Foreign Agricultural 
Service, United States 
Department of Agriculture
Benin McGovern-Dole Project
BASELINE EVALUATION 
September 2019
DISCLAIMER: This publication was produced at the request of the United States Department of 
Agriculture. It was prepared independently by Et4d. 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.
Baseline Report Catholic Relief Services – United States 
Department of Agriculture (USDA) McGovern-Dole 
International Food for Education and Child Nutrition Program 
Evaluation Award # FFE-680-2017/027-00 
Program: McGovern-Dole International Food for Education and Child Nutrition 
Program 
Agreement Number: FFE-680-2017/027-00
Funding Year: Fiscal Year 2017 
Project Duration: 2017 - 2022 
Implemented by: Catholic Relief Services 
Evaluation Authored by: Evaluation Technology for Development (Et4d)
Isabelle McMahon
Brenda Sinclair
Michel Rousseau
 
 
Baseline Report
Catholic Relief Services – United States Department of Agriculture 
(USDA) McGovern-Dole International Food for Education and Child 
Nutrition Program Evaluation Award # FFE-680-2017/027-00
February 2019
Conducted by:
Evaluation Technology for Development (Et4d) 
Isabelle McMahon, Brenda Sinclair, Michel Rousseau
Acknowledgments
The evaluation team would like to thank Catholic Relief Services (CRS) and the U.S. Department 
of Agriculture (USDA) for their financial support. We would also like to thank the Ministry of 
Education of Benin, its regional and provincial offices in the Alibori and Department of Benin, as 
well as CRS’s staff for supporting and facilitating the rollout of the evaluation. We would also like 
to acknowledge Ms. Isabelle McMahon, Ms. Alice Michellazzi, Dr Michel Rousseau and Dr Brenda 
Sinclair from Evaluation Technology for Development (Et4d), and M. Emmanuel Ahouansinkpo, 
Ms. Emmanuelle Gbènakpon, M. Ousmane Maiga, M. Thierry Olatoundé, Dr Stacy Prieto, and 
Ms. Marie Rongeard, from Catholic Relief Services for their support and advice.
The views, opinions, and/or findings contained in this report are those of the author(s) and should 
not be construed as an official CRS or Government position, policy, or decision, unless so 
designated by other documentation. 
 
Table of Contents 
1 Introduction.............................................................................................................................. 1
1.1 Context............................................................................................................................. 1
1.2 Project description ........................................................................................................... 2
1.3 Program activities ............................................................................................................ 3
1.4 Evaluation purpose .......................................................................................................... 3
2 Methodology ............................................................................................................................ 5
2.1 Evaluation design............................................................................................................. 5
2.2 Data collection sources.................................................................................................... 5
2.3 Sampling strategy ............................................................................................................ 9
2.4 Data Analysis ................................................................................................................. 16
2.5 Challenges, issues and limitations................................................................................. 18
3 Field Work.............................................................................................................................. 20
3.1 Enumerator recruitment and training ............................................................................. 20
3.2 Data collection ............................................................................................................... 20
3.3 Quality control and Review Board.................................................................................. 21
3.3.1 Review Board......................................................................................................... 21
3.3.2 Field quality control ................................................................................................ 21
3.3.3 Data quality control ................................................................................................ 21
4 Results of Questionnaires and Observation, and PMP ......................................................... 22
4.1 Strategic objective 1: Improved literacy of school-aged children ................................... 22
4.1.1 Improved quality of literacy instruction................................................................... 22
4.1.2 Improved attentiveness.......................................................................................... 23
4.1.3 Improved student attendance................................................................................. 25
4.1.4 Foundational results............................................................................................... 29
4.2 Strategic Objective 2: Increased use of health and dietary practices ............................ 31
4.2.1 Improved knowledge of health, hygiene and nutrition practices ............................ 31
4.2.2 Increased knowledge of safe food preparation and storage practices................... 37
4.2.3 Increased knowledge of nutrition ........................................................................... 37
4.2.4 Increased access to clean water and sanitation services ...................................... 38
4.2.5 Increased access to preventative health intervention ............................................ 41
4.2.6 Foundational results............................................................................................... 41
5 Results of Early Grade Reading Assessment........................................................................ 42
5.1 Improved literacy indicator............................................................................................. 42
5.2 Snapshot of mean and max scores by school type ....................................................... 42
5.2.1 Mean scores by school type and gender ............................................................... 44
5.2.2 Mean scores by school type and location .............................................................. 44
5.3 Zero scores by school type............................................................................................ 45
5.3.1 Zero scores by school type and gender................................................................. 45
5.3.2 Zero scores by school type and location................................................................ 46
5.4 EGRA by reading categories ......................................................................................... 46
5.5 Literacy achievement index ........................................................................................... 48
5.5.1 Literacy achievement index by school type............................................................ 48
5.5.2 Literacy achievement index by gender................................................................... 49
5.5.3 Literacy achievement index by geographic location (Urban/Rural)........................ 49
5.6 Proposed targets............................................................................................................ 49
5.7 Factors associated with reading fluency ........................................................................ 50
5.7.1 Student characteristics........................................................................................... 50
5.7.2 Reading frequency at school.................................................................................. 51
5.7.3 Reading books at home ......................................................................................... 51
5.7.4 Family literacy ........................................................................................................ 52
5.7.5 Home study and reading habits ............................................................................. 52
Parental support .................................................................................................................... 53
6 Research Question #1: What are the participants’ capacities, knowledge and level of 
application of dietary, hygiene and health practices?.................................................................... 54
6.1 Context and methodology .............................................................................................. 54
6.2 Analysis and findings on hygiene................................................................................... 55
6.3 Analysis and findings on nutrition .................................................................................. 60
7 Research Question #2: What is the parents’ knowledge, attitude and practices regarding 
supporting their children to be successful at school? .................................................................... 65
7.1 Context and methodology .............................................................................................. 65
7.2 Findings and analysis .................................................................................................... 65
8 Research Question #3: Based on lessons learned from community participation, what 
strategies should be adopted to make this a success for McGovern-Dole FY17 .......................... 73
8.1 Context and methodology .............................................................................................. 73
8.2 Findings and analysis .................................................................................................... 73
9 Recommendations................................................................................................................. 77
9.1 Performance Monitoring Plan ........................................................................................ 77
9.2 Information to collect at midline ..................................................................................... 79
9.3 Intervention design......................................................................................................... 79
Tables
Table Executive Summary 1: Summary of completed surveys ......................................................... i
Table Executive Summary 2: Indicators from current PMP..............................................................ii
Table Executive Summary 3: Proposed new indicators .................................................................. iii
Tables 
Table 1: Baseline values and proposed targets for project indicators ............................................. 4
Table 2: Research questions........................................................................................................... 4
Table 3: FGDs with Child WASH Club ............................................................................................ 7
Table 4: FGDs with parents............................................................................................................. 7
Table 5: FGDs with teachers ........................................................................................................... 8
Table 6: Key informant interview ..................................................................................................... 8
Table 7: School sample ................................................................................................................... 9
Table 8: Summary statistics for schools ........................................................................................ 10
Table 9: Summary statistics for teachers....................................................................................... 11
Table 10: Summary statistics for principals ................................................................................... 11
Table 11: Summary statistics for the number of teachers interviewed per grade.......................... 12
Table 12: Summary statistics for interviewed teachers ................................................................. 12
Table 13: Summary statistics for interviewed parents ................................................................... 13
Table 14: Summary statistics for assessed students (Attentiveness indicator)............................. 15
Table 15: Summary statistics for students EGRA.......................................................................... 16
Table 16: Other improved literacy of school-age children indicators............................................. 22
Table 17: Attentiveness ................................................................................................................. 23
Table 18: Food consumption and attentiveness............................................................................ 24
Table 19: Other improved attentiveness indicators ....................................................................... 24
Table 20: Student attendance ....................................................................................................... 25
Table 21: Attendance per gender and grade................................................................................. 26
Table 22: Attendance per commune ............................................................................................. 26
Table 23: Percentage of parents who report spending time on literacy activities with their students 
the last three days ......................................................................................................................... 27
Table 24: Percentage patents who report that their child received help of somebody else the last 
three days...................................................................................................................................... 27
Table 25: Student reporting receiving from someone else in the past 3 days............................... 28
Table 26: Other improved attendance indicators........................................................................... 28
Table 27: Other improved attendance indicators – Suggested by Et4d ........................................ 28
Table 28: APE and AME active ..................................................................................................... 30
Table 29: Hygiene committee........................................................................................................ 30
Table 30: School feeding committee ............................................................................................. 31
Table 31: Percent of correct answers per question (Parent Test) ................................................. 32
Table 32: Percentage of parents with a passing score.................................................................. 33
Table 33: Average parent score .................................................................................................... 33
Table 34: Percent of correct answers per question (Student test) ............................................... 34
Table 35: Percentage of students with a passing score of 70%.................................................... 35
Table 36: Average student score................................................................................................... 35
Table 37: Percentage of children receiving a minimum acceptable diet........................................ 36
Table 38: Other increased knowledge of nutrition indicators......................................................... 38
Table 39: School with soap (or ash) and water at hand washing stations commonly used by 
students ......................................................................................................................................... 39
Table 40: Increased access to clean water and sanitation services indicators ............................. 39
Table 41: Availability of water........................................................................................................ 40
Table 42: Number of washing stations .......................................................................................... 40
Table 43: Number of latrines ......................................................................................................... 40
Table 44: Proportion of students meeting the improved literacy indicator..................................... 42
Table 45: Distribution of reading comprehension scores by school type....................................... 42
Table 46: EGRA mean scores by school type............................................................................... 43
Table 47: EGRA max scores by school type ................................................................................. 44
Table 48: EGRA mean scores by school type and gender............................................................ 44
Table 49: Mean scores by school type and urban/rural location ................................................... 45
Table 50: Zero scores by school type............................................................................................ 45
Table 51 Zero scores by school type and gender.......................................................................... 46
Table 52 : Zero scores by school type and urban/rural location.................................................... 46
Table 53: Reading category definitions ......................................................................................... 47
Table 54: Differences between percentage of program and control students in each reading level
....................................................................................................................................................... 48
Table 55: Literacy achievement index point scale......................................................................... 48
Table 56: Literacy index by school type ........................................................................................ 49
Table 57: Literacy index by school type and gender ..................................................................... 49
Table 58: Literacy index by commune by school type and location (urban/rural).......................... 49
Table 59: Regression coefficient for student characteristics ......................................................... 51
Table 60: Regression coefficient for reading in class .................................................................... 51
Table 61: Regression coefficient for French reading books at home ............................................ 51
Table 62: Regression coefficient for family literacy ....................................................................... 52
Table 63: Regression coefficient for home study and reading habits............................................ 52
Table 64: Regression coefficient for good grades ......................................................................... 53
Table 65: Regression coefficient for bad grades ........................................................................... 53
Table 66: Scale for proportion ....................................................................................................... 54
Table 67: Students’ knowledge and practices of hygiene rules based on teachers’ answers ....... 55
Table 68: Important hygiene practices mentioned by parents....................................................... 56
Table 69: Parent and child hand washing frequency per day........................................................ 57
Table 70: Occasion for hand washing for Parents......................................................................... 57
Table 71: Occasion for hand washing for children......................................................................... 58
Table 72: Water treatment used .................................................................................................... 58
Table 73: Parents’ response regarding where their children defecate when they are not in school
....................................................................................................................................................... 59
Table 74: Children’s response regarding the last place they defecated........................................ 59
Table 75: Knowledge and practices of nutrition’s rules based on teachers’ answers.................... 60
Table 76: Parental knowledge of well-balanced diet ..................................................................... 61
Table 77: Proportion of students receiving help based on teachers’ perception ........................... 66
Table 78: Proportion of parents reporting that someone tells stories to their children at home..... 66
Table 79: Children reporting being told stories.............................................................................. 67
Table 80: Proportion of parents reporting talking to their children about their school experience . 68
Table 81: Children reporting being asked what they did at school ................................................ 68
Table 82: McGovern-Dole activity involvement ............................................................................. 73
Figures
Figure 1: Departments (Regions) and Communes of Benin ........................................................... 1
Figure 2: Parents’ test scores........................................................................................................ 33
Figure 3: Students’ test scores ...................................................................................................... 35
Figure 4: Distribution of program and control group scores at each reading level ........................ 47
Figure 5: Reason for not practicing proper hygiene rules.............................................................. 56
Figure 6: Reasons for not having a balanced diet ......................................................................... 61
Figure 7: Reasons for not being able to provide a balanced diet .................................................. 62
Figure 8: Types of feedback parents report giving when their child gets good grades.................. 69
Figure 9: Types of feedback students report receiving when they get good grades ..................... 69
Figure 10: Types of feedback parents report giving when their child gets bad grades.................. 70
Figure 11: Types of feedback students report receiving when they get bad grades ..................... 71
Figure 12: Difficulty finding volunteers.......................................................................................... 74
Figure 13: Number of years working as cook ................................................................................ 75
Abbreviations and Acronyms
AAM Assessor Accuracy Measurement
AME Mother Association (Association des Mères d'Élèves) 
APE Parent Association (Association des Parents d'Élèves) 
ATPC Community Led Total Sanitation (Assainissement Total Par la Communauté) 
CCS Head of School Districts (Chef de circonscription) 
CRP Education District Officer (Chef de Région Pédagogiques) 
CI First grade (Cours d’Initiation) 
CLTS Community Led Total Sanitation (idem ATPC)
CP Second grade (Cours Préparatoire) 
CP Educational Adviser (Conseiller pédagogique)
COGES Canteen Management Committee (Comité de Gestion des Cantines 
scolaires) 
CRS Catholic Relief Services
CWPM Correct Words per Minute
DANA Direction of Food and Applied Nutrition
DAS Direction of School Feeding
DDHAS Direction of Basic Hygiene and Sanitation (Direction département de 
l'hygiène et de l'assainissement de base) 
DEP Direction of Primary Education
DEMP Direction of Preschool and Primary Education
DIIP Direction of Pedagogical Inspection and Innovation
EGRA Early Grade Reading Assessment
FFE Food for Education
ICC Intraclass Correlation Coefficient
INFRE National Institute for Training and Research in Education
KII Key Informant Interviews 
MAD Minimum Acceptable Diet
MEMP Ministry of Preschool and Primary Education
MT Metric Tons
ORF Oral Reading Fluency 
PfD Partners for Development
PHAST Participatory Health and Sanitation Transformation
PMP Performance Monitoring Plan
PPS Probability Proportion Sample
SFCM Soy-Fortified Corn Meal 
SO Specific Objective
THR Take Home Ration
TOR Terms of Reference
UP Unité Pédagogique
USG United State Government
URP Union Régionale des Producteurs
USDA United States Department of Agriculture
WASH Water Sanitation and Hygiene
WEI World Education Inc
i
Executive Summary
Context and intervention
Catholic Relief Services (CRS) is implementing the second phase of a United States Department 
of Agriculture (USDA) McGovern-Dole International Food for Education and Child Nutrition 
Program, hereafter referred to as McGovern-Dole project, in two districts in Benin. CRS works in 
partnership with sub-partners World Education International (WEI), Caritas Kandi, Partners for 
Development (PfD) and Union Régionale des Producteurs (URP), as well as coordinates with 
government ministries and departments, principals, teachers, and parent groups.
The key objective of the McGovern-Dole program is to reduce hunger and improve literacy and 
primary education. The program provides school meals, as well as teacher training, health, 
sanitation, and hygiene interventions and other activities. The intended beneficiaries of the 
McGovern-Dole program are 50,965 primary school students in targeted districts of Borgou and 
Alibori Departments, their families, and communities.
Evaluation
This baseline study is the first major activity of the program’s evaluation plan. The baseline 
evaluation serves three main objectives. First, it provides baseline values for key program 
indicators that will support CRS Benin to refine program targets and eventually modify or add 
custom indicators. Secondly, the study presents evaluation data and findings to inform the 
following research questions: 
1. What are the participants’ capacities, knowledge and level of behavioral change for
dietary, hygiene and health practices (KAP Survey)? 
2. What are the parents’ knowledge, attitude and practices regarding supporting their children 
to be successful at school? (KAP Survey)?
3. Based on lessons learned from community participation in McGovern-Dole FY14, what 
strategies should be adopted to make this (program or intervention) a success under 
McGovern-Dole FY17? 
Thirdly, it includes an improved theory of change and provides recommendations. The evaluation 
baseline is a quasi-experimental, mixed-methods design. Using probability proportionate to size, a 
sample of program and control schools were randomly selected from the CRS program 
implementing area. To provide baseline values for each of the indicators and responses to the 
three research questions, Et4d employed several quantitative and qualitative methods. The 
quantitative data collection tools included: An Early Grade Reading Assessment (EGRA), multiple 
questionnaires of students, parents, principals, teachers, cooks and store keepers, as well as 
administration of two observation tools – student attentiveness observation and school 
observation.
The qualitative component of the evaluation consisted of in-depth interviews with local 
government authority figures as well as non-governmental organizations (NGO) partners and 
focus groups discussions (FGDs) and key stakeholders, such as teachers, parents, and children 
who were also members of a WASH Club. Data was collected in schools which will receive the 
intervention (program schools), and in schools which do not have canteens (control schools). 
ii
Table Executive Summary 1: Summary of completed surveys
Type of data Control Project
EGRA + Student Survey 773 748
Observation + Student Survey 964 957
Parent Survey 513 517
Teacher Survey 98 96
Principal Survey 49 50
Cooks Survey N/A 112
Storekeeper Survey N/A 38
School observation 49 50
The control schools were in communes neighboring the intervention areas (Banikoara, 
Bembereke, Nikki and Sinende). They were selected to serve as a control group for the baseline, 
midline and end line evaluations; however— if a canteen is being established in any of the control 
schools, they will to be replaced.
Balance tests were performed to demonstrate the adequacy of the control group as a comparison. 
Statistically significant differences were noted when detected, but no weighting was applied to 
correct imbalances for the baseline results. If necessary, we will add propensity score matching to 
the difference-in-difference analysis at end line. Data has been weighted to represent the entire 
population of students, teachers, principals, or parents, within school clusters and strata of urban 
or rural schools. 
Indicators
Results of the baseline study revealed that in some areas such as hygiene practices there were 
significant differences between control schools and program schools, which were most likely due 
to the lasting effect of the previous intervention, McGovern-Dole FY14. The study results 
suggested a need to revise the target values for the literacy indicator. 
The baseline values and proposed targets for each indicator are illustrated below. The evaluation 
team included additional proposed custom indicators that would be useful to evaluate the impact 
of the program. 
Table Executive Summary 2: Indicators from PMP
Indicator Baseline 
Value
Proposed 
Target
Comments
Percentage of students who, by the 
end of two grades of primary 
schooling, demonstrate that they can 
read and understand the meaning of 
grade level text 
Benchmark: 60% (3/5 correct 
answers)
Original: 
2% 
Reality: 
1%
Original:
32%
Suggestio
n: 20% 
The initial target included in the 
performance indicators (Attachment D) 
of 32% does not seem realistic, based 
on the very low EGRA score. Et4d 
propose to lower it to 13 %, which is 
still very ambitious. Currently 1%, to 
which 3% should be added every year.
iii
Indicator Baseline 
Value
Proposed 
Target
Comments
Percentage of parents who report 
spending time on literacy activities 
with their students in the preceding 
three days
31.1% N/A This indicator might be subject to 
social bias and might give a partial 
view of the assistance received by 
students. Et4d suggests removing the 
indicator and replacing it by percent of 
students who report receiving help 
outside of school in the preceding 
three days
Percentage of students in target 
schools who are identified as 
attentive during class/instruction
67.6% 75% Target could remain identical to 
McGovern-Dole FY14
Percentage of schools with soap and 
water at a hand washing station 
commonly used by students
54.4% 68% Target is based on 25% increased 
practice.
Percentage of students in target 
schools who achieve a passing 
score on a test of good health and 
hygiene practices
Passing Score: 70%
34% 40%
Percentage of Children receiving a 
minimum acceptable diet
68.9% 75 Et4d suggest keeping this indicator. 
The current values are high because 
the schools are still providing food 
Et4d proposed target values which we believe to be achievable and passing score values for the 
test of good health and hygiene practices, while CRS confirmed that the benchmark for the EGRA 
reading comprehension test should be 60% correct answers.
The evaluation team is proposing to collect data on the following six (6) custom indicators which 
will help evaluate new activities that we been added to project. 
We would like to include some additional custom output indicators in the PMP to evaluate new 
activities included in the intervention, and to provide greater insights into some outcome 
indicators.
Table Executive Summary 3: Proposed new indicators
Indicator Baseline 
Value
Proposed 
Target
Comments
Literacy Achievement Index 35 45 Et4d proposes this indicator, which gives a 
global indication of students’ performance 
distribution on letter reading, reading fluency 
and reading comprehension subtasks, 
providing a broader picture of reading 
competence
Average student attendance 
rate in USDA supported 
classrooms/schools
88.2% 93% This indicator was calculated based on USDA 
new requirements.
iv
Indicator Baseline 
Value
Proposed 
Target
Comments
Control school had an attendance rate of 
92%, project schools should have better 
results.
Percentage of students who 
report receiving help outside 
school in the preceding three 
days
64.8% 70% Values are already high, but improvements 
can be achieved
Percentage of parents in 
target schools who achieve a 
passing score on a test of 
good health and hygiene 
practices
Passing Score: 80%
45.9% 60% Et4d suggests adding this indicator. Parents 
in control schools had a 55% passing rate. 
Therefore, 60% seems achievable.
Percentage of schools 
meeting WASH-Friendly 
School standards
N/A 80% This indicator needs to be defined based on 
criteria which defines a WASH-Friendly school
Percentage of puberty-aged 
children who achieve a 
passing score on the test of 
puberty health and hygiene 
test
N/A Target and a passing score could be 
established during the midline evaluation
Add specific custom output 
indicators for the activity 
undertaken by World 
Education International (WEI) 
based on the updated 
program design, they could 
include literacy events 
organized, library created, 
community of practice 
meeting held, etc.
N/A WEI activities were agreed upon and defined 
precisely during early January discussion. 
Therefore, the updated PMP could be more 
specific in terms of output indicators.
Question #1: What are the participants’ capacities, knowledge and level of 
application of dietary, hygiene and health practices?
To assess the level of knowledge and behavior change related to proper hygiene and health 
practices, the evaluation team collected data from teachers, parents and students. Overall, the 
data showed that hygiene and health prevention best practices are well known by children and 
parents alike across both control and intervention schools. Concerning the level of practice, it 
appears that students and parents report washing their hands with a higher propensity in program
schools than in control schools. For water treatment practices, parents in program schools are 
more likely to treat the water before drinking it. These results are very encouraging, and it is highly
likely that these levels of practice are due to the program activities implemented during 
McGovern-Dole FY14. 
Based on teachers’ feedback, practices could still be notably improved by increasing children’s 
awareness of the risk factors involved if not applying improved/best hygiene practices. The 
WASH-friendly schools and the Community-Led Total Sanitation interventions should help 
reinforce the importance of hygiene practices and behaviors. 
v
Unfortunately, among parents, knowledge regarding the importance of proper dietary balance 
remains very low. Mother Association members (AME) have been trained on balanced diet issues, 
but their acquired knowledge and practices did not spillover to other community members. 
Practices are also very poor, partially because of the lack of awareness, but mostly because 
parents have no access to proper food or no means to acquire it. 
Question #2: What is the parents’ knowledge, attitude and practices regarding 
supporting their children to be successful at school?
EGRA scores revealed that parental involvement had the strongest association with improved 
student reading. Therefore, it appears essential to encourage parents to play an active role in their 
children’s education.
