zEvaluation of Leer Juntos, Aprender Juntos Early-Grade Reading Intervention in Peru
Final report
October 2019
This report was produced for review by the United States Agency for International Development under task order AID-OAA-M-12-00020. It was prepared by Julieta Lugo-Gil, Nancy Murray, Steven Glazerman, Camila Fernandez, 
and Ivonne Padilla of Mathematica. Photo credit: Save the Children.
Evaluation of Leer Juntos, Aprender Juntos Early-Grade Reading Intervention in Peru 
Final report
Disclaimer
The authors’ views expressed in this publication do not necessarily reflect the views of the 
United States Agency for International Development or the United States Government. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
i 
ABSTRACT
This report estimated the impact of the two main components of Leer Juntos, Aprender 
Juntos—an early-grade reading instruction program for linguistically diverse communities. Save 
the Children developed the program based on its Literacy Boost model and implemented it in 
Apurímac, Peru. The in-school component aims to train and coach teachers to be better equipped 
for reading instruction in the early grades. The community action component aims to strengthen 
parental and community involvement and increase children opportunities to practice reading 
outside school. 
We randomly assigned schools to one of three evaluation groups: Group A schools 
implemented both program components; Group B schools implemented only the in-school 
component; and Group C schools did not implement the program. To assess the impact of the 
community action component, we compared Group A with Group B. To assess the impact of the 
in-school component, we compared Group B with Group C. Within each school, we followed a 
group of children from 1st grade through the end of 3rd grade. 
We found a positive impact of Leer Juntos, Aprender Juntos’ in-school component on 
reading comprehension, and on some, but not all, of the measures on decoding and fluency. The 
positive impacts were mostly driven by girls. 
We found no impacts of Leer Juntos, Aprender Juntos’ community component on reading 
outcomes. Children’s reading skills were similar between schools implementing both program 
components and schools implementing only the in-school component. 
We estimated a program cost effectiveness of $136 per 0.10 standard deviation increase in 
the basic reading comprehension rate. 
This page has been left blank for double-sided copying.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
iii
ACKNOWLEDGEMENTS
The authors gratefully acknowledge numerous people and organizations that contributed to 
the implementation of all evaluation activities in Peru and the production of this report. 
The staff at the regional government of Apurímac and the Andahuaylas, Chincheros, and 
Huancarama provincial education units (known as UGELs, or Unidades de Gestión Educativa 
Local) provided a warm welcome and essential institutional support for the evaluation. Wilber 
Venegas, regional governor of Apurímac, and Fernando Acuña, education advisor to the regional 
governor and director of Education in the Andahuaylas UGEL, provided important support in 
coordinating evaluation activities in the Apurímac region. The UGEL directors, education 
specialists, and administrative staff—Wilfredo Siancas, José Porfirio Vivanco Hermoza, 
Washington Mendoza, Abraham Arenas, Neyser Tello, Kenny Monteagudo, and Celia Reynaga 
in Andahuaylas; Gary Huamán Tarco in Andahuaylas and Chincheros; Rolando Ochoa 
Talaverano, Ramiro Ortiz, Francisco Chipana, Luz Elva Villanueva Zevallos, and Tessy 
Geovanna Camiña Berrocal in Chincheros; and Santiago Benitez in Huancarama—shared key 
information on the schools and communities in the Apurímac region and offered continuous 
assistance during our data collection efforts. 
The staff from Save the Children and Kallpa provided us with key information on program 
implementation and valuable support in coordinating data collection activities. They include, 
Katy Anis, Chief of Party for Leer Juntos, Aprender Juntos; Brizza Zuazo, Leer Juntos, 
Aprender Juntos-Peru project manager; Marízabel Alagón, Save the Children/Peru’s coordinator 
of monitoring and information; José Carmen Reynaga and Eutropía Delgado, project 
coordinators, Kallpa; Francisco Zúñiga, Kallpa’s in-school actions coordinator, and Edison 
Palomino, Kallpa’s community actions coordinator. Kallpa’s technical and administrative project 
team in the Apurímac region also provided valuable information and assistance in the 
coordination of the evaluation activities.
USAID/Washington and USAID/Peru staff guided and supported the evaluation design and 
its implementation. They include, from USAID/Washington, Barbara Knox-Seith, senior 
education advisor, and Michael Lisman, education advisor, Bureau for Latin America and the 
Caribbean; and while at USAID/Peru, Clifford Brown, deputy chief, Governance & Institutional 
Strengthening; Natalia Petroni, monitoring and evaluation specialist; Miriam Choy, development 
assistance specialist; Evelyn Rodriguez-Perez, chief, Health and Education Office; Marisol 
Pérez, office director, Governance and Institutional Strengthening; Jo Jean Elenes, acting chief, 
Health and Education Office; Patricia Perez, education advisor, Health and Education Office; 
and Brindis Ochoa, education project management specialist, Health and Education Office. 
We thank our local evaluation partner, Grupo de Análisis para el Desarrollo (GRADE) and 
its staff for high quality data collection in the evaluation schools and communities. Miguel 
Jaramillo ensured careful planning of evaluation activities in Peru and provided keen oversight 
of all data collection activities. Elsa Bardalez developed a collaborative relationship with staff at 
the UGELs and has been instrumental in obtaining local support for the evaluation in the 
Apurímac region. Juan José Diaz provided overall project management support, oversaw the 
production of the data sets, and ensured consistency of data programming. Gabriela Guerrero 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
iv
coordinated the design and pilot of the follow-up reading assessments and supervised the 
enumerator training. María Balarín coordinated the qualitative data collection on program 
implementation with assistance of Kristell Benavides. Mariel Bedoya, Carla Pajares, Licet 
Infantas Díaz, Julio Almonacid, Luciana de la Flor, and Milagros Maldonado coordinated and 
supervised essential data collection and data entry activities. 
We are also indebted to Sylvia Linan-Thompson for her thorough and incisive input at 
different stages of the evaluation. Her expertise in models of reading interventions in bilingual 
classrooms and ample experience with measurement of early-grade reading skills were essential 
in designing the evaluation and interpreting baseline findings. 
At Mathematica Policy Research, Galina Lapadatova, Seth Morgan, Edith Felix, Jennifer 
Brown, Sarah Chan, Colleen Fitts, and Kimberly A. Moody-Ruffin provided critical support in 
data cleaning, programming, editing, and presentation. 
We are enormously appreciative of all the school directors and teachers for opening the 
doors of their classrooms and allowing us to observe the valuable work they do with students. 
Last, but not least, we thank the students and parents who provided us with the important 
information we are able to share in this report.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
v 
CONTENTS
ABSTRACT ………………………………………………………………………………………………………..i
ACKNOWLEDGEMENTS .............................................................................................................................iii
ACRONYMS ..............................................................................................................................................xi
EXECUTIVE SUMMARY ........................................................................................................................ ES-1
A. Introduction.................................................................................................................................... ES-1
B. Evaluation questions and design .................................................................................................. ES-2
C. Summary of findings...................................................................................................................... ES-3
D. Conclusions, limitations, lessons learned, and recommendations................................................ ES-4
I. INTRODUCTION................................................................................................................................. 1
A. Early grade literacy education in linguistically diverse contexts.......................................................... 1
B. The intervention and evaluation .......................................................................................................... 1
C. Overview of Leer Juntos, Aprender Juntos evaluation ....................................................................... 8
II. EVALUATION DESIGN AND DATA ................................................................................................... 9
A. Evaluation questions .......................................................................................................................... 9
B. Evaluation design ................................................................................................................................ 9
C. Leer Juntos, Aprender Juntos logic framework................................................................................. 11
D. Types of data collected for the evaluation......................................................................................... 12
E. Analytic approach.............................................................................................................................. 16
F. Characteristics of the sample at endline ........................................................................................... 17
G. Baseline equivalence of the analysis sample.................................................................................... 22
III. WAS LEER JUNTOS, APRENDER JUNTOS IMPLEMENTED AS INTENDED? ............................ 25
A. School participation in education interventions ................................................................................. 25
B. Implementation challenges in Peru ................................................................................................... 26
C. Implementation of the in-school component ..................................................................................... 28
D. Implementation of the community action component........................................................................ 31
IV. WHAT ARE THE IMPACTS ON INSTRUCTIONAL PRACTICES AND LITERACY 
ENVIRONMENT?.............................................................................................................................. 35
A. Classroom instruction........................................................................................................................ 35
B. Reading environment and behaviors ................................................................................................ 37
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
vi
V. WHAT ARE THE IMPACTS ON CHILDREN’S EARLY-GRADE READING SKILLS? ..................... 45
A. How we measured early-grade reading skills ................................................................................... 45
B. Impacts on reading skills ................................................................................................................... 46
C. Impacts by gender............................................................................................................................. 52
D. Cost-effectiveness............................................................................................................................... 57
VI. CONCLUSIONS, LIMITATIONS, LESSONS LEARNED, AND RECOMMENDATIONS.................. 59
A. Conclusions ....................................................................................................................................... 59
B. Limitations ......................................................................................................................................... 61
C. Lessons learned ................................................................................................................................ 62
D. Recommendations ............................................................................................................................ 63
REFERENCES............................................................................................................................................ 65
APPENDIX A: TIMELINE OF EVALUATION ACTIVITIES IN PERU, 2013–2016......................................A.1
APPENDIX B: QUALITATIVE DATA COLLECTION .................................................................................B.1
APPENDIX C: CONTROL VARIABLES IN CHILD-LEVEL REGRESSION ANALYSES ..........................C.1
APPENDIX D: SCHOOL INFRASTRUCTURE CHARACTERISTICS.......................................................D.1
APPENDIX E: CLASSROOM AND HOUSEHOLD ENVIRONMENTS......................................................E.1
APPENDIX F: STANDARD DEVIATIONS OF CHILDREN’S READING SKILLS MEASURES................F.1
APPENDIX G: SENSITIVITY ANALYSES................................................................................................ G.1
APPENDIX H: EFFECTS OF THE COMMUNITY COMPONENT FOR BOYS AND GIRLS.....................H.1
APPENDIX I: THE COSTS OF LEER JUNTOS, APRENDER JUNTOS.................................................... I.1
APPENDIX J: EMERGENT LITERACY SKILLS OF GIRLS AND BOYS IN THE EVALUATION AT 
BASELINE........................................................................................................................................ J.1
APPENDIX K: DATA COLLECTION INSTRUMENTS...............................................................................K.1
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
vii
TABLES 
II.1. Response rates to data collections in the evaluation..................................................................... 14
II.2. Characteristics of evaluation schools in the final follow-up year ................................................... 19
II.3. Background characteristics of the teachers of the evaluation students in the final follow￾up year ........................................................................................................................................... 20
II.4. Experience and tenure characteristics of the teachers of the evaluation students in the 
final follow-up year ......................................................................................................................... 20
II.5. Teacher-reported language skills in the final follow-up year (percentages of teachers) ............... 21
II.6. Grade and school at the time of the final follow-up data collection (percentage of 
students) ........................................................................................................................................ 22
II.7. Baseline characteristics of children in the evaluation sample........................................................ 23
III.1. School participation in education programs, reported by principals or other school 
administrators................................................................................................................................. 24
IV.1. Materials to facilitate early reading instruction that are visible to students in the classroom 
(percentage of classrooms) ........................................................................................................... 37
IV.2. Characteristics of the books observed in the classrooms (percentage of classrooms)................. 38
IV.3. Availability of books for children at home....................................................................................... 39
IV.4. Dedicated time for reading at home............................................................................................... 41
IV.5. Children’s activities at home that are not related to reading.......................................................... 42
V.1. In-school component effects on literacy outcomes in standard deviation units............................. 51
V.2. Impacts on the literacy outcomes of girls and boys in the evaluation............................................ 53
V.3. In-school component effects on literacy outcomes for girls and boys, in standard 
deviation units ................................................................................................................................ 54
B.1. Respondents, methods, topics of inquiry, and location in the qualitative data collection .............B.1
C.1. Control variables included in the analyses of the reading skills of 3rd grade children in the 
evaluation......................................................................................................................................C.1
D.1. Available utilities in evaluation schools (percentages of schools) ................................................D.1
D.2. Other facilities available in schools ...............................................................................................D.2
D.3. Physical and health hazards observed in schools ........................................................................D.3
E.1. Teacher-reported participation in reading instruction professional development (PD) 
activities in the endline year (percentages of teachers)................................................................E.2
E.2. Teacher-reported participation in reading instruction professional development (PD) 
activities in the year before endline survey administration (percentages of teachers).................E.3
E.3. Teacher-reported participation in reading instruction professional development (PD) 
activities focused on specific topics in the endline year................................................................E.5
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
viii
E.4. Teacher-reported methods and strategies to teach reading.........................................................E.6
E.5. Students’ use of different types of text, as reported by teachers..................................................E.7
E.6. Number of minutes students spend on the following activities during a typical language 
and reading class, as reported by teachers ..................................................................................E.8
E.7. Teacher-reported needs and challenges (percentages of teachers)............................................E.9
E.8. Observed instructional practices to teach foundational literacy skills.........................................E.12
E.9. Observed general instructional practices....................................................................................E.13
E.10. Observed teachers’ use of language in the classroom...............................................................E.13
E.11. Observed students’ use of language in the classroom ...............................................................E.14
E.12. Books and materials observed in the classroom ........................................................................E.14
F.1. Standard deviations of children’s reading skills outcome measures ............................................F.1
G.1. Differences between treatment groups in children’s literacy outcomes, accounting for 
nonresponse to the 3rd grade assessment ................................................................................. G.1
G.2. Differences between treatment groups in children’s literacy outcomes, based on sample 
with complete data ....................................................................................................................... G.2 
G.3. Impacts on the literacy outcomes of children in Phase I and Phase II schools ........................... G.3
G.4. Impacts of the full intervention ..................................................................................................... G.4
H.1. Community component effects on literacy outcomes for girls and boys, in standard 
deviation units ...............................................................................................................................H.1
J.1. Baseline emergent literacy skills of girls and boys in the evaluation ............................................ J.1
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
ix
FIGURES
II.1. Leer Juntos, Aprender Juntos evaluation design........................................................................... 11
II.2. Leer Juntos, Aprender Juntos logic framework ............................................................................. 12
II.3. The Apurímac region in Peru ......................................................................................................... 18
IV.1. Teacher-reported participation in reading instruction training and coaching in the endline 
year ................................................................................................................................................ 36
IV.2. Percentages of evaluation children whose parents or other household members read with 
them ............................................................................................................................................... 40
IV.3. Reading-related activities at home (percentage of children) ......................................................... 42
V.1. No statistically significant differences between treatment groups in the number of pseudo￾words read correctly in the decoding task (of 50) .......................................................................... 47
V.2. No statistically significant differences between treatment groups in the number of words 
read correctly in the fluency task (of 154)...................................................................................... 48
V.3. Positive program impacts on decoding and fluency accuracy scores ........................................... 48
V.4. Positive program impacts on the percentage of children who achieved a basic level of 
reading comprehension skills......................................................................................................... 50
V.5. No differences between treatment groups in the number of questions answered correctly 
in the reading comprehension task (of 15)..................................................................................... 50
V.6. Growth in girls decoding skills........................................................................................................ 55
V.7. Growth in boys decoding skills....................................................................................................... 55
A.1. Timeline of Leer Juntos, Aprender Juntos activities 2012–2016 in Peru......................................A.1
E.1. Teachers in intervention groups reported higher rates of participation in professional 
development (PD) activities focused on specific literacy skills .....................................................E.4
E.2. Observed allocation of instruction time in a typical language or reading class ..........................E.10
This page has been left blank for double-sided copying.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
xi
ACRONYMS
DIGEIBIR General Office of Intercultural, Bilingual, and Rural Education 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ECE Evaluaciones Censales de Estudiantes
EGRA Early Grade Reading Assessment
GRADE Grupo de Análisis para el Desarrollo
IBE Intercultural and bilingual education 
INEI Instituto Nacional de Estadística e Informática 
LAC Latin America and the Caribbean 
LAC Reads Latin America and the Caribbean Reads
LJAJ Leer Juntos, Aprender Juntos 
NGO non-governmental organization 
PD Professional development activities
PELA Strategic Learning Outcomes Program
REDES National Rural Education Network Program 
SPI Intercultural Pedagogical Support 
UGEL Unidades de Gestión Educativa Local 
USAID U.S. Agency for International Development
This page has been left blank for double-sided copying.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
ES-1 
EXECUTIVE SUMMARY
A. Introduction 
This report presents results of the impact evaluation of the Leer Juntos, Aprender Juntos
program—an approach to early-grade reading instruction in communities with linguistically 
diverse populations. Save the Children developed the Leer Juntos, Aprender Juntos program 
based on its Literacy Boost model, which includes teacher training and community involvement, 
and implemented the program in Quechua-speaking Apurímac region of Peru and the K’iche’-
speaking region of Guatemala. Mathematica designed a rigorous evaluation of Leer Juntos, 
Aprender Juntos in both Peru and Guatemala. This report focuses on the impact evaluation 
findings for Peru. A separate report presents the impact evaluation findings for Guatemala 
(Lugo-Gil et al. 2021b).
