regression with programme participation as the dependent va coded as follows; • WFPSMP = 1, versus Control = 0. • HGSMP = 1, versus WFPSMP = 0. Comparison of key learning outcomes was adjusted for, using the propensity score quintiles. Table 10a: Propensity score quintiles distributed by CONTROL and WFPSMP stratified by gender of the child Variables Male (n=1254) Female (n=1286) Total (N=2540) CONTROL (n=675) WFPSMP (n=579) p value CON (n TROL =721) WFPSMP (n=565) p value CON (n= TROL 1396) WFPSMP (n=1144) p value Propensity score quintiles First 249 36.9% 1 0.2% <0.001 257 35.6% 0 0.0% <0.001 507 36.3% 1 0.1% <0.001 Second 243 36.0% 8 1.4% 254 35.2% 4 0.7% 494 35.4% 14 1.2% Third 162 24.0% 89 15.4% 181 25.1% 76 13.5% 344 24.6% 164 14.3% Fourth 18 2.7% 233 40.2% 29 4.0% 228 40.4% 49 3.5% 459 40.1% Fifth 3 0.4% 248 42.8% 0 0.0% 257 45.5% 2 0.1% 506 44.2% Table 10b: Propensity score quintiles distributed by WFPSMP and HGSMP stratified by gender of the child Variables Male (n=1306) Female (n=1284) Total (N=2590) WFPSMP (n=593) HGSMP (n=713) p value WF (n PSMP =541) HGSMP (n=743) p value WF (n= PSMP 1134) HGSMP (n=1456) p value Propensity score quintiles First 258 43.5% 4 0.6% <0.001 252 46.6% 4 0.5% <0.001 513 45.2% 5 0.3% <0.001 Second 241 40.6% 20 2.8% 220 40.7% 37 5.0% 451 39.8% 67 4.6% Third 84 14.2% 177 24.8% 65 12.0% 192 25.8% 158 13.9% 360 24.7% Fourth 10 1.7% 251 35.2% 4 0.7% 253 34.1% 11 1.0% 507 34.8% Fifth 0 0.0% 261 36.6% 0 0.0% 257 34.6% 1 0.1% 517 35.5% 149 Annex 10 - UWEZO 2016 results by county 1. UWEZO Midline evaluation has just been done (2016) following a baseline in 2015 (Table 6). The key results of the evaluation are that: • The Tusome approach is having a strong, positive influence on reading outcomes, with relationships between project implementation and reading outcomes. • Reading outcomes for Class 1 and 2 pupils greatly improved during the one-year period between the baseline and midline evaluations. While impressive gains have been made, continuing with the Tusome approach will be critical to sustaining or improving on those gains. 2. The table below shows the results of the 2016 Tusome project, with counties covered by this baseline shown in yellow. Tusome (2016) County Ranks: Class 3 who can do Class 2 Work County Rank County Name OUTCOMES (Percentage) Class 3 who can do class 2 Work Teacher presence (%) Pupil Presence (%) 1. Nyeri 51.8 88.3 88.9 2. Nairobi 50.8 86.7 92.2 3. Mombasa 49.9 88.1 93.8 4. Nyandarua 46.3 91.3 85.5 5. Kajiado 42.3 83.9 88.1 6. Homa bay 39.6 88.8 81.9 7. Kiambu 39.5 87.6 90.5 8. Laikipia 39.2 90.4 86.0 9. Nandi 37.8 89.4 81.4 10. Kirinyaga 36.1 90.1 94.8 11. Uasin Gishu 35.3 81.4 88.8 12. Taita Taveta 35.1 82.8 87.3 13. Meru 35.0 90.6 87.2 14. Muranga 33.1 92.1 91.3 15. Tharaka Nithi 32.7 92.1 87.0 16. Nyamira 31.8 85.9 88.2 17. Elgeyo Marakwet 31.0 91.0 84.0 18. Nakuru 30.9 82.4 89.6 19. Kisumu 30.2 88.2 87.6 20. Embu 29.5 86.7 86.6 21. Kericho 29.0 92.7 88.5 22. Migori 28.7 