In our inquiries, we looked at parents’ or caregivers’ involvement broadly. We expanded our
inquiries beyond structured help such as assisting students with homework and practicing 
lessons. We included different aspects of parents’ engagement in education such as telling 
stories, discussing students’ school experience, and encouraging their children to do well in 
school. These components of parents’ engagement are related to some of the parents’
involvement in improvement activities planned for McGovern-Dole FY17, such as radio story-time, 
as well as community library and awareness campaigns. 
Teachers expressed that parents were not providing enough support and attention to their 
children’s homework. They communicated to us that parents often barely knew which grade their 
children were enrolled in and whether they were able to follow instruction in class.
Overall, the proportion of parents who reported that their children were told stories is relatively 
high; and the information reported by students was significantly higher in program schools, where 
44% of students reported that someone told them stories versus 24% in control schools. Based on 
these data, it seems that McGovern-Dole FY14 intervention might have produced some lasting 
behavioral changes on parents of early grade children. 
The second aspect of engagement considered was whether parents talk daily with their 
children about their school experience. Parents overwhelmingly reported that they asked their 
child what they had learned that day. The analysis shows that it was more likely to be the case in 
program schools than in control schools. However, the parents’ education appears to be a major 
influential factor and, unsurprisingly, literate parents are more likely to inquire about their child’s 
day than illiterate parents.
The last aspect investigated was parents’ feedback and attitude when their child received good 
or bad grades. The parents overwhelmingly responded that they encouraged their child, while 
children reported that their parents did nothing when they received good grades. Parents and 
children alike reported that the most common reactions to bad grades were scolding, punishment 
or being hit. In Benin, there is still a widely held belief that harsh parental action leads to higher 
academic achievement. Some anti-punishment or positive reinforcement awareness campaigns 
might be useful to encourage parents to change their attitude and provide more encouragement to 
their children.
Despite the empirically proven benefits of parent engagement in their children’s education, there 
are many factors that impact families’ ability or willingness to support their children’s learning. For 
instance, one of the key aspects to keep in mind is that about 70% of parents in northern Benin 
are illiterate (see table 13). Therefore, most of them report that they feel a sense of inadequacy 
when it comes to helping their children with their homework.
vi
Question #3: Based on lessons learned from community participation, what 
strategies should be adopted to make this a success on McGovern-Dole FY17? 
Achieving a high level of community participation is necessary to operate a program smoothly and 
successfully, as well as to ensure sustainability. CRS identified three main areas of possible 
concern for McGovern-Dole FY17: 1) difficulties to find volunteers, 2) frequent complaints from the 
cooks, and 3) challenges to collect parents’ financial contribution for school meals. 
During our evaluation, principals did not report volunteer recruitment challenges, even if it was 
the case when McGovern-Dole FY14 was initiated. Most cooks work every day and have been 
doing so for the past four years. Cooks complain mostly about not receiving adequate 
compensation for their work, not having soap to wash their apron, and not receiving a medical 
check-up. Nonetheless, the evaluation team did not feel that the complaints led to any serious 
decline in their motivation to remain active with the program. 
Several principals mentioned difficulties encountered with collecting parents’ financial 
contribution. We identified four modes for collecting the contribution, depending on the 
community, and source of the parents’ revenues. 
1. Collecting the contribution from the students daily via the teachers
2. Collecting the contribution from the students weekly via the teachers
3. Collecting the contribution from the parents monthly or per term via the AME/APE
4. Collecting the contribution from the harvest fund yearly 
Parent Association (Association des Parents d'Élèves) (APE) members reported that the 
difficulties initially encountered to collect the contributions for the canteen have lessened over the 
course of the McGovern-Dole FY14 program. However, it might be beneficial for CRS to renew its 
guidance on best practices, based on the latest practices experienced by different participating 
schools. 
1
1 Introduction
1.1 Context
Benin is one of the world’s poorest countries, with an annual 
gross domestic product per capita of US$2,266, which is
below the Sub-Saharan Africa average of US$3,806.1 The 
United Nations Development Program (UNDP) ranks it 163 
out of 189 countries on the Human Development Index. 
Poverty is highest in northern Benin, where Catholic Relief 
Services (CRS’s) McGovern-Dole program is implemented.
French is the official government language of Benin and the 
only language of instruction. However, indigenous languages 
such as Fon and Yoruba are commonly spoken, and Benin 
has over 50 indigenous languages in total. In the northern 
region, where CRS is implementing the McGovern-Dole
project, the four most common languages are: Bariba, Dendi, 
Peulh and Mokole; it is important to note that children 
entering primary school do not speak any French. 
Social services are lacking in the north, as exemplified by an 
insufficient number of primary schools. Although the four
targeted departments (regions)2 in Northern Benin represent 
almost 75% of Benin’s land mass, fewer than half (48%) of 
the country’s primary schools3 are located there- resulting in
schools being very remote for many school-aged children to 
attend. Local customs and beliefs place little value on 
education; and parents—most of whom are poorly educated 
themselves—are generally not involved in their children’s 
education4. Moreover, poor nutrition due to lack of availability 
of food and dietary diversity—compounded by a severe, annual dry season (December to April)—
affects children’s ability to learn and grow. Borgou and Alibori Departments in particular receive 
little assistance from either the government or non-governmental organizations (NGOs) to support 
their school canteens.
Recently, the Government of Benin (GoB) has made an important political commitment to 
support5 school canteens. For the 2016-17 Government Budget, the budget for school feeding 
programs has increased from 1 billion West African Franc (CFAF6) to 7 billion CFAF, and reached 
the goal of 51% of all primary schools implementing a school feeding initiative
1 The statistics provided in this section are from the latest UNDP Human Development Report (2018) and the World Bank Database 
(2017) World Bank, International Comparison Program database. 
2 Northern Benin includes 4 departments (or regions): 1) Atakora, 2) Alibori, 3) Donga, and 4) Borgou
3 Benin’s primary school system has three cycles featuring six types of introductory, preparatory, elementary and middle-level 
classes with the following levels: Grades 1-2 (CP-CE1), Grades 3-5 (CE2- CM1-CM2).
4 MCGOVERN-DOLE FY14 – Final evaluation - Advisem
5 Canteen will be created, and food provided in the selected schools, similar to the model implemented by CRS
6 CFCF is Communauté Financière d'Afrique which is the West African Franc (1 billion =1.7 Millions USD)
Figure 1: Departments (Regions) 
and Communes of Benin
2
1.2 Project description
Catholic Relief Services (CRS) Benin implemented a United States Department of Agriculture 
(USDA) McGovern-Dole International Food for Education and Child Nutrition Program, entitled 
Food for Education (McGovern-Dole FY14) from 2014 to 2017, hereafter referred to as 
McGovern-Dole. In 2017, CRS was awarded funding from the USDA for the second phase (2018-
2022) of the Food for Education project in Benin, referred to as McGovern-Dole FY17. 
Unfortunately, due to a teachers’ strike which took place from mid-January to April 2018, the 
project launch date was postponed until March 2019. However, since not all the USDA 
commodities had been used under McGovern-Dole FY14, the school feeding program from 
McGovern-Dole FY14 was sustained during the first term of the 2018-2019 school year
(September to December). 
For McGovern-Dole FY17, CRS will continue to work in the same communities, with the same 144 
schools in Gogounou, Malanville, and Kandi Communes (Alibori Department), and Kalalé 
Commune (Borgou Department)
McGovern-Dole Phase II, similar to McGovern-Dole Phase I, focuses on improving the literacy, 
health and dietary practices of school-age children (grade 1 and 2) in northern Benin. The 
objectives of McGovern-Dole FY17 project are: 
• To improve the literacy of school-age children by improving the quality of literacy instruction, 
improving attentiveness, and improving student attendance;
• To improve student attentiveness by providing school meals;
• To improve student attendance through raising awareness of the importance of education, 
enrollment campaigns, and training parent-teacher associations (PTAs); 
• To increase the use of health and dietary practices by improving knowledge of health and 
hygiene practices, increasing knowledge of safe food preparation and storage practices, 
increasing knowledge of nutrition, increasing access to clean water and sanitation services, 
and increasing access to requisite food preparation and storage tools and equipment.
McGovern-Dole FY17 will build upon the achievements of McGovern-Dole FY14, with plans to
complete or transfer activities to the communities and government by the end of the project. With 
the McGovern-Dole FY17 project, CRS wishes to sustain a healthy learning environment for 
children at home, in the community, and at school. The project will strengthen local, regional, and 
national institutions such as the Mothers Associations (AMEs), school districts, and the National 
Institute for Training and Research in Education (INFRE). McGovern-Dole FY17 will strengthen
literacy activities by expanding and improving teacher training, establishing bi-monthly story clubs, 
as well as training parents on students’ literacy progress and weaknesses with devised 
appropriate action plans. The project will also expand activities from the schools into the 
communities. For example, there will be community outreach through Participatory Health and 
Sanitation Transformation (PHAST) activities to improve water, sanitation and hygiene behaviors 
through Community-led Total Sanitation (CLTS) implemented along with literacy activities via 
community story hours and children’s radio literacy programs. Finally, McGovern-Dole FY17 will 
strengthen the enabling environment to improve the quality of education through advocacy with 
government to prioritize primary education and health-related initiatives in its national strategies 
and agendas.
McGovern-Dole FY17’s school feeding program will require 6,610 metric tons (MT) of USG 
donated Soy-Fortified Corn Meal (SFCM), green split peas, rice, lentils, and vegetable oil 
commodities. McGovern-Dole FY17 will distribute take-home rations (THR) consisting of 4 liter of 
vegetable oil per student, to male and female students in grades 4 to 6 who have achieved 90% 
or above attendance during the previous three months in year one (2019) of the project. In years 
two, three, and four (2020-2022), a THR, consisting of 4 liter of vegetable oil per student, will be 
distributed to male and female students in grade 6 who achieved at least 95% attendance rate 
3
over three months. In order to phase out the McGovern-Dole project, school feeding will be 
sustained through gradual community contributions and capacity building. The project will 
establish community farms that will contribute food to school meals and sell food products in local 
markets as a source of income and revenue for school activities. McGovern-Dole FY17 will 
continue community nutrition education and outreach activities such as media programs, mothers’ 
clubs, listening and discussion groups established under McGovern-Dole FY14 to improve dietary 
diversity and nutritional status. The project will also continue to support the school gardens
established under McGovern-Dole FY14 for use as demonstration plots for growing nutritious and 
dietary diverse foods. As in McGovern-Dole FY14, the McGovern-Dole FY17 project will promote 
WASH-clubs and WASH-friendly schools, rehabilitate or construct water systems and latrines, and 
promote menstrual hygiene awareness and management7. 
CRS Benin will work with its partners and sub-recipients, including World Education Inc. (WEI), 
Caritas Kandi, Regional Union of Producers (Union Régionale des Producteurs or URP) and 
Partners for Development (PfD). CRS Benin will also collaborate with the Ministry of Pre-school 
and Primary Education (Ministère de l’Enseignement Maternelle et Primaire or MoE), the 
Administration of School Feeding (Direction de l'Alimentation Scolaire or DAS), the Administration 
of Food and Nutrition (Direction de l'Alimentation et de la Nutrition Appliquée or DANA), the 
National Institute for Training and Research in Education (INFRE), the Department of Public 
Education, the Direction of Pedagogical Inspection and Innovation, the national council on 
Education, the Ministry of Secondary Education and Vocational Training, Directorate of the Family 
and Social Affairs (DGFAS) of the Ministry of work, Public Function and Social affairs, the Ministry 
of Plan and Development, the Ministry of Agriculture Livestock and Fisheries, the Ministry of 
Water and the Ministry of Health.
1.3 Program activities
To achieve the aforementioned strategic objectives and expected results, the project will carry out 
the following activities: 
1. Building and rehabilitation of improved latrines 
2. Building and rehabilitation of wells and water stations
3. Capacity building at local, regional, and national levels 
4. Scaling-up of WASH-friendly schools
5. Distribution of school supplies and materials 
6. Establishment of activities to promote literacy 
7. Establishment and maintenance of school gardens 
8. Establishment of community farms
9. Promotion of teacher attendance 
10. Provision of school meals (lunch)
11. Provision of take-home rations of food
12. Raise awareness of the importance of education 
13. Training on food preparation and storage
14. Training on good health and sanitation 
15. Training for parents-teachers association
16. Training for teachers and school administrators
1.4 Evaluation purpose
7 CRS Evaluation Plan for McGovern-Dole 2017-2022 
4
The baseline study will first be used to refine program targets for indicators included in the 
Performance Monitoring Plan (PMP) and eventually modify or add custom indicators. The results 
will serve as the reference point to measure McGovern-Dole FY17 progress in the midterm and 
final evaluations. The baseline evaluation will establish baseline values for the following indicators 
(Table 1). All results will be disaggregated by gender.
Table 1: Baseline values and proposed targets for project indicators
Indicator Baseline 
Value
Proposed 
Target
1 Percent of students who, by the end of two grades of primary schooling, 
demonstrate that they can read and understand the meaning of grade level 
text 
Benchmark: 60% (3/5 correct answers)
1% 20% 
2 Percent of parents who report spending time on literacy activities with their 
students in the preceding three days
31.1% N/A8
3 Percentage of students in target schools who are identified as attentive 
during class/instruction 67.6% 75%
4 Percent of schools with soap and water at a hand washing station 
commonly used by students
54.4% 68%
5 Percentage of students in target schools who achieve a passing score on a 
test of good health and hygiene practices
Passing Score: 70%
34% 40%
6 Percentage of Children receiving a minimum acceptable diet 68.9% 75%
Secondly, it will respond to the following research questions (Table 2): 
Table 2: Research questions
1 How will the targeted interventions increase the participants’ (students and parents) 
capacities, knowledge and uptake of dietary, hygiene and health practices and 
behaviors? (measured through a KAP Survey)
2 What is the parents’ knowledge, attitude and practices regarding supporting their children 
to be successful at school? (measured through a KAP Survey)
3 Based on lessons learned from the previous McGovern-Dole FY14, what strategies 
should be adopted to increase community participation?
Thirdly, the evaluation will be used to refine the projects theory of change and provide recommendations. 
 
8 Et4d suggested replacing this indicator, therefore we did not propose a target value
5
2 Methodology
2.1 Evaluation design
The baseline evaluation is grounded in a quasi-experimental, mixed-methods design. Using
probability proportionate to size (PPS)9 sampling method a sample of both program and control 
schools were randomly selected from the CRS project implementation area. 
To provide baseline values for each of the indicators and responses to the three research 
questions, quantitative and qualitative methods were employed. The quantitative data collection 
tools include: An Early Grade Reading Assessment (EGRA), questionnaires for students, parents, 
principals, teachers, cooks and store keepers, as well as two observational tools – 1) a student 
attentiveness observation and 2) school observation.
The qualitative component of the evaluation consisted of 1) key in-depth interviews with authority 
figures and NGO partners and 2) focus groups discussions (FGD) with key project stakeholders
including teachers, parents, and children who were also members of WASH clubs.
To control for any important effect on the results due to schools’ characteristics, an equivalent 
sample of schools from both urban and rural populations were selected. The difference between 
the program schools and the control schools is described in the school sampling section (Section 
2.3) Any differences will be taken into consideration at the end line by using the propensity scores 
matching (PSM)10 approach to correct for group non-equivalence. Before running the difference￾in-difference analysis for the outcomes, we will compute propensity scores, based on any 
significant differences at baseline, that will be added to the difference-in-difference regression 
analysis to control differences between groups.
2.2 Data collection sources
As described above, the evaluation design features five data collection methods: 1) a 
document/literature review, 2) the EGRA assessment, 3) six survey questionnaires, 4) focus 
group discussions (FGDs) and 5) key informant interviews (KIIs).
Document review
The document/literature review consisted of a thorough examination of program documentation 
provided by CRS relevant to McGovern-Dole (FY14 and FY17) project activities in Benin. This 
review led to the development of the inception report and the data collection tools. The documents 
included:
 The three reports related to McGovern-Dole FY14 (Baseline, Midline, Endline)
 The data collection instruments used for McGovern-Dole FY14
 CRS’ McGovern-Dole FY17 agreement with USDA and the 
 Performance Monitoring Plan (PMP)
9 PPS: Probability Proportionate to Size is a sampling method in which the probability of selecting 
a unit is directly proportional to the unit's measure of size. 
10 PSM: Propensity Score Matching is a statistical matching technique that attempts to estimate 
the effect of a treatment, policy, or other intervention by accounting for the covariates that predict 
receiving the treatment
6
The TOR was drafted in 2017 and since then while CRS asked Et4d to work with an updated 
version of the PMP. As a result, some indicators were modified and needed to be clarified for the 
evaluation team during a kick-off meeting in Cotonou with the M&E team and the Chief of Party. 
EGRA 
In preparation for the EGRA, the evaluation team reviewed the existing EGRA tool designed 
under McGovern-Dole FY14. Since it had already been used in three prior evaluations and was 
also used as a monitoring tool by World Education Inc (WEI), it was not possible to use the 
original tool for the McGovern-Dole FY17 baseline. Moreover, the tool was designed without 
considering the latest EGRA standards (Second Edition of the EGRA Toolkit11). Thus, Et4d 
adapted the tool with CRS’ consent, particularly the listening and reading passages, as well as the 
comprehension questions. Unfortunately, WEI or INFRE did not validate the instrument, since the 
timing of the evaluation did not allow for an instrument design workshop or piloting. The 
instrument’s items are featured in Annex K. The Early Grade Reading Assessment (EGRA) was
administered one-on-one to students in grade 3, which was considered an acceptable proxy for 
the assessment of reading and comprehension skills in French at the end of grade 2. It is planned, 
as part of MGD II to hold a workshop with INFRE in April 2019 to update the EGRA tool used in 
Benin. However, in order for the new tool to be used for midline and endline evaluation, WEI will 
need to equate the tools.
Surveys - questionnaire
To establish baseline values for indicators and to collect data related to the three research 
questions, the evaluation team used six survey questionnaires.
• 1. Principal questionnaire: To collect school level data, teacher attendance, and evaluate 
parent’s involvement in school activities. See survey tool in Annex A. 
• 2. Parent questionnaire: To assess WASH and nutrition knowledge and practices, and their 
involvement in students learning and in school activities. See survey tool in Annex B. 
• 3. Teacher questionnaire: To determine teacher motivation to stay in the same school and
extent of parent’s involvement in students learning. See survey tool in Annex C. 
• 4. Student questionnaire #1: To assess WASH and nutrition knowledge, practices and 
behaviors and to measure students’ dietary diversity. See survey tool in appendices Annex D. 
• 5. Student questionnaire #2: To understand parents’ and families’ involvement in learning 
activities. See survey tool in appendices Annex E. 
• 6. Cook and storekeeper questionnaire: To gauge knowledge of WASH practices and 
behaviors in food storage and preparation. See survey tool in appendices Annex F. 
Student attentiveness observation
At each of the 100 schools visited, the attentiveness of students during classroom instruction was 
assessed in two randomly selected classrooms (grades 4, 5 or 6). Within each classroom, 10 
randomly selected students were deemed as attentive or inattentive at the time of observation, 
following a method similar to what was performed for McGovern-Dole FY14. Girls and boys were 
11 RTI International. (2015). Early Grade Reading Assessment toolkit, Second Edition, prepared for USAID under the Education Data 
for Decision Making (EdData II) project, Research Triangle Park, NC: RTI. https:// www.eddataglobal.org/documents/index. 
cfm?fuseaction=pubDetail&id=929
7
differentiated in the data recorded, which showed that an equal number of girls and boys were 
observed.
School observation
The evaluation team conducted two field observations. The first was to observe at the school if 
water and soap (or the locally-accepted substitute ash) were available at hand-washing stations 
and whether hand-washing stations were properly used. The second was a general observation of 
the physical school environment (e.g. examining the condition of kitchens, storage rooms and 
latrines). These observation tools are featured in Annex G.
Focus group discussions
Focus group discussions (FGDs) were held in each of the four communes with parents and
teachers as well as children from the WASH clubs. Each of the FGD was help separately. As 
illustrated in Table 3, 12 children from three communes participated in the WASH club focus 
group discussions. The number of children in each FGD ranged from one in Kandi to six in 
Gogounou.
Table 3: FGDs with Child WASH Club
Commune Number of 
Participants
Number of 
FGD 
Gogounou 6 2 
Kalale 5 1 
Kandi 1 1 
Total 12 4 
A total of 34 parents participated in the parent FGDs (Table 4). The number of parents FGDs 
varied per commune ranging from one in Kalale to four in Malanville. Applying the principle of 
gender sensitivity, Et4d held same-sex FGDs when puberty issues were to be discussed. This 
approach helped women to feel comfortable expressing their views and enabled them to share 
issues and concerns not typically discussed in mixed groups due to gender norms.
Table 4: FGDs with parents
Commune [# FGD] Number of 
Participants
Number of 
FGD
Gogounou [2] 8 2
Kalale [1] 5 1
Kandi [2] 8 2
Malanville [4] 13 4
Total 34 9
Twenty-nine teachers from four communes participated in FGDs (Table 5). The number of 
teachers in each FGD ranged from 5 to 10, falling within an acceptable standard for ensuring all 
participants’ voices are equally represented.
8
Table 5: FGDs with teachers
Commune Number of 
Participants
Number of 
FGD
Gogounou 9 2
Kalale 10 2
Kandi 5 2
Malanville 5 2
Total 29 8
Overall, Et4d conducted 16 FGDs with a total of 75 participants. Key Discussion Guides 
developed to lead the focus groups are featured in Annexes H, I and J. 
Key informant interviews
Key informant interviews (KIIs) took place with partners and authority figures during fieldwork in 
the four communes involved in the project as well as the locality of Parakou and Cotonou, where 
some partners have their headquarters. For the KIIs with the partners, the evaluation team did not 
follow a strict guide, but focused on gaining a better understanding of the activities planned for 
McGovern-Dole FY17. A deeper comprehension of the activities enabled us to check that the 
PMP captured all the relevant outputs, and that the activities were planned in alignment with each 
indicator. During the meetings, we inquired about the implementation plan and coordination with 
other partners. This helped us identify possible risk factors for the implementation and/or potential 
opportunities for strengthening collaboration, which could have been missed during the planning 
phase. As presented in Table 6, there were a total of 17 key informant interviews with 30 
participants. Unfortunately, we had to cancel the meeting in Porto Novo with INFRE and DAS 
because of some health issues.
We employed a purposeful sampling methodology to select participants for FGDs and KIIs. The 
selection criteria for parents included those who had different levels of involvement (some parents 
were actively involved such as APE or AME members, and some were not involved in any activity 
of the McGovern-Dole FY14 project), or mothers whose daughter had recently reached puberty. 
For teacher FGDs, we sought diversity among teachers interviewed in terms of classes taught 
(From 1 to 6), gender and zone type (Urban/Rural). 
Table 6: Key informant interview
Organization Location Number
Authority-CRP Malanville 1
Authority-CRP + CP Kandi 3 
Authority-DANA Kandi 1 
Authority-DAS Porto Novo 0 
Authority-DDEMP Kandi 3 
Authority-DDHAS Kandi 3 
Authority-DHS Malanville 1 
Authority-INFRE Porto Novo 0 
Community Health worker Malanville 1
9
Organization Location Number
Partner for development Cotonou 4 
Partner-Alafia Kandi 2 
Partner-CARITAS Kandi 1 
Partner-Derana Parakou 1 
Partner-URP Parakou 5 
Partner-WEI Parakou 2 
Staff - CRS Kandi 1 
Staff - CRS Kandi 1 
2.3 Sampling strategy
Sampling design for schools
We applied a two-stage cluster sampling with project and control schools as the primary sampling 
unit and pupils as the secondary unit. The sample included a total of 50 program schools and 50 
control schools plus an additional 10 replacement schools in each category.