The Leer Juntos, Aprender Juntos intervention has two main components, the in-school 
component and community action component. 
In-school component: teacher training and coaching. The in-school component of Leer 
Juntos, Aprender Juntos was based on the Literacy Boost model. The goal of this component 
was to train and coach teachers to be better equipped for mother tongue and/or Spanish reading 
instruction in the early primary grades. This component entailed six main activities intended to 
increase reading instruction class time and improve the quality of reading instruction: 
1. Training trainers in reading instruction techniques in both Spanish and mother tongue 
2. Training teachers in the five core skills of reading: alphabet knowledge, phonological 
awareness, vocabulary, fluency, and comprehension 
3. Creating materials for print-rich classrooms in both Spanish and mother tongue 
4. Mentoring and coaching teachers in reading instruction practices 
5. Training teachers in conducting formative assessments to track progress of children’s reading 
skills
6. Guiding teachers to incorporate five core skills of reading and related reading instruction 
techniques into daily school activities
Teacher training consisted of nine modules on a range of reading instruction topics, 
including introduction to reading acquisition and instruction in the early primary grades, use of 
formative assessments, and five core reading skills. Following the teacher training, the program’s 
technical staff (coaches) visited classrooms at least once (and in most cases, two or three times) 
every three months. These visits aimed to support teachers by observing their work, 
demonstrating teaching techniques, and suggesting improvements in the use of reading 
instruction strategies. Coaches also moderated teacher learning groups to strengthen and practice 
topics discussed in the training workshops, which included conducting activities to develop the 
five core reading skills, designing lesson plans incorporating core reading skills, and planning 
strategies to develop literacy skills in students’ mother tongue. The training and coaches did not 
dictate or favor one language of instruction over the other. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
ES-2 
Community action component. The community action component of the program, also 
based on the Literacy Boost model, aimed to strengthen parental and community involvement in 
building children’s reading abilities and increasing their opportunities to practice reading outside 
regular class time. This component was delivered primarily by trained community volunteers and 
included engaging group activities. Specifically, the community action component included the 
following main activities:
• Creating printed materials in the children’s mother tongue, Spanish, or both, to build portable 
libraries known as book banks 
• Promoting the use of book banks among community members 
• Conducting reading activities in the community such as story hours, reading camps, and 
reading festivals or fairs
• Coordinating peer assistance through reading buddies 
• Conducting school–community accountability meetings and reading awareness workshops 
with parents and community members 
B. Evaluation questions and design
The evaluation was implemented as a randomized controlled trial. The evaluation team 
randomly assigned schools to one of three evaluation groups: Group A, schools that implemented 
both components of Leer Juntos, Aprender Juntos; Group B, schools that implemented only the 
in-school component of Leer Juntos, Aprender Juntos; and Group C, schools that did not 
implement the program—we refer to these schools as prevailing practice. The evaluation aimed 
to answer the following two primary evaluation questions: 
● What is the impact of the teacher training and support component of Leer Juntos, Aprender 
Juntos on early-grade reading and other outcomes relative to prevailing practice? This 
question contrasts schools in Group B to those in Group C. 
● What is the impact of the community action component of Leer Juntos, Aprender Juntos on 
early-grade reading and other outcomes relative to an intervention that does not have the 
community action component? This question contrasts schools in Group A to those in 
Group B. 
In addition to these two primary questions, the evaluation addresses the following question about 
implementation:
● Were the program components implemented as intended? This question relates to whether 
each program component’s services were offered as originally intended, whether 
participants took part, and whether these program components had intermediate impacts on 
teaching and the availability and use of reading materials. It also considers any barriers or 
challenges to effective implementation.
In Peru, we recruited 145 schools. Within each school, we followed a group of children from 
1st grade through the end of 3rd grade. The implementation of Leer Juntos, Aprender Juntos in 
Peru started in May 2013 and ended in December 2015. Because Save the Children planned to 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
ES-3 
roll out the program in phases, we also carried out the evaluation in two phases. We randomly 
assigned eligible schools and collected baseline data in the first half of the evaluation schools in 
2013 (Phase I), and the second half of schools in 2014 (Phase II). We conducted two follow-up 
data collections after the baseline: one at the end of the second year of implementation (midline, 
in 2014 for Phase I and 2015 for Phase II) and another at the end of the third year of 
implementation (endline, in 2015 for Phase I and 2016 for Phase II). Implementation of the 
program started in May 2013 for Phase I schools and March 2014 for Phase II schools. We 
present results for students pooled across the two phases because we did not find any significant 
pattern of differences between phases.
As part of the evaluation, we also conducted an implementation evaluation and a cost 
effectiveness analysis. This involved data from several sources, described in Chapter II, 
including school observations, classroom observations, teacher surveys, focus groups, and 
interviews. The observations took place at baseline (first year of the evaluation) and midline 
(second year). Teacher surveys took place at baseline, midline and endline, as described in 
Chapter II. For the cost-effectiveness analysis, described in Chapter V, we relied on program 
implementation and cost data provided to us by Save the Children. 
C. Summary of findings
The results for Peru suggest that the program’s teacher training and coaching 
component did improve children’s reading skills. By the end of 3rd grade, about 83 percent of 
the children in schools assigned to the basic Leer Juntos-school only intervention could read a 
simple passage and answer at least one question correctly—an indication of basic reading 
comprehension (defined as be able to complete both the decoding and fluency tasks and answer 
at least one question correctly in the reading comprehension task)—compared to only 74 percent 
of children in schools using prevailing practices. This positive impact of nearly 9 percentage 
points on basic reading comprehension was statistically significant. Impacts on other reading 
skills, such as decoding accuracy and fluency accuracy, told a similar story. Subgroup analysis 
suggests that all of these impacts were driven by improvements in girls’ reading skills. However, 
we found no statistically significant impacts on non-accuracy reading measures, such as fluency 
and decoding measured by number of correct words read per minute and reading comprehension 
when we measured it by number of questions answered correctly, instead of whether they were 
able to answer at least one question correctly. 
Students in the in-school component-only group had better reading outcomes than 
students in the control group. The sizes of some of the impacts were substantial, equivalent to 
effect sizes of 0.19, 0.21, and 0.13 for decoding accuracy, fluency accuracy, and reading 
comprehension, respectively. These effect sizes represent 15 to 24 percent of the gap between 
indigenous and non-indigenous Peruvian children in Spanish language skills (Hernandez-Zavala 
et al. 2006). Impacts on decoding and fluency in terms of words read per minute were also 
positive, but as noted above, not statistically significant. 
Impacts in Peru were driven by impacts on girls’ emergent reading skills. There were 
positive differences in the means of the outcomes of interest between Group B and Group C for 
boys and girls, but the differences were statistically significant only for the girls. For example, 
the impact of the in-school component (difference between Groups A and B) on the percentages 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
ES-4 
of boys and girls who showed emergent reading comprehension skills is statistically significant 
for the girls. The results are similar for the decoding and fluency accuracy scores (statistically 
significant impacts for girls). 
The estimated cost-effectiveness of the in-school component was $136 per student per 
0.1 standard deviation increase in emergent reading skills. We estimate that three years of 
providing the full program (the community and the in-school components) cost $527 per child, 
and the cost of providing the in-school component only is $434 per child. If we use the impact 
estimate for the percentage of children who could read a simple passage and answer at least one 
question correctly (the children who had a basic level of reading comprehension), the cost of the 
in-school component per 0.1 standard deviation improvement on that outcome is $136 per child. 
Policymakers need to know whether these impacts are large enough to justify investing resources 
in Leer Juntos, Aprender Juntos compared with other ways to improve early-grade literacy
The evidence of favorable impacts of the program’s in-school component is consistent 
with some of our findings generated by classroom observations and teacher survey data. 
The schools that implemented only the in-school component of the program provided a more 
print-rich environment in their classrooms than did schools that did not implement any of the 
components of the intervention. In particular, the evaluation team observed materials to facilitate 
early-grade reading instruction in Spanish and Quechua in more classrooms implementing the in￾school component of the program than in classrooms in the prevailing practice group. In 
addition, picture books that do not have any text were observed in more classrooms in the 
prevailing practice group than in the schools implementing the in-school component. On 
instructional practices, either reported by teachers or observed by evaluators, we found no 
meaningful pattern of statistically significant program effects. 
We found no evidence of impacts on reading outcomes from the community component 
of Leer Juntos, Aprender Juntos. We did not find any meaningful pattern of statistically 
significant differences in children’s reading skills between the group that implemented both 
components of the program and the group that implemented only the in-school component of the 
program. This finding is consistent with the qualitative evidence: implementing the community 
component activities was challenging. In Peru, recruiting and retaining volunteers for the 
community activities was difficult, attendance at parent workshops was lower than expected, and 
creating spaces dedicated to reading activities at the home was difficult for some families. 
Despite those challenges, however, program stakeholders had positive opinions of the 
community component activities. In focus groups, parents and teachers said they thought reading 
camps and book availability through the program’s book banks did inspire children’s interest in 
reading. 
D. Conclusions, limitations, lessons learned, and recommendations
1. Conclusions
Leer Juntos, Aprender Juntos was implemented largely as intended, although there 
were some challenges. Teachers in schools assigned to implement the in-school component of 
Leer Juntos, Aprender Juntos did receive the program’s training and coaching activities. 
Teachers reported establishing positive and supportive relationships with the program trainers, 
expressed that the program’s methodology and training approach are practical and feasible to 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
ES-5 
apply in the classroom, and reported observing increased student engagement. However, 
challenges in the implementation of this component included difficulty retaining qualified 
program staff for training and coaching. 
The activities that were part of the community action component were delivered mostly as 
intended in the schools that had been assigned to receive that component. Those activities 
included access to book banks, story hour, reading camps, and reading festivals. Challenges in 
the implementation of this component included difficulty recruiting and retaining staff to lead the 
activities of the community action component, and logistical challenges for teachers, coaches, 
trainers, and volunteer coordinators due to the geography of the region and the limited access to 
public transportation.
The in-school component of Leer Juntos, Aprender Juntos had favorable impacts in the 
classroom literacy environment. Teachers in schools implementing only the in-school 
component of Leer Juntos, Aprender Juntos participated in the program training and coaching 
activities, and more classrooms in these schools than in prevailing practice schools had a 
complete alphabet in Spanish and Quechua that was visible to students and more books with text. 
Our findings also show that teachers in schools implementing only the in-school component 
established positive and supportive relationships with the program specialists who conducted 
training workshops and coaching sessions with them. In addition, teachers in schools 
implementing Leer Juntos, Aprender Juntos did use the reading instruction practices they learned 
from the program, and teachers reported observing increased student engagement. 
The in-school component of Leer Juntos, Aprender Juntos had favorable impacts on 
some but not all measures of reading skills. Three years after the program started, about 83 
percent of the children in schools assigned to only the in-school intervention component could 
read a simple passage and answer at least one question correctly—an indication of basic reading 
comprehension—compared with only 74 percent of children in schools using prevailing 
practices. This was a statistically significant positive impact on basic reading comprehension, 
driven mainly by improvements in girls’ emerging reading comprehension skills. We also found 
positive, statistically significant impacts of the in-school component on decoding and fluency 
accuracy. The sizes of some of the impacts were substantial, equivalent to effect sizes of 0.19, 
0.21, and 0.13 for decoding accuracy, fluency accuracy, and basic reading comprehension, 
respectively. However, we found no impacts on non-accuracy reading measures, such as fluency 
measured by number of correct words read per minute. 
Three years of providing the in-school component costs an estimated $434 per child. If 
we use the impact estimate for the percentage of children who could read a simple passage and 
answer at least one question correctly (basic level of reading comprehension), the cost of the in￾school component per 0.1 standard deviation improvement on that outcome is $136 per child. 
The community action component of Leer Juntos, Aprender Juntos did not have an 
impact on students’ reading skills beyond what the in-school component attained. We did 
not find evidence that students in schools implementing both Leer Juntos, Aprender Juntos
components had better reading skills or spent more time on reading activities at home than 
students in schools that only implemented the Leer Juntos, Aprender Juntos in-school 
component. Implementation challenges could help explain this result. Program staff reported 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
ES-6 
challenges in recruiting and retaining volunteers, which resulted in delayed implementation of 
the community action component relative to the implementation of the in-school component. In 
addition, the reading buddies activity was eventually folded into the in-school component, with 
teachers delivering the activity in their classrooms during regular school hours once a week 
instead of volunteers conducting the activity in the community outside of school hours. Another 
possible explanation is that it was challenging for the community action component to create an 
effect in a context where parents were already engaging in reading-related activities with their 
children. In Peru, most parents in all treatment groups reported being literate and reading and 
looking at books with the children. 
2. Limitations
The baseline of the evaluation was not a “true” baseline. The baseline assessment of 
students’ early literacy skills was administered about three months after random assignment, 
which was about two months after the rollout of the program’s teacher training activities. It was 
not feasible to measure baseline student outcomes before random assignment because several 
evaluation activities needed to be completed before data collectors could begin baseline data 
collection; these included identifying a local partner that could assist in recruiting and training 
field workers, implementing the data collection plan, and supervising data collection activities. 
As a result of the late baseline, the children’s skills that we captured in the baseline assessment 
(when students in the evaluation were in first grade) reflect the ability of the students after up to 
three months of potential exposure to different conditions caused by assignment of schools to 
different intervention groups. 
Despite this limitation, it is still appropriate to consider the evaluation’s baseline assessment 
as the reference point for the evaluation. Under most circumstances, a late baseline will still be 
useful for estimating the impact of that intervention as long as the impact on skill development is 
slow in the early period of program implementation (Schochet 2010). That is the case in the 
current evaluation because the teachers in the groups receiving the program (Groups A and B) 
had only been practicing their newly acquired skills for at most two months when the baseline 
data collection took place, and literacy acquisition in children is a process that happens gradually 
over many months or even years. 