86.8 81.9 23. Machakos 28.5 87.1 91.5 24. Kisii 27.7 87.7 84.0 25. Trans Nzoia 26.8 81.5 68.2 26. Kitui 26.1 91.3 81.4 27. Busia 25.9 88.5 84.1 28. Kilifi 25.9 84.5 83.3 29. Marsabit 24.5 90.5 92.5 30. Makueni 24.1 91.7 89.0 31. Siaya 23.9 85.7 83.5 32. Kakamega 22.0 88.4 82.1 33. Narok 21.4 89.3 85.1 34. Kwale 21.1 89.1 82.5 35. Vihiga 19.3 88.5 76.2 36. Bomet 19.1 85.6 84.9 37. Lamu 18.7 90.0 90.7 38. Tana River 18.2 86.4 86.6 39. Samburu 16.7 89.2 67.3 40. Baringo 16.6 88.9 81.5 41. Bungoma 15.4 89.6 83.0 42. West Pokot 15.4 88.5 79.3 43. Isiolo 15.4 88.5 89.8 44. Garissa 15.3 86.1 83.9 45. Turkana 12.9 83.9 76.7 46. Mandera 10.1 89.0 77.7 47. Wajir 9.9 89.2 83,6 Source: Are Our Children Learning (2016)? UWEZO KENYA SIXTH LEARNING ASSESSMENT REPORT DECEMBER 2016 Control or HGSMP Counties WFSMP/MGD Counties 150 Acronyms AOR – Adjusted Odds Ratio ASALs-Arid and Semi-Arid Lands BOM – Board of Management CI – Confidence Interval CSI – Coping Strategy Index DID – Difference- in - Difference DTL – Deputy Team Leader DtWI – Deworm the World Initiative ECDE – Early Childhood Development Education ECD – Early Childhood Development EMIS – Education Management Information Systems FCS – Food Consumption Score FEWS NET – Famine Early Warning Systems Network FGD – Focus Group Discussion GAIN –Global Alliance for Improved Nutrition GoK – Government of Kenya GPE – Global Partnership for Education GPS – Global Positioning System HGSMP – Home Grown School Meals Programme HH – Household IR – Inception Report JKUAT – Jomo Kenyatta University of Agriculture and Technology MGD – Mc Govern Dole MoA, L&F- Ministry of Agriculture, Livestock and Fisheries MOE – Ministry of Education MoH – Ministry of Health MS- Excel – Microsoft Excel NGO- Non Governmental Organization NSBDP- National School Based Deworming Programme. ODK- Open Data Kit PCD – Partnership for Child Development PMF- Performance Measurement Framework PMP- Performance Measurement Plan PRIEDE - Kenya Primary Education Development Programme PPS – Probability Proportionate to size PSM- Propensity Score Matching PSU- Primary Sampling Unit 151 PTA-Parents/Teachers Association SO -Strategic Objective SSU-Secondary Sampling Unit SMC- School Management Committee SMP- School Meals Programme SNM- School Nutrition and Meals SNMP- School Nutrition and Meals Programme STH –Soil Transmitted Helminthes TL- Team Leader TOR- Terms of Reference Tusome – (Let’s Read in Kiswahili – refers to USAID/UKAID funded Early Grade Reading Activity) UNICEF – United Nations Children’s Education Fund UKAID – United Kingdom Agency for International Development USAID – United States Agency for International Development USD – United States Dollars USDA – United States Department of Agriculture US – United States UWEZO – Kiswahili for ‘Capability’ VAM-Vulnerability Assessment Matrix WASH – Water Sanitation and Health WFP- World Food Programme WFPSMP-World Food Programme School Meals Programme WHO – World Health Organization