We excluded from the population the schools with less than 120 students enrolled. These schools 
were likely to be new schools with only a few active grades or the numbers of students required 
for the sample were not likely to be available. CRS provided a list of potential control schools, 
among which the schools were randomly selected using a Probability Proportion Sample (PPS), 
while trying to be aligned with the urban/rural ratio of the project. The control schools were all 
located in different communes than project schools, where the local population speak different 
languages, and the socio-economic condition also differed slightly, which could explain some 
differences in the baseline results. 
Among the 50 control schools surveyed, one school in Nikki commune had a school feeding 
program and was therefore excluded from the final sample resulting in a total of 49 control 
schools. The school was not replaced, as the evaluation team realized this was the case while 
analyzing the data. Currently, the Beninese government in collaboration with NGO and 
government agencies is providing canteens services in about 50% of the schools, but this ratio 
should increase in the upcoming years. Therefore, some of the control schools we have sampled
for our baseline could be selected in the future to be part of the government target for the school 
feeding program. This could result in a modification of the sampling for the midline and end line 
evaluations. The final sample, as shown below, consisted of 99 schools (49 control and 50 
program schools). There was a slightly higher proportion of rural schools versus urban (Table 7). 
Table 7: School sample
COMMUNE RURAL URBAN Total
Control 27 22 49
BANIKOARA 7 8 15
BEMBEREKE 6 5 11
NIKKI 9 9 
SINENDE 14 14
Program 27 23 50
10
COMMUNE RURAL URBAN Total
GOGOUNOU 4 7 11
KALALE 4 3 7 
KANDI 9 10 19
MALANVILLE 10 3 13
Total 54 45 99
Table 8 presents the mean number of students enrolled in each grade. In some cases, there was 
more than one classroom per grade, or more than one grade per classroom; therefore, the values 
are not strictly equivalent to the number of students per classroom. At the school level, the 
samples are fairly well balanced, with no significant differences in the number of students per 
grade as revealed by the independent sample t-tests conducted. 
Table 8: Summary statistics for schools 
Control
Mean (S.D)
Program
Mean (S.D.)
Difference
Number of enrolled 
students in kindergarten
65.9 (44.7) 51.7 (15.9) 14.2
Number of enrolled 
students in grade 1
65.9 (18.5) 72.4 (31.2) -6.5
Number of enrolled 
students in grade 2
57.5 (16.4) 64.4 (29.7) -6.9
Number of enrolled 
students in grade 3
55.5 (14.8) 54.2 (24.3) 1.3
Number of enrolled 
students in grade 4
48.3 (14.5) 53.5 (24.9) -5.2
Number of enrolled 
students in grade 5
42.2 (14.4) 45.3 (19.4) -3.1
Number of enrolled 
students in grade 6
33.5 (11.0) 37.8 (13.1) -4.3
Number of total 
enrolled students
300.5 (66.2) 317.1 (129.0) -16.6
Number of levels in the 
school
6.1 (0.5) 5.9 (0.8) 0.2
Girls ratio in the school 0.53 (0.07) 0.49 (0.08) 0.04
*p<0.05; **p<0.01, ***p<0.001
As revealed in Table 9, independent sample t-test shows that program schools have a 
significantly lower percentage of female teachers than control schools. The teacher per pupil ratio 
is higher in program schools than control schools, however the differences are not statistically 
significant. The percentage of community teachers between the two groups is fairly similar.
11
Table 9: Summary statistics for teachers
Control
Mean (S.D)
Program
Mean (S.D.)
Difference
Proportion of male teachers in % 56.9 (22.8) 71.6 (20.9) -14.7**
Proportion of female teachers in % 43.0 (22.8) 28.4 (20.9) 14.6**
Proportion of community teachers in % 28.0 (17.5) 34.1 (21.5) -6.1
Teacher/pupil ratio 49.8 (11.6) 58.6 (17.4) -8.8
*p<0.05; **p<0.01, ***p<0.001
The samples are well balanced in terms of principal characteristics. There are no statistically 
significant differences between the two groups (Table 10) as reported by independent sample t￾test for age and years of experience and chi-square test for other variables. Data for principals 
from all schools are available for a response rate of 100%.
Table 10: Summary statistics for principals 
Control
Mean (S.D)
Program
Mean (S.D.)
Difference
Average age 45.8 (8.5) 44.1 (9.1) 1.7
Average years of experience 17.3 (6.9) 17.3 (7.9) 0.0
Female director 18.4% 8.0% 10.4
Education level
BEPC 71.4% 68.0% 3.4
BAC, DEAT 18.4% 20.0% -1.6
DEUG, DUES 6.1% 4.0% 2.1
Licence, Maitrise, Master/DEA 4.1% 8.0% -3.9
*p<0.05; **p<0.01, ***p<0.001
Sampling design for teachers 
Teachers were randomly selected in each of the sampled schools, using a random number 
generator12. The population sampled included teachers from all six grades plus the kindergarten 
teacher. Principals were generally excluded, unless there was no other option, since they were 
interviewed as principal. Two teachers were sampled in each school, for a total targeted number 
of 200. Precision of this sample is computed considering an intraclass correlation coefficient (ICC) 
of 0.2 and an expected observed proportion of 0.50. With those parameters, results will be 
generalized at 95% CL with a margin of error of 7.1%.
For this specific sample, we also use an observed proportion of 50% at baseline since it is the 
proportion where the standard error will be the largest. Probability of success for 95% of the 
teachers will fall between .2 and .8 and level of significance is set at 0.05. Using these 
12 The random number generator used is an android app installed on the enumerator tablet.
12
parameters, we will have 80% power to detect a change from 50% to 69%, a difference of 19% 
between the baseline and end line.Thus, all variables/ indicators from which we seek teacher 
input will be able to detect a 19% change. This change is relatively small; yet, it is realistic to 
expect a change of this magnitude at end line.
A total of 194 teachers were interviewed (98 from control schools and 96 from program schools)
for a response rate of 97%. Many of the teachers reported teaching more than one grade. The 
sample was relatively well balanced in terms of teachers’ ages and years of experience. More 
female teachers were interviewed in control schools, which is consistent with the fact that there 
are more female teachers in our sample of control schools (Table 11 and 12). 
Table 11: Summary statistics for the number of teachers interviewed per grade
Control
n 
Program
n 
Difference
Total number of teachers interviewed 98 96 2 
Kindergarten [Maternelle] 1 1 0 
First Grade [CI] 21 15 6 
Second Grade [CP] 7 12 -5 
Third Grade [CE1] 17 18 -1 
Fourth Grade [CE2] 27 27 0 
Fifth Grade [CM1] 32 26 6 
Sixth Grade [CM2] 5 2 3 
*p<0.05; **p<0.01, ***p<0.001
Table 12: Summary statistics for interviewed teachers
Control
Mean (S.D)
Program
Mean (S.D.)
Difference
Average age 32.7 (6.4) 32.3 (8.1) 0.4
Average years of experience 7.0 (4.7) 7.3 (5.0) -0.3
Female 40.8% 29.2% 11.6
Male 59.2% 70.8% -11.6
*p<0.05; **p<0.01, ***p<0.001
Sampling design for parents
For each school, the target sample size was 8 to 10 parents. Fifteen parents were chosen by 
selecting 15 children among the students present the day before the survey and informing them 
that they had been selected, and that they should come to school with one of their parents, who 
would be interviewed on the next day. Parents of students from any of the 6 grades and 
kindergarten were selected, using a random number generator. Precision of this sample is 
computed considering an ICC of 0.2 and an expected observed proportion of 0.50. With those 
parameters, we will have a 95% CI with margin of error of 3.2%.
13
For this specific sample, we also used an observed proportion of 50% at baseline since it is the 
proportion where the standard error will be the largest. Probability of success for 95% of the 
program will fall between .2 and .8 and level of significance is set at 0.05. Using these 
parameters, we will have 80% power to detect a change from 50% to 66% (a difference of 16% 
between the baseline and end line) if we sample 5 parents by school and a change from 50% to 
63% (a difference of 13% between baseline and end line) if we sample 10 parents by school. This 
change is relatively small, and it is realistic to expect a change of this magnitude at end line. For 
indicator 2, the observed value is 31.1%, given the sample size, at endline if the observed 
proportion is above 43.6%, the difference will be statistically significant.
The total number of parents interviewed was 1,030 for a response rate of 69.3%. The analysis 
showed that the samples of parents is not well balanced in terms of language. This is due to the 
communes in which the schools were selected. In control schools, 81.1% of parents were Bariba 
speakers and 6.2% were Dendi speakers, while in program schools 38.7% of parents were Bariba 
speakers and 37.2% were Dendi speakers. We also interviewed significantly more female parents 
in program schools (46.4%) versus control schools (35.3%). 
In terms of educational background and literacy level, the samples are well balanced. However, 
control schools have a significantly higher proportion of parents speaking French, with a ratio of 
41.7% for control schools and 34.2% for program schools (Table 13). 
Table 13: Summary statistics for interviewed parents
Control
Mean (S.D)
Program
Mean (S.D.)
Difference
Total number of Parents Surveyed 513 517 4 
Average age 38.6 (10.9) 38.4 (10.4) 0.2
Parents’ Gender **
Female 35.3% 46.4% -11.1
Male 64.7% 53.6% 11.1
Language at home **
Bariba 81.1% 38.7% 42.4
Boo 0.2% 5.4% -5.2
Dendi 6.2% 37.2% -31
French 0.8% 4.5% -3.7
Mokole 0.2% 7.5% -7.3
Peulh 2.7% 4.1% -1.4
Yorouba 2.3% 0.2% 2,1
Fon 1.4% 1.6% -0.2
Other 5.1% 1.9% 3.2
Level of Literacy
Spoke French 41.7% 34.2% 7.5* 
Read French 31.9% 27.3% 4.6
Level of education
14
Control
Mean (S.D)
Program
Mean (S.D.)
Difference
None 49.7% 49.7% 0.0
Can read 10.3% 15.9% -5.6
Primary 16.0% 14.7% 1.3
Secondary Cycle 1 15.0% 14.5% -0.5
Secondary Cycle 2 5.7% 2.7% 3.0
Superior Cycle 1 1.9% 1.6% 0.3
Superior Cycle 2 1.4% 1.0% 0.4
Main occupation **
Unemployed 3.1% 1.9% 1.2
House wife 4.3% 11.0% -6.7
Farm worker 48.0% 51.5% -3.5
Day laborer 5.3% 0.6% 4.7
Seller 11.5% 15.9% -4.4
Worker 2.9% 2.3% 0.6
Craftsman 18.9% 12.4% 6.5
Retired 1.0% 0.4% 0.6
Other 5.1% 4.1% 1.0
*p<0.05; **p<0.01, ***p<0.001
Sampling design for classroom observation and student attentiveness
In each school, two classes from each of the grades 4, 5 and 6 were sampled used Randomized 
Number Generator (RNG). Students in grade 3, were excluded because they were assessed for 
EGRA, and children in grades 1 and 2 were deemed too young to be able to answer the nutrition 
and hygiene questions.
Enumerators sampled five girls and five boys in each class using a random number generator, 
and attentiveness was evaluated by observing each student for three minutes.
The students sampled for student attentiveness were also given the WASH and nutrition test (10 
true and false questions) and asked to answer some questions regarding their hygiene practices. 
Power analysis computed for the student attentiveness will also apply for the sanitation, hygiene 
and nutrition test. The design will have a probability of 80% to detect changes of 11% or between 
baseline and end line. 
For student attentiveness, we also used an observed proportion of 50% at baseline since it is the 
proportion where the standard error will be the largest. Probability of success for 95% of the 
students will fall between .2 and .8 and level of significance is set at 0.05. Using these 
parameters, we will have 80% power to detect a change from 50% to 61%, a difference of 11% 
between the baseline and end line. This change is relatively small, and it is realistic to expect a 
change of this magnitude at end line. With this sample size, for indicator 1, a proportion of 78.9% 
or higher will be identified as statistically significant.
15
The total number of students observed for student attentiveness and WASH and Hygiene Test 
was 1,921 for a response rate of 97%. The student samples for the classroom observation are 
fairly well balanced in terms of age and gender, with no significant differences. In terms of 
languages, the differences are significant, particularly for Bariba, Dendi, and Mokole. The 
languages students speak at home are dependent on the commune and resemble the languages 
used by parents, as reflected in the parent survey (Table 14). 
Table 14: Summary statistics for assessed students (Attentiveness indicator)
Control
Mean (S.D)
Program
Mean (S.D.)
Difference
Total number of students observed 964 957 7 
Average age 11.3 (1.8) 10.9 (1.9) 0.4
Student Gender
Female 50.4% 48.5% 1.9
Male 49.6% 51.5% -1.9
Language at home**
Bariba 78.6% 40.3% 38.3
Boo 0.1% 7.2% -7.1
Dendi 5.6% 28.1% -27.9
French 0.5% 3.5% -3.0
Mokole 0.0% 8.7% -8.7
Peulh 3.6% 5.7% -2.1
Yorouba 2.5% 0.7% 1.8
Fon 3.5% 2.3% 1.2
Other 5.6% 3.4% 2.2
Student level*
Fourth Grade [CE2] 33.0% 38.8% -5.8
Fifth Grade [CM1] 42.1% 39.2% 2,9
Sixth Grade [CM2] 24.8% 22.1% 2.7
*p<0.05; **p<0.01, ***p<0.005
Sampling design for EGRA
The target sample size for EGRA was 16 students (8 boys/8 girls) per school for all 99 schools, 
but in each class, we sampled 24 students, 4 boys and 4 girls served as replacement. Precision of 
this sample is computed considering an ICC of 0.2 and an expected observed proportion of 0.50. 
With those parameters, this student sample size will allow generalizing results at 95% with a 
margin of error of 5%. Power analysis was conducted to detect differences with an expected 
power of 0.8 and type-I error of 0.05, an ICC of 0.2 was also considered. At end line, the minimum 
detectable effect size is 0.25 standard deviation of any continuous outcome will be detected as 
statistically significant. For example, if a standard-deviation of 12 words per minute is observed at 
baseline, the design will have a probability of 80% to detect changes as small as 4 words per 
minute between baseline and end line. 
16
For any dichotomous outcome, we use an observed proportion of 50% at baseline since it is the 
proportion where the standard error will be the largest. Probability of success for 95% of the 
students will fall between .2 and .8 and level of significance is set at 0.05. Using these 
parameters, we will have 80% power to detect a change from 50% to 63%, a difference of 13% 
between the baseline and end line. This change is relatively small, and it is realistic to expect a 
change of this magnitude at end line. With this sample size, for indicator 3, a proportion of 5.7% or 
higher will be identified as statistically significant.
It was decided between the consultant team and CRS that grade 3 student repeaters would be 
excluded from the sample, because they could not be considered as an acceptable proxy for ‘end 
of grade 2’. To avoid stigmatizing the repeaters, students were treated as “no consent” and were 
asked to go back to the classroom. The eight (4 boys/4 girls) extra sampled students were used 
as replacement for the repeaters. However, in some cases the enumerators had to sample again, 
because more than four repeaters per gender had been initially sampled. The total number of 
students assessed for EGRA was 1,521. The response rate was 96%. As shown in Table 15, the 
population of EGRA students is well balanced apart from the language.
Table 15: Summary statistics for students EGRA 
Control
Mean (S.D)
Program
Mean (S.D)
Difference
Total number of students assessed 773 748 25
Average age 8.4 (1.6) 8.4 (1.8) 0.0
Student Gender
Female 49.0% 46.5% 2.5
Male 51.0% 53.5% -2.5
Language at home
Bariba 85.4% 45.2% 40.2**
Boo 0.7% 6.2% -5.5**
Dendi 4.9% 24.5% -19.6**
French 3.2% 2.0% 1.2
Mokole 0.0% 8.3% -8.3**
Peulh 3.9% 6.0% -2.1
Yorouba 1.0% 1.2% 0.2
Fon 2.1% 2.5% -0.4%
Other 7.5% 11.4% -3.9%*
*p<0.05; **p<0.01, ***p<0.005
2.4 Data Analysis
Sampling weights
In order to correct estimates for unequal probability of selection caused by clustering of the 
sample, we applied a two-step procedure for weighting the data. In the first step, base weights 
were computed for each dataset; and, in the second step, adjustment factors were applied to 
correct for non-participation of the selected subjects as well as selection within school. 
17
The probability of inclusion of each student in strata s = {control urban, control rural, program urban, 
program rural} is:
π𝑠𝑠 = �
𝑚𝑚𝑖𝑖
𝑠𝑠 ∗ 𝑛𝑛𝑠𝑠
𝑀𝑀𝑠𝑠 � ∗ �
𝑠𝑠𝑠𝑠
𝑚𝑚𝑖𝑖
𝑠𝑠� = 𝑠𝑠𝑠𝑠 ∗ 𝑛𝑛𝑠𝑠
𝑀𝑀𝑠𝑠
Where
- gl
ab is the total enrollment of school i in strata s
- 𝑀𝑀𝑠𝑠is the total enrollment of all schools in strata s
- 𝑛𝑛𝑠𝑠 is the total number of schools sampled in strata s
- 𝑠𝑠𝑠𝑠 is the number of students sampled per school in strata s 
Thus, the school weight (inverse probability of selection) for strata s is: 
𝑊𝑊𝑠𝑠 = 1
𝜋𝜋� 𝑠𝑠 = 𝑀𝑀𝑠𝑠
𝑠𝑠𝑠𝑠 ∗ 𝑛𝑛𝑠𝑠
Adjustment factors
The probability of inclusion of participant type k = {teacher, parent, student (Grades 4-6)} in the 
teacher, parent, student attentiveness and WASH surveys, respectively is: 
𝜋𝜋𝑘𝑘 = 
𝑠𝑠𝑘𝑘,𝑖𝑖
𝑀𝑀𝑘𝑘,𝑖𝑖
Where
- 𝑠𝑠𝑘𝑘,𝑖𝑖 is the number of participant type k sampled in school i 
- 𝑀𝑀𝑘𝑘,𝑖𝑖
is the total number of all participant type k in school i
Thus, the adjustment factor (inverse probability of selection) for participant type k is:
𝐴𝐴𝑘𝑘 = 1
𝜋𝜋
� 𝑘𝑘 = 𝑀𝑀𝑘𝑘,𝑖𝑖
𝑠𝑠𝑘𝑘,𝑖𝑖
Finally, for the EGRA reading assessment, the probability of selecting a non-repeating student of 
gender g = {male, female} from the 3rd grade of school i is:
𝜋𝜋𝑔𝑔,𝑖𝑖 = 𝑠𝑠𝑔𝑔,𝑖𝑖
𝑀𝑀𝑔𝑔,𝑖𝑖
Where
- 𝑠𝑠𝑔𝑔,𝑖𝑖 is the number of non-repeating students of gender g sampled from the 3rd grade of 
school i 
- 𝑀𝑀𝑔𝑔,𝑖𝑖 is the total number of non-repeating students of gender g in school i
Thus, the adjustment factor (inverse probability of selection) for EGRA reading scores are:
𝐴𝐴𝑔𝑔,𝑖𝑖 = 1
𝜋𝜋
� 𝑔𝑔,𝑖𝑖 = 𝑀𝑀𝑔𝑔,𝑖𝑖
𝑠𝑠𝑔𝑔,𝑖𝑖
18
Adjustment factors are multiplied by the respective school weight when weighting each 
observation.
Statistical analysis
For performing comparisons between control and program schools independent sample t-test was 
performed with continuous outcome and chi-square test with categorical outcome. For the 
comparisons between two groups for indicators, linear regression was used for continuous 
outcome and logistic regression for dichotomous outcome.
2.5 Challenges, issues and limitations 
The following challenges, issues and limitations were identified for this baseline evaluation:
Time constraints
The contract was signed only a few days before the consultants’ departure date, and the 
evaluation team had only a couple of days in country to pilot the instruments and translate the 
instruments into local languages. However, the evaluation team made all efforts to mitigate this 
challenge.
Lack of feedback during instrument design 
Only the CRS M&E team gave limited feedback on the instruments, while no subject matter 
experts such as health and nutrition or education staff were consulted as they were not made 
available to the evaluation team. If they had been given the opportunity to provide feedback, they 
could have made sure we had captured all the information they needed to improve the specifics of 
the program and evaluate it.
Limited project information
Because McGovern-Dole FY17 officially started in January 2019 after the baseline data collection, 
the design of each activity was not completed prior to the inception of the evaluation. Discussions 
were still taking place with the partners, who had not hired their staff yet. As a result, it was 
difficult to make sure that every aspect of the program was included in the survey tools. The 
quantitative survey tools do not cover the menstrual health intervention among the activities.
EGRA Instrument design 
As presented in section 2.2, the EGRA tool for McGovern-Dole FY14 could not be used for 
McGovern-Dole FY17. However, an instrument design workshop was unable to be performed by 
WEI and INFRE prior to the evaluation due to a lack of authorization. As a result, Et4d designed a 
new EGRA instrument, which did not receive INFRE’s approval.
Issue with repeaters
The evaluation was supposed to be administered at the end of grade 2, but because of the 
teacher strikes it was not possible. Therefore, it was decided that an acceptable proxy was the 
beginning of grade 3. To limit the sample to the children who have completed 2 years of school 
and not 3, the third-grade repeaters were excluded from the assessment. To avoid stigmatization, 
repeaters were sampled, but not tested. This resulted in some complications, since the teacher 
had to know how many repeaters were present, so that the weights could be accurately 
estimated.
19
Enumerator’s languages
While recruiting the enumerators, we made sure that we hired individuals who spoke the local 
languages, and if possible, several of them. Unfortunately, we did not know that most children in 
our sample, especially the control schools, spoke Bariba, while most of our enumerators spoke 
Dendi. This issue was partially mitigated by the fact that the student and parent questionnaires 
were translated into these languages.
Parent attrition
The response rate of parents was 69%. We had over sampled 15 children in order to get about 10
parents, which we did. However, we can’t exclude that the parents who did not participate are
different from other parents, so representativeness may be biased
Attentiveness Evaluation
The approach for measuring student’s attentiveness seem as though the surveyors could become 
an item of distraction for the students in the room. 
20
3 Field Work
3.1 Enumerator recruitment and training
The enumerator training workshop for McGovern-Dole FY17 project was conducted in Kandi from 
the 19th to 23rd of November 2018. The workshop training was led by Et4d from CRS and 47 
enumerators who had been pre-selected to participate. The purpose of the workshop was to train 
enumerators on the EGRA tool and associated data collection tools, including the student 
questionnaire, student attentiveness observation tool, school observation tool and the six surveys 
for students, teachers, principals, parents, cook and shopkeepers. The workshop also aimed to 
train enumerators on the fieldwork protocol to ensure high-quality and reliable data was collected. 
The enumerators were trained to conduct one-on-one in-person surveys using tablets or 
smartphones, and on how to submit the surveys electronically. 
EGRA enumerators were trained over the course of five days, while the survey enumerators and 
the supervisors were trained in four days. All trainings were held in a classroom environment. An 
additional day was dedicated to a practice session which took place in a nearby primary school. 
Assessor Accuracy Measurement (AAM) simulations were conducted twice on EGRA in order to 
familiarize the enumerators with the process and examine their accuracy level. The 
comprehension of the survey tools was evaluated with a quiz as well as EGRA protocol. 