The evaluation design assumes that the impacts of in-school and community action 
components of the intervention are additive. Specifically, the evaluation design assumes that 
the effect of the community action component on its own relative to prevailing practice is equal 
to the sum of the impact of the combined in-school and community action components relative to 
the in-school component on its own. This assumption could be violated if implementing the in￾school component makes it easier or harder to simultaneously do community activities. We did 
not find clear evidence of such a violation, but it should be taken into account by any 
policymakers who wish to apply the lessons of this evaluation to a future implementation of 
community action on its own. 
Findings from this evaluation may not necessarily apply to other parts of the LAC 
region. It is always tempting to generalize from the experience of one or two evaluations to other 
contexts, but caution is warranted. In Peru, the evaluation included schools from just 3 of 7 
provinces in Apurímac, which is just one of 24 regions in the country. While a parallel 
evaluation was conducted in the El Quiche region of Guatemala, it too included schools from a 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
ES-7 
small area of the country. In both countries, communities were selected with the following 
criteria in mind: the locations had to be within reasonable driving distance from each other to 
facilitate intervention and evaluation activities. They had to contain a high percentage of families 
that spoke one language besides Spanish, in this case Quechua (in Peru) or K’iche’ (in 
Guatemala). Results could differ if an approach like Leer Juntos, Aprender Juntos was taken in 
other communities with different characteristics and challenges than those communities 
included in this evaluation. 
3. Lessons learned
Teacher training and coaching on how to teach literacy skills can be effective when 
they are distinct from other approaches already in use. In Peru, the in-school component was 
effective, generating positive impacts on some reading skills, whereas in Guatemala this was not 
the case. In Peru, the prevailing practice approach (recommended by the Ministry of Education) 
advocates a communicative and textual approach aimed at educating students to understand, 
speak, read, and write in their mother tongue and in Spanish, and to use both languages in 
communicative contexts. The communicative and textual approach is very different from the 
approach of the Leer Juntos, Aprender Juntos program, in which teachers are trained in specific 
strategies to promote foundational reading skills and reading comprehension in multilingual 
contexts; monitor students’ mastery of those skills; and increase students’ access to and use of 
reading materials in the mother tongue. In Guatemala, the prevailing practice approach to early 
reading instruction promotes foundational reading skills and reading comprehension. The 
national reading curriculum for the early grades emphasizes children’s alphabet knowledge, 
phonological awareness, vocabulary, and decoding skills, and seeks to gradually improve 
children’s reading fluency and reading comprehension skills, similarly to the Leer Juntos, 
Aprender Juntos approach. Therefore, the Leer Juntos, Aprender Juntos approach is more 
distinct to the prevailing practice in Peru than to the prevailing practice in Guatemala, which 
could explain why the in-school component had impacts in Peru but not in Guatemala. 
Relying on volunteer work has substantial risks for fidelity of implementation. This 
evaluation demonstrated that implementing and evaluating a program component that relies on 
volunteer work has substantial risks. In the Apurímac region where the program and evaluation 
took place, doing volunteer work related to children’s schooling is not common practice and 
people expected payment for their work. This made recruiting and retaining volunteers 
challenging and caused delays and disruptions in implementing the community action 
component. 
4. Recommendations
Policymakers should continue implementing the in-school component of Leer Juntos, 
Aprender Juntos and possibly expand to other schools in the Apurímac region if they wish 
to improve literacy skills and do not have an alternative intervention that can deliver these 
impacts for this cost. The in-school component showed evidence of positive impacts on some 
reading outcomes, so sustaining and possibly expanding this component of the program might 
help further the region’s progress in reading achievement. In addition, the region already has the 
capacity and experience to support the sustainability of this component. Nevertheless, the costs 
of the program are important to consider when thinking about continuing and expanding the 
program. We estimated that the cost of the in-school component of the program in Peru was 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
ES-8 
approximately $434 per child. This translates to an estimated cost effectiveness of the Leer 
Juntos, Aprender Juntos program of $136 per child per 0.10 standard deviation increase in the 
basic reading comprehension rate. This places the intervention on the high end of education 
interventions that were catalogued by Evans and Ghosh (2008) as being effective. It is not 
surprising that to improve reading comprehension in multilingual populations that are 
geographically harder to reach would be at the higher end of the range of cost-effective 
interventions. 
Policymakers and donors should consider alternative ways to conduct activities that 
increase the time children spend reading outside of school hours. If there is a desire to 
continue testing activities that aim to increase time reading outside of school hours, we 
recommend exploring an alternative way to implement the activities that does not rely primarily 
on volunteer work. We also recommend assessing the feasibility and potential take-up of the 
alternative approach to ensure it suits the cultural context of the communities for which it is 
intended. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
1 
I. INTRODUCTION
A. Early grade literacy education in linguistically diverse contexts
The past decade has seen outstanding progress toward the Millennium Development Goals 
of universal primary school completion and gender parity in education, both around the world 
and in the Latin America and Caribbean (LAC) region in particular. Yet deficits related to 
learning and access to education for poor and marginalized populations persist, particularly for 
groups whose mother tongue is not the same as the societal or official language of their country 
or region (Benavides et al. 2010; Gove and Cvelich 2011; Kim et al. 2016). In Peru, for example, 
among dual-language students1
 in grade 4 in 2016, only 25 percent achieved the Spanish reading 
comprehension level expected for their grade (Ministerio de Educación 2017). 
In LAC countries with linguistically diverse populations, such as Peru and Guatemala, 
children enter school with a wide variation in abilities to understand and speak the country’s 
societal or official language (Cueto et al. 2012). In the case of Peru and Guatemala, the societal 
or official language is Spanish. In these countries, some children have no knowledge of Spanish, 
whereas others speak and understand it fluently, even if they live and attend school in a region 
where the predominant language is the same as their mother tongue. Therefore, the challenge for 
educational programs aimed at improving reading skills is to work with a range of instructional 
approaches in a context of significant linguistic diversity. Such programs must focus on helping 
children develop the underlying skills that will ultimately result in improved reading skills.
Studies have demonstrated that programs involving instructional practices and other 
supports in children’s mother tongue improve academic outcomes in some developing countries 
(Chesterfield and Abreu-Combs 2011; Crouch et al. 2009; Friedlander and Goldenberg 2016; 
Hernandez-Zavala et al. 2006; Patrinos and Velez 2009; Piper et al. 2016). However, no studies 
based on a rigorous evaluation design have been implemented in the LAC region. Moreover, 
existing studies have not focused on assessing the effectiveness of comprehensive reading 
programs that incorporate both transitional instructional approaches in the classroom and mother 
tongue supports for building reading abilities outside the classroom. Models aimed at increasing 
instruction time in the classroom and reading practice time at home (such as community 
engagement models, volunteer training to teach low-performing students, and remediation 
provided by locally trained teachers) have shown promise (Banerjee et al. 2010, 2007; Bruns and 
Luque 2014). Again, such models have not been rigorously evaluated in the LAC region, and 
their cost-effectiveness is unknown. 
B. The intervention and evaluation
To address the need for high quality evidence on interventions that improve student 
outcomes, particularly in reading, the U.S. Agency for International Development (USAID) 
contracted with Mathematica Policy Research as its independent evaluator to design and conduct 
impact evaluations of promising reading interventions and education-access interventions. The 
first program evaluated under the contract was the Leer Juntos, Aprender Juntos program. 
Hypothesizing that the community action component might be an effective way to improve 
1 This refers to students whose mother tongue is a language other than Spanish (Ministerio de Educación 2017).
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
2 
learning by increasing time on task, USAID funded Save the Children to implement this 
promising approach to early-grade reading instruction in LAC communities with linguistically 
diverse populations. Save the Children based the program on its Literacy Boost model, which 
includes teacher training and community involvement, and implemented the program in the 
Quechua-speaking Apurímac region of Peru and the K’iche’-speaking region of Guatemala. 
Mathematica worked with Save the Children to design a rigorous evaluation of Leer Juntos, 
Aprender Juntos in both Peru and Guatemala. This report focuses on the impact evaluation 
findings for Peru, and a similar report presents the impact evaluation findings for Guatemala 
(Lugo-Gil et al. 2021b; Lugo-Gil et al, 2021).
1. Leer Juntos, Aprender Juntos 
Leer Juntos, Aprender Juntos means Read Together, Learn Together. Save the Children, an 
international nongovernmental organization focused on promoting children’s rights and 
providing relief and support to children in developing countries, developed this program based 
on its Literacy Boost model. Save the Children has implemented Literacy Boost in at least 24 
countries (Save the Children 2013a), including El Salvador, Haiti, Ethiopia, Pakistan, and more 
recently, Rwanda (Friedlander and Goldenberg 2016). Literacy Boost has three main 
components: (1) an in-school component that includes teacher training and coaching, (2) a 
community component that relies on volunteers to engage with young students and promote a 
culture of reading outside of regular instructional periods at school, (3) and a formative 
assessment component to track children’s progress in reading. The assessment component of 
Literacy Boost was incorporated as part of the in-school component of Leer Juntos Aprender 
Juntos. The combined in-school and community components represent the full Leer Juntos, 
Aprender Juntos intervention. 
The in-school and community components of the Leer Juntos, Aprender Juntos program 
align with the elements of reading instruction that existing research (Comings 2012) has 
identified as effective when implemented simultaneously: (1) text and materials, (2) teacher 
training and support, (3) community and parental support, and (4) assessment and tracking. 
Therefore, in addition to targeting important changes in the classroom, the Leer Juntos, Aprender 
Juntos program sought to engage community members and parents in the learning process. The 
community outreach component is a potentially low-cost, effective way to increase children’s 
time on task.
The Leer Juntos, Aprender Juntos program adapted the teacher training and community 
action components of Literacy Boost to K’iche’- and Quechua-speaking populations2
 in 
Guatemala and Peru, respectively. The program’s primary implementers—Save the Children in 
Guatemala and Kallpa, its local partner in Peru—trained teachers to teach and monitor students’ 
mastery of core reading skills. Save the Children and Kallpa also worked to strengthen parent 
and community involvement in building children’s reading abilities and opportunities to practice 
reading skills in their mother tongue. Those efforts included creating and providing context￾2 In Peru, the program was implemented in two provinces (Andahuaylas and Chincheros) in the southern-central 
Apurímac region of the country. Within those two provinces, the evaluation was implemented in three school 
districts (Andahuaylas, Chincheros, and Huancarama). 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
3 
relevant reading materials in children’s mother tongue and conducting community activities that 
promoted reading engagement and a culture of literacy. 
The Leer Juntos, Aprender Juntos implementation team consisted of staff from two 
organizations; a management team from Save the Children/Peru based in Lima and Kallpa’s field 
team primarily based in Apurímac. The implementation team was led by the project director 
based in the United States, in collaboration with a country project coordinator, a monitoring and 
evaluation coordinator, a finance officer, and an education adviser based in Lima. Save the 
Children’s management team coordinated and monitored program activities implemented 
directly by Kallpa’s staff in the region. Kallpa’s field team consisted of two coordinators, who 
each oversaw a small team of specialists. Save the Children and Kallpa held weekly meetings; 
both teams met every two weeks with Save the Children’s project director. In addition, Save the 
Children/Peru’s country project manager conducted supervision visits in the region about five 
times per year. 
Planning for implementation of Leer Juntos, Aprender Juntos in Peru began in fall 2012 and 
continued through the first quarter of 2013 with the development of a work plan and project 
strategy, recruitment and training of program implementation staff, meeting with stakeholders, 
development of a program monitoring and evaluation system, and preparation of program 
materials (Save the Children and Kallpa 2012, 2013). Save the Children rolled out the program 
in two phases. Implementation of Phase I started in 2013 and included half of the schools 
assigned to receive the full intervention, the teacher training and coaching component and the 
community action component, and half of the schools assigned to receive the teacher training 
and coaching component. In 2014, Save the Children continued implementing the program with 
Phase I schools and began implementing it with the remainder of the schools assigned to the 
groups where the full intervention and the teacher training and coaching component were 
offered. The community component rolled out in August 2013 with the training of community 
volunteers who would serve as ‘reading promoters.’ In December 2015, Save the Children 
completed the implementation of Leer Juntos, Aprender Juntos; closed out field activities; and 
prepared the sustainability strategy and materials.
In-school component: teacher training and coaching. The in-school component of Leer 
Juntos, Aprender Juntos, which is based on the Literacy Boost model, followed the 
recommendations of the National Reading Panel for best approaches in reading instruction 
(National Reading Panel 2000). The goal of this component was to train and coach teachers so 
they are better equipped to provide mother tongue and/or Spanish reading instruction focusing on 
five foundational reading skills (alphabet knowledge, phonological awareness, vocabulary, 
fluency, and comprehension), to monitor students’ mastery of these skills, and to increase 
students’ access to and use of reading materials in the mother tongue. Teachers learned to 
incorporate instruction in the mother tongue and adapt this instruction to the linguistic 
background of the students in a particular school. 
The school-based component entailed six main activities, intended to increase class time on 
reading instruction and improve the quality of reading instruction: 
1. Training trainers in reading instruction techniques in both Spanish and mother tongue 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
4 
2. Training teachers in the five core skills of reading: alphabet knowledge, phonological 
awareness, vocabulary, fluency, and comprehension
3. Creating materials for print-rich classrooms in both Spanish and mother tongue 
4. Mentoring and coaching teachers in reading instruction practices 
5. Training teachers in conducting formative assessments to track progress of children’s reading 
skills
6. Guiding teachers to incorporate five core skills of reading and related reading instruction 
techniques into daily school activities
The teacher training activities (for more details, see the teacher training toolkit [Save the 
Children 2012a]) consisted of nine modules focused on the following topics: 
1. Introduction to reading development and instruction for young children 
2. Formative assessment 
3. Addressing language issues in the literacy classroom 
4. Letter knowledge/alphabetic principles 
5. Phonological awareness 
6. Reading fluency 
7. Vocabulary 
8. Reading comprehension 
9. Reflecting on and applying best practices for teaching reading comprehension and formative 
assessment
Teachers in 1st, 2nd, and 3rd grades in the intervention schools (schools implementing the 
full intervention and the teacher training and coaching component) received most of the training 
content at the beginning of the school year and completed their training throughout the school 
year.3
 Training in the nine program models in Peru was provided in eight one-day sessions in 
2013 (to Phase I teachers) and six one-day sessions in 2014 and 2015 (to Phase I and Phase II 
teachers). Following training, the program’s technical staff (coaches) visited classrooms at least 
once (and in most cases, two or three times) every three months. These visits aimed to support 
teachers by observing their work, interacting in the classrooms, demonstrating teaching 
techniques, and suggesting improvements in applying program teaching strategies (Save the 
Children and Kallpa 2013, 2014, 2015). Coaches also supported open discussion sessions (called 
Teacher Circles) with groups of teachers during the afternoons or evenings. These sessions 
focused on (1) strengthening and practicing the topics learned in the training, (2) conducting 
activities to develop the five core reading skills, (3) designing lesson plans that incorporate the 
five core reading skills, and (4) identifying language problems in different contexts and 
proposing strategies to develop literacy skills in the mother tongue. For more details on the 
implementation of the teacher training and coaching component in the evaluation schools, such 
3 The school year in Peru begins in March and ends in December.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
5 
as frequency of activities, please consult the Leer Juntos, Aprender Juntos midline report (Lugo￾Gil et al. 2017a). 