Enumerators demonstrated good performance for administration of the EGRA. The results of the 
quiz were acceptable, and enumerators continued to improve with further practice and proper 
supervision.
3.2 Data collection
Data collection for the baseline of CRS 
McGovern-Dole FY17 project was conducted in 
Alibori and Borgou departments from November 
26th to December 7, 2018. A total of 10 teams 
comprised of four enumerators and five 
supervisors were engaged in data collection 
activities. Supervisors visited schools the day 
before the team arrived to conduct the following 
activities: 1) inform the school principal of the 
team visit, 2) sample parents to be interviewed, 3) 
interview the school principal, 4) conduct the 
school observation and 5) collect school-wide 
attendance. On the following school day, the data 
collection team performed the EGRA, interviewed 
teachers, observed student attentiveness, 
interviewed parents, cooks and storekeepers and 
collected school-wide attendance.
We organized the enumerators into sub-teams of 
four individuals and assigned each supervisor two 
teams to manage. Et4d facilitators closely 
followed the teams of enumerators daily to 
oversee the quality of the data that enumerators 
collected and provide them with technical support. 
Et4d
Supervisor A
Team 1 
Team 2
Supervisor B
Team 3
Team 4
Supervisor C
Team 5
Team 6
Supervisor D
Team 7
Team 8
Supervior E
Team 9
Team 10
21
All enumerators regrouped with their supervisors in their respective community daily to debrief, 
submit their data collection logs and electronic surveys, and to plan for the next day of data 
collection. Teams completed fieldwork in 10 days.
3.3 Quality control and Review Board
3.3.1 Review Board
CRS completed the review board with the Benin administrative authority while Et4d ensured 
compliance with international best practices in the way that monitoring, evaluation and research is 
planned and conducted. This is particularly critical with regards to activities involving children and 
vulnerable groups. Et4d ensured that provisions were made to protect the confidentiality of those 
people involved. Et4d was attentive to the administrative, technical and physical safeguards to 
protect the confidentiality of the data in its storage and disposal.
Et4d has trained the enumerator to ensure compliancy with the Ethical Conduct for Research 
Involving Humans, including respect for human dignity, respect for free and informed consent, 
respect for vulnerable persons, respect for privacy and confidentiality, respect for justice and 
inclusiveness, recognizing the potential for harm and maximizing benefits for all who are involved.
3.3.2 Field quality control
Enumerators were observed during the first two days of data collection and when Et4d visited field 
teams for spot checks. Et4d observed that the sampling strategy defined by Et4d and CRS was
enforced, and that quality assurance processes described in the training manuals were followed. 
Et4d assisted the teams throughout the data collection process to ensure that the process was
running smoothly and to address any issues the teams encountered, especially in terms of 
managing the schedule, due to unplanned school closures for Unité Pédagogique (UP). Et4d 
performed spot checks with school principals to ensure the evaluation was performed, and no 
issues were encountered. CRS’s M&E team was also present in the field and notified Et4d any 
time an enumerator was not following the proper protocol so that Et4d could efficiently address 
the issue and reinforce the importance of following consistent protocol in the sampling or 
administration of the survey.
3.3.3 Data quality control
Tangerine13 data was downloaded daily and analyzed to make sure that the 3-second rule14 was 
respected. 
Data from Survey CTO was downloaded daily for quality checks. The main objective was to make 
sure that no data was missing by comparing the school summary sheet, which showed all surveys 
and assessments completed by the team, and the actual data available on the server.
Other quality checks included: 
- Cross checking the attendance with sampling sheets
- Making sure that we had collected attendance data on two different dates 
- Checking for congruence between the number of days the school was open and the data 
collection date
- Checking for agreement between the number of groups sharing the same latrine
13 Tangerine® is a mobile data collection app developed by RTI, used to collect EGRA data
14 The student should be asked to go to the next item after 3 seconds with no response 
22
4 Results of Questionnaires and Observation, and PMP 
4.1 Strategic objective 1: Improved literacy of school-aged children
4.1.1 Improved quality of literacy instruction 
MGD 1: Percent of students who, by the end of two grades of primary schooling, 
demonstrate that they can read and understand the meaning of grade level text
This indicator is described in detail in section 5.
Other Quality of Literacy Instruction Indicators
The following indicators (Table 16), which are dependent on project interventions, have “zero” 
baseline values:
Table 16: Other improved literacy of school-age children indicators
No. Indicator Baseline 
Values 
(2019)
Target
(2022)
1 Percentage of teachers receiving merit certificates 
for their attendance in the previous semester
0 
2
MGD 3 
Number of teaching and learning materials 
provided as a result of USDA assistance
0 187,350
3
MGD 4 
Number of teachers/educators/teaching assistants 
in target schools who demonstrate use of new and 
quality teaching techniques or tools as a result of 
USDA assistance
0 240
4
MGD 5 
Number of teachers/educators/ teaching 
assistants trained or certified as a result of USDA 
assistance
0 312
5 Percentage of teachers who demonstrated the 
use of good teaching
0 
6
MGD 6 
Number of school administrators and officials in 
target schools who demonstrate use of new 
techniques or tools as a result of USDA 
assistance
0 222
7
MGD 7 
Number of school administrators and officials 
trained or certified as a result of USDA assistance
0 288
These output indicators are all linked to the project’s interventions or activities. However, some 
indicators might need to be precisely defined prior to collecting data in the future.
23
A classroom observation tool was developed during McGovern-Dole FY14 to allow the school 
counselor (Conseiller pédagogique) to evaluate the percentage of teachers demonstrating good 
teaching practices. However, the evaluation team did not find the tool used to evaluate the school 
administrators’ and officials’ use of new techniques, and we are not sure who performs the 
evaluation. For instance, the school’s counselors (CP) have been trained to coach the teachers 
and could be further trained in McGovern-Dole FY17. However, we did not identify a tool 
evaluating the appropriate use of the new coaching techniques. It will be important to incorporate 
the new techniques of the administrators and officials in the list of activities planned for WEI.
4.1.2 Improved attentiveness 
Percentage of students in target schools who are identified as attentive during 
class/instruction
Attentiveness was evaluated through classroom observation. The criteria for attentiveness were 
identical as criteria used for the tool developed to evaluate McGovern-Dole FY14: 
A student was considered attentive if: 
- The student is actively doing the task assigned by the teacher;
- The student actively listens to the teacher or other students while they are participating; 
- The student takes notes that appear to be related to the content of the class; or 
- The pupil raises his/her hand to answer the question in a sincere way.
A student was considered inattentive if: 
- The student is reminded to act orderly by the teacher to remind him/her of the purpose of 
the class; 
- The student makes comments that disturb the class; 
- The student speaks to other students for a period when he/she should be listening; 
- The student moves in his/her seat to the point that he/she disturbs the others around 
him/her; 
- The student is sleeping; 
- The student seems distracted or forgets obvious things in classroom activities
Enumerators observed the level of attentiveness from each of the 10 children randomly selected 
individually for about 3 minutes to establish a firm classification. As described in section 2.3, two 
grades were randomly selected among grades 4 to 6 for the student attentiveness observation. To 
evaluate the degree of attentiveness per gender, we computed the distribution of responses on 
attentiveness for girls and boys in control and program schools. 
Based on the evaluation, 67.6% of students were deemed attentive in program schools, and 71% 
in control schools, with no statically significant differences between the two groups. The target for 
this indicator is 75%. Thus, control schools are closer to achieving the target than program
schools. In terms of gender, girls showed a higher degree of attentiveness in both school types, 
particularly in program schools. Boys had a markedly higher level of attentiveness in control 
schools, but the results were not statistically significant (Table 17). 
Table 17: Attentiveness
Control Program Difference
Proportion of attentive students 71.0% 67.6% 3.4
Proportion of attentive students by gender
Girls 72.0% 71.8% 0.2
24
Control Program Difference
Boys 70.1% 63.4% 6.7
Each child was also asked a few questions related to their food consumption, in order to see 
whether we could establish some correlation between the food consumption and the degree of 
attentiveness. The results did not show any statistically significant differences among attentive 
child based on questions related to food consumption. (Table 18). 
Table 18: Food consumption and attentiveness
Control
% 
Program
% 
Difference
Did the child eat before coming to class?
No 72.1% 76.2% -4.1
Yes 70.4% 65.2% 5.2
Did the child eat during the morning break?
No 73.6% 63.5% 10.1
Yes 70.7% 68.1% 2.6
Does the child eat a lunch provided by the school
No - 64.2%
Yes - 68.1%
*p<0.05; **p<0.01, ***p<0.001
Other Improved attentiveness indicators
The following indicators have “zero” values at baseline, as they can only be measured post￾intervention (Table 19). 
Table 19: Other improved attentiveness indicators
No. Indicator Baseline 
values (2019)
Target
(2022)
1
MGD 18
Number of social assistance beneficiaries
participating in productive safety nets as a result of 
USDA assistance
0 50,965
2
MGD 16
Number of daily school meals (breakfast, snack, 
lunch) provided to school age children as a result 
of USDA assistance
0 7,789,096
3 
MGD 17
Number of students receiving daily school meals
(breakfast, snack, lunch) as a result of USDA 
assistance
0 50,965
4
MGD 15
Number of individuals receiving take home rations 
as a result of USDA assistance
0 7,281
25
These output indicators are expected to have a value of zero at baseline; however, there should 
be activities aimed at impacting this output.
4.1.3 Improved student attendance 
MGD 2 - Average student attendance rate in USDA supported classrooms/schools 
USDA does not require more than one data point at baseline. However, Et4d misunderstood, and 
headcount was taken twice. Since multiple data points are always better than one, we decided to 
keep the average of the two dates.
To record attendance, both the supervisor and the lead enumerator (on the following day) went 
into each classroom and performed a headcount of boys and girls present that day and asked the 
teacher for the current enrollment figures. We did not include children in kindergarten in the 
results, because school is not mandatory, and enrollment as well as attendance is usually erratic. 
In addition, children are often sick or might stay with their parents when they are moving 
temporarily to another location, especially during the harvest season. Surprisingly, attendance in 
program schools is significantly lower than in control schools. There is no clear explanation for this 
phenomenon, since providing lunches usually increases school attendance. During our discussion 
with the school principals, some of them mentioned that the school attendance was low because it 
was harvest season, and some parents move temporarily with their children to be close to the field 
during this period. 
Next, we examined gender differences in attendance rates and found that boys in program
schools had the lowest attendance rates and that the differences were significant (Table 20 and 
21). The geographic differences between the communes selected for control and program schools 
could explain partially the differences in levels of attendance. Some communes such as Malanville 
had a much lower attendance rate (83.9%) than others such as Sinende (93.2%) (Table 22). 
Communes located in the north, such as Malanville and Kandi are also more affected by the 
seasonal relatively cold weather, which often results in health-related absenteeism during the 
month of December. This could explain the relatively low attendance levels in these communes. 
The attendance will be taken again by CRS bi-annually. We recommend waiting until the harvest 
season is over to see if the differences between control and program schools remain significant, 
and also exploring the causes behind these findings.
This is a new indicator for USDA, and the evaluation team will be keen to watch it during the 
midline evaluation. This may also be an indication that the control schools do not serve as an 
adequate comparison group, because they are not located in the same communes. 
Table 20: Student attendance
Control
Mean
Program
Mean
Difference
Overall attendance 92.2% 88.3% 3.9***
First Grade [CI] 89.3% 84.6% 4.7**
Second Grade [CP] 90.2% 86.2% 4.0
Third Grade [CE1] 90.1% 86.7% 3.4
Fourth Grade [CE2] 92.9% 88.7% 4.2**
Fifth Grade [CM1] 94.6% 89.3% 5.3**
Sixth Grade [CM2] 96.2% 94.4% 1.8
26
*p<0.05; **p<0.01, ***p<0.001
Table 21: Attendance per gender and grade
Control
Mean
Program
Mean
Difference
Overall attendance Boys 92.9% 89.9% 3.0**
Overall attendance Girls 91.9% 90.0% 1.9*
First Grade [CI] - Boys 90.6% 85.0% 5.6**
First Grade [CI] - Girls 88.4% 86.1% 2.3
Second Grade [CP] - Boys 91.8% 88.2% 3.6**
Second Grade [CP] - Girls 89.7% 90.1% -0.4
Third Grade [CE1] - Boys 90.6% 90.6% 0.0
Third Grade [CE1] - Girls 90.1% 86.8% 3.3*
Fourth Grade [CE2] - Boys 93.0% 91.4% 1.6
Fourth Grade [CE2] - Girls 93.1% 92.7% 0.4
Fifth Grade [CM1] - Boys 95.2% 91.0% 1.8
Fifth Grade [CM1] - Girls 93.8% 90.3% 3.5*
Sixth Grade [CM2] - Boys 95.9% 93.4% 2.5
Sixth Grade [CM2] - Girls 96.4% 93.4% 3.0
*p<0.05; **p<0.01, ***p<0.001
Table 22: Attendance per commune
Mean overall attendance
Banikoara [C] 90.4% 
Bembereke [C] 92.9% 
Gogounou [C] 93.0% 
Kalale [C] 89.6% 
Kandi [P] 88.6% 
Malanville [P] 83.9% 
Nikki [P] 93.1% 
Sinende [P] 93.2% 
Percent of parents who report spending time on literacy activities with their students within 
the last three days
McGovern-Dole FY17 intends to impact the literacy results by increasing the number of parents 
who help their children with school-related activities at home. The indicator is measured by the 
‘proportion of parents who declared having helped their child during the past three days’. The 
information was collected through the parent questionnaire. If the parent had more than one child, 
27
s/he was asked to give the answer for the youngest child in elementary school, since the focus of 
the intervention is on grades 1 and 2.
The results demonstrated that parents from program schools were significantly more likely to 
support their children (31.1%) than parents from control schools (23.4%) (Table 23). This is likely 
to be the result of McGovern-Dole FY14, since a strong focus was placed on getting parents 
involved in their child’s education through activities such as word building or study groups as well 
as awareness-raising campaigns. The difference in performance is especially notable in urban 
areas, where 39.3% of program school parents reported spending time on literacy activities with 
their students in the last three days compared to 19.9% of parents in control schools. There was 
no statistically significant difference based on the parents’ gender.
Table 23: Percentage of parents who report spending time on literacy activities with their students the last 
three days
Control Program Difference
Help student in the last 3 days 23.4% 31.1% -7.7*
Gender of the parent
Female 21.6% 31.4% -9.8
Male 24.3% 30.8% -6.5
Type of zone
Rural 24.9% 24.4% 0.5
Urban 19.9% 39.3% -19.4**
*p<0.05; **p<0.01, ***p<0.005
However, asking the parent whether they helped their child with their homework presents a very 
partial view of the situation. In most families in developing nations, the task of raising children is 
widely shared between family members and even the community. Therefore, we also asked the 
following question, “Has anyone else in your home helped your children (with homework) in the 
last 3 days?”. The results demonstrated that children are more likely to receive help from 
somebody other than their parent in program schools than in control schools (see Table 24). 
Table 24: Percentage patents who report that their child received help of somebody else the last three days
Control Program Difference
Students receiving help from someone else in the 
last 3 days
48.5% 57.7% -9.2*
Type of zone
Rural 48.6% 52.4% -3.8
Urban 48.5% 64.2% -15.7
*p<0.05; **p<0.01, ***p<0.005
To triangulate parents’ responses, we also asked children who performed the EGRA test whether 
they received any help during the past three days. It is important to consider the fact that for the 
parents, the children could potentially be in any grade, while all children interviewed are in third 
grade. The results showed a higher proportion of students receiving help from someone in the 
28
program schools (64.8%) than in the control schools (51.1%) (Table 25). These results are likely 
to be the result of the McGovern-Dole FY14 intervention and are very encouraging.
Table 25: Student reporting receiving from someone else in the past 3 days
Control Program Difference
Someone helped the student in the last 3 days 51.1% 64.8% -13.7**
Type of zone
Rural 43.8% 61.9% -18.1*
Urban 64.8% 67.2% -2.4
*p<0.05; **p<0.01, ***p<0.005
For the indicator, our recommendation would be not to limit the help received to the parent 
interviewed but consider the help received by the child in general. Since the activities are focused 
on first and second grades, we believe that the information reported by the student might be a 
more reliable indicator of the outcome of the activities undertaken with McGovern-Dole FY17. The 
parents’ responses should, however, still be collected from the parents in order to triangulate the 
results. Considering the stark difference between control and program schools (13.7 points), the 
program might already be very close to maximizing the effect. Therefore, we recommend a target 
value of 70% for the students reporting receiving assistance from someone outside of the school 
within the past 3 days. This target would necessitate an increase of 8 percent.
Other improved attendance indicators
The performance indicators below for attendance have “zero” baseline values, as they can only be 
measured post-intervention (Table 26 and 27). 
Table 26: Other improved attendance indicators
No. Indicator Baseline 
Values (2019)
Target
(2022)
1
MGD 9 
Number of students (boys and girls) enrolled in 
schools receiving USDA assistance
Females: 
18,792
Males: 19,988
Total: 38,780
64,580
Table 27: Other improved attendance indicators – Suggested by Et4d
No. Indicator Baseline 
Values (2019)
Target
(2022)
2 Output indicators include: number of radio 
messages broadcasted, number of participants 
within listening groups, number of story hours 
organized and number of participants in story hours 
and number of stories broadcasted
0 
The PMP might need to be more specific regarding the output indicators, and make sure that it 
matches the recently updated list of activities more closely. Et4d has identified some additional 
custom indicators that could be included to cover all of the project activities. For instance, number 
29
of community libraries created, number of books distributed in the community libraries, percentage 
of parents or children who have participated in the story hours or borrowed books at the 
community library. 
To evaluate the impact of the education awareness campaign on school attendance, particularly
for girls, an outcome indicator, such as the student the drop-out rate could be included. However, 
to calculate drop-out rate, a longitudinal study would be necessary.
During our focus group discussions, several parents mentioned that the main reason they do not 
keep their children in school is because the children have not managed to learn the basic skills. 
Therefore, they do not see the benefit of paying for school if a child who has not learned to read 
with fluency by the end of third grade will most likely never learn and will eventually drop out. 
Therefore, the best intervention to keep children in school might not be to go door-to-door to 
convince parents to send their children to school or to broadcast an awareness campaign, but to 
improve students’ learning outcomes.
4.1.4 Foundational results 
Number of classrooms assessed using the Early Grade Reading Assessment
The number of schools assessed using the Early Grade Reading Assessment (EGRA) is 
considered a foundational result because it is a way to raise awareness on the achievement level 
of children in northern Benin. Some members of the ministry might have an inaccurate perception 
of the students’ actual learning outcome. 
The base line value is “0” because it should include only monitoring use (WEI and MoE EGRA). 
Other Foundational Result Indicators
The number of active Parent-Teacher Associations (PTAs) or similar "school" governance 
structures supported as a result of USDA assistance is null at baseline. However, some data was 
collected on the various school committees, which could inform program activities.
Parent and Mothers Associations
Data on parents and mothers’ associations were gathered from the school principal 
questionnaires. Principals were asked if their schools had a parents’ association (APE) and 
mothers’ association (AME). All sampled schools had an APE and AME. It is important to note 
that the control schools could have been part of previous WEI interventions, which could explain 
the existence of AME in all the schools. WEI has been active in northern Benin for several 
decades, and the establishment of AME was one of their key achievements. An APE/AME was 
deemed active if they hold regular meetings and keep minutes.
The results (Table 28) demonstrated that the activity level of the AME varied substantially for 
control and program schools, while the differences for APE program and control schools were 
small and insignificant.
30
Table 28: APE and AME active
Control Program Difference
APE 100% 100%
Not at all active 0% 3.9% -3.9
Mildly active 26.7% 32.0% -5.3
Very active 73.3% 64.1% 9.2
AME ** 100% 100%
Not at all active 26.7% 1.9% 24.9
Mildly active 26.7% 34.1% -7.4
Very active 46.6% 62.0% -14.5
*p<0.05; **p<0.01, ***p<0.001
Hygiene Committee
Data on hygiene committees were gathered from school principals. Surprisingly, all of the schools 
had a hygiene committee (Table 29). The committees in control schools seemed to be more 
active15 than in program schools, with 88.5% of principals in control schools reporting a very 
active committee versus 78% in program schools, but the difference was not significant.
Table 29: Hygiene committee
Control Program Difference
Hygiene Committee 100% 100%
Not at all active 0.0% 0.0% 0.0
Mildly active 11.5% 22.0% -9.5
Very active 88.5% 78.0% 10.5
*p<0.05; **p<0.01, ***p<0.001
It is also important to note that the data for hygiene committees are self-reported by the principal, 
who is often the president of the hygiene committee. Considering this potential bias, it might be 
useful for CRS to assess the efficiency of the hygiene committees by establishing standards 
against which to evaluate the schools and to conduct focus group discussions with committee 
members. 
School Feeding Committees 
Data on the activity level of school feeding committees were also gathered from school principals. 
Overall, 80% of the committees were considered very active (Table 30). 
15 A committee is considered active if it holds meetings and keep minutes
31
Table 30: School feeding committee
Program
School Feeding Committee 100%
Not at all active 0.0%
Mildly active 20.0%
Very active 80.0%
4.2 Strategic Objective 2: Increased use of health and dietary 
practices
4.2.1 Improved knowledge of health, hygiene and nutrition practices 
Percent of parents in target communities who achieve a passing score on a test of 
knowledge of good health, hygiene and nutrition practices
The PMP did not include a test for assessing knowledge of good health, hygiene and nutrition 
practices. However, this information was useful for one of the research questions and was
therefore collected. We also believe that it could be included in the PMP as a custom indicator for 
the midline and end line evaluation. The test consisted of the following 10 statements and parents 
were asked to state whether they were true or false.
1. Washing your hands before eating can help prevent diarrhea [True]
2. Walking barefoot can cause illnesses [True]
3. There is no way to prevent the death of a child because of diarrhea. [False]
4. We wash our hands with soap to remove germs and prevent them from getting on 
food [True]
5. A diet consisting solely of rice and egg is balanced. [False]
6. It is sufficient to rinse the can that contains water to drink with water to clean it. [False]
7. The best way to avoid diseases is to wash your hands with water and soap before eating 
and after going to the toilet. [True]
8. To improve the quality of the drinking water, a little bleach can be added. [True]
9. We brush our teeth only so that our mouth smells good. [False]
10. Balanced diet is important to ensure the good health of children. [True]
The test questions were translated from English to Bariba, Dendi, Peulh and Mokole to prevent 
any language barriers and ensure inclusive participation of all parents tested. Because no 
statements were answered correctly or incorrectly by 100% of respondents, there is no ceiling or 
floor effect. For Question 6, a higher proportion of parents in program schools provided correct 
answers, while for questions 9 and 10, parents in the control group scored higher (Table 31). 
32
Table 31: Percent of correct answers per question (Parent Test)
Questions Control Program
1. Washing your hands before eating can help prevent 
diarrhea
96.8% 95.4%
2. Walking barefoot can cause illnesses 96.4% 95.6%
3. There is no way to prevent the death of a child because of 
diarrhea
64.0% 59.5%
4. We wash our hands with soap to remove germs and 
prevent them from getting on food 
94.6% 94.0%
5. A diet consisting solely of rice and egg is balanced 25.2% 28.8%
6. It is sufficient to rinse the can that contains water to drink 
with water to clean it
67.9% 58.8%
7. The best way to avoid diseases is to wash your hands with 
water and soap before eating and after going to the toilet
96.7% 97.1%
8. To improve the quality of the drinking water, a little bleach 
can be added
84.7% 85.3%
9. We brush our teeth only so that our mouth smells good 40.0% 25.1%
10. Balanced diet is important to ensure the good health of 
children
95.0% 96.8%
The histogram in Figure 2 features the distribution of scores, which is similar for both groups. A 
mean score of 8 correct answers, based on the median value of scores, could be considered the 
passing score. As a result, the passing rate is significantly higher in control schools (54.9%) than 
in program schools (45.9%). If this indicator was to be maintained in the PMP, the target could be 
60% of parents passing the test (Table 32 and 33). 