Community action component. The community action component of the program, also 
based on the Literacy Boost model, aimed to strengthen parental and community involvement in 
building children’s reading abilities and increasing their opportunities to practice reading outside 
of school. (For more details, see the Literacy Boost community action toolkit [Save the Children 
2012b].) This component drew on the many ways that parents, older siblings, and community 
members can support children in becoming readers. Community activities were designed to be 
enjoyable, to engage students, and to empower adults even if they themselves are not fully 
literate. Parents and community members who do not read can engage children in oral language 
through storytelling and still contribute to children’s listening skills, oral comprehension, and 
vocabulary development. This component was delivered through entertaining group activities, 
such as reading camps, reading buddy sessions (peer assistance), reading festivals, and reading 
contests led by community volunteers. Specifically, the community action component included 
the following main activities (Save the Children and Kallpa 2013, 2014, 2015): 
1. Creating printed materials in the children’s mother tongue, Spanish, or both, to build 
portable libraries known as book banks. In Peru, the program team selected, collected, 
adapted, and produced reading materials covering different literary genres (such as legends, 
poems, and stories) in Spanish and Quechua. 
2. Promoting the use of book banks among community members. Each community had at 
least one book bank. 
3. Conducting reading activities in the community such as story hours, reading camps, 
and reading festivals or fairs. These activities were implemented during the school year 
(March through December) so that invitations for children and their families to participate in 
the activities could be distributed in the classrooms, and community volunteers could have 
time off during the school vacation period. The community activities were as follows: 
a. Story hour. This activity, for the most part, was conducted in parallel with the 
distribution of books through community book banks. Story hour activities took place 
about once a week. Children could attend the activity with their parents and siblings. 
During story hour in Peru, community volunteers, parents, and other community 
members shared stories about the community or acted out local legends and myths. 
b. Reading camps. In this activity, teachers read stories to children. This structured 
activity took place outside of school hours about once a week in each school community 
in Group A. The purpose of the reading camps was to stimulate the development of all 
five core reading skills (alphabet knowledge, phonological awareness, vocabulary, 
fluency, and comprehension) in the mother tongue or in Spanish, and incorporate 
storytelling and reading games to engage children. The reading camps also served as an 
opportunity for program staff to engage local youth to participate in the community 
volunteering activities. 
c. Reading buddies (peer assistance). In this activity, older children (in grades 4 to 6) 
read together with the younger children (grades 1 to 3) to help them improve their 
reading skills. This activity took place mostly within schools, to maximize the number of 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
6 
children participating in the activity and to facilitate pairing younger children with older 
children, one or two times per week in each Group A community. 
d. Reading festivals. Children participated in games and activities in which they could 
practice, develop, and strengthen their reading skills. In Peru, three to six reading 
festivals took place each year of program implementation. The reading festivals involved 
all the education stakeholders in the community: students, teachers, principals, parents, 
community volunteers (reading promoters), and local education authorities. 
e. School–community accountability meetings. Program staff met with community 
leaders, government officials from the Apurímac region, and authorities from the 
provincial education units (known as UGELs, or Unidades de Gestión Educativa Local), 
which are similar to school districts, to discuss funding sources of reading activities, 
progress with the reading activities implemented in schools and the communities, 
resources used and their purpose, and future sustainability of the activities. In Peru, these 
meetings took place in each community during the second half of 2015, the last year of 
intervention implementation.
4. Coordinating peer assistance through reading buddies. The purpose of this training 
was to enable volunteers to implement the community action activities (such as 
promoting the use of book banks, reading buddies, and story hour) with children in their 
communities. During the training, volunteers learned how to lead the community reading 
activities and developed a weekly schedule of activities for their community to guide the 
rollout of implementation. 
5. Conducting school–community accountability meetings and reading awareness 
workshops with parents and community members. The purpose of these workshops 
was to promote awareness and understanding among parents and community members 
about how the community reading activities help children learn to read. The workshops 
also aimed to raise awareness about how important it is for children to start learning to 
read in their mother tongue and to have a dedicated place for reading (‘reading corners’) 
at home. 
2. Prevailing practice in Peru
In the region of Apurímac, where the evaluation was conducted, the prevailing practice was 
the early-grade reading instruction approach promoted by the Ministry of Education, known as 
“communicative-textual”, and. was also potentially influenced by other national policies and 
programs to promote literacy among the Quechua-speaking population. In addition, schools were 
subject to departmental and local decisions about education priorities and practices. 
Peruvian authorities have a history of designing reading instruction with attention to 
students’ cultural heritage. Since the 1970s, the government of Peru has supported intercultural 
and bilingual education (IBE) service provision in selected schools. The Ministry of Education, 
through its General Office of Intercultural, Bilingual, and Rural Education (known by its 
acronym in Spanish, DIGEIBIR), recommends that teachers in schools offering IBE services 
should know and value the cultural heritage of their students. Teachers in IBE schools should 
also demonstrate oral and written language skills in the prevalent native language of the region 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
7 
as well as in Spanish, must have received training in IBE, and must implement and practice IBE 
approaches in the classroom (Ministerio de Educación 2013). In addition, IBE schools should 
have a curriculum and a pedagogical approach that considers knowledge, skills, history, and 
cultural heritage of the students as well as knowledge of other cultures, and should provide 
educational materials for different content areas both in the prevalent indigenous language of the 
region and in Spanish. 
The Ministry of Education in Peru advocates a communicative and textual approach aimed 
at educating students to understand, speak, read, and write in different contexts within and 
outside their communities (Ministerio de Educación 2008). DIGEIBIR recommends that IBE 
programs initiate instruction of literacy skills in the mother tongue but also provide oral 
instruction in Spanish. According to DIGEIBIR, children should first acquire literacy skills in 
their mother tongue and then gradually transition to learning those skills in Spanish. Therefore, 
in Peru, there is no specific grade level at which instruction should transition fully from mother 
tongue to Spanish; rather, the transition is expected to be gradual, and the extent of instruction in 
Spanish should increase as children progress to higher grades. The expectation for bilingual 
programs is that in grades 4, 5, and 6, the amount of instruction provided in the mother tongue 
and in Spanish in bilingual education programs is about the same (Ministerio de Educación 
2013). 
Two key initiatives from the Peruvian Ministry of Education to strengthen children’s 
literacy skills are the National Rural Education Network Program (REDES, or Redes Educativas 
Rurales) and the Strategic Learning Outcomes Program (PELA, or Programa Estratégico Logros 
de Aprendizaje). REDES is a model of quality rural education that offers teacher support on 
classroom practices, teacher training, access to information technologies, curricular development 
for multigrade and single-teacher schools, educational materials in the mother tongue and in 
Spanish, and improvement in school infrastructure (Ministerio de Educación 2012). The PELA 
approach to educational policy change uses a performance-based budget strategy aimed at 
strengthening educational management and administration, teacher training, school 
infrastructure, and access to quality education (Vargas Dávila 2009). Participation in these 
programs is voluntary. Among the schools participating in the evaluation, participation in 
REDES and PELA was similar across the three groups of schools implementing each of the 
approaches that we are comparing in this evaluation: about 19 to 29 percent of the schools in 
each treatment group participated in REDES and PELA (see Table III.1 for more details on 
evaluation schools’ participation in these programs). 
In 2015, the last year of implementation of Leer Juntos, Aprender Juntos, the Ministry of 
Education began two new initiatives: Soporte Pedagógico (Pedagogical Support) targeting urban 
areas and Soporte Pedagógico Intercultural (SPI; Intercultural Pedagogical Support) aiming to 
serve bilingual schools located primarily in rural areas. The goal of these initiatives is to improve 
student learning in elementary schools in four ways: (1) providing training workshops, peer-to￾peer study groups for teachers, and virtual educational support for teachers and principals; (2) 
implementing learning sessions designed to develop students’ communication, math, and social 
studies skills that require additional support; (3) executing efficient mechanisms for timely 
allocation and use of appropriate teaching materials and educational resources; and (4) building 
capacity of principals, teachers, and parents to ensure adequate learning conditions for students. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
8 
In addition, these initiatives employ teachers who work after regular school hours (docentes 
fortaleza) to provide remedial support to students who are falling behind (Consejo Nacional de 
Educación 2016). Although the two initiatives aim to improve the quality of education and 
student achievement, SPI emphasizes the implementation of the Ministry of Education’s 
bilingual intercultural education model with strong bilingual literacy acquisition (in mother 
tongue and in Spanish) and cultural diversity components. 
C. Overview of Leer Juntos, Aprender Juntos evaluation
As noted in the introduction, the Leer Juntos, Aprender Juntos evaluation was conducted to 
contribute to the evidence base on what works in LAC to improve early grade reading. Knowing 
the most cost-effective approaches to improving reading skill acquisition can guide policy 
makers and educators in their efforts to promote child literacy. Specifically, this evaluation 
sought to understand whether the Leer Juntos, Aprender Juntos program was more cost effective 
than prevailing practice, and whether the community action component of the program was a 
cost effective approach to improve learning. It did so by comparing the impacts on learners and 
teachers of the full program, including the in-school and community components, the in-school 
program only, and the usual education services provided to schools (which we refer to as 
prevailing practice). In order to have a better understanding of the program in different contexts, 
we conducted separate but parallel impact evaluations of Leer Juntos, Aprender Juntos for Peru 
and Guatemala. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
9 
II. EVALUATION DESIGN AND DATA
As described in Chapter I, the goal of this evaluation is to determine the relative impact and 
cost effectiveness of the in-school and community action components of the Leer Juntos, 
Aprender Juntos program. This chapter details the evaluation design, the program’s logic 
framework, the types of data collected, the analytic methods, and the evaluation sample. 
A. Evaluation questions
The evaluation sought to answer the following two primary questions about program 
impacts:
● What is the impact of the teacher training and support component of Leer Juntos, Aprender 
Juntos on early-grade reading and other outcomes relative to prevailing practice? This 
question contrasts schools in Group B to those in Group C. 
● What is the impact of the community action component of Leer Juntos, Aprender Juntos on 
early-grade reading and other outcomes relative to an intervention that does not have the 
community action component? This question contrasts schools in Group A to those in 
Group B. 
In addition to these two primary questions, the evaluation addresses a question about 
implementation: 
● Were the program components implemented as intended? This question relates to whether 
each program component’s services were offered as originally intended, whether 
participants took part, and whether these program components had intermediate impacts on 
teaching and the availability and use of reading materials. It also considers any barriers or 
challenges to effective implementation.
B. Evaluation design
In Peru, we recruited 145 schools.4
 Within each school, we followed a group of children 
from 1st grade through the end of 3rd grade. The evaluation was implemented as a randomized 
controlled trial, in which the evaluation team randomly assigned schools to the three treatment 
groups (we explain the assignment process to treatment groups in detail in Chapter II): 
1. Group A: These schools implemented the full intervention (the teacher training and coaching 
component and the community action component). We refer to this group as Leer Juntos, 
Aprender Juntos, or Leer Juntos for ease of presentation. 
2. Group B: These schools implemented only the in-school component of the intervention. We 
refer to this group as Leer Juntos-school only). Including this group in the evaluation enabled 
us to isolate the effects of the in-school component from the effects of the community action 
component. 
4 In Guatemala, we recruited 150 schools.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
10
3. Group C: The schools assigned to this group did not implement any of the components 
provided by Leer Juntos, Aprender Juntos, but instead implemented the prevailing reading 
instruction approach and supports in the early-grades in each country, in this case, Peru. We 
refer to this group as the prevailing practice group. 
To provide rigorous evidence on the relative impacts of the in-school and community action 
components of the program to improve early-grade reading, we used a randomized controlled 
trial design in which we randomly assigned schools to three intervention groups (Figure II.1). 
We conducted random assignment at the school level because the intervention was implemented 
at this level. Aligned with program rollout, the evaluation was carried out in two phases. We 
randomly assigned 74 eligible schools to intervention groups and collected baseline data in the 
first half of the evaluation schools in 2013 (Phase I), and then assigned 71 schools in 2014 
(Phase II) 5
. After random assignment, the evaluation team randomly selected one 1st grade 
classroom, one teacher, and 10 1st grade students from each school to serve as the analysis 
sample. We sampled this way to minimize data collection cost and burden while still obtaining 
data that were representative of the schools and students in the evaluation. We conducted two 
follow-up data collections after the baseline: one at the end of the second year of implementation 
(midline, in 2014 for Phase I and 2015 for Phase II) and another at the end of the third year of 
implementation (endline, in 2015 for Phase I and 2016 for Phase II). Implementation of the 
program started in May 2013 for Phase I schools and March 2014 for Phase II schools. For a 
more detailed timeline of the data collection and program implementation activities, see Figure 
A.1 in Appendix A. 
We present results for students pooled across the two phases because we did not find any 
significant pattern of differences (see Table G.3 in Appendix G) between phases. This is the 
same approach we have taken in earlier reports (Lugo-Gil et al. 2016b; Lugo-Gil et al. 2017a). 
5 In Guatemala, the evaluation team assigned a similar number of schools in the same manner: 75 schools in Phase I 
and another 75 in Phase II.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
11
Figure II.1. Leer Juntos, Aprender Juntos evaluation design
Source: Leer Juntos, Aprender Juntos evaluation plan (Glazerman et al. 2013)
The primary outcomes of interest for the impact analysis presented in this report are early￾grade reading skills, such as fluency and comprehension, but we also examined other outcomes, 
such as improved classroom practices and culture of reading at home and in schools. We answer 
the first question of the evaluation, about the impact of the teacher training and support 
component of Leer Juntos, Aprender Juntos on early-grade reading and other outcomes relative 
to prevailing practice, by comparing outcomes for schools in Group B with those in Group C. 
And by comparing the outcomes for schools in Group A with those in Group B, we answer the 
second evaluation question, which is to assess the community action component of Leer Juntos, 
Aprender Juntos on early-grade reading and other outcomes relative to an intervention that does 
not have the community action component. To understand the implementation context and 
address the program implementation question, we adopted a mixed-methods approach 
integrating survey data, interviews, focus groups, and a qualitative review of program 
documents. These data came from implementers, teachers, and parents of students. 
C. Leer Juntos, Aprender Juntos logic framework
Figure II.2 illustrates the logic framework for the program. As described in Chapter I, Leer 
Juntos, Aprender Juntos has two components: (1) a teacher training and coaching component 
that was implemented within schools (Groups A and B), and (2) a community action component 
that was implemented outside of school hours with children and their families within the 
community (Group A). We expect the program to have effects in three key areas: (1) the 
instructional practices that teachers implement in their classrooms, (2) children’s access to and 
use of instructional materials in their mother tongue and in Spanish, and (3) participation in 
community reading activities after school hours, among other intermediate outcomes. The 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
12
expected final outcomes for the program are children’s improved foundational and reading 
comprehension skills. 
Figure II.2. Leer Juntos, Aprender Juntos logic framework
Source: Leer Juntos, Aprender Juntos Performance Management Plan (Save the Children 2013b).
D. Types of data collected for the evaluation
We partnered with the Grupo de Análisis para el Desarrollo (GRADE) to collect data for 
the evaluation. We collected data from schools, classrooms, teachers, and students, following 
what the program’s logic framework indicates as the expected intermediate and final outcomes 
of Leer Juntos, Aprender Juntos (Figure II.2). The survey instruments, data documentation, and 
data files used for this evaluation are available online upon request as restricted use data files at 
USAID’s Development Data Library website (https://data.usaid.gov/). We describe the 
evaluation’s data collection activities in more detail in this section of the report.
1. Data collection in schools and households
The evaluation team collected data at baseline and again at the midline and endline points of 
program implementation. Data collection for the baseline occurred when the children 
participating in the evaluation were in 1st grade. The midline (first follow-up) data collection 
took place during the second year of program implementation for children in both Phase I and II, 
when most of the children in the evaluation would have progressed to 2nd grade and have had 
about one year and a half of potential exposure to the intervention. The endline (final follow-up) 
data collection took place during the third year of program implementation, when most of the 
children in the evaluation would have progressed to 3rd grade and have had about two to two and 
a half years of potential exposure to the intervention. See Figure A.1 in Appendix A for a 
detailed timeline of Leer Juntos, Aprender Juntos activities in Peru. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
13
At the baseline, midline, and endline data collections, we observed schools and classrooms 
and interviewed teachers. The school observation used a checklist that the evaluation team could 
complete based on an environmental scan of the school premises. In the classroom observation, 
the evaluation team observed and recorded information on teachers’ instructional practices, 
teachers’ and students’ language use in the classroom and time on task, and other school and 
classroom characteristics during a one-hour class period. The evaluation team also administered 
an in-person survey of teachers about their instructional practices; education and experience; 
participation in professional development activities; occupational needs; career expectations; and 
background characteristics, such as Quechua proficiency and income. For more information 
about the development, administration, and contents of the instruments for the school and 
classroom observations, and the teacher survey, refer to the Leer Juntos, Aprender Juntos
baseline report (Lugo-Gil et al. 2016b). 
In addition to the observations and teacher surveys administered in schools, the evaluation 
team administered individual assessments to children in the baseline and endline data collections. 
During the baseline data collection, when evaluation children were in the 1st grade, the 
evaluation team assessed children’s oral language proficiency and emergent literacy skills (letter 
identification, emergent writing, emergent reading, phonemic awareness, pseudo-word decoding, 
and passage comprehension skills) in Spanish and Quechua. The Leer Juntos, Aprender Juntos
baseline report (Lugo-Gil et al. 2016b) presents a detailed description of the process to develop 
those assessments and of their contents, and reports on the findings from the examination of the 
baseline data. In the endline data collection, the individual assessments administered to children 
focused on reading skills in Spanish. This assessment was administered when most of the 
children in the evaluation attended the 3rd grade. The endline reading skills assessment was 
adapted for Peru following the guidelines of the Early Grade Reading Assessment (EGRA) 
Toolkit (RTI International 2015). The endline assessments were conducted in Spanish because 
Spanish is the language of instruction in the 3rd grade classrooms in the evaluation (see Table 
E.10 in Appendix E and the baseline report [Lugo-Gil et al. 2016a])).
In the midline data collection, we observed the evaluation schools’ classrooms and 
administered a teacher survey, but we did not assess children’s reading skills. Instead, the 
evaluation team conducted a household survey. This survey was administered in person to the 
main caregivers (usually the mother or father) of the children in the evaluation at their homes. 
The purpose of this survey was to learn about the household composition, family socioeconomic 
status, household assets, children’s schooling background and routines at home, and children’s 
and families’ participation in reading activities offered in their communities. Findings from the 
midline data relevant to understanding intermediate outcomes are also presented in Chapter IV of 
this report. 
2. Response rates
Table II.1 shows that the response rates to each round of data collection were high, mostly 
more than 95 percent for most data collections. Of the 145 schools that were randomly assigned 
to treatment groups, one school assigned to Group B closed and one school assigned the same 
group refused to participate in the baseline and midline data collections. The latter school 
allowed data collectors to enter the school for the endline data collection. Therefore, the impact 
analysis presented in this report includes 144 schools (49 in Group A, 47 in Group B, and 48 in 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
14
Group C). The number of children in the sample who we randomized into treatment groups is 
1,074, and the number who completed a reading skills assessment in the 3rd grade is 1,022. The 
low levels of attrition across all three groups give us confidence that the evaluation sample 
continues to be balanced and representative across all treatment groups at endline and that our 
experimental design has been maintained.
Table II.1. Response rates to data collections in the evaluation
Data collection round
Leer
Juntos
(A)
Leer Juntos￾school only
(B)
Prevailing 
practice
(C) All
Baseline data collection (2013 in Phase I and 2014 in Phase II)
School infrastructure observation, classroom observation, and 
teacher survey
100 95.8 100 98.6
Emergent literacy skills assessment 96.3 96.0 96.7 96.4
Midline data collection (2014 in Phase I and 2015 in Phase II)
School infrastructure observation, classroom observation, and 
teacher survey
100 95.8 100 98.6
Household survey 96.3 95.5 97.9 96.6
Endline data collection (2015 in Phase I and 2016 in Phase II)
School infrastructure observation, classroom observation, and 
teacher survey
100 97.9 100 99.3
Reading skills assessment 94.2 95.5 95.9 95.2
Number of children who were randomized into treatment 
groups
380 354 340 1,074
Number of children who completed endline assessments 358 338 326 1,022
Source: Leer Juntos, Aprender Juntos School Infrastructure Observation, Classroom Observation, Teacher Survey, 
Household Survey, Emergent Literacy Skills Assessment, and Reading Skills Assessment—Baseline 2013 
and 2014, Midline 2014 and 2015, and Final Follow-ups 2015 and 2016.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
15
3. Qualitative data collection 
The evaluation team also collected qualitative data from stakeholders to understand how the 
school and community components were implemented, the facilitators and barriers to 
implementation, and teachers’ and parents’ perceptions of the Leer Juntos, Aprender Juntos
program. We partnered with GRADE to collect data for the evaluation. Qualitative data 
collection took place over a two-week period in May 2015— at the end of the second year of 
program implementation for Phase I and of the first year for Phase II (see Appendix A, Figure 
A.1). The data are used to understand how program services were delivered, how services were 
received by stakeholders, and what services were received in each experimental group. They also 
illustrate some of the challenges to implementation. GRADE, in collaboration with the UGEL, 
Save the Children and Mathematica, recruited a purposive sample of parents of students in the 
treatment groups (Groups A and B), 1st through 3rd grade teachers in treatment and control 
schools, community volunteers, and local and implementation specialists from Save the Children 
to participate in focus groups and interviews. Using a convenience sampling approach, the team 
identified and selected participants that were easily accessible and could provide rich information 
relevant to the goals of the evaluation. Participant inclusion criteria were flexible, and took into 
account potential constraints and barriers to participation in interviews and focus groups like 
distance, access to transportation, and work schedules. Participant inclusion criteria were: 
• For parents, to have a child enrolled in first, second or third grade, in a school assigned to 
Group A or B (phase I and phase II schools included) 
• For teachers, to teach first, second and third grade, in a school assigned to Group A, B, or C 
(phase I and phase II schools included) 
• For volunteers, to be and active volunteer for the community action component 
GRADE staff carried out interviews and focus groups with directors in the provincial 
education units (UGELs) and education specialists, Save the Children staff, Kallpa staff, 
community volunteers, parents, and teachers from Phase I and Phase II schools located in the 
three provincial education units (Andahuaylas, Chincheros, and Huancarama) of the Apurímac 
region. All interviews and focus groups were audio recorded and transcribed verbatim. Field 
staff also prepared interpretative summaries synthesizing key themes that emerged during the 
focus groups, which were later used to triangulate findings from the analysis of verbatim 
transcripts.
The team faced two main challenges with the recruitment of participants for qualitative 
data collection in Peru. First, a teacher strike during the time when data collection activities were 
scheduled made it necessary to shift the timing of data collection, and the teacher strike also 
made it more difficult to obtain consent from control schools which were already somewhat 
reluctant to participate in the proposed interviews because they were not receiving program 
services. Second, the geographic dispersion of the target communities and transportation 
constraints for parents required some logistical accommodations. Instead of asking parents to 
travel to a central location for the focus groups, the field staff team travelled to a site in the 
community, closer to the families’ homes, to conduct the focus groups following a small group 
interview format. (See Appendix B, Table B.1 for a summary of respondent types, methods, 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
16
topics of inquiry, and location for all qualitative data collection activities, and Appendix L for 
the qualitative instruments used in these data collection activities.) 
E. Analytic approach
We examined differences between treatment groups in measures at the school level 
(characteristics of schools, teachers, and classrooms) and at the child level (emergent literacy and 
reading skills and home literacy environment). To assess the differences between treatment 
groups in those measures, we conducted regression analyses that accounted for the design of the 
random assignment of schools and the data cohort (Phases I and II). The design used stratified 
random assignment, in which we first grouped schools with similar characteristics (within phase) 
into strata and randomly assigned to the three groups within these strata. This approach increases 
the likelihood that certain observable and unobservable characteristics are evenly balanced 
across the three treatment groups. For more details on how we conducted the stratified 
randomization of schools and selected the sample, refer to the Leer Juntos, Aprender Juntos
baseline report (Lugo-Gil et al. 2016b). 
With those regressions, we computed regression-adjusted means of the relevant measures 
for each treatment group (A, B, and C) and tested whether the differences in those adjusted 
means between Groups A and B, and between Groups B and C, are statistically different from 
zero. We also computed effect sizes, which are the differences in adjusted means between two 
treatment groups expressed in standard deviation units. We used the p-values from the tests of 
differences in group means to assess statistical significance of the differences in means. We 
defined the difference in means between treatment groups as statistically significant whenever 
the p-value of the test was lower than 0.05. In determining the statistical significance of mean 
differences between treatment groups in measures at the child level, we accounted for the 
clustering of children within schools. 
For analyses of children’s reading skills measures in the 3rd grade, we defined the analytic 
sample as the sample of children who completed a reading skills assessment in the final follow￾up (endline) data collection (also known as “respondents”). Therefore, our strategy to estimate 
program impacts follows an intent-to-treat approach.6
 
Our regression analyses of children’s reading skills measures in the 3rd grade also control 
for several variables at the child, household, and school levels measured at the baseline. We 
included these variables in our analyses to account for any initial, observed differences between 
treatment groups and to increase the precision of our impact estimates. These control variables 
include (1) child characteristics such as age, sex, results from the baseline emergent literacy 
assessments, and language spoken at home at the time of baseline data collection; (2) household 
characteristics such as mother’s education level and household income and assets; and (3) school 
characteristics at the time of baseline data collection such as participation in other programs, 
number of physical and health hazards observed in the school, and school infrastructure 
resources. Table C.1 in Appendix C presents the complete list of control variables included in the 
6 With the intent-to-treat approach, all the evaluation participants in the schools that were randomly assigned to the 
three treatment groups (Groups A, B, and C) are included in the impact analysis and are analyzed in the groups to 
which they were randomized.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
17
analyses of children’s reading skills measures. We imputed missing values in the control 
variables using a regression-adjusted average of the nonmissing values for the variable. We 
describe the process to impute missing values in control variables in more detail in Appendix C. 
The approach for the reduction and analysis of qualitative data sought to meet the goals of 
the implementation research questions efficiently, and in keeping with the resources available for 
the qualitative component of this evaluation. Mathematica staff reviewed all transcripts, field 
notes, and interpretative summaries submitted by GRADE’s field staff team for quality 
assurance. Then, we conducted a thematic analysis of the transcripts and other sources of 
qualitative data in three steps. First, we parsed meaningful segments of transcripts (or field notes) 
linked to the key implementation research questions. Second, we extracted memos of key 
findings and illustrative quotes, and organized them in synthesis tables. Finally, we triangulated 
finding from these tables with field notes, and interpretative summaries submitted by GRADE’s 
field staff team.
F. Characteristics of the sample at endline
This section provides a snapshot of the characteristics of the schools, teachers, and students 
in the evaluation sample during the third year of program implementation, when most of the 
students in the sample attended the 3rd grade. 
1. Schools 
The schools in the evaluation are located in the Apurímac region, in the Andes of southern￾central Peru. Apurímac was selected because its population is linguistically diverse: about 70 
percent of residents report that their mother tongue is Quechua (Instituto Nacional de Estadística 
e Informática [INEI] 2007),7
 which provided the opportunity to examine how the Leer Juntos, 
Aprender Juntos intervention works for that population. The evaluation schools were selected 
from three provincial education units within Apurímac’s two western provinces, Andahuaylas 
and Chincheros. Figure II.3 presents a map of Peru that highlights the location of the Apurímac 
region and the provincial education units where the evaluation schools are located. 
7 The other 29 percent of the population in the Apurímac region report their mother tongue is Spanish, and about 1 
percent report it is another language, such as Aymara (INEI 2007).
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
18
Figure II.3. The Apurímac region in Peru
As shown in Table II.2, most schools in the sample are located in the Andahuaylas school 
district, followed by the Chincheros and Huancarama districts. The schools in all three 
intervention groups average approximately six primary classrooms and about nine 3rd grade 
students. About 50 to 63 percent of the schools in each group have classrooms with more than 
one grade (multigrade or single-teacher schools8
), meaning that the children in 3rd grade in these 
schools are taught in classrooms together with children from other grades. The classrooms in 
each group have, on average, an equal number of female and male students. 
The average school characteristics were similar for each treatment group (Table II.2). Fifty￾eight of the 59 differences between treatment groups in the school infrastructure characteristics 
that we examined were not statistically significant (see Appendix D, Tables D.1 through D.3). 
Nevertheless, our approach for estimating program impacts controls for baseline characteristics 
of schools, such as available utilities and facilities (see Appendix C, Table C.1 for the complete
list of variables for which we controlled in calculating program impacts. For details on the 
infrastructure characteristics of evaluation schools, refer to the Leer Juntos, Aprender Juntos
baseline report (Lugo-Gil et al. 2016b)). 
8 In schools with multigrade classrooms, several grades are taught in the same classroom. In single-teacher schools, 
one teacher provides instruction to all the primary grades offered in the school, usually in the same classroom. In 
schools with single-grade classrooms, only one grade is taught in each classroom.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
19
Table II.2. Characteristics of evaluation schools in the final follow-up year
Characteristic
Leer
Juntos
(A)
Leer Juntos￾school only
(B)
Prevailing 
practice
(C) (A–B) (B–C)
Percentage of evaluation schools in school district (UGEL):
Andahuaylas 56.5 52.3 54.5 4.2
(0.161)
-2.2
(0.477)
Chincheros 27.3 31.4 30.4 -4.2
(0.268)
1.1
(0.778)
Huancarama 16.2 16.2 15.2 -0.0
(0.994)
1.1
(0.644)
Percentage of schools with multigrade 
classrooms
50.4 57.6 63.0 -7.2
(0.255)
-5.4
(0.401)
Number of primary grade classrooms in use 
(average)
5.9 5.7 5.6 0.2
(0.613)
0.1
(0.839)
Number of students enrolled in observed 3rd grade classrooms (average)
Total 9.2 9.1 9.2 0.1
(0.884)
-0.1
(0.898)
Female 4.5 4.6 4.6 -0.1
(0.811)
0.0
(0.938)
Number of schools 49 47 48
Source: Leer Juntos, Aprender Juntos School Infrastructure Observation Form and Classroom Observation Form—
Final Follow-ups 2015 and 2016.
Note: Regression-adjusted means. P-values in parentheses. None of the differences between treatment groups 
presented in this table is statistically significant at the 0.05 level.
2. Teachers
The teachers surveyed in the 3rd grade follow-up have similar background characteristics 
across the three treatment groups. In particular, we did not find any statistically significant 
differences between treatment groups in terms of teachers’ sex, age, and highest level of 
education (see Table II.3). 