33
Figure 2: Parents’ test scores
Table 32: Percentage of parents with a passing score
Control Program Difference
Score of 8 correct answers or more 54.9% 45.9% 9.0* 
*p<0.05; **p<0.01, ***p<0.001
Table 33: Average parent score
Control
Mean
Program
Mean
Difference
Knowledge Mean score 76.1 73.6 2.5
*p<0.05; **p<0.01, ***p<0.001
Percent of children in target communities who achieve a passing score on a test of 
knowledge of good health, hygiene and nutrition practices
Similarly to the parent’s test, the children’s test included 10 true/false questions, but it was slightly 
different from the parents’ version. The children who had been evaluated for attentiveness were 
selected to take the test. Therefore, all children tested were in grades 4-6.
The 10 true/false questions on the children’s test were:
1. A dirty body can cause disease. [True]
2. Walking barefoot can cause illnesses. [True]
3. Diarrhea cannot cause death. [False]
4. Food protects us from diseases. [True]
1.20%
5.50%
13.50%
24.20%
25.90%
21.60%
8.20%
1.40% 1.70%
27.90%
22.80%
25.20%
15.90%
5.20%
0%
10%
20%
30%
40%
50%
4 5 6 7 8 9 10
Control Project
34
5. A diet that contains only bread, rice and corn is balanced. [False]
6. It is enough to wash the bottle that contains the water to drink once a year. [False]
7. The best way to avoid diseases is to wash your hands with water and soap before eating 
and after going to the toilet. [True]
8. Students can make the school clean. [True]
9. We brush our teeth only so that our mouth smells good. [False]
10. To be strong, you must just eat a lot of corn. [False]
Because no statements were answered correctly or incorrectly by 100% of respondents, there is 
no ceiling or floor effect. There was no significant difference between the two groups (Table 34). 
Table 34: Percent of correct answers per question (Student test)
Questions Control Program
1. A dirty body can cause disease 91.8% 89.7%
2. Walking barefoot can cause illnesses 87.2% 85.4%
3. Diarrhea cannot cause death 32.8% 39.3%
4. Food protects us from diseases 70.5% 66.7%
5. A diet that contains only bread, rice and corn is balanced 26.0% 30.7%
6. It is enough to wash the bottle that contains the water to 
drink once a year.
59.0% 54.2%
7. The best way to avoid diseases is to wash your hands with 
water and soap before eating and after going to the toilet
96.1% 96.7%
8. Students can make the school clean 98.0% 97.5%
9. We brush our teeth only so that our mouth smells good 16.5% 12.9%
10. To be strong, just eat a lot of corn 26.3% 31.2%
The histogram (Figure 3) illustrates how students’ scores are distributed. The distribution is similar 
for both control and program groups, with most students averaging between 5 to 7 correct 
answers. With 7 correct answers as the passing score, the results indicate near equal pass rates 
for control schools (35%) and program schools (34%). For this indicator, the target could be 40% 
of students passing the test (Table 35 and 36). 
35
Figure 3: Students’ test scores
Table 35: Percentage of students with a passing score of 70%
Control Program Difference
Score of 7 correct answers or more 35.0% 34.1% 0.9
*p<0.05; **p<0.01, ***p<0.001
Table 36: Average student score
Control
Mean 
Program
Mean
Difference
Knowledge Mean score 60.4 60.4 0.0
*p<0.05; **p<0.01, ***p<0.001
Minimum acceptable diet
The MAD was not part of the TOR. However the evaluation team was asked to collect data 
regarding food intake from the previous day from all the children selected for the attentiveness 
observation test. Therefore, our sample included 1,838 students from grades 4-6. The child was 
asked whether he had eaten the following food:
1. Grains, roots and tubers 
2. Legumes and nuts
3. Dairy products
4. Animal flesh foods (meat, fish, poultry and liver/giblets)
0.0% 0.4%
2.3%
8.0%
22.6%
31.6%
22.4%
8.6%
3.3%
0.1% 0.4% 0.7%
2.3%
7.5%
27.6% 27.9%
21.1%
10.3%
2.2%
0.5%
0%
10%
20%
30%
40%
50%
1 2 3 4 5 6 7 8 9 10
Control Project
36
5. Eggs
6. Vitamin A fortified foods, including vegetable oil, fruits and vegetables
7. Other fruits and vegetables
The tool was reviewed by a nutrition expert (Noreen Mucha) after the study had been conducted, 
since initially CRS had asked us to remove this indicator. The expert identified several issues 
which will need to be addressed for the next evaluation. The issues are: 
- Grains, white roots and tubers should have been divided into two categories: 1) Pulses 
(beans, peas and lentils) and 2) Nuts and Seeds—they should not have been combined.
- Animal flesh foods should have been three categories when asking the students: 1) Flesh 
meats: beef, poultry, pork and 2) Organ meats: beef organ meat, chicken organ meat, pork 
organ meat and 3) fish/seafood
- Other fruits and vegetables should have been two categories while asking the questions
- There is a missing category: Dark leafy vegetables
- There is a missing category: Vitamin A-Rich Vegetables, Roots and Tubers
- There is a missing category Vitamin A-Rich Fruits
To collect the data, the enumerator also should perform a recall of what the child has eaten the 
previous day, and not ask whether s/he has eaten a given item.
The Minimum Acceptable Diet (MAD) indicator is one of eight core indicators for assessing infant 
and young child feeding (IYCF) practices developed by the World Health Organization. This 
indicator reflects both the minimum dietary diversity (MDD) and minimum meal frequency (MMF) 
indicators. We have adapted this indicator for school age children. A minimum acceptable diet 
was defined as eating 4 or more foods of the 7 main food groups, and a minimum frequency of 3 
times a day. For each type of food group, the enumerators had a list of examples of local food or 
typical dishes, which they could use to describe a food category (See questionnaire in Annex D). 
The McGovern-Dole FY14 test was replicated for McGovern-Dole FY17 with no modifications, 
only the administration procedures were altered, since the test was administered directly to 
students rather than parents in order to collect more accurate data. Unfortunately, the number of 
meals intake per day was not collected, consequently the results are not reliable.
The indicator is the proportion of children, which reported eating 4 out of the 7 food groups. 
The results indicate that a higher proportion of children in program schools (68.9%) are receiving 
a minimum acceptable diet compared to students in control schools (60.4%); and this difference 
was statistically significant (See Table 37). 
Table 37: Percentage of children receiving a minimum acceptable diet
Control Program Difference
4 or more food groups 60.4% 68.9% -8.5%**
*p<0.05; **p<0.01, ***p<0.001
Other improved knowledge of health, hygiene and nutrition practices indicators
As the following activities have not taken place yet, these indicators have a zero-baseline value:
- Number of communities using Community-Led Total Sanitation approach
- Number of female community health workers trained in menstrual hygiene
- Number of individuals who demonstrate knowledge in Menstrual Hygiene Management 
(MHM) 
37
For “Improved Knowledge of Health, Hygiene and Nutrition Practices,” focus group discussions 
with participants revealed a high degree of variance in quality of implementation. In some 
communities, nobody had heard of the community health worker, while in others, everybody knew 
him/her and praised the work accomplished under McGovern-Dole FY14. Therefore, it seems 
necessary to report on the actual number of school trainings performed.
To capture data for the “number of individuals who demonstrate knowledge in menstrual hygiene 
management” indicator, CRS designed a test for the community health workers to use with 
students. In order to assess the impact at midline and end line, we should have implemented a 
similar test at baseline, but it was not requested. Our recommendation is to develop a true/false 
test with 10 questions that will be taken by students aged 10 and above at midline and end line. 
Since boys and girls received the training, we should design two gender-based versions of the 
test.
For capturing data on the number of communities using the Community-Led Total Sanitation 
Approach, additional information will be needed in order to assess whether the approach is really 
being implemented. An observation test could be developed to that end, such as existence open 
defecation.
4.2.2 Increased knowledge of safe food preparation and storage 
practices 
Increased knowledge of safe food preparation and storage practice indicators
All of the following performance indicators have “zero” at base line:
- Number of individuals trained in safe food preparation and storage as a result of USDA 
assistance
- Number/Percentage of individuals who demonstrate use of new safe food preparation and 
storage practices as a result of USDA assistance
- Number of government staff in relevant ministries /offices implicated in canteen/commodity 
management training
Our understanding is that data on the percentage of individuals who demonstrate use of new safe 
food preparation and storage practices as a result of USDA assistance will be collected by CRS. 
However, we recommend developing a version of the test or observation tool that could be used 
by the external evaluator at midline and end line, in order to triangulate the monitoring results 
provided by CRS’s staff.
4.2.3 Increased knowledge of nutrition
All of the performance indicators below have “zero” baseline values (Table 38). 
38
Table 38: Other increased knowledge of nutrition indicators
No. Indicator Baseline 
Values 
(2019)
Target
(2022)
1
MGD
23
Number of individuals trained in child health and 
nutrition as a result of USDA assistance
0 
2
MGD
19
Number of individuals who demonstrate use of new 
child health and nutrition practices as a result of 
USDA assistance
0 
3 Custom: Number of individuals trained on 
improved agricultural techniques and new 
technologies
0 
4 Custom: Number of school gardens established 
and maintained
0 
5 Number of community farms contributing to school 
feeding
0 
6 Custom: Percentage of farmers using best 
practices
0 
Our understanding is that data on the number of individuals who demonstrate use of new child 
health and nutrition practices as a result of USDA assistance will be collected by CRS. However, 
we recommend developing a version of the test that could be used by the external evaluator at 
midline and end line, in order to triangulate the monitoring results provided by CRS’s staff. The 
child health and nutrition practices also need to be clearly defined.
Criteria for what constitutes ‘contributing to school feeding’ should be clearly defined and, for the 
percentage of farmers using best practices, additional information will be needed in order to 
assess whether best practices are really being implemented. An observation test could be 
developed to that end if the season allows or, if not, through FGDs with parents involved in 
farming. 
4.2.4 Increased access to clean water and sanitation services 
Percent of schools with soap and water at hand washing stations commonly used by 
students
Data related to the percentage of schools with soap and water at hand washing stations 
commonly used by students was collected by supervisors through the school observation tool. 
Items observed were as follows:
1. Do hand washing stations have water? (Options are 100%-80%-60%-40%-20%-0%)
2. Do hand washing stations have soap (or replacement ash locally accepted)? (Options are 
100%-80%-60%-40%-20%-0%)
3. Do you think hand washing stations are commonly used? (Options are: Yes, often or Yes, 
sometimes, Never)
39
To be considered “commonly used", the answers to all three questions had to meet the following 
criteria:
1. Question 1: More than 40% percent 
2. Question 2: More than 40% and above
3. Question 3: Yes, often or Yes, sometimes
Table 39 illustrates that over half of the program schools are maintaining good hygiene practices 
learned under McGovern-Dole FY14. On the other hand, less than 20% of control schools have 
access to soap and water at hand washing stations. For this indicator, a reasonable target would 
be a 25% increase over the course of the program to achieve a 68% rate of proper 
implementation.
Table 39: School with soap (or ash) and water at hand washing stations commonly used by students
Control Program Difference
School with soap and water at hand washing 
stations commonly used by students
19.6% 54.4% 34.8***
*p<0.05; **p<0.01, ***p<0.001
Other increased access to clean water and sanitation services indicators
All of the other performance indicators under this objective have “zero” values at baseline (Table 
40). 
Table 40: Increased access to clean water and sanitation services indicators
No. Indicator Baseline 
Values 
(2019)
Target
(2022)
1 Number of schools using an improved water source 80 144
2 Custom: Number of schools with WASH-friendly 
action plan
0 
3 Number of educational facilities (i.e. school 
buildings, classrooms, and latrines) 
rehabilitated/constructed as a result of USDA 
assistance (latrines)
0 
Regarding the “WASH-friendly action plans,” it is unclear whether this activity is purely related to 
McGovern-Dole FY17, since there are already a lot of WASH activities taking place in the school. 
The WASH-Friendly approach is prescriptive, and CRS will use WASH-friendly standards. 
Criteria will include hand washing station in place, a set of improved latrines, an improved water 
source, hygienic cooking utensils, etc.. The indicator would then be the number of schools who 
meet the WASH friendly standards. A competition between schools could also be organized to 
reinforce the messages about the importance of Hygiene.
To provide some guidance and establish some targets for what could be expected in a “WASH￾friendly school, some observation data was collected, including:
1. Access to a clean, safe water source
2. Number of hand-washing stations with soap and clean water near latrines and the cooking 
area per school 
40
3. Number of students per basic (improved) latrine (according to UNICEF/JMP standards)
Access to water is currently better in program schools, which all have water, even if in some cases 
the water source is far away from the school, or the water is not potable (Table 41). As a result of 
McGovern-Dole FY14 interventions, the number of hand-washing stations available is also much 
higher in program schools, with an average of 7.9 hand washing stations versus 2.2 in control 
schools (Table 42). 
The number of students per latrine was calculated as the school enrollment divided by the number 
of latrines. The standard recommended by UNICEF is 25 students per latrine. Unfortunately, the 
average number of students per latrine is 105 for the program schools and 73 for control schools. 
However, this number should be interpreted with caution, as some schools were part of a schools’ 
complex (several schools grouped together)—and it was not always easy to assess the number of 
latrines related to the sampled school only.
The percentage of latrines specifically designated for girls is less than 5% overall, which is very 
low. Considering the high pupil to latrine ratio, principals recommended increasing the overall 
percentage of latrines for all. However, all of the evidence for sexual abuse/harassment and 
school attendance (due to menstruation, etc.) points to having more latrines and dedicating some 
latrines for girls (Table 43). 
Table 41: Availability of water
Control Program
Where is the nearest water 
source 
Available in the school 45.6% 65.9%
Available next to the 
school
29.7% 26.0%
Available far from the 
school
5.2% 8.1%
Not available 19.4% 0%
Is the water drinkable? 66.3% 78.0%
Table 42: Number of washing stations
Control
Mean S.E
Program
Mean S.E
Number of washing stations per school 2.2 0.4 7.9 1.2
Table 43: Number of latrines
Control
Mean S.E
Program
Mean S.E
Number of students per latrine 72.8 7.3 105.1 28.5
Number of students per functional latrines 111.27 10.6 134.8 32.9
41
Average proportion of latrines for girls 4.0 % 1.3 4.7% 1.2
4.2.5 Increased access to preventative health intervention
Number of students receiving deworming medication
The value for this base line indicators is “zero”.
4.2.6 Foundational results
Number of government staff involved in the creation of WASH friendly school
As mentioned previously, it is not clear what the criteria for a WASH-friendly school is and how the 
government is involved— whether they will participate in the intervention or control the oversight 
of the implementation. 
Through our discussions with stakeholders and school observations, we learned that government 
staff had their own agenda regarding health and safe food preparation services, which applied to 
the women selling food on the school premises during the break. The government request which 
was most often mentioned was the doctor consultation required to be allowed to sell food. The 
government-led health and sanitation services is also authorized to inspect and conduct some 
monitoring of food safety and make sure that the food is properly stored, and the bowls used by 
children are properly washed. 
To be considered WASH-friendly one of the criteria could be having a common hygiene policy for 
the food safety standards including food selling stands and the school feeding program. In 
addition, possible program guidelines that follow national guidelines for food safety with school 
canteen saleswomen for school feeding could be introduced.
42
5 Results of Early Grade Reading Assessment 
5.1 Improved literacy indicator
According to the CRS Benin McGovern-Dole FY17 Performance Monitoring Plan (PMP), students 
have met the SO1 Improved Literacy of School-Age Children impact indicator if they can read and 
understand the meaning of grade-level text. This indicator is measured by the percentage of 
second grade children who obtain three correct answers on five reading comprehension 
questions. As displayed in Table 44, 1% of students in program schools and 0.9% percent in 
control schools met this indicator. These results indicate that 99% of students in both school types 
cannot read and understand a simple grade 2 reading passage in French. 
Table 44: Proportion of students meeting the improved literacy indicator
Control Program Difference
% of students meeting indicator 0.9% 1.0% 0.1 
Table 45 further illustrates that the majority of students in both program and control schools could 
not read and comprehend any grade-level text, with over 90% scoring zero. While roughly 2% of 
students in both school types comprehended 40% of text, only 1% understood between 60-100%. 
Poor reading comprehension is often an indication of low reading fluency and poor decoding 
skills.
Table 45: Distribution of reading comprehension scores by school type
Reading Comprehension Score Control Program Difference
0% correct 93.7% 92.0% 1.7
20% correct 3.9% 5.0% -1.1
40% correct 1.5% 2.0% -0.5
60% correct 0.5% 0.5% 0.0
80% correct 0.4% 0.2% 0.2
100% correct 0% 0.3% -0.3
5.2 Snapshot of mean and max scores by school type
The mean scores per sub-task are presented in Table 46 by school type. The sub-tasks are listed 
according to difficulty beginning with the pre-reading vocabulary sub-task. For this sub-task, 
students in both control and program schools were able to provide the correct answers for fewer 
than 6 vocabulary words out of a total possible of 10. With such limited vocabulary skills in 
French, their listening comprehension skills were also poor, with students only able to answer 
17.5% of listening comprehension questions in control schools and 20.4% in program schools. 
This is quite surprising considering pre-reading skills should be mastered long before third grade. 
Low listening comprehension, however, is indicative of poor language skills. For students to read 
fluently and with comprehension, they must have sufficient vocabulary in the language of
instruction to recognize familiar words and extract meaning. According to the Simple View of 
43
Reading—language comprehension coupled with decoding skills are essential components of 
reading comprehension.16 
As the EGRA tasks became more complex first with the identification of letter names and sounds, 
to then recognizing and decoding words and reading comprehension, the mean scores declined 
considerably for both school types. For the letter identification sub-tasks, students in both groups 
were able to identify more letter sounds than letter names, but no more than 12.8 letter sounds in 
program schools and 11.3 in control schools. Students in both school types could read an average 
of 5 correct words per minute (cwpm) on the familiar word-reading sub-task and decode an 
average of only 3 non-words. This result demonstrates students’ limited ability to recognize 
familiar words and decode unfamiliar words in the French language. The key reason for students’ 
low reading fluency is that most students speak Bariba, Dendi, or another mother tongue (Table 
15) and their parents are not fluent in French, as evidenced in Table 13. The Benin national 
language of instruction policy requires all students to learn in French. Thus, students are learning 
in a language they do not understand or speak at home. 
Due to the lack of pre-requisite language and reading skills in French, students scored the lowest 
on oral reading fluency and reading comprehension sub-tasks. Students in control schools read 
an average of 1.5 cwpm and comprehended 2% of the questions asked while program school 
students read 1.9 cwpm and answered 2.6% of the questions correctly. In order to improve 
student reading comprehension, teachers must focus on improving language comprehension 
(e.g., vocabulary and oral listening comprehension) and apply phonics to help children learn how 
to decode words and extract meaning from text.
Overall, students performed similarly across all sub-tasks. The low scores imply that students in 
control and program schools face similar challenges with learning how to read in a second 
language. 
Table 46: EGRA mean scores by school type
Control
Mean
Program
Mean
Difference
Vocabulary (%) 54.7 58.6 -3.9
Listening comprehension (%) 17.5 20.4 -2.9
Letter name (clpm) 8.4 10.6 -2.2
Letter sound (cspm) 11.3 12.8 -1.5
Familiar word (cwpm) 3.7 4.6 -0.9
Non-word (cwpm) 2.7 3.3 -0.6
Reading passage (cwpm) 1.5 1.9 -0.4
Reading comprehension (%) 2.0 2.6 -0.6
16 Hoover, W and Gough, P. (1986). The simple view of reading. Reading and Writing: An Interdisciplinary Journal 2: 127-160. Kluwer 
Academic Publishers: Netherlands
44
Table 47: EGRA max scores by school type
Control
Max Score
Girls Boys
Program
Max Score 
Girls Boys
Vocabulary (%) 100 100 100 100
Listening comprehension (%) 100 100 100 100
Letter name (clpm) 46 52 44 58
Letter sound (cspm) 63 59 61 63
Familiar word (cwpm) 47 47 46 51.6
Non-word (cwpm) 29 35 41 41
Reading passage (cwpm) 58.9 54.3 73.9 73.9
Reading comprehension (%) 80 60 100 100
5.2.1 Mean scores by school type and gender 
According to EGRA results by gender in Table 48, boys performed better than girls in control 
schools on five sub-tasks, but both genders scored equally on reading comprehension, the 
highest-level subtask. In program schools, girls scored higher on five subtasks, including reading 
comprehension, but the differences were small and insignificant. 
Table 48: EGRA mean scores by school type and gender
Control
Girls Boys
Program
Girls Boys
Overall Gender
Difference
Vocabulary (%) 51.6 58.0 57.7 59.5 6.4***
Listening comprehension (%) 15.2 20.1 19.9 20.8 4.9**
Letter name (clpm) 8.0 8.8 10.1 11.0 0.8
Letter sound (cspm) 11.4 11.2 12.9 12.8 -0.2
Familiar word (cwpm) 3.7 3.6 4.8 4.4 -0.1
Non-word (cwpm) 2.6 2.8 3.4 3.2 0.1
Reading passage (cwpm) 1.4 1.5 1.9 1.8 0.3
Reading comprehension (%) 2.0 2.0 2.9 2.3 -0.1
*p<0.05; **p<0.01, ***p<0.001
5.2.2 Mean scores by school type and location 
As illustrated in Table 49, control schools in urban areas outperformed all other schools on 
reading comprehension and most sub-tasks. Urban control schools scored higher than rural 
control schools on all sub-tasks and higher than urban program schools on seven of the eight sub￾tasks. It is not surprising that urban schools outperformed rural schools in both groups, since 
parents have higher French literacy rates and children have more exposure and access to French 
reading materials in urban areas. However, it is disconcerting that urban control schools 
outperformed urban program schools, despite interventions from the McGovern-Dole FY14
program. More research is necessary to better understand this result.
45
Table 49: Mean scores by school type and urban/rural location
Control
Rural Urban
Program
Rural Urban
Overall
Difference
Vocabulary (%) 49.8 64.4 55.9 60.9 14.6***
Listening comprehension (%) 11.7 29.1 17.8 22.6 17.4***
Letter name (clpm) 7.2 10.8 10.0 11.2 3.7**
Letter sound (cspm) 10.5 12.9 14.6 11.3 2.4
Familiar word (cwpm) 3.3 4.5 4.9 4.3 1.1
Non-word (cwpm) 2.5 3.1 3.7 2.9 0.6
Reading passage (cwpm) 1.3 1.9 2.1 1.7 2.1
Reading comprehension (%) 1.4 3.2 2.8 2.3 1.9
*p<0.05; **p<0.01, ***p<0.001
5.3 Zero scores by school type
The percentage of students scoring zero for both school types is displayed in Table 50. Overall, 
control students had a higher percentage of zero scores when compared to students in program
schools. The differences ranged from 0.5% on the vocabulary sub-task to 10.7% on listening 
comprehension. This result is most likely attributed to the effects of the McGovern-Dole FY14
literacy interventions.