We also did not find any statistically significant differences in the composition of the sample 
of teachers in the evaluation in terms of years of teaching experience and tenure in the early 
grades. As shown in Table II.4, there were no statistically significant differences across treatment 
groups in the teachers’ number of years of experience—overall, at the current school, or teaching 
3rd grade. About 30 percent of the 3rd grade teachers in each treatment group taught the same 
evaluation children as they moved through 1st, 2nd, and 3rd grades. Another 32 to 47 percent of 
the 3rd grade teachers in the evaluation have taught at the current school for less than two years. 
This is equivalent to five to seven fewer teachers in Group A than in Groups B or C taught at the 
current school for less than two years. However, these differences between treatment groups are 
not statistically significant.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
20
Table II.3. Background characteristics of the teachers of the evaluation 
students in the final follow-up year
Characteristic
Leer
Juntos
(A)
Leer Juntos￾school only
(B)
Prevailing 
practice
(C) (A–B) (B–C)
Female (percentage) 48.6 48.9 60.7 -0.3
(0.974)
-11.8
(0.252)
Age (average) 47.9 45.1 43.9 2.8
(0.149)
1.2
(0.535)
Highest level of education (percentage)
Incomplete associate, technical, or college 
degree
3.8 2.2 2.2 1.6
(0.658)
0.0
(1.000)
Associate or technical degree 43.5 47.8 39.2 -4.3
(0.688)
8.6
(0.427)
College degree 48.6 37.5 46.1 11.1
(0.260)
-8.6
(0.386)
Graduate degree (master’s or doctorate) 4.2 12.5 12.5 -8.3
(0.180)
0.0
(1.000)
Number of teachers 49 47 48
Source: Leer Juntos, Aprender Juntos Teacher Survey—Final Follow-ups 2015 and 2016.
Note: Regression-adjusted means. P-values in parentheses. None of the differences between treatment groups 
presented in this table is statistically significant at the 0.05 level.
Table II.4. Experience and tenure characteristics of the teachers of the 
evaluation students in the final follow-up year
Characteristic
Leer
Juntos
(A)
Leer Juntos￾school only
(B)
Prevailing 
practice
(C) (A–B) (B–C)
Years of teaching experience (average) 21.6 17.7 16.5 3.9
(0.063)
1.2
(0.575)
Years of teaching at current school (average) 9.6 6.9 6.5 2.7
(0.097)
0.4
(0.814)
Years teaching 3rd grade (average) 4.9 5.0 4.4 -0.1
(0.923)
0.6
(0.570)
Percentage of teachers who have taught 
evaluation children for three years (when they 
were in the 1st, 2nd, and 3rd grades)
30.0 29.0 27.9 1.0
(0.911)
1.1
(0.907)
Percentage of the teachers in 3rd grade in the 
evaluation who have less than two years of 
experience at current school
32.1 43.5 46.7 -11.4
(0.236)
-3.2
(0.740)
Number of teachers 49 47 48
Source: Leer Juntos, Aprender Juntos Teacher Survey—Final Follow-ups 2015 and 2016.
Note: Regression-adjusted means. P-values in parentheses. None of the differences between treatment groups 
presented in this table is statistically significant at the 0.05 level.
Finally, we found that although nearly all teachers in all treatment groups reported that they 
understand Quechua, fewer than 40 percent of the teachers in the sample reported that they speak 
very good Quechua, and none reported an advanced level of written Quechua (see Table II.5). 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
21
However, we did not find any statistically significant difference across treatment groups in 
teachers’ reported language skills. 
Table II.5. Teacher-reported language skills in the final follow-up year 
(percentages of teachers) 
Skills
Leer
Juntos
(A)
Leer Juntos￾school only
(B)
Prevailing 
practice
(C) (A–B) (B–C)
Teacher’s mother tongue
Spanish only 29.0 24.9 20.6 4.1
(0.643)
4.3
(0.630)
Quechua only 27.3 36.2 29.8 -9.0
(0.339)
6.5
(0.495)
Both Spanish and Quechua 43.7 38.9 49.6 4.9
(0.645)
-10.8
(0.315)
Teacher is capable of understanding 
Quechua
100.0 97.9 97.9 2.1
(0.396)
1.0
(1.000)
Teacher’s proficiency level in spoken Quechua. 
Very poor to fair 26.5 20.3 21.3 6.3
(0.477)
-1.1
(0.904)
Good 46.8 41.9 54.8 4.9
(0.635)
-12.9
(0.218)
Very good 26.6 37.8 23.8 -11.2
(0.244)
14.0
(0.150)
Teacher’s proficiency level in written Quechua. 
Very poor to fair 38.6 33.3 40.8 5.3
(0.592)
-7.5
(0.454)
Good 61.4 66.7 59.2 -5.3
(0.592)
7.5
(0.454)
Number of teachers 49 47 48
Source: Leer Juntos, Aprender Juntos Teacher Survey—Final Follow-ups 2015 and 2016.
Note: Regression-adjusted means. P-values in parentheses. None of the differences between treatment groups 
presented in this table is statistically significant at the 0.05 level.
In summary, we did not find noteworthy differences across treatment groups in the terms of 
teacher characteristics.
3. Students
As shown in Table II.6, there is no significant difference in retention rates between 
treatment and prevailing practice schools. The majority of students in the evaluation (the students 
in the two treatment groups, Groups A and B, and the students in the prevailing practice group, 
Group C) had progressed to the 3rd grade by the time the evaluation team administered the 
endline (final follow-up) reading skills assessment.9
 Only about 10 to 12 percent of the students 
9 One of the students in the sample (in Group C; this result is not in Table II.6) was not enrolled in school at the time 
of the endline reading skills assessment.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
22
in each treatment group were retained in 2nd grade when assessed for the final follow-up. That 
rate is lower than the rate of the country’s overall grade repetition in elementary school of 16 
percent (Ikeda and Garcia 2014).
We also found that at the time of the endline reading skills assessment, the majority of the 
children in the sample attended the same school they attended at the time of the baseline data 
collection. Therefore, the majority of the children in the treatment groups had been enrolled in a 
school that implemented the in-school component of the intervention for about three years. We 
did not find any statistically significant differences across treatment groups in the grade or school 
that children in the evaluation sample attended at the time of the endline data collection.
Table II.6. Grade and school at the time of the final follow-up data collection 
(percentage of students)
Grade and school
Leer
Juntos
(A)
Leer Juntos￾school only
(B)
Prevailing 
practice
(C) (A–B) (B–C)
In 3rd grade 88.3 89.5 89.3 -1.2
(0.686)
0.2
(0.947)
Current school is the same as the school 
attended at the time of the evaluation’s 
baseline data collection
86.7 90.9 90.5 -4.2
(0.207)
0.4
(0.883)
Number of children 358 338 326
Source: Leer Juntos, Aprender Juntos Students’ Reading Skills Assessment—Final Follow-ups 2015 and 2016.
Note: Regression-adjusted means. P-values in parentheses. None of the differences between treatment groups 
presented in this table is statistically significant at the 0.05 level.
G. Baseline equivalence of the analysis sample
The random assignment process yielded similar treatment groups in most of the baseline 
characteristics examined. Because the evaluation is based on an experimental design (schools 
were randomly assigned to the treatment groups), we did not expect to find systematic 
differences across all the baseline characteristics of the treatment groups that we measured. 
However, differences in some characteristics could exist because there might be no way to assign 
144 schools to three treatment groups such that the groups are identical. Random assignment 
facilitates ensuring the three groups are, on average, similar and that any unusual characteristic 
has an equal chance of representation in any of the three groups. 
We examined differences across treatment groups in the characteristics of students in the 
analytic sample that are unlikely to be influenced by the program and therefore can indicate 
whether our evaluation design produced similar treatment groups. As shown in Table II.7, the 
treatment groups are similar in student baseline characteristics, with a few exceptions: children in 
schools in Group B were marginally older at baseline and had slightly lower scores in the 
baseline assessments of emergent reading, emergent writing, and pseudo-word reading skills 
than children in schools in Groups C. To account for these differences across treatment groups, 
we included the variables presented in Table II.7, among other student- and school-level 
variables, as controls in the regression models we used to calculate program effects.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
23
Table II.7. Baseline characteristics of children in the evaluation sample
Characteristic
Leer
Juntos
(A)
Leer Juntos￾school only
(B)
Prevailing 
practice
(C) (A–B) (B–C)
Average age at baseline (in months) 75.4 75.9 75.0 -0.5
(0.135)
-0.9*
(0.010)
Female (percentage) 49.7 51.7 51.3 -2.0
(0.503)
0.4
(0.874)
Baseline letter identification score (0 to 24) 14.6 14.2 14.3 0.4
(0.477)
-0.1
(0.926)
Baseline emergent reading score (0 to 9) 1.9 1.7 2.5 0.2
(0.631)
-0.8*
(0.010)
Baseline emergent writing score (0 to 22) 10.1 9.5 11.1 0.6
(0.348)
-1.6*
(0.014)
Baseline phonemic awareness score (0 to 10) 3.0 2.9 3.1 0.1
(0.775)
-0.2
(0.432)
Baseline number of pseudo-words read in one 
minute (0 to 50)
1.7 1.6 3.1 0.1
(0.832)
-1.5*
(0.007)
Baseline passage comprehension score (0 to 6) 2.0 2.1 2.2 -0.1
(0.719)
-0.1
(0.573)
Child assessed in Spanish only at baseline 89.1 86.6 88.2 2.5
(0.603)
-1.6
(0.744)
Child assessed in both Spanish and in Quechua 
at baseline
7.0 5.7 5.6 1.4
(0.551)
0.1
(0.952)
Child attended preschool or kindergarten 98.7 99.4 99.5 -0.7
(0.186)
-0.1
(0.913)
Child’s mother completed at least one year of 
formal schooling
87.9 92.0 98.8 -4.1
(0.367)
-6.8 
(0.139)
Child’s parents are able to read 73.6 73.2 76.9 0.4
(0.918)
-3.7
(0.373)
Language spoken at home is Quechua only 30.8 27.1 27.4 3.7
(0.414)
-0.3 
(0.946)
Number of childrena 358 338 326
Source: Leer Juntos, Aprender Juntos Students’ Emergent Literacy Skills Assessment—Baseline 2013 and 2016, 
and Household Survey—Midline 2014 and 2015.
Note: Regression-adjusted means. P-values in parentheses.
aThis is the number of children who completed an endline reading skills assessment and who have complete data on 
age and sex. The number of children in the analytic sample who completed a baseline assessment is 989 (349 in 
Group A, 324 in Group B, and 316 in Group C).
* Difference in group means is statistically significant at the 0.05 level.
24
This page has been left blank for double-sided copying.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
25
III. WAS LEER JUNTOS, APRENDER JUNTOS IMPLEMENTED AS INTENDED?
As described in Chapter I, the Ministry of Education in Peru had been implementing two 
programs to strengthen children’s literacy skills: Redes Educativas Rurales (REDES) and the 
Programa Estratégico Logros de Aprendizaje (PELA). In 2015,10 the Ministry of Education 
developed two additional initiatives that aim to improve student learning in elementary schools: 
Soporte Pedagógico (Pedagogical Support) and Soporte Pedagógico Intercultural. These 
initiatives aim to improve the quality of education and student achievement through activities 
and components similar to the activities carried out by Leer Juntos, Aprender Juntos. Therefore, 
the implementation of Leer Juntos, Aprender Juntos must be understood in a setting where 
several practices and programs aiming to improve early-grade literacy coexisted across the 
treatment groups. This chapter presents findings on how services were delivered, how services 
were received by participants, and what services were received in each evaluation group. It also 
presents findings on challenges to implementation. The implementation findings help to 
contextualize and explain impact findings. 
A. School participation in education interventions
The practices and programs that coexisted with Leer Juntos, Aprender Juntos were 
present in equal measure in all three experimental groups. This finding is based on reports by 
principals (Table III.1) and teachers (Appendix E). Less than a third of the schools in the 
treatment group (fewer than 16 schools each group) participated in education programs with 
similar goals to Leer Juntos, Aprender Juntos (REDES, PELA and, Soporte Pedagógico). The 
differences in participation between treatment groups were small and not statistically significant. 
School principals in both Groups A and B (73 and 69 percent of schools, respectively) 
reported higher participation in the Leer Juntos, Aprender Juntos program. This finding is 
based on reports by principals (Table III.1) and teachers (Appendix E). However, one quarter of 
the school leaders in the prevailing practice group reported participating in Leer Juntos, 
Aprender Juntos, whereas none of the teachers in those same prevailing practice schools reported 
receiving any component of Leer Juntos, Aprender Juntos. We believe this reflects differences in 
respondents’ knowledge about the programs that are being implemented in the schools. It is 
likely that the school leaders are less familiar than teachers about the program activities being 
rolled out in classrooms. Although school leaders may be more aware of the general activities 
occurring in the school, they may be less informed about the specific teacher training activities or 
the identity of the implementer, program name, or funding source for those activities. 
10 In 2015, when the Phase I schools were in the second year of Leer Juntos Aprender Juntos implementation and 
the Phase II schools were in their first year, Soporte Pedagógico was implemented in 15 regions of the country 
(3,180 schools), benefiting 52,563 teachers and 1,144.213 students (Chinen and Bonilla 2017). 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
26
Table III.1. School participation in education programs, reported by 
principals or other school administrators
. . 
Program.
Leer
Juntos
(A)
Leer Juntos
School-Only
(B)
Prevailing 
practice
(C)
Impact of 
community 
action
(A–B)
Impact of
in-school 
compone
nt
(B–C)
Percentage of schools participating in the following education-focused programs
Leer Juntos, Aprender Juntos 73.2 69.2 25.2 4.0
(0.674)
44.1*
(0.000)
National Rural Education Networks 
Program (REDES)
24.9 29.2 20.6 -4.3
(0.626)
8.6
(0.337)
National Strategic Learning Achievement 
Program (PELA)
23.8 26.0 18.5 -2.2
(0.731)
7.5
(0.247)
Soporte Pedagógico 26.9 20.5 27.0 6.4
(0.344)
-6.5
(0.342)
One Laptop per Child 63.5 73.6 63.9 -10.0
(0.288)
9.7
(0.312)
Education materials from the Ministry of 
Education
100.0 97.9 97.9 2.1
(0.396)
0.0
(1.000)
Other school programs 28.6 16.5 22.9 12.1
(0.155)
-6.5
(0.451)
Number of schools 49 47 48
Source: Leer Juntos, Aprender Juntos School Infrastructure Observation Form—Final Follow-ups 2015 and 2016.
Note: Regression-adjusted means. P-values in parentheses.
* Difference in group means is statistically significant at the 0.05 level.
B. Implementation challenges in Peru
During the first year of implementation, management and staffing challenges with 
Kallpa hindered the fidelity of implementation. While new to the education sector, Kallpa had 
been a strategic partner for Save the Children/Peru for many years before implementing Leer 
Juntos, Aprender Juntos. However, decisions from Kallpa’s staff in the field during the first year 
of implementation resulted in deviations from the Literacy Boost model, hiring of staff who did 
not meet specified requirements, and frequent disagreements with regional government staff and 
with personnel from intervention schools. As a course correction measure, one of the field 
coordinators was replaced for the second year of implementation. According to interviews with 
Save the Children Kallpa staff, that change in the field leadership team led to improved fidelity 
to the Literacy Boost model, more frequent and harmonious interactions with regional 
government representatives, and innovations to the interventions in response to contextual needs 
(for example, rotating home visits to parents to implement reading corners at home). 
The implementation team had to adapt the Literacy Boost methodology to the 
Peruvian context, taking into account the Ministry of Education’s reading curriculum 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
27
Rutas del Aprendizaje.
11 Save the Children/Peru and Kallpa worked together to introduce the 
program to regional educational authorities and to align the program’s approach with the 
regional reading curriculum. After their staff received training on the reading instruction 
methodology by Save the Children’s project director, Save the Children/Peru and Kallpa worked/ 
together to adapt the methodology to the regional sociocultural context and to integrate it with 
the national strategy for reading instruction. To align the program’s instructional approach with 
the national curriculum, program coordinators and specialists participated in the training 
workshops on the learning units of the national curriculum Rutas del Aprendizaje, offered by the 
Ministry of Education. Several teachers had also participated in training workshops for Rutas del 
Aprendizaje in their corresponding school district (UGEL); however, program specialists 
identified substantial gaps in teachers’ knowledge about how to implement Rutas del 
Aprendizaje in the classroom, particularly how to conduct the ‘learning projects.’12 
Qualified project staff could not serve out a full three years. Most of the hired specialists 
were active teachers or principals who were granted a leave of absence to participate in the
project. Because these leaves of absence were typically limited to one or two years, the program 
experienced substantial turnover of specialists for its second and third years. This generated a 
need for annual training cycles for program staff who delivered coaching sessions to Group A 
and Group B intervention schools. 
Cooperation and acceptance from the local school districts varied for the three districts 
where Leer Juntos, Aprender Juntos was implemented. The Chincheros UGEL was optimistic 
about the complementarity of the Leer Juntos, Aprender Juntos program with other education 
programs from the Ministry of Education. In contrast, the Huancarama UGEL was somewhat 
skeptical about the feasibility of integrating the Leer Juntos, Aprender Juntos pedagogical 
approach with the Ministry of Education’s approach to reading instruction. The resistance to the 
program in the Huancarama UGEL was in part due to changes in management staff and 
misunderstanding about the program’s approach. However, program staff worked closely with 
the UGEL’s leadership to show them that there was no conflict between Leer Juntos, Aprender 
Juntos and the recommended approach of the Ministry of Education. The Andahuaylas UGEL 
was open to the Leer Juntos, Aprender Juntos program, but it faced organizational challenges 
coordinating different education programs in its area of operation. Effective coordination of 
program activities with the UGEL was crucial because about one-fourth of intervention schools 
were also receiving services from other programs. Twenty-three percent of schools in Group A 
also received services from PELA, and 26 percent received Soporte Pedagógico. Similarly, 26 
percent of schools in Group B received services from PELA and 20 percent from Soporte 
Pedagógico. 
The geography of Andahuaylas and the limited access to public transportation to 
intervention schools imposed logistical challenges for the teacher coaches, trainers, and 
volunteer coordinators. Recruitment and transportation of Kallpa team members was difficult 
11 Ensuring teachers’ take-up of reading instruction strategies and the sustainability of the program was essential to 
Save the Children’s approach. To that end, it aligned the program framework with the national curriculum Rutas del 
Aprendizaje. 
12 Known as Proyectos de Aprendizaje.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
28
due to the geography of the Andahuaylas province with unpaved mountainous roads and long 
distances between schools. Longer travel times and the overnight stays required to visit some 
communities made coordination of schedules for coaching and community visits challenging. 
Additionally, there was limited telephone land-line and cell phone reception in the field. 
C. Implementation of the in-school component
According to teachers from intervention schools who participated in focus groups, 
Kallpa specialists established positive and supportive relationships with them. For the in￾school component, Kallpa specialists conducted training workshops and coaching sessions with 
teachers. Their approach to training and program coordination was cordial, respectful, and 
democratic. Teachers reported a good working relationship with Kallpa specialists and valued 
their innovative approach to training and coaching. Teachers highlighted marked differences 
between Kallpa specialists and their counterparts 
from other programs. Teachers described 
coaching visits (acompañamiento) from personnel 
from programs sponsored by the central 
government as supervisory, whereas they 
described visits from Kallpa specialists as 
motivating and supportive. 
The teacher who visits came as a friend, gained 
our trust; supported us ...This helps you perform 
better in the classroom.
—Andahuaylas teacher focus group participant
Teachers described the program’s methodology and training approach as practical 
and feasible to apply in the classroom, and highlighted advantages of the Leer Juntos, 
Aprender Juntos training compared to trainings imparted by Ministry of Education.
Teachers who participated in focus groups stated that they were able to apply the program’s 
teaching strategies with ease, not only in the target classrooms but in other classrooms and 
subject areas as well. They thought the program’s reading instruction strategies were clear and 
had well-defined guidelines for applying in the classroom. Teachers also shared that the 
program’s approach to reading instruction differed from regional or national approaches, yet was 
compatible. 
About one-third of teachers interviewed at midline in Group A schools and 42 percent of the 
teachers in Group B schools reported having participated in trainings offered by the Ministry of 
Education or other institutions that calendar year (2015 for Phase I and 2016 for Phase II). 
Teachers stated that, compared to the Ministry of Education’s teacher training programs, 
Kallpa’s teacher training was less theoretical and included more explicit demonstration of 
pedagogical strategies and sequential steps for using those strategies in classroom practice. 
Similarly, control group teachers pointed out that Ministry of Education teacher trainings tend to 
be too theoretical and lack concrete tools and strategies for classroom practice (52 percent of the 
teachers interviewed at midline in control schools 
had participated in trainings imparted by the 
Ministry of Education or other institutions). 
Teachers who participated in trainings offered by 
Kallpa specialists expressed that the demonstration 
and step-by-step layout of Leer Juntos, Aprender 
Juntos strategies for reading instruction helped them 
apply with their students what they had learned during Kallpa’s trainings. 
Before I taught very differently from how I 
teach now. Kallpa supported me on how I 
should teach children to write and read, and 
gave me concrete strategies.
— Chincheros teacher focus group
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
29
While the program calls for the implementation of reading instruction in mother 
tongue and in Spanish, incorporating mother tongue reading instruction was challenging 
for Kallpa specialists and for teachers in intervention schools. Teachers and school principals 
were resistant to mother tongue instruction. Some schools displayed explicit signs forbidding the 
use of Quechua. Program specialists noticed that, even if teachers were proficient in Quechua, 
they did not have mother tongue instruction skills and did not feel well prepared to use Quechua 
for instruction in the classroom (73 percent of the teachers interviewed at midline in Group A or 
B schools reported having good or very good 
oral language skills in Quechua). Program 
implementers worked around this challenge by 
explaining the benefits of mother tongue 
instruction with students who speak primarily 
Quechua at home, and by providing specific 
guidelines and materials to help teachers 
conduct classroom activities in Quechua. Parents also expressed mixed views about Quechua 
instruction for their children. Although some parents explicitly opposed it, others stated that their 
children should learn it and preserve their linguistic heritage (more than 80 percent of the parents 
interviewed at midline reported speaking Spanish and Quechua or only Quechua most of the 
time). Parents who participated in focus groups conveyed mixed views on Quechua instruction 
for their children and expressed the tension between their desire to preserve their cultural and 
linguistic heritage and the need for Spanish proficiency for opportunities in the labor market. 
[The Teacher Circles] were not in the model, but 
they are part of the coaching system which works 
very well given the context. It also provides a 
space for feedback; a space where teachers can 
not only be observed, but also participate actively.
—Save the Children staff
Teachers in intervention schools used the reading instruction practices they learned 
from the program in their classrooms, and they reported observing increased student 
engagement and class participation among their students. Teachers in intervention schools 
stated that they have changed how they teach reading and have incorporated the reading 
instruction strategies they learned in the program into their regular classroom practice (more than 
60 percent of the teachers interviewed at midline in schools receiving the program reported 
applying what they learned in training to their classroom practice every day). Teachers value the 
program materials and learning to prepare their own materials with a specific pedagogical intent. 
Teachers reported feeling that the instructional practices they have implemented promote student 
engagement, and they have noticed more active student participation in their classes. However, 
teachers still struggle with heterogeneity in students’ reading skills during the early primary 
grades. 
Coaching visits were key for supporting teachers, but some teachers felt they did not 
receive enough of these visits. The teacher 
coaching of Leer Juntos, Aprender Juntos (known 
as acompañamiento) consisted of two steps. 
During the first coaching visit, the program 
specialist modeled the lesson to demonstrate core 
reading instruction strategies. During the second 
coaching visit, the teacher delivered the lesson 
while the program specialist observed and provided 
feedback, after which they jointly discussed how the teacher could improve. Specialists from 
Kallpa visited teachers in intervention schools every two months. Fluctuations in the number of 
active program specialists resulted in very tight coaching visit schedules; for example, there was 
Of course I would like [my child] to take 
Quechua in upper grades as a general culture, 
to be able to communicate in the region. But 
[my child] should also [learn in] other 
languages because when they leave [the 
school] Quechua will not help much in a job.
— Chincheros parent focus group
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
30
a time during the project when seven specialists had to cover coaching visits for approximately 
500 teachers. As a result, some teachers who participated in focus groups felt they did not 
receive enough coaching visits. Two factors—capacity constraints in the number of coaching 
visits program specialists could conduct and teachers’ interest in participating in and observing 
other teachers’ coaching visits—led the Kallpa to modify the original in-class coaching strategy. 
Kallpa added three sessions of the Teacher Circles per year (known as círculos de 
interaprendizaje colaborativo) to complement the coaching visits. This addition to the original 
coaching component was well suited to the local context, creating a space for teachers to discuss 
what they were practicing in their classrooms, create pedagogical materials with other teachers, 
and receive targeted assistance from program specialists. 
The formative evaluation component of the program was implemented in the teacher 
trainings but was not always used systematically to track students’ progress at the 
individual level. Leer Juntos, Aprender Juntos includes a formative evaluation component that 
enables teachers to monitor students’ acquisition of foundational reading skills and to inform 
differentiated instruction. Program specialists provided assessment forms for teachers to monitor 
students’ phonemic awareness, reading fluency, vocabulary, and comprehension. They also 
showed teachers how to display assessment results for students to see their progress. 
Approximately three-quarters of the teachers interviewed at midline in intervention schools 
reported being trained on the use of assessments (compared to only 40 percent of teachers in 
control schools); however, teachers in treatment and control schools reported spending nearly the 
same number of hours per week (approximately three hours) planning, administering, and 
grading students’ assessments or homework. The implementing team’s reflections on the 
formative assessment component suggested that formative assessments were not used 
consistently in all intervention schools. As a 
result, the implementation team developed 
detailed protocols to facilitate teachers’ use of 
formative assessment. However, as interview 
respondents indicated, it is possible that this 
component might have not been implemented 
with the same thoroughness as the rest of the 
in-school intervention activities. 
[The coaching] makes you see where you are 
failing, what you lack, because when a person is 
observing your weaknesses; you can improve 
according to that (feedback).
—Andahuaylas teacher focus group
High teacher turnover and rotation of teachers to upper grades were barriers to 
implementation. Teachers who were trained and subsequently left made it necessary to train 
their replacements or have untrained teachers. Rotating teachers to different grade levels (other 
than 1st, 2nd, or 3rd) even when they stayed in the same school was a barrier to strengthening 
new reading instruction strategies in the classroom. Kallpa’s team made efforts to engage 
educational authorities and school principals with the goals of the programs and emphasized the 
importance of keeping the cohorts of trained teachers assigned to the early elementary grades (1 
through 3); however, ultimately grade-level assignments were beyond their control and teacher 
rotation could not be prevented. 
The duration of the nine training modules made implementation difficult. This 
commitment made it difficult to secure the support of the teachers’ union to support teachers’ 
participation in the training. The original training model from Literacy Boost, which called for 
nine training modules (one day each), was not feasible in the local context. When tailoring the 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
31
Literacy Boost training modules for implementation in Peru, Save the Children/Peru and Kallpa 
adapted the content for delivery in three and a half days, prioritizing the information most 
relevant for teachers. 
Teachers valued the training and coaching they received and reported an increase in 
student motivation to read. Teachers who participated in focus groups shared that the reading 
instruction strategies they learned from the program had motivated students in their classes and 
that the students seemed more engaged and interested in reading activities. Teachers believed the 
program had provided them with pedagogical materials and with an array of activities to teach 
reading that are feasible to implement in their classrooms. Also, teachers stated that the one-on￾one support they received during coaching sessions has driven them to make specific efforts to 
improve their teaching practice. The conversations, self-reflection, and agreements that took 
place during the coaching visits added a layer of accountability to teachers’ commitment to 
improve their practice. 
D. Implementation of the community action component
Recruiting and retaining volunteers was a major challenge to implementation. The 
implementation of community action activities to promote reading relied on the participation of 
volunteers—known in the program as “reading promoters.” Initially, Kallpa recruited literate 
adults to join the cadre of volunteers (one per community). Illiterate adults who wanted to 
participate as volunteers were included, but only to help the lead volunteers who had sufficient 
reading skills. Volunteers were expected to conduct the reading camps, lead the storytelling hour, 
and manage the book banks. 
During the first year, almost half of the volunteers quit the program. According to 
interviews with Kallpa and Save the Children staff, even though volunteers were motivated to 
help children practice reading, they struggled to find the time for the community activities and 
wanted to be paid for their time. In addition, volunteering is not common practice in the region, 
so the implementation team had difficulties finding candidates who had previous experience 
leading and/or participating in volunteering activities. 
In response to these challenges, the implementing team changed the recruitment profile and 
sought students in the upper grades to become volunteers. The change in the profile of volunteers 
was favorable— students participating as community volunteers felt they were making a positive 
contribution to their school and received recognition by teachers and peers. Young volunteers 
also felt they gained leadership and communication skills. 
Despite these adjustments to the recruitment of volunteers, barriers to implementing the 
community action component persisted. Turnover of community volunteers was high, and some 
stakeholders perceived deficiencies in the coordination of community activities. Volunteers felt 
they needed more training to conduct community activities and more support from program 
specialists. Some parents also shared concerns about the skill level of volunteers. Although 
Kallpa specialists excelled in their work with teachers, parents expressed that Kallpa specialists 
seemed to be less equipped to work successfully with parents and volunteers. 
The reading buddies activity, which as part of the community action component was 
designed to be implemented outside of school hours, did not work as expected and was 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
32
instead adapted for implementation during regular school hours. The reading buddies 
activity (known as Amigos de la Lectura) was originally part of the community action 
component that volunteers would implement outside the regular school schedule. The 
implementing team noticed that the after-school activity was not working well in the 
community—volunteers had difficulty managing reading banks, planning reading camps, and 
simultaneously coordinating the reading buddies. As a result, the implementation team modified 
the delivery format of the reading buddies activity, while preserving its pedagogical intent. This 
activity was shifted to the school context and organized by teachers in their classrooms during 
school once a week. It was implemented only in group A schools (schools implementing the 
community action component). 
Despite challenges recruiting and retaining volunteers, children in the evaluation were 
offered the activities that corresponded to the group to which they were assigned.
Parents/caregivers of at least 60 percent of the children in Group A reported that they were 
indeed invited to participate in reading camps, story hour, reading buddies activities, and reading 
fairs. In comparison, parents/caregivers of a smaller percentage of children in Groups B and C 
reported their children were invited to participate in those community activities. In addition, the 
parents/caregivers of 46 percent of children in Group A reported their children were invited to 
borrow books from a book bank or library, but more than 50 percent said that children in that 
group actually borrowed books or printed materials.13
Children in Group A were exposed to the community reading activities of the 
intervention, and most of the children in Groups B and C were not. In the month before the 
household survey was administered, 56 to 64 percent of the children in Group A participated in 
reading camps outside school hours, story time activities in which volunteers read aloud to the 
children, activities in which older children provided tutoring and played reading games with the 
evaluation children (reading buddies), and reading fairs/festivals in which they read with other 
children in the community. Six to 11 percent of the children in Groups B and C reported 
participation in reading activities and games in their communities in the previous month. 