Table 50: Zero scores by school type
 Control 
% SE
 Program
% SE
Overall
Difference
Vocabulary 0.6 0.3 0.1 0.1 0.5
Listening comprehension 51.7 2.9 41.0 2.4 10.7 **
Letter name 18.9 2.2 10.4 1.5 8.5 **
Letter sound 10.1 1.7 5.6 1.3 4.5 *
Familiar word 66.0 3.3 58.9 3.2 7.1
Non-word 70.0 3.3 61.0 2.9 9
Reading passage 67.2 3.5 58.6 3.0 8.6
Reading comprehension 93.7 0.9 91.9 1.1 1.8
*p<0.05; **p<0.01, ***p<0.001
5.3.1 Zero scores by school type and gender 
In control schools, girls had a higher percentage of zero scores on every subtask (Table 51). 
Meanwhile, in program schools, boys scored zero more frequently than girls. Thus, the 
performance of girls and boys differed in each school type, but the differences were not 
statistically significant except for listening comprehension, which demonstrate that boys have a 
better understanding of French than girls.
46
Table 51 Zero scores by school type and gender
Control
% 
Girls Boys
Program
% 
Girls Boys
Overall
Gender
Difference
Vocabulary 0.2 1.0 0.2 0 -0.4
Listening comprehension 57.7 45.2 40.0 41.9 7.5**
Letter name 20.0 17.7 10.8 10.0 2.1
Letter sound 10.5 9.6 5.5 5.6 0.7
Familiar word 67.0 64.8 56.5 61.1 -0.1
Non-word 71.9 67.9 58.7 63.1 1.3
Reading passage 69.4 64.8 56.0 60.9 1.4
Reading comprehension 94.1 93.2 90.8 93.0 -0.2
*p<0.05; **p<0.01, ***p<0.001
5.3.2 Zero scores by school type and location 
In control schools, rural students had a greater percentage of zero scores than urban students, 
and this pattern was consistent across all sub-tasks. Within program schools, more urban 
students scored zero than rural students across most sub-tasks. More research is necessary to 
determine why the situation of rural and urban in control school versus program school is reversed
(Table 52).
Table 52 : Zero scores by school type and urban/rural location
Control
Rural Urban
Program
Rural Urban
Overall
Difference
Vocabulary 0.9 0 0.2 0 0.7
Listening comprehension 63.1 29.1 42.9 39.3 22.8***
Letter name 22.2 12.2 9.0 11.5 6.3*
Letter sound 13.2 4.0 3.7 7.2 4.7*
Familiar word 67.0 63.9 55.7 61.7 0.7
Non-word 72.1 65.8 56.5 65.0 2.0
Reading passage 71.2 59.3 55.7 61.1 6.3
Reading comprehension 95.3 90.4 91.5 92.4 2.8
*p<0.05; **p<0.01, ***p<0.001
5.4 EGRA by reading categories
In order to get a better sense of students’ actual reading levels, we developed a scale that 
measures reading fluency across five levels, from non-reader to fluent reader. Then, we computed 
students’ scores for each level based on the results of three sub-tasks: letter name, oral reading 
fluency, and reading comprehension. The five levels are as follows (Table 53): 
47
Table 53: Reading category definitions
Level Category Definition
1 Non-Reader The child does not read any letter. 
2 Emerging Pre-Reader The child reads letters but no words. 
3 Novice Reader The child reads less than 20 words per minute. 
4 Intermediate Reader The child reads more than 20 words per 
minute but does not meet the 60% reading 
comprehension threshold
5 Fluent Comprehending 
Reader
The child reads more than or equal to 60% for 
reading comprehension.
As illustrated in Figure 4 and Table 54, the majority of students in this study were in the emerging 
pre-reader level. Nearly half of students in each school type could read some letters but no words. 
About one-third of students fell into the novice reader category. Less than 5% of students in the 
control group and 7% in the program group were in the intermediate reader category. 
Unfortunately, only one percent of students in both school types were fluent comprehending
readers, meeting the McGovern-Dole FY17 benchmark. Unfortunately, nearly 20% of students in 
control schools and 10% in program schools were non-readers. 
The differences in the percentage of control and program students in each level is greatest for 
non-reader and novice reader categories. There were 8% more control students who were non￾readers versus program students. Moreover, the percentage of novice readers was 6.6 
percentage points higher for program students as opposed to control students.
As noted earlier, the improved literacy benchmark is 60% correct on the reading comprehension test. 
Considering 99% of children are reading below the required level, the reading categories offer 
McGovern-Dole FY17 a way to track student progress as they advance through the program. The 
scale can be used to measure achievement at midline and end line to determine program impact.
Figure 4: Distribution of program and control group scores at each reading level
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Control
Project
Level 1 Level 2 Level 3 Level 4 Level 5
48
Table 54: Differences between percentage of program and control students in each reading level
Control Program Difference
Level 1: Non-Reader 18.0% 10.0% -8.0 *
Level 2: Emerging Pre-Reader 49.2% 48.5% 0.7
Level 3: Novice Reader 27.3% 33.9% -6.6
Level 4: Intermediate Reader 4.6% 6.6% -2.0
Level 5: Fluent Comprehending Reader 0.9% 1.0% 0.1
*p<0.05; **p<0.01, ***p<0.005
5.5 Literacy achievement index
Apart from tracking reading performance using the McGovern-Dole FY17 strategic indicators and levels 
noted above, another way to measure progress is to develop a literacy achievement index, which is a point 
system that is used to measure how well a sample of students are performing. An index system measures 
achievement and changes in achievement along the performance distribution rather than the very narrow 
focus of a proficiency indicator. 
In the literacy achievement index below (Table 55), each reading level is assigned points based on a scale 
ranging from 0 to 100, with 100 representing the fluent comprehending reader who meets the 
McGovern-Dole FY17 benchmark (60% correct on Reading Comprehension) and 0 representing the non￾reader. The point value for each reading level is illustrated below and increases in equal increments as 
students advance through the levels.
Table 55: Literacy achievement index point scale
Level Category Points
1 Non-Reader 0 
2 Emerging Pre-Reader 25
3 Novice Reader 50
4 Intermediate Reader 75
5 Fluent Comprehending Reader 100
To determine the literacy achievement index of a given sample, we assigned points to each child for the 
highest level achieved and then computed a weighted average for the overall sample. For example, if none 
of the children could read any letters and were in the non-reader category, the literacy index would be 
zero. On the other hand, if all students met the reading comprehension benchmark of 60%, the literacy 
index value would be 100. For illustrative purposes, we provide the literacy indexes by school type, 
gender, commune and geographic location (urban/rural) in the subsequent sections.
5.5.1 Literacy achievement index by school type
Per the results below, the literacy achievement index is almost 5 points (15%) higher in program
schools than in control schools (Table 56). The difference is statistically significant and likely due 
49
to previous McGovern-Dole FY14 interventions. Overall, more students in program schools are 
making progress.
Table 56: Literacy index by school type
Control 
Mean
Program
Mean
Difference
Literacy Achievement Index 30.3 35.0 4.7 (15%)* 
*p<0.05; **p<0.01, ***p<0.001
5.5.2 Literacy achievement index by gender
As illustrated below (Table 57), boys and girls had similar literacy indexes within each school type. 
Both boys and girls are reading at novice levels.
Table 57: Literacy index by school type and gender
Control
Girls Boys
Program
Girls Boys
Literacy Achievement Index 29.7 30.9 35.7 34.4*
*p<0.05; **p<0.01, ***p<0.001
5.5.3 Literacy achievement index by geographic location (Urban/Rural) 
The literacy achievement index by location presents an interesting finding (Table 58). For the 
control group, urban schools had higher literacy indexes than rural schools. However, within the 
program area, rural schools slightly outperformed urban schools. Perhaps, the rural schools, 
which are often neglected, embraced the literacy interventions under McGovern-Dole FY14 or had 
higher teacher retention rates. Albeit minor differences, it would be interesting to investigate these 
trends, which are consistent with earlier results, to identify any best practices or lessons learned in 
rural areas. The difference might be explained by the effects of last year’s teachers’ strike. 
Community-hired teacher (more prevalent in rural and more prevalent in project schools) did not 
strike like government-hired teachers
Table 58: Literacy index by commune by school type and location (urban/rural)
Control
Urban Rural
Program
Urban Rural
Literacy Achievement Index 28.1 34.7 36.7 33.5*
*p<0.05; **p<0.01, ***p<0.001
5.6 Proposed targets
The McGovern-Dole FY17 improved literacy of school-age children indicator requires that 
students achieve 60% correct (3 out of 5 questions) on the reading comprehension subtask, which 
was increased from 40% (2 out of 5 questions) under McGovern-Dole FY14. The target, as stated 
in the TOR, is 32% of students in program schools. The baseline results showed that only 1 % of 
the students in program schools met this target at baseline, and 0.9% in control schools. 
Therefore, this is an ambitious target. 
50
Given that 1) the results of MCGOVERN-DOLE FY14 need to be taken with caution and cannot 
be strictly compared with MCGOVERN-DOLE FY17, as the tools were not identical, and the 
calculation rule differed 2) the McGovern-Dole FY14 end line results showed an improvement of 
13% for boys and 11% for girls and 3) that the bar has been raised under McGovern-Dole FY17, 
the evaluation team proposes that McGovern-Dole FY17 consider lowering the target to 13%, 
which correspond to an additional 3% reaching the targets for each of the 4 years of 
implementation. This assume the program interventions will be implemented as planed. 
Additionally, CRS Benin may choose to set a custom indicator and target for the literacy 
Achievement index. In this case, we propose to set the target at 45 points by end line, which is a 
realistic target above the baseline value of 35 points.
5.7 Factors associated with reading fluency
Potential factors related to oral reading fluency were identified and collected through teacher, 
principal and student questionnaire (See Annex D). The findings for the following topics are 
presented in this section:
• Student characteristics
• Reading frequency at school
• Reading books at home
• Family literacy
• Study and reading habits at home
• Parental Support
• Socio-economic status
Each section reports the results of regression analyses conducted to explore the relationship of 
school- and home-related factors with Oral Reading Fluency (ORF) scores. To account for 
clustering of the students within schools, mixed model regression was used for all these analyses. 
In each model, students’ results were considered as a random effect and each factor was tested 
for their prediction independently. In order for the two variables to have a strong relationship, they 
must be statistically significant (<0.05) and show a notable increase or decrease in ORF scores. 
Please note that a strong correlation between two variables does not indicate a causal 
relationship. Thus, any significant regression coefficient should be seen as potentially having an 
impact on reading competency, and warrants further research to verify a true causal effect.
5.7.1 Student characteristics 
Based on the regression coefficients in Table 59, there are no student characteristics that were 
strongly associated with ORF scores. In other words, there was no or minimal significant 
differences in student performance whether they had attended kindergarten, ate before or at 
school, or if they were absent from school the last week. Further analysis would be needed to see 
if food intake has an impact on lower level tasks.
51
Table 59: Regression coefficient for student characteristics
Student Characteristics % of students Regression 
coefficient
p-value
Attended kindergarten 36.5% 0.21 0.252
Ate before going to school 66.9% -0.42 0.042
Ate during recess 88.3% 0.46 0.068
Ate at the canteen 38.3% 0.31 0.121
Was absent from school last week 25.8% -0.35 0.057
5.7.2 Reading frequency at school 
Table 60 demonstrates how frequently students read in class, according to stakeholders 
interviewed. Reading daily had a strong association with reading fluency scores. However, only 
13% of schools read on a daily basis. 
Table 60: Regression coefficient for reading in class
Reading in Class % of students Regression 
coefficient
p-value
Never (base level) 3.9% - - 
Once every month 4.3% 0.10 0.774
Once every week 27.7% 0.62 0.055
Several times per week 51.2% 1.05 0.002
Every day 12.9% 3.53 <0.000
5.7.3 Reading books at home
The regression coefficients in Table 61 suggest that having French reading books or other French 
reading materials at home had little effect on reading fluency scores. Students with French 
reading materials at home read one more correct word per minute compared to students with no 
reading materials in French. Thus, although 30% of students had French reading books at home, 
it was not necessarily an advantage in this EGRA study. However, other factors could have 
influenced this result such as the type and level of the book, parents’ literacy, and whether 
students read at home – all of these factors are explored and discussed in the following sections.
Table 61: Regression coefficient for French reading books at home
Books at home % of students Regression 
coefficient
p-value
French reading book at home 29.9% 1.06 <0.000
Other French reading materials at 
home
25.4% 1.44 <0.000
52
5.7.4 Family literacy 
Family literacy results in Table 62 indicate that parent literacy is low, but most students have 
somebody at home who can read. Given that 74% of students have a sibling who can read, this is 
likely the person who most often helps them with homework and reading. However, because their 
knowledge of reading may also be low, there is practically no correlation with improved reading 
scores. On the other hand, students who had a mother or father who can read, read 1.3 more 
cwpm than students without a literate parent. Therefore, parent literacy may be an important 
factor in improving student reading fluency.
Table 62: Regression coefficient for family literacy
Family literacy % of students Regression 
coefficient
p-value
Somebody can read at home 83.7% 0.98 <0.000
Father can read 18.6% 1.29 <0.000
Mother can read 9.5% 1.29 <0.000
Brother or sister can read 74.2% 0.16 0.419
5.7.5 Home study and reading habits 
The assertion above that parent literacy is an important factor related to students’ reading 
performance is strengthened with the results in Table 63, which demonstrates a strong positive 
association with parents’ help with homework and improved fluency scores. Unfortunately, less 
than 10% of fathers and mothers help their children with their homework likely due to challenges 
with low literacy. Other factors that were strongly correlated with improved reading fluency scores 
were reading or studying French at home, doing homework after school, and someone asking 
them to read letters or words. However, the influence on reading fluency rates was small with only 
one to two additional correct words per minute.
Table 63: Regression coefficient for home study and reading habits 
Reading habits % of 
schools
Regression 
coefficient
p-value
Read or study in French at home 64.1% 1.52 <0.000
Students do homework after school 21.0% 1.45 <0.000
Someone at home reads or tells stories 32.2% 1.07 <0.000
Someone at home helps with homework 64.4% 0.83 <0.000
Father helps with homework 7.1% 1.81 <0.000
Mother helps with homework 4.1% 1.74 <0.000
Brother or sister help with homework 51.9% 0.43 0.018
Someone helped in the last three days 56.9% 1.81 <0.000
Someone asks student to read letters or words 70.7% 1.79 <0.000
53
Parental support 
When parents congratulate students, give a reward or say they are happy when students get good 
grades, there was a strong positive effect on ORF scores. Alternatively, when parents did nothing 
or were not abreast of students’ grades, there was a negative association. Nearly 40% of students 
said their parents do nothing when they get a good grade. Although giving an incentive had the 
strongest positive effect on student grades, only 13.9% of students said their parents give them a
reward (Table 64). 
Table 64: Regression coefficient for good grades
What parents do when student has good 
grades
% of 
schools
Regression 
coefficient
p-value
Congratulate 22.2% 1.10 <0.000
Give a reward 13.9% 1.71 <0.000
Say that they are happy 21.8% 1.16 <0.000
Do nothing 39.3% -1.14 <0.000
They don’t know student grades 6.4% -0.75 0.006
When students received bad grades, most parents did nothing, some scolded them, hit them or 
punished them, and very few tried to help them. The most effective response was trying to help 
them, while doing nothing or punishing them had a negative effect (Table 65). 
Table 65: Regression coefficient for bad grades
What parents do when student has bad
grades
% of 
schools
Regression 
coefficient
p-value
Try to help me 3.2% 2.93 <0.000
Say that I must work harder 10.4% 1.13 0.001
Scold me 23.6% 0.75 0.001
Punish me 5.6% -0.32 0.338
Hit me 13.2% 1.23 <0.000
Do nothing 44.6% -0.98 <0.000
They don’t know student grades 6.0% -0.89 0.001
 
54
6 Research Question #1: What are the participants’ 
capacities, knowledge and level of application of 
dietary, hygiene and health practices? 
6.1 Context and methodology
CRS’s program strives to improve hygiene, health and dietary practices through McGovern-Dole
activities, recognizing that “access to safe drinking water, sanitation and hygiene (WASH) services 
is a fundamental element of healthy communities and has an important positive impact on 
nutrition.”17 
CRS’s intervention during McGovern-Dole FY14 initially focused on the schools through the 
creation or rehabilitation of latrines, installation of hand washing stations and improved access to 
safe drinking water. CRS also reached out to children through the school WASH clubs (Club 
d’hygiène), whose role is to ensure that proper sanitary regulations are in place and respected 
within the school. Some parents were trained on sanitation and nutrition best practices through 
the food preparation training. With McGovern-Dole FY17, CRS intends to broaden its outreach to 
the community by implementing community-led total sanitation programs (CLTS or ATPC in 
French). The expectation is that CLTS will systematize the use of good hygiene practice beyond 
the school premises. This section of the report examines the findings from the study and offers 
insights and suggestions to enhance CRS’s WASH activities. 
To assess the level of knowledge and proper hygiene and health practices among schools and 
students, we collected data from teachers, parents and students using questionnaires. To assess 
community attitudes towards hygiene and health, we organized focus groups with teachers, 
parents and children from WASH clubs. The qualitative component was essential to understand 
how McGovern-Dole FY17 could have an impact on some factors influencing their attitudes and 
improve the communities’ knowledge and practices. We blended the results from both the 
quantitative data and findings from the qualitative data, and mapped out participants’ knowledge 
and practices of hygiene, health, as well as nutrition.
Several of the KAP survey questions used a 5-point scale, which corresponds to a percentage as 
evaluated by a respondent. Using this scale, we calculated mean values for a defined population
(Table 66). 
Table 66: Scale for proportion
5 Almost all of them (80-100%)
4 Most of them (60-80 %)
3 About half (40-60%)
2 A small portion (20-40%)
1 A few (0-20%)
0 None
17 https://www.spring-nutrition.org/sites/default/files/events/05b_integratingwashandnut_who_unicef_usaid_nov2015.pdf
55
6.2 Analysis and findings on hygiene
Teachers were asked whether they thought that students knew the rules and practices of proper 
hygiene and nutrition, whether children generally put the rules into practice and what was the 
main reason for not applying the rules.
According to teachers surveyed, the percentage of students who know the rules of good hygiene 
is moderate (2.74: a small proportion for control schools, and 3.18: about half for program
schools) and the difference between the control schools and the program schools was significant. 
The proportion of students who practice good hygiene habits was significant as well and very 
close to the evaluated knowledge (Control: 2.66- a small proportion; Program: 3.23 - about half) ( 
Table 67). The difference in program schools could be attributed to the previous McGovern-Dole
FY14 intervention, which included implementing a WASH club in every school, building hand 
washing stations, installing improved latrines, and ensuring access to clean water.
Table 67: Students’ knowledge and practices of hygiene rules based on teachers’ answers
Control
Mean SE
Program
Mean SE
Overall
Mean SE
Which proportion of children 
know the rules of good 
hygiene? 
2.74 0.14 3.18* 0.12 2.90 0.10
Which proportion of children 
apply the rules of good 
hygiene? 
2.66 0.12 3.23* 0.09 2.87 0.09
*p<0.05; **p<0.01, ***p<0.005
The results in Figure 5 further illustrates that children do not know the rules and do not understand 
the health risks associated with poor hygiene. These results indicate that a health and hygiene 
awareness intervention is still needed, and that the program’s intervention could have a strong 
impact on improving awareness of the risks and increase good hygiene practices.
56
Figure 5: Reason for not practicing proper hygiene rules
To triangulate the information collected from the teachers, we collected information from parents
and students about their own health knowledge and hygiene practices. Beginning with health 
knowledge, Table 68 presents results regarding the number of times a parent mentioned a 
specific practice as being important for maintaining good hygiene and preventing illness. For most 
practices deemed important, there seems to be no statistically significant difference between 
parents in control schools and program schools, except for deworming, which was mentioned 
more often in control schools. 
Washing hands, drinking safe water and using soap were considered most important by over 60% 
of the parents in both school types, while washing the drinking water container, using the latrine, 
purifying water and using deworming medication were considered important for less than 27% of 
parents. The results were similar for control and program schools, with two exceptions: use of 
purified water was higher in program schools, while deworming medication was mentioned nearly 
twice as often in control schools as being important.
Table 68: Important hygiene practices mentioned by parents
Hygiene Practices Control Program Overall
Wash your hands 84.1% 82.5% 83.5% 
Consume drinking water 77.6% 79.8% 78.4% 
Use soap or ashes 61.6% 66.7% 63.4% 
Wash drinking water container 27.3% 25.4% 26.6% 
Use Latrine 22.9% 20.3% 21.9% 
Use of purified water 18.5% 23.7% 20.5% 
Deworming 15.3% 8.1%* 12.6% 
*p<0.05; **p<0.01, ***p<0.005
Additionally, we asked students and their parents a series of questions to assess their own health 
and hygiene practices. The first question concerned the number of times the parent or the child 
washed their hands. For both groups, the frequency of hand washing was notably higher in 
11.3
12.9
20.9
21.9
26.5
43.7
0 5 10 15 20 25 30 35 40 45 50
4. They have no access to water
5. They have no access to soap or ashes
3. They don't think that it's useful
2. They are not interested
1. They do not know the rules
6. They are not aware of the risks
57
program schools (Table 69). These results indicate that the McGovern-Dole FY14 program might 
have impacted children’s and parents’ hygiene behavior in program schools, as well as at home.
Table 69: Parent and child hand washing frequency per day
Questions Control Program Overall 
Parent: Number of times you wash your hands per day *
1 to 2 times 4.0% 2.9% 3.6%
3 to 5 times 69.7% 48.6% 61.9%
6 and more 26.3% 48.5% 34.5%
Child: Number of times you wash your hands per day * 
1 to 2 times 16.2% 13.2% 15.1%
3 to 5 times 77.2% 69.2% 74.0%
6 and more 6.5% 17.6% 10.9%
*p<0.05; **p<0.01, ***p<0.005
The second question was asked when parents and children washed their hands throughout the 
day at critical times—such as before eating, after working, or using the latrine, etc. Most parents in 
both school types wash their hands after defecation, but more parents in program schools do so 
compared to parents in control schools. The majority of parents in both school types washed their 
hands before and after eating; however, it is more common in control schools (Table 70). 
Table 70: Occasion for hand washing for Parents
Control Progra
m
Overall
Before eating 99.9%* 96.9% 98.8% 
After eating 81.8%* 74.2% 79.0% 
After defecation 59.7% 69.6%* 63.3% 
After preparing a meal 20.7% 21.4% 20.9% 
After working in the field 22.7% 20.7% 21.9% 
After washing my children or their panties 4.8% 8.7%* 6.2% 
After cleaning the latrines 6.1% 7.1% 6.4% 
After cleaning the pot 4.5% 6.7% 5.3% 
Before selling my products 2.3% 3.2% 2.6% 
*p<0.05; **p<0.01, ***p<0.005
58
Image 1: Tippy tap with soap
For the children, most wash their hands before and after 
eating, with a higher percentage doing so in control schools. 
More children in control schools also washed their hands 
after working in the field compared to program schools, 
mirroring the parents’ results. However, a lower percentage in 
control schools washed their hands after defecation, which 
was mentioned by 52 % of the children in program schools 
versus 41% in control schools, and ‘after cleaning the latrine’, 
which was mentioned by 9% of the children in program
schools versus 4% in control schools (Table 71). 
Table 71: Occasion for hand washing for children
Control Program Overall
After defecation 41.7% 52.1%* 45.8%
Before eating 98.0%* 95.5% 97.0%
After eating 77. 8%* 64.5% 72.5%
After cleaning the latrines 4.9% 9.3%* 6.6%
After working in the field 7.5% 5.2% 6.6%
*p<0.05; **p<0.01, ***p<0.005
The third inquiry concerned the water treatment method used to purify drinking water. The results 
clearly show a stark difference between the method used by parents in program versus control
schools. Over 60% of parents in program schools use bleach, chloride solution, aqua tabs or alum 
compared to 37.8% of parents in control schools. Moreover, 12.5% of parents in program schools 
boil water versus only 8.4% in control schools. No treatment of drinking water is performed by 
22% of parents in program schools and 41% of parents in control schools (Table 72). 
Table 72: Water treatment used
Control Program Overall
Nothing 41.6%* 22.0% 34.4%
Boil water 8.4% 12.5%* 9.9%
Add bleach, chloride solution, aqua tabs or alum 37.8% 62.7%* 50.0%
*p<0.05; **p<0.01, ***p<0.005
Finally, the last question regarding hygiene practices concerned where children defecated. The 
parents were asked where their child usually defecates when not at school. The children were 
asked where they defecated the last time that they did so. Nearly half of the parents reported that 
their children usually defecate in the house latrine. Based on parents’ responses, defecation in the 
bush happens less often with children in program schools (39%) than with children in control 
schools (44%) (Table 73). The children themselves reported similar results with nearly half of 
them stating that they last defecated in the house latrine. Results for defecating in the bush were 
less similar to parents’ responses, with 26% reporting they had defecated in the bush in control 
schools, versus 32% in program schools. The children in program schools also seemed to be 
more inclined to use the school latrine; 23% versus 15% in control schools, and parents were less 
aware of this, with only 3.5% of parents providing similar reports. This result is somewhat 
59
surprising since the school observations showed a higher number of children per latrine in 
program schools than in control schools (Table 74). 
Table 73: Parents’ response regarding where their children defecate when they are not in school
Defecation Control Program Global
In the house latrine 46.3% 49.2% 47.4%
In the bush 44.9% 39.3% 42.9%
In a pot 3.5% 4.8% 4.0%
In the school latrine 2.3% 5.5%* 3.5%
In the latrine shared with the neighbors or the community 5.9%* 2.7% 4.7%
In courtyard and garden 1.4% 5.4%* 2.9%
*p<0.05; **p<0.01, ***p<0.005
Table 74: Children’s response regarding the last place they defecated
Control Program Global
In the house latrine 47.7% 46.3% 47.2%
In the bush 32.9% 26.0% 30.3%
In the school latrine 15.6% 23.8% 18.8%
In the latrine shared with the neighbors or the community 3.1% 1.1% 2.3%
In the garden or the court yard 0.6% 2.7% 1.4%
*p<0.05; **p<0.01, ***p<0.005
To understand the menstruation hygiene component of the project, we met with a community health 
supervisor, a health worker, and several mothers whose daughters had reached puberty or were 
about to. The participants who were directly or indirectly involved with the program unanimously 
praised its quality and effectiveness. They asserted that the instructional materials are very well 
designed and meet the needs of both boys and girls. Even among the Muslim population, the weekly 
class was very well accepted, according to a Muslim community health worker who was well aware
of the culture. The program seemed to target specific communities. The communities where the 
activity was not implemented expressed their desire to be included in the McGovern-Dole FY17
program. An indicator should be included in the PMP to assess the level of participation in this 
activity and the impact of the intervention.
60
6.3 Analysis and findings on nutrition
Image 2: School garden in Gogounou
To determine stakeholders’ level of 
knowledge regarding what constitutes a 
diverse, balanced diet, and to what
extent the communities applied these 
concepts, we collected data from 
teachers, parents and students through 
a teacher survey and focus group 
discussions with parents and a separate 
FGD with children from the WASH club. 
We asked similar questions to all 
stakeholders. This approach enabled us 
to gain a deep understanding of 
community attitudes towards nutrition, 
the types of difficulties they face, and to 
determine where McGovern-Dole FY17
could potentially have an impact in 
improving diet and nutrition.
Using a teacher survey, we collected data on whether teachers thought that children could 
demonstrate knowledge about good nutrition and apply this knowledge into practice. With respect 
to the knowledge of a balanced diet, teachers scored the children’s knowledge of proper nutrition 
at 2.42 which corresponds to ‘a small portion’ on the 5 points scale. Although the mean score 
was slightly higher in program schools, the difference between control schools and program
schools was small and not statistically significant. Meanwhile, the percentage of students applying 
good nutrition practices was even lower for both groups, with 2.5 (a small portion) in program
schools and 1.96 (a few) in control schools, which is a statistically significant difference. These 
encouraging results can most likely be attributed to the canteen in place in the McGovern-Dole
schools, which resulted in an improvement of the children’s diet (Table 75). 
Table 75: Knowledge and practices of nutrition’s rules based on teachers’ answers
Control Program Global
Mean SD Mean SD Mean SD 
In which proportion children 
demonstrate good nutrition 
knowledge
2.35 1.19 2.58 1.43 2.42 1.29
In which proportion children follow
good nutrition rules 
1.96 0.99 2.50** 1.52 2.14 1.19
*p<0.05; **p<0.01, ***p<0.005
Based on the teachers’ survey, the two mains reasons for children not eating a nutritious diet are 
1) that their parents cannot afford to buy nutritious foods, and 2) that both children and parents do 
not know what constitutes a good diet (Figure 6). McGovern-Dole FY17 should include activities 
on improving parents’ and students’ understanding of what constitutes a balanced, dietary diverse
diet through targeted behavior change activities.
61
Figure 6: Reasons for not having a balanced diet
During focus group discussions, parents were asked what constitutes a balanced diet. The 
enumerators did not propose any answer, but they classified the answers received into one of the 
four categories below:
1. Includes the five main food groups (dairy, meat and fish, fats, fruit and vegetables)
2. Includes food containing vitamins
3. Includes food containing vitamin A (such as red palm oil)
4. Includes fruits and vegetables
If any response did not fall into one of the four categories, enumerators selected “other” and 
recorded the answer (Table 76). 
Table 76: Parental knowledge of well-balanced diet
Rules Control Program Overall
Include the five main food families 70.4%*** 58.1% 65.9%
Include foods containing vitamins 25.1% 28.4% 26.3%
Include foods containing vitamin A 12.9%** 6.5% 10.6%
Include foods containing vegetables 17.7% 18.9% 18.2%
*p<0.05; **p<0.01, ***p<0.005
11%
11%
24%
33%
37%
40%
0% 5% 10% 15% 20% 25% 30% 35% 40% 45%
2. They are not interested
6. Children have no influence on what they eat
5. The priority is to have a full belly
1. They do not know the rules
4. Their parents do not know the rules
3. Their family cannot buy the proper food
62
Image 3: Child Eating Fruit
Based on the results and understanding that a 
balanced diet includes the five main food groups, 
parents in control schools seemed to be more 
knowledgeable than parents in program schools. 
However, these results are limited since very often 
even if parents understand that there are different 
types of foods, they do not know which food belongs to 
which category/food group. For instance, for the 
True/False test, about 73% of parents said that a diet 
consisting solely of rice and egg was balanced. During 
the focus groups with parents, we asked more specific 
questions and often received wrong answers, 
especially concerning which food contained vitamins. 
Parents overall seemed to value the importance of 
including meat more often than vegetables and fruits. 
We also observed that parents who were involved in 
the canteen had an excellent knowledge of what 
constitutes a balanced diet, while most other parents 
did not. The knowledgeable parents also seemed to 
feel that they should share their understanding of 
nutrition only with other cooks. 
We also asked parents why they could not provide a 
well-balanced diet to their family. An overwhelming 
majority of parents mentioned that they could not afford to buy the food necessary to prepare 
balanced meals (69.5%) or that the specific food items were not available either because they 
lived in a remote area or because it was not in season. (23.2%) (Figure 7). Even if only 5.8 % of 
parents mentioned that they did not know the rules, we feel that McGovern-Dole FY17 should 
include activities on improving parents’ understanding of what constitutes a balanced diet, and 
how it can impact their children’s health.
Figure 7: Reasons for not being able to provide a balanced diet
0.3%
1.9%
1.9%
2.2%
5.8%
18.9%
23.2%
69.5%
0% 10% 20% 30% 40% 50% 60% 70% 80%
2. I am not interested
3. I think that it is useless
7. It takes too much time
8. I never think about it
1. I do not know the rules
6. The priority is to have a full belly
5. I have no access to the proper food
4. I can not afford to buy the proper food
63
The students’ questionnaire, unlike the parents’ questionnaire, did not include any questions 
related to their knowledge of nutrition apart from the question related to what they ate the day 
before (24-hour recall) and the nutrition question included in the true/false test, for which about 
72% of students said that a diet that contains only bread, rice and corn was balanced, which 
shows their low level of understanding.
Most of the children surveyed (Grades 4 to 6) are too young to have been taught what a balanced 
diet consists of, because in Benin, the lessons on nutrition and food groups is part of the sixth￾grade curriculum. Therefore, none of the students interviewed were expected to be familiar with 
these concepts. Children themselves usually have limited influence on their food consumption. 
However, they are remarkably open to new types of food, and their willingness to try new food can 
help promote new practices within the community. For instance, the moringa plant, which has 
recently been introduced and promoted in Benin because of its high nutrient content, has been 
successfully introduced in some of the McGovern-Dole schools and has become part of the family 
diet.
For McGovern-Dole FY17, we encourage CRS and its 
partners to introduce nutrient-dense crops including 
iron-rich leafy greens and vegetables, Vitamin-A rich 
vegetables and fruits, legumes and nuts as well as 
fruit trees into the children’s diets via school gardens 
and school farms. They can rely on the children’s 
willingness to try new food since the participants 
reported that most common complaint was children’s reluctance to eating corn.
“Children are more open to what's 
new in terms of diet, and sometimes 
they bring back a portion of their 
meals home to allow their brothers to 
taste.” 
T h M l ill
64
Conclusion
Overall, the data demonstrates that hygiene best practices are fairly well known by children and 
parents alike across all schools, even if it could still be notably improved by increasing children’s
knowledge regarding the risk factors involved in not following the best hygiene practices.
Regarding the extent of good hygiene practices, we have found small, yet statistically significant 
differences between control schools and program schools on some practices. These results are 
very encouraging, and it is very likely that these levels of practice are due to the program activities 
implemented during McGovern-Dole FY14, but, unfortunately, because we are establishing 
baseline data, we cannot use mathematical modeling to establish causation. Another interesting 
result is that some practices do not seem to improve only at school, but also in the home 
environment. For instance, children from program schools use the latrine more frequently than 
the bush to defecate whether they are in school or at home.
During the discussion with the parents, the teachers and the children, the participants all shared 
similar views. Basically, the knowledge is mostly in place, but adoption of practices and 
behaviors—and the measurement of this behavior change could still be improved. Some of the 
most interesting discussions we had were with children from the WASH Clubs. They shared with 
us that they were very honored to have been selected as members of the WASH club, but 
sometimes felt threatened by the children who do not follow the hygiene rules, and they do not 
dare to tell them to act properly because they fear retaliation if they report them to the teacher. 
Their role seems to be focused on acting as “hygiene police,” rather than as “hygiene advocates 
and promoters”. The children of the WASH club were trained, and they are very knowledgeable. 
However, they did not seem to be inclined to share their knowledge with their peers, and they 
have no instructional material support to do so.
Regarding improved nutrition, unfortunately, among parents, knowledge regarding the 
importance of a properly balanced, diverse diet remains very low. Mother Association members 
(AME) have been trained on nutrition education, but their acquired knowledge did not spread to 
other community members. Practices are also very poor, partially because of lack of awareness, 
but mostly because parents have no access to nutrient-dense, dietarily diverse food, or no means 
to acquire them.
For McGovern-Dole FY17, we encourage CRS and its partners to introduce new nutrient-rich 
crops18 into children’s diets via school gardens and school farms19 , to identify locally nutrient rich 
food and to introduce them to recipes. To initiate behavior change, CRS should also encourage 
AAM members, who have been trained in nutrition, to share their knowledge with the communities 
to increase parents’ awareness of how nutrition impacts their children’s health and the importance 
of a giving their children a diverse diet.
18 A crop is defined as nutrient-rich if it meets any of the following criteria: 1. Is bio-fortified 2. Is a legume, nut, or some seeds such 
as sesame, sunflower, pumpkin seeds, wheat germ, or sprouted legume seeds 3. Is a dark yellow or orange-fleshed root or tuber 4. 
Is a fruit or vegetable that meets the threshold for being a “high source” of one or more micronutrients on a per 100 calorie and per 
100 gram basis.
19 “USAID Multisectoral Nutrition Strategy 2014–2025. Technical Guidance Brief. Nutrition-Sensitive Agriculture: Nutrient-Rich 
Value Chains.” Guidance. USAID, December 3, 2015. https://www.usaid.gov/sites/default/files/documents/1864/nutrition-sensitive￾agriculture-508.pdf.
65
7 Research Question #2: What is the parents’ 
knowledge, attitude and practices regarding 
supporting their children to be successful at school? 
7.1 Context and methodology
Families who engage with their children’s learning have the potential to encourage their children 
to spend more time learning and increase their motivation to learn.20 Family and community 
engagement is a key factor contributing to children’s educational achievement.21 
In our survey, we looked at parents’ or caregivers’ involvement with the children broadly. We did 
not limit our inquiries to structured help such as assisting students with homework and practicing 
reading lessons, but we also included activities such as telling stories, discussing school 
experiences or encouraging their children to do well in school.
First, to estimate the extent to which parents are engaged in their children’s education, we 
surveyed principals, teachers, parents and students. To complement our understanding of 
parents’ attitudes towards supporting their children to be successful in school, we organized focus 
groups with teachers and parents, and conducted in-depth interviews with CRS’s partners, who 
had been involved in the implementation of the school-related activities during McGovern-Dole
FY14. The information collected through focus group discussions was essential for deepening our 
understanding of how McGovern-Dole FY17 could influence parents’ attitudes and behaviors, 
which in turn, could improve students’ academic achievement.
When analyzing parents’ participation in their children’s literacy activities at home, we tried to see 
if we could identify differences based on the parents’ characteristics, such as gender, education, 
language and location (urban/rural), as well as control schools versus program schools.
For McGovern-Dole FY14, WEI worked with two local NGOs: Alafia and Derana, who 
implemented a series of interventions with support from parent associations (AME, APE). Such 
activities included study groups, word-building workshops, parent discussions on the importance 
of education, and reading competitions. Most of these activities had been discontinued when we 
performed the evaluation; however, we sporadically found some study groups that remained 
active and some parents who were still using the letter cards received from McGovern-Dole FY14
to practice word-building activities with their children. 
7.2 Findings and analysis
The information provided by parents on their level of involvement in their children’s education was 
triangulated with information provided by the children, teachers and school principals.
In section 4.1.3, we presented the results for the custom indicator ‘percentage of parents who 
report spending time on literacy activities with their students within the last three days, which 
demonstrated that parents in program schools are more likely to provide support to their children 
than parents in control schools. The same is also true for help provided with homework by other 
family members. The findings from parents were triangulated with student responses, and both 
20 Banerji, R., Berry, J., & Shotland, M. (2017). The impact of mother literacy and participation programs on child learning: Evidence 
from a randomized evaluation in India. American Economic Journal: Applied Economics, 9(4).
21 McNeal, R. B. J. (1999). Parental involvement as social capital: Differential effectiveness on science achievement, truancy, and 
dropping out. Social Forces, 78(1), 117-144.
66
were consistent. This demonstrates that the previous intervention had a lasting impact on parents’ 
behavior. There was no difference in terms of support received by mothers or fathers. In program
schools, parents were more likely to help if they lived in an urban setting.
During our focus group discussions with teachers, we were often 
under the impression that teachers did not feel that parents were 
providing enough support and attention to their children’s 
schoolwork. They felt that parents often barely knew which grade 
their children were enrolled in and whether they were able to 
follow instruction appropriately in class. In the teacher survey, we 
asked teachers what percentage of children were most likely to 
receive help with their homework. The answers were based on a 
5-point scale, identical to the previous research question (See 
section 6.1). 
Teachers in program schools reported a significantly higher proportion of students received help 
in program schools than in control schools which may be attributed to McGovern-Dole FY14. 
However, it seems that teachers underestimate the help received by students, since 64% 
(approximately 4 on the 5-point scale) of students in program schools and 51% (approximately 4 
on the 5-point scale) in control schools reported having received help in the last 3 days, while by 
teachers’ estimations only a few students (1.74) in control schools and a small portion (2.35) in 
program schools received help (Table 77). 
Table 77: Proportion of students receiving help based on teachers’ perception
Control
Mean SD
Program
Mean SD
Overall
Mean SD
Which proportion of children receive help? 1.74 1.09 2.35*** 1.57 1.96 1.30
*p<0.05; **p<0.01, ***p<0.005
The three types of parents’ engagement for which we collected data are telling stories, discussing 
school experience and encouraging their children to do well in school. These components of 
parents’ engagement are related to some of the activities planned for McGovern-Dole FY17 such 
as radio story time and awareness campaigns on the importance of being involved in children’s 
education.
We asked the parents and children whether someone at home read or told stories to children.
For this question, there was no difference between control and program schools, or rural and 
urban schools. Overall, about 60% of parents reported that someone in their home told stories to 
the children.
Unsurprisingly, parents who are literate are more likely to tell stories to their children (70%), than 
parents who cannot read (55%). We also found a significant difference based on the languages 
spoken at home. Sixty-two percent of Bariba speakers reported telling stories, while only 45.7% of 
Dendi speakers did. This might be partially explained by the fact that many Dendi speakers are 
located in the commune of Malanville, which also has the highest proportion of illiterate parents 
(over 80% in our sample) (Table 78). 
Table 78: Proportion of parents reporting that someone tells stories to their children at home
Control Program Overall
"Parents do not ask their 
children anything about their 
school activities. Children 
are left wandering by 
themselves without any 
supervision."
Teacher, Malanville
67
Does someone read or tell 
stories to your child at home? 
59.9% 59.7% 59.8%
Location type
Rural 61.3% 60.8% 61.1%
Urban 56.9% 58.3% 57.5%
Parent literacy
Parent can read 66.6% 75.2% 69.5%
Parent cannot read 56.9% 53.9% 55.7%
Home Language
Bariba 60.1% 68.4% 61.8%
Dendi 58.0% 43.7%* 45.7%*
Other 59.9% 69.2% 65.2%
*p<0.05; **p<0.01, ***p<0.005
Overall, the proportion of parents who reported that their children were told stories is higher than 
we would have expected, based on the discussion we had in focus groups with teachers, who did 
not believe that parents told stories to their children. The results were triangulated with the 
children’s responses to the same question. Unfortunately, the results were very different, with only 
32% of children reporting that they were being told stories. Interestingly, the results were 
significantly higher in program schools, where 44% of students reported that someone told them 
stories versus 24% in control schools (Table 79). Based on these data, it seems that awareness 
campaigns from McGovern-Dole FY14 might have produced some lasting behavioral changes for 
parents of children in early grades. However, more work needs to be done to improve reading or 
storytelling at home. McGovern-Dole FY17 planned interventions should identify ways for literate 
and illiterate parents to get involved in their children’s literacy development and results should be 
monitored.
Table 79: Children reporting being told stories
Control Program Overall
Does someone read or tell you stories at home 22.8% 47.7%*** 32.2%
*p<0.05; **p<0.01, ***p<0.005
The second aspect of engagement we considered was whether parents talk with their children 
about their experience at school on a daily basis. Parents overwhelmingly said that they asked 
their child what they had learned that day. The analysis shows that this was more likely to be the 
case in program schools (76%) than in control schools (69%). Parents’ education is also a major 
factor, with 89% of literate parents inquiring about their child’s day versus 64% of non-readers
(Table 80). 
68
Table 80: Proportion of parents reporting talking to their children about their school experience
Control Program Overall
When your children come back from school, do 
you ask them what they have learned
69.7% 76.0% 72.1%
Location type
Rural 70.1% 77.4% 72.4%
Urban 68.9% 74.4% 71.5%
Parent literacy 
Parent reads 87.1%*** 95.1%*** 89.8%***
Parent does not read 61.7% 68.9% 64.5%
Home Language 
Bariba 69.7% 72.5% 70.3%
Dendi 67.7% 73.2% 72.1%
Other 71.1% 85.4% 79.3%
*p<0.05; **p<0.01, ***p<0.005
Since parents are likely to report desirable behaviors, we suspected some social desirability bias 
could influence their answer, and therefore we asked children the same question. Over half of 
children (44.5%) reported that they were asked what they did in school (Table 81), which is much 
lower than the percentage (72.1%) reported by parents (Table 80). Children in control schools 
were more likely to report higher engagement of their parents (Table 81). 
Table 81: Children reporting being asked what they did at school
Control Program Overall
When you come back from school, does 
someone ask you what you did today? 
39.3% 52.9%*** 44.5%
Home Language 
Bariba 36.2%* 42.6%* 37.6%*
Dendi 56.3% 61.1% 60.1%
Other 58.4% 59.3% 59.0%
*p<0.05; **p<0.01, ***p<0.005
While there is no consensus among researchers upon the benefit of praising children when they 
get good grades, it is widely recognized that rewarding positive behavior, such as hard work, 
reinforces and encourages children to repeat the same behavior. Thus, parents should be 
proactive with setting expectations and rewarding the behaviors they want their children to learn. 
The regression analysis of the EGRA results (See section 5.7.6) also showed that when parents 
congratulate students, give a reward, or say they are happy when students get good grades, there 
was a strong positive effect on ORF scores. Alternatively, when parents did nothing or were not 
abreast of students’ grades, there was a negative association. Nearly 40% of students said their 
69
parents do nothing when they get a good grade. Although giving an incentive had the strongest 
positive effect on student grades, only 13.9% of students said their parents give them a reward.
In order to evaluate parents’ behavior, we asked them what type of feedback they gave to their 
children when they had good grades and when they had bad grades. We then asked children 
what type of feedback they received in order to triangulate our data since parents’ responses are 
likely to be biased. 
Among parent responses, the most common was “I encourage him/her” for 58.7% of respondents, 
followed by “I give him/her a reward” for 33.7% of parents surveyed. Only 7.8% of parents 
reported doing nothing (Figure 8). 
Figure 8: Types of feedback parents report giving when their child gets good grades 
As we suspected, the children’s responses were quite different. Only 22% of children reported 
receiving encouragement and 13% reported receiving a reward, while 39% reported that their 
parents did nothing (Figure 9). 
Figure 9: Types of feedback students report receiving when they get good grades 
0.2%
5.5%
7.8%
28.4%
29.5%
33.7%
58.7%
0% 10% 20% 30% 40% 50% 60% 70%
6. S/he never has good grades
7. I don’t know his/her grades
5. I don’t do anything
4. I smile, I'm happy
2. I compliment him/her
1. I give him/her a reward
5. I encourage him/her
39.3%
22.2%
21.8%
13.9%
6.4%
0.7%
0% 10% 20% 30% 40% 50%
4. They don't do anything
1. They compliment me
3. They tell me that they are happy
2. They give me a reward
6. They do not know my grades
5. I never have good grades
70
A large body of research has demonstrated that punitive parenting 
strategies are likely ineffective in promoting achievement when it does 
not directly address the underlying problem that is causing academic 
underperformance.22 However, in Benin, there is still a widely held belief 
that harsh parental action leads to higher academic achievement. While 
being asked what their reaction was when their child received bad 
grades, parents reported punishing and/or hitting their child in 30.8% of 
the cases, scolding him/her in 43% of the cases, and only in 6% of the 
cases did a parent report taking action to help the child to learn (Figure 
10). 
Figure 10: Types of feedback parents report giving when their child gets bad grades 
Children reported receiving similar feedback. When they receive bad grades, only 3% reported 
receiving help from their parents and 13% reported being hit. The main discrepancy in parent and 
child responses was that 44% of students reported their parents doing nothing, while only 13% of 
parents reported doing nothing (Figure 11). During the discussion with the teachers, the general 
feeling was that a lot of parents did not really know how their child was performing at school, and 
only reacted at the end of the year if the child had to repeat the class.
22 Tang, Sandra & Davis-Kean, Pamela. (2015). The Association of Punitive Parenting Practices and Adolescent Achievement. Journal 
of Family Psychology. 29. 10.1037/fam0000137.
https://www.researchgate.net/publication/280628899_The_Association_of_Punitive_Parenting_Practices_and_Adolescent_Achiev
ement
43.1%
40.6%
20.8%
13.9%
10.0%
6.3%
5.8%
1.9%
0.7%
0% 10% 20% 30% 40% 50%
3. I scold him/her
5. I tell him/her, s/he has to work more
2. I punish him/her
4. I'm not doing anything
1. I hit him/her
6. I help him/her learn
9. I don't know his/her grades
7. I fin him/her a tutor or a study group
8. S/he never has bad grades
“When children do not do
well in school the parents 
are angry with the 
children but also at the 
teacher”. 
Caregiver, Kandi
71
Figure 11: Types of feedback students report receiving when they get bad grades 
Image 4: Mothers of students and interviewer in focus group discussion
1.5%
3.2%
5.6%
5.9%
10.4%
13.2%
23.6%
44.6%
0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50%
7. I never have bad grades
1. They are trying to help me work
8. They don't know my grades
4. They punish me
2. They tell me I have to work more
5. They hit me
3. They scold me
6. They don't do anything
72
Conclusion
Despite the empirically proven benefits of family and 
community engagement, there are many factors that impact 
families’ ability or willingness to support their children’s 
learning. These factors include: 
 Parents’ motivational beliefs, including their sense of 
responsibility for their children’s educational 
outcomes
 Parents’ sense of efficacy, perceptions of invitations 
to engage, including those from schools, teachers, 
their children, or other actors; and 
 Life-context variables, including parents’ understanding of their own skills and 
knowledge, time or energy, and local culture of family involvement in education.23
One of the key aspects to keep in mind is that about 70% of parents in northern Benin 
are illiterate. Therefore, most of them feel a sense of inadequacy when it comes to helping 
their children with their homework. During our focus group discussions, some parents 
shared the belief that their involvement would not have a positive influence on their 
children’s educational outcomes because of their own lack of education. Illiterate parents 
may be embarrassed about their lack of reading and writing skills; however, many such 
parents have a strong desire for their children to become literate.
In order to increase parents’ engagement, CRS’s interventions should not be limited to 
advocacy but should also include activities helping or even teaching illiterate parents how 
to help their children. A toolkit could be designed on how parents can develop early 
literacy skills, assist children with their homework, and encourage them even when they 
are not performing well. In order to nurture literacy, children should also be given more 
opportunities to listen to stories, which can be provided by the radio, or told by story tellers 
or caregivers using wordless picture books in the local language or French, since fluency 
in the first language builds the foundation for second language learning. Given the 
inconsistency of results among teachers, parents and students, CRS Benin should 
develop a monitoring system for tracking if literacy activities are taking place at home, 
working with all stakeholders involved. For instance, teachers can send books home with 
students or provide homework that can done orally, along with a sheet for the parent and 
student to sign to shows that the activity has taken place. Other ways to confirm activities 
have taken place are to involve multiple families or form community reading groups.
23 Hoover-Dempsey, K., Walker, J., Sandler, H., Whetsel, D., Green, C., Wilkins, A., & Closson, K. (2005). Why Do Parents Become 
Involved? Research Findings and Implications. The Elementary School Journal, 106(2), 105-130.
“Living conditions make 
work at home difficult. 
The huts have no light, 
and there is no table or 
chair. Children have 
nothing to write with.”
Teacher Kalale
73
8 Research Question #3: Based on lessons learned 
from community participation, what strategies should 
be adopted to make this a success for McGovern￾Dole FY17 
8.1 Context and methodology
CRS’s staff mentioned that in some localities during McGovern-Dole FY14, the staff confronted 
some challenges to achieve the level of community participation necessary to operate the 
program smoothly and successfully. CRS identified three main areas of possible concern: 1) 
difficulties finding volunteers, 2) frequent complaints from the cooks, and 3) challenges to collect 
parents’ financial contribution for the school meals. 
Data was collected from parents’ and principals’ questionnaires to evaluate the level of 
participation. Unfortunately, no systematic data was collected in regard to the parents’ financial 
contribution.
During focus group discussions with the APE and AME members, we explored these issues in 
order to identify strategies to enhance community participation. 
8.2 Findings and analysis
First, we asked the parents whether they participated in McGovern-Dole’s activities, and then in 
which activity they were involved. Of the total respondents, 21.3% of parents mentioned 
participating in McGovern-Dole activities. This number is to be taken with caution, since we feel 
that the notion of “participation” was not understood by all respondents in the same manner. 
Some parents understood it as intending to volunteer for the program, while others thought 
participation meant assisting during an event or presentation. 
Table 82 illustrates the program-related activities mentioned by parents. When we only consider 
the parents, who could mention one of the six program activities below, then the percentage of 
parent participation drops to 11.6%. Because of the low numbers, we cannot assert that this 
represents the actual distribution of activities among parents who are directly involved in the 
McGovern-Dole program, but it gives us a sense of the activities they are familiar with.
Table 82: McGovern-Dole activity involvement
Activities Quantity
n 
Frequency
Canteen 25 23%
Store 6 6%
Garden/Farm 4 4%
Construction/Maintenance 9 8%
Literacy Event 9 8%
WASH Club 10 9%
Next, the evaluation team asked principals how often they experienced challenges with 
recruiting volunteers to get a sense of the magnitude of problem. As illustrated in Figure 12, 
74
72% of school principals stated they have never had issues with recruiting volunteers, while 26% 
said it was sometimes difficult. Only one school principal said that it was always difficult. In this 
case, parents had contributed materials for constructing teacher accommodations and the 
construction was never completed, so they were reluctant to participate in other activities. In 
general, however, recruitment of volunteers was not a key concern of school principals. 
Figure 12: Difficulty finding volunteers
Image 5: Cook distributing food
Based on the quantitative and qualitative 
data collected, Et4d does not anticipate 
that community participation will be a 
challenge for McGovern-Dole FY17. 
However, it is advised to exercise 
caution in communities where farms or 
CLTS will be implemented, as these 
activities might require a higher level of 
community engagement leading to 
volunteer fatigue.
We interviewed cooks to find out their 
level of satisfaction with their work. All 
the cooks are women and most of the 
cooks interviewed shared similar 
complaints, the first one being the lack of 
compensation for their work. They 
mentioned that in other programs cooks 
receive a salary. However, after 
investigating this presumption, we 
learned that this is no longer the case. 
For McGovern-Dole FY17, cooks will 
receive take home rations, which could improve their level of satisfaction.
A few cooks complained about the perceived level of effort and consider that the cooking activity 
is taking too much of their time. Very few schools seem to have a rotation of cooks, and about 
90% of the cooks reported working 5 days per week. A few schools have two teams who rotate, 
which might to be a better approach, granting mothers more free time to spend on personal and 
household tasks.
Some cooks compared their working conditions to the vendors, who are required by law to have a 
yearly medical check-up, while McGovern-Dole cooks are not. If budget allows, Et4d believes that 
it would make sense to harmonize practices. It was also mentioned on several occasions that 
72%
26%
2%
Never
Sometimes
Always
0% 10% 20% 30% 40% 50% 60% 70% 80%
75
cooks have to pay themselves for the soap to wash their apron, and they suggested that 
detergent could be provided by the program. 
Overall, despite a few complaints, recruitment and retention of cooks did not seem to be a 
challenge. Most cooks have been working since the beginning of the program and have shown no 
intention to quit (Figure 13). The only cooks who are being replaced are the ones who no longer 
have children enrolled in the school.
Figure 13: Number of years working as cook
Several principals mentioned difficulties encountered in collecting parents’ financial 
contributions. The contribution was initially established at 25cts/day, but very few schools 
manage to collect this amount from all parents. We discussed this issue with principals and 
parents, but unfortunately no systematic data was collected with the survey tools. What resulted 
from the discussions was that different modes of collecting the contribution are being implemented 
based on the community and parents’ revenues. We have identified four main modes of collection:
1. Collecting the contribution from the students daily via the teachers
2. Collecting the contribution from the students weekly via the teachers
3. Collecting the contribution from the parents monthly or per term via the AME/APE
4. Collecting the contribution from the harvest fund yearly 
The principals, in agreement with the APE, had often tried different modes over the course of 
McGovern-Dole FY14 to see which one was easier and yielded the best return. Very few 
communities seemed to collect the contribution daily, even if many of them started that way. In the 
rural communities, the yearly collection after the harvest seems to be emerging as the best 
practice because parents are mostly farmers, and they have little to no cash during most of the 
school year. The community as a whole sells the harvest, and part of the revenue is retained by 
the school to pay the community teachers. With McGovern-Dole FY17, an additional amount of 
money will be contributed to the canteen for the procurement of the condiments.
Based on discussions held with several APE members and school principals, it seems that the 
difficulties encountered in collecting contributions from parents for the canteens have lessened 
over the course of the program. However, Et4d believes that some APEs could benefit from the 
9%
1%
4%
16%
70%
0% 10% 20% 30% 40% 50% 60% 70% 80%
Less than 1 year
1 year
2 years
3 years
4 years or more
76
experience of others, and CRS could provide some guidance on the different possible ways to 
collect financial contributions from parents.
Conclusions
Et4d does not anticipate that community participation will be a challenge for McGovern￾Dole FY17 for the activities which were already part of McGovern-Dole FY14, such as 
the canteen. However, we advise some caution in communities where farms or CLTS will 
be implemented, as these activities might require a higher level of community 
engagement leading to volunteer fatigue.
If budget allows, some of the cooks’ complaints could be addressed, such as the 
procurement of soap and ensuring cooks receive an annual health check-up.
Regarding parents’ financial contribution, it might be beneficial for CRS to provide some 
guidance on best practices, as experienced by different participating schools.
To improve community participation, CRS might also want to strengthen the dynamic of 
the collaboration between its partners. Indeed, during discussions with participants, one 
of the recurrent comments was the lack of perceived coherence of the “McGovern-Dole 
Program”. The different components, such as literacy intervention and school canteen, 
are often perceived by participants as being implemented by different organizations. For 
instance, an AME could get a visit from a CRS staff person related to the food preparation, 
and on the same day, a visit from one of the NGOs working on parental involvement, 
without coordinating the visit or activities. With McGovern-Dole FY17, three new partners 
will be involved, Caritas for CLTS, and URP and PfD for the creation of community farms. 
Therefore, we recommend that CRS pays close attention to improving the coordination 
between partners. The lack of coordination between partners can create confusion and 
frustration among the participants and could hinder community engagement.
 
77
9 Recommendations 
9.1 Performance Monitoring Plan
The performance monitoring plan is not perfectly aligned with the interventions, as some activities 
do not have indicators. In addition, some indicators do not have a clear set of activities which 
could impact the results. Et4d’s suggestions include modifications or clarification of the indicators
and target values, and clarification of tools used for assessment. 
1. Clarify what tool will be used to evaluate the schools’ administrators’ use of new 
techniques.
2. For the indicator “Percent of parents who report spending time on literacy activities with 
their students in the last three days” our recommendation would be not to limit the help 
received to the parent interviewed but consider the help received by the child in general, 
and since the activities are focused on grades 1 and 2, we believe that the information 
reported by the student will be a better indicator of the outcome of the activities undertaken 
with McGovern-Dole FY17. The target value could be 70%.
3. If new EGRA instruments are designed in 2019, Et4d recommends equating the subtasks
with the instruments used for the baseline of McGovern-Dole FY17, which would enable us 
to use the new instruments for midline and endline, unless INFRE validates the McGovern￾Dole FY17 baseline instruments. Et4d does not recommend using the same instrument for 
evaluation and monitoring purposes.
4. The target of 32% of students reaching the reading comprehension target (3/5) seems high 
considering the current achievement is less than 1%. We would advise to lower it to 13% 
(3% per year)
5. Add a custom indicator to measure the progress on the EGRA lower level tasks, which 
could be the literacy achievement index as outlined in the current report. 
6. Add specific custom output indicators for the activity undertaken by WEI based on the 
updated program design. They could include literacy events organized, library created, 
community of practice meeting held, etc.
7. Add a custom indicator to evaluate the level of implementation of the activities of the 
WASH clubs, not only their existence.
8. Include the indicator “Percentage of Children Receiving a Minimum Acceptable Diet” in the 
updated PMP. MAD is an indicator measuring the proportion of children age 6‐23 months 
who receive the minimum acceptable diet. This indicator will need to be modified for school 
age children24
9. The indicator “Percent of parents in target communities who achieve a passing score on a 
test of knowledge of good health and hygiene practices” is currently not in the PMP. 
However, we feel that parents outside of the AME benefit from some activities or 
awareness campaign on hygiene and nutrition. The test used for the baseline evaluation 
could be included in the PMP with a passing score of 80% correct answers and a target of 
60% of parents passing the test.
10. The indicator “Percent of children in target communities who achieve a passing score on a 
test of knowledge of good health and hygiene practices” could have a passing score of 
70% correct answers and a target of 40% of students passing the test.
11. If the activity related to menstruation training is maintained, an output indicator should be 
included to the PMP, such as the number of training sessions, and an impact indicator 
which could be a knowledge test with 10 true/false questions, for instance. Inadequate and 
24 see: http://www.fantaproject.org/monitoringand‐evaluation/iycf‐indicators
78
unsafe sanitation facilities in schools-combined with insufficient menstrual hygiene 
materials, fear of menstrual accidents and limited access to extra uniforms can lead to up 
to 20% of girls missing school.25 Menstruation hygiene management is defined by 
WHO/UNICEF Joint Monitoring Program as “women and adolescent girls using clean 
menstrual management material to absorb menstrual blood that can be changed in privacy 
as often as necessary for the duration of the menstruation period, using soap and water for 
washing the body as required and having access to facilities to dispose of used menstrual 
management materials.” Future activities and indicators to measure could include: 1) 
Number of students trained on making re-usable sanitary pads, 2) other activities to 
increase hygiene materials, 3) increased safe and improved latrines.
12. For the indicator “Percent of schools with soap and water at hand washing stations 
commonly used by students” a reasonable target would be a 25% increase over the 
course of the project to achieve a 68% rate of proper implementation.
13. For the indicator “Number of community farms contributing to school feeding”, criteria for 
what constitutes ‘contributing to school feeding’ should be clearly defined.
14. For the indicator “Percentage of farms using best practice”, additional information will need 
to be collected to assess whether best practices are being implemented. Also, these best 
practices will need to be clearly defined. An observation test could be developed to that 
end, if the season allows, or another evaluation tool can be developed. 
15. Values might need to be set to declare that a community farm contributes to school 
feeding.
16. For the communities using the Community-led Total Sanitation approach, additional
information will be needed in order to assess whether the approach is really implemented. 
An observation test could be developed to that along with an associated indicator.
17. Regarding the indicator “Number of schools with WASH-friendly action plans” it is unclear 
whether this activity is purely related to McGovern-Dole FY17, since there are already a lot 
of WASH activities taking place in the school. A WASH-friendly criterion, in alignment with 
national standards, needs to be clearly defined with guidance developed and adhered to. 
25 Sommer, Marni, Jennifer S. Hirsch, Constance Nathanson, and Richard G. Parker. “Comfortably, Safely, and Without Shame: 
Defining Menstrual Hygiene Management as a Public Health Issue.” American Journal of Public Health 105, no. 7 (July 2015): 1302–
11. https://doi.org/10.2105/AJPH.2014.302525.
79
9.2 Information to collect at midline
In order to improve our analysis or triangulate some of the data collected directly by CRS, some
additional information should be collected at mid-line and end line. This list should be considered 
as a reminder for the development of the future instruments. 
1. Gender: In each school, for each class collect the gender of each teacher, and whether 
s/he is a community teacher, an intern or a government teacher
2. Materials: Collect information on material distributed by WEI (Activity sheets, booklet, big 
books)
3. Coaching: Teachers’ feedback on coaching performed by “conseiller pédagogique” 
4. Safe Food Preparation and Storage: Our understanding is that the indicator “Percentage 
of individuals who demonstrate use of new safe food preparation and storage practices as 
a result of USDA assistance” will be measured by CRS. However, we recommend 
developing a version of the test which would be used by the external evaluator at midline 
and end line in order to triangulate the results provided by CRS’s staff. One indicator to 
measure would be “Knowledge of Food Preparation Safety Practices: Percentage of 
School Principals/Cooks Knowledge of Food Preparation Safety Practices”. This could 
include: 1. Wash hands with clean water and soap before preparing food or before eating, 
2. Protect food from flies, cockroaches and dust, 3. Respect the shelf-life dates, 4. Avoid 
food with mold, 5. Avoid breaking the cold chain, 5. All raw consumed food (fruits and 
vegetables) must be cleanly washed before consumption.
5. Child Health and Nutrition Practices: Our understanding is that the indicator 
“Percentage of individuals who demonstrate use of new child health and nutrition practices 
as a result of USDA assistance” will be measured by CRS. However, we recommend 
developing a version of the test which would be used by the external evaluator at midline 
and end line in order to triangulate the results provided by CRS’s staff. These health and 
nutrition practices need to be clearly defined. For example-what are the nutrition-related 
practices? Consumption of iron-rich foods? Consumption of nutrient-dense foods? Eating a 
balanced, dietarily diverse diet? Deworming?
6. Meal Frequency: Include a question on the number of meals eaten for the child nutrition 
survey. “How many times did (CHILD NAME) eat foods other than liquids yesterday?”
7. Food type: Update the list of food types to follow the Minimum Acceptable Diet (MAD) 
indicator developed by the World Health Organization
8. WASH Clubs: Collect systematic information from the WASH Club, such as the frequency 
to which latrines are washed, playgrounds swept, water stations filled and cleaned, etc.
9. Teacher Language: Capture the language spoken by the teacher in the EGRA, so that we 
can see whether the fact that the teacher does or does not speak the language spoken by 
the child affects his/her reading outcome.
9.3 Intervention design
The baseline evaluation was not designed to provide feedback on the intervention design. 
However, we feel that we should share a few findings and some recommendations for CRS’s 
feedback, which are mostly the result of in-depth interviews with the partners involved in 
McGovern-Dole FY17. However, we are aware that new activities under McGovern-Dole
FY17cannot start until the baseline report is approved by USDA. This is a limitation to take into 
consideration when recommendations regarding the timing of activities.
1. Teachers’ turnover is a major challenge because the training provided is lost when the 
teacher is transferred. In order to mitigate this challenge, we suggest designing a video 
training kit, which could be built using the best practice videos which were created during 
80
McGovern-Dole FY14. This could be especially useful for interns, who are usually young 
and likely to be familiar with video training techniques. 
2. To encourage schools to be more WASH friendly, CRS might consider creating a WASH￾friendly label/sticker or badge. A competition between schools could also be organized to 
reinforce WASH messages and motivate participants to improve their school.
3. CRS could systematize knowledge-sharing about nutrition and WASH activities, currently 
the focus for only the AME and WASH Club members, but there is no mechanism in place 
to ensure the knowledge is shared with other students and with the community. Mobilizing 
that knowledge-sharing within the school and across the community would have some
important multiplier effects.
4. Some schools have implemented successful practices in regard to parents’ contributions, 
school cleanliness, or water sanitation. CRS could be more proactive in sharing success 
stories between the schools, so that the good practices could be used in more schools. 
5. The role of the children in the WASH-Club seems to be focused on acting more like 
hygiene police rather than advocating and promoting good hygiene practices. The children 
were trained, and they are very knowledgeable, however they don’t seem to be inclined to 
share their knowledge with their peers, and they have no didactic support to do so.
6. Regarding CLTS activities, the planning shared with us by Caritas did not seem realistic. 
One of the challenges is that the implementation of such a program takes about 9 months, 
and it is recommended to start working with the villagers on the diagnosis around the end 
of the harvest season so that the most time-consuming activities take place when the 
farmers’ activity is very low. Because McGovern-Dole FY17 started in January, the timing 
will be off. Our recommendation is to quickly select only a few communities and not 25 as 
originally planned and to use these communities as learning pilots. Caritas has never 
implemented the CLTS activity before, so it might also be helpful to ask for assistance from 
the local government, which has conducted it recently in the region and has national 
standards that need to be followed. 
7. McGovern-Dole FY17 has two partners working on the school farm program, URP and 
PfD. Based on our interviews, it seems that the two partners have very different ideas 
about how the farm project should be implemented in terms of planning, responsibilities, 
size of the farm, type of culture etc. During the workshop to be organized between the two 
partners, they will come to an agreement on a common approach while clarifying the 
responsibility. Home gardens are classified into three types: “traditional”, “improved” and 
“developed” gardens. Our recommendation is to promote “developed gardens”, which 
produce a wide variety of vegetables and fruit on a fixed plot of land throughout the year. 
This model relies on the dedication of participating schools/households. The model also 
builds linkages with local health and agriculture structures to provide additional services to 
participating schools/households and builds the capacity of local partners to provide 
sustained technical assistance on both agriculture and nutrition, to both schools and 
beneficiary homestead food producers after the life of the program. The school gardens 
should serve as a place for training and demonstrations on improved agricultural 
techniques and technologies, as well as a source for inputs for the school 
breakfasts/lunches. This includes looking at quality inputs such as low-cost quality seeds, 
seedlings, saplings and access to small scale irrigation. Nutrition education is an essential 
component, and leaders such as school teachers should facilitate sessions for parents to 
emphasize the importance of eating nutritious food, the food groups and dietary diversity, 
while teaching parent groups how to improve their households’ dietary diversity and 
micronutrient consumption through cooking demonstration sessions using locally grown 
foods.
8. The WEI component of the project seems to be for the most part re-conducting McGovern￾Dole FY14 intervention and tools such as the activity sheets. The activity sheets were 
designed to help the children practice at home, and participants responded very positively. 
81
Nonetheless, the activity sheets are not reusable and not sustainable. We recommend 
developing a version of the activity sheets as laminated posters which could be used in the 
classroom.
9. To improve knowledge sharing, CRS could develop some documentation which could be 
similar to the ones developed for the menstruation training. This could help the parents 
and children who have been trained to share their knowledge with other people.
10. It might be useful to harmonize hygiene practice and collaboration with the school vendors. 
More collaboration could take place and sharing of knowledge could be encouraged. This 
could be a way to ensure sustainability. In many schools the vendors could become the 
canteens’ cooks when the funding from the USDA comes to an end.
11. Some cooks compared their working conditions to the vendors, who are required by law to 
have a yearly medical check-up, while McGovern-Dole cooks are not. If budget allows, 
Et4d feels that it would make sense to harmonize practices. It was also mentioned on 
several occasions that cooks have to personally pay for the soap to wash their aprons, and 
they suggested that detergent could be provided by the program. 
12. It is not clear how building accommodations for teachers could be an incentive for teachers 
to stay in some of the remote schools. In fact, teachers do not decide whether they stay or 
not in a given school, and 90% of teachers express the desire to remain in the same 
school, despite their lack of accommodation.