Table III.2. Parents’ reports on children’s exposure to and participation in 
reading activities in the community
 
Leer
Juntos
(A)
Leer Juntos￾school only
(B)
Prevailing 
practice
(C) A–B B–C 
Percentage of children invited to participate in:
Reading camps 61.9 10.7 13.6 51.1*
(0.000)
-2.9
(0.499)
Story time with community volunteers 59.5 10.4 11.8 49.1*
(0.000)
-1.4
(0.737)
Tutoring and reading games with older 
children (reading buddies)
64.6 8.6 13.1 56.0*
(0.000)
-4.5
(0.238)
13 In all groups, the percentage of households that reported they borrowed a book is larger than the percentage of 
households that reported they were “invited” to borrow a book. It is likely that families borrow books if they are 
available without waiting to receive a formal invitation to do so. The survey questions were not structured in such a 
way as to be able to determine this, however.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
33
. .
Leer
Juntos
(A)
Leer Juntos￾school only
(B)
Prevailing 
practice
(C) A–B B–C 
Reading fairs 67.0 11.1 11.2 55.9*
(0.000)
-0.2
(0.962)
Borrowing books from a book bank or a 
library
45.5 22.3 15.5 23.2*
(0.000)
6.7
(0.133)
Percentage of children who participated in:
Reading camps 58.2 8.7 11.4 49.5*
(0.000)
-2.7
(0.505)
Story time with community volunteers 55.6 8.6 10.1 47.1*
(0.000)
-1.5
(0.690)
Tutoring and reading games with older 
children (reading buddies)
60.9 6.4 11.2 54.4*
(0.000)
-4.8
(0.185)
Reading fairs 63.5 9.4 9.5 54.1*
(0.000)
-0.2
(0.967)
Borrowing books from a book bank or a 
library
55.3 36.1 25.2 19.3*
(0.001)
10.9
(0.055)
Number of childrena 366 338 333
Source: Leer Juntos, Aprender Juntos Household Survey—Midline 2014 and 2015.
Note: Column A–B presents differences in the regression-adjusted group means between the Leer Juntos group and the Leer 
Juntos-school only group, and Column B–C presents differences in the regression-adjusted group means between the 
Leer Juntos-school only group and the prevailing practice group. The p-values from tests of differences between group 
means are presented in parentheses.
a Number of children for whom a household survey was completed.
* Difference in group means is statistically significant at the 0.05 level.
This page has been left blank for double-sided copying.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
35
IV. WHAT ARE THE IMPACTS ON INSTRUCTIONAL PRACTICES AND 
LITERACY ENVIRONMENT? 
As indicated in the program’s logic framework presented in Chapter II (Figure II.2), we 
expected that participation in the in-school component of the program would result in trained 
teachers who use the strategies learned in the training effectively, track children’s reading 
achievement with formative assessments, and create a print-rich classroom environment where 
reading materials are available in mother tongue and Spanish. We also expected that the 
community action component of the program would provide children with more opportunities to 
practice reading outside of school hours and help parents support their children’s reading. 
The teacher survey and classroom observation data provided evidence that the children in 
the schools implementing the in-school component of the intervention (Group B) experienced 
improved classroom environments when compared with the children in the schools 
implementing the prevailing practice (Group C). This is consistent with reports from intervention 
teachers in focus groups (see Chapter III): that the Leer Juntos, Aprender Juntos training and 
coaching provided them with concrete instructional strategies that they could apply in their 
teaching practice, as well as with tools to create pedagogical materials. Regarding children’s 
activities outside of school hours, our findings from the analysis of household survey data are 
consistent too with the findings from the analysis of qualitative information collected. In 
particular, we did not find impacts of the community action component of the intervention on the 
reading activities that children do at home. This finding reflects the challenges experienced in 
implementing this component of the intervention and the difficulties that parents in the 
evaluation faced to attend intervention workshops and creating reading spaces at home. 
A. Classroom instruction
1. Teacher participation in professional development activities 
The Leer Juntos, Aprender Juntos intervention significantly increased the percentage of 
teachers receiving coaching. Both in the focus groups conducted in the qualitative evaluation 
(see Chapter III) and in the teacher survey, teachers in the treatment groups (Groups A and B) 
reported they actively participated in professional development activities focused on reading 
instruction. More teachers in the Leer Juntos, Aprender Juntos intervention groups reported 
receiving training and coaching than did teachers in the prevailing practice group (Group C) 
during the calendar year when the teacher survey was administered (2015 for Phase I teachers 
and 2016 for Phase II teachers). We expected this finding, given that one of the components of 
the intervention focused on providing teachers with training and coaching. The differences in 
training rates (53 percent in Groups A and B versus 39 percent in Group C) were not large 
enough to be statistically significant, but the differences in coaching between Groups B and C 
(71 versus 47 percent, respectively) were (see Figure IV.1). 
In terms of the sources for training and coaching, essentially all Group C teachers reported 
that the Ministry of Education provided their training and coaching. Teachers in Groups A and B 
reported that their training and coaching were provided by a combination of Save the Children 
(for Leer Juntos, Aprender Juntos) and the Ministry of Education (see Tables E.1 through E.3 
and Figure E.1 in Appendix E). 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
36
Figure IV.1. Teacher-reported participation in reading instruction training and 
coaching in the endline year
2. Teacher-reported and observed practices
We measured classroom practices using both teacher reports and live classroom 
observations. In the endline teacher survey, the evaluation team asked teachers about the 
methods, strategies, materials, and allocation of instruction time they use to teach reading, and 
about the needs and challenges they face in teaching their students to read. The evaluation team 
also conducted structured observations of 3rd grade classrooms during a typical language or 
reading class to measure the proportion of observed instruction time that teachers allocated to 
literacy instruction activities. Observers noted features of the print environment and recorded 
teachers and students use of mother tongue and Spanish language in the classroom. Observers 
also completed a checklist on teachers’ use of instructional practices focused on teaching the five 
foundational reading skills targeted in the program’s training (alphabet knowledge, phonological 
awareness, vocabulary, fluency, and comprehension) and of general practices such as providing 
feedback to students. 
We found no meaningful pattern of statistically significant program effects on 
instructional practices as reported by teachers or observed by evaluators. One possible 
reason for this finding is teacher exits. As presented in Chapter III, the implementation team did 
not control grade-level assignments, so it was impossible to prevent school leaders from 
reassigning Leer Juntos, Aprender Juntos-trained teachers to grades 4 to 6, which were not part 
of the evaluation. (See Appendix E, Tables E.4 through E.10 and Figure E.2 for the full set of 
detailed findings on teacher-reported and observed practices.) 
Another possible explanation for the lack of impacts on classroom practices is that training 
time was compressed to adapt to the teachers’ busy schedules and the regulations of local 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
37
teachers’ unions (see Chapter III), and may have been insufficient. If teachers did not have 
enough time to process and practice what they learned from the training, they might not have 
been able to internalize and practice what they had learned and fully incorporate those new 
strategies into their instruction practices. 
B. Reading environment and behaviors 
1. Environment in the classroom 
Classroom observers recorded the presence of a reading area in the classrooms, the location 
and accessibility of books, and the availability of physical print resources for students, noting 
their language and physical condition. 
Intervention schools provided greater exposure to print materials than schools 
assigned to implement prevailing practice. As shown in Table IV.1, more classrooms in Group 
A than in Group B had books placed in multiple locations within the classroom. Also, more 
classrooms in Group B than in Group C had complete alphabets in Spanish and in Quechua that 
were visible to students. Finally, the percentage of classrooms that displayed familiar words 
written in Quechua was higher in classrooms in Group A than in classrooms in Group B (74 and 
54 percent, respectively). The differences mentioned here were all statistically significant.
Table IV.1. Materials to facilitate early reading instruction that are visible to 
students in the classroom (percentage of classrooms)
Materials
Leer
Juntos
(A)
Leer 
Juntos￾school
only
(B)
Prevailing 
practice
(C)
Impact of 
community 
action
(A–B)
Impact of in￾school 
component
(B–C)
Reading area equipped with reading materials 
for students
41.0 43.6 44.7 -2.6
(0.809)
-1.1
(0.921)
Classroom with books located in areas that are 
accessible to students
97.9 90.5 95.9 7.4
(0.100)
-5.4
(0.234)
Classroom with books placed in a number of 
locations
57.5 37.1 43.5 20.4*
(0.042)
-6.5
(0.521)
Materials to facilitate early reading instruction
Complete alphabet in Spanish visible to 
students
87.3 72.1 44.1 15.2
(0.081)
28.0*
(0.002)
Complete alphabet in Quechua visible to 
students
56.9 46.4 22.7 10.6
(0.259)
23.7*
(0.014)
Familiar words written in Spanish visible to 
students
100.0 95.7 100.0 4.2
(0.081)
-4.3
(0.080)
Familiar words written in Quechua visible to 
students
73.7 53.9 56.0 19.8*
(0.030)
-2.2
(0.813)
Number of classrooms 49 47 48
Source: Leer Juntos, Aprender Juntos Classroom Observation Form—Final Follow-ups 2015 and 2016.
Note: Regression-adjusted means. P-values in parentheses.
* Difference in group means is statistically significant at the 0.05 level.
In terms of the type of books observed in the classrooms (Table IV.2), we found one 
statistically significant difference between treatment groups: more classrooms in Group C (29 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
38
percent) had picture books without text than in Group B (10 percent). This result is consistent 
with teachers’ reports on the type of texts their students use (Table E.5 in Appendix E): more 
teachers in Group C (32 percent) than in Group B (10 percent) reported their students use picture 
or art books every day. 
Table IV.2. Characteristics of the books observed in the classrooms 
(percentage of classrooms)
Difficulty and genre
Leer
Juntos
(A)
Leer 
Juntos￾school 
only
(B)
Prevailing 
practice
(C)
Impact of
community 
action
(A–B)
Impact of in￾school 
component
(B–C)
Difficulty level of books
Simple books with figures or the alphabet 9.7 14.3 15.4 -4.7
(0.512)
-1.1
(0.881)
Picture books 19.6 10.4 28.7 9.1
(0.201)
-18.3*
(0.013)
Picture books with some text 73.5 74.1 68.7 -0.6
(0.950)
5.4
(0.582)
Simple books (obras sencillas) 34.5 28.7 29.8 5.7
(0.564)
-1.1
(0.915)
Textbooks 97.9 97.9 100.0 0.0
(0.989)
-2.2
(0.386)
Other books 17.9 20.9 29.5 -3.0
(0.726)
-8.6
(0.320)
Book genre
Reference books 49.3 37.8 38.9 11.5
(0.250)
-1.1
(0.915)
Narrative books 75.2 77.4 68.8 -2.2
(0.812)
8.6
(0.363)
Workbooks 97.9 91.5 98.0 6.4
(0.132)
-6.5
(0.131)
Number of classrooms 49 47 48
Source: Leer Juntos, Aprender Juntos Classroom Observation Form—Final Follow-ups 2015 and 2016.
Note: Regression-adjusted means. P-values in parentheses.
* Difference in group means is statistically significant at the 0.05 level.
Observers also recorded the number of books and other instructional texts in Spanish and 
Quechua that they observed in the classroom, but we did not find any statistically significant 
difference between treatment groups in the availability of those materials in the classroom. We 
present these results in Table E.12 in Appendix E. 
The findings presented in Tables IV.1 and IV.2 are consistent with the teacher reports we 
received during the focus groups conducted for the qualitative evaluation (Chapter III): teachers 
in the treatment groups were enthusiastic about using what they learned in the intervention 
training and creating a print-rich environment in their classrooms. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
39
2. Environment in the home
As explained in Chapter II, we conducted a household survey to obtain information on the 
characteristics of the literacy environment in the homes of the children participating in the 
evaluation. The evaluation team collected information on the availability of books for children in 
the household, the frequency with which the children use those books, the family members who 
read with the child, the existence of dedicated time for reading at home for children, and the 
activities—related to reading and not—that children do when they are at home. 
a. Availability of reading materials at home 
We found no statistically significant differences between treatment groups in the 
availability of books for children at home. However, as shown in Table IV.3, books for 
children were available in most of the households in the evaluation (97 to 99 percent in each 
treatment group). Also, households in each treatment group had, on average, six to seven books 
for children. In 50 to 57 percent of the households in each treatment group, books were available 
only in Spanish, and about 40 percent of the households in each treatment group reported the 
availability of children’s books written in either Spanish or in Quechua. Those differences 
between treatment groups were not statistically significant (Table IV.3). 
Table IV.3. Availability of books for children at home
Blank cell
Leer
Juntos
(A)
Leer 
Juntos￾school only
(B)
Prevailing 
practice
(C)
Impact of 
community 
action
(A–B)
Impact of 
in-school 
component
(B–C)
Percentage of households in which books for 
children are available
99.2 98.6 97.0 0.6
(0.482)
1.5
(0.192)
Number of books for children in the household 
(average)
6.3 5.5 6.5 0.8
(0.093)
-1.0
(0.067)
Percentage of households with books for children in the following languages:
Spanish only 56.0 56.5 49.8 −0.5
(0.916)
6.7
(0.190)
Quechua only 0.2 0.4 0.3 -0.2
(0.603)
0.1
(0.887)
Spanish and Quechua 43.0 41.8 47.0 1.3
(0.755)
−5.2
(0.265)
Percentage of children using the books for children in the household with the following frequency:
Never 1.4 1.8 3.1 −0.4
(0.673)
−1.3
(0.295)
One or two times per week 31.0 32.2 21.8 -1.2
(0.776)
10.4*
(0.012)
Three or four times per week 38.1 40.8 41.8 −2.7
(0.490)
-0.9
(0.824)
Every day 29.5 25.2 33.3 4.4
(0.313)
-8.2
(0.057)
Number of children 358 338 326
Source: Leer Juntos, Aprender Juntos Household Survey—Midline 2014 and 2015.
Note: Regression-adjusted means. P-values in parentheses.
* Difference in group means is statistically significant at the 0.05 level.
We also found that in each treatment group, 22 to 32 percent of the children use the 
children’s books one or two times per week, and 25 to 30 percent of the children in each 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
40
treatment group use the children’s books every day. Across all the comparisons presented in 
Table IV.3, we found one statistically significant difference between treatment groups in 
frequency of use of the books for children available in the home: more children in Group B than 
C use the children’s books one or two times per week. More children in Group C than in Group 
B use the children’s books every day, but this difference is not statistically significant at the 0.05 
level (it is statistically significant at the 0.10 level; p = 0.057). 
b. Reading at home 
Most children in the evaluation communities already read at home with a parent or 
other family members. There was little room for the intervention to make a difference, because 
most children attending schools in the prevailing practice group already have someone who 
looks at books with them (95 percent) or reads books to them (74 percent), as shown in Figure 
IV.2. This finding is consistent with the household survey respondents’ reports about their 
literacy skills: about three-quarters of survey respondents (76 percent in Groups A and C, and 73 
percent in Group B) reported they can read (data not shown). 
Figure IV.2. Percentages of evaluation children whose parents or other 
household members read with them 
76
95
41
58
72
96
45
51
74
95
42
54
0
10
20
30
40
50
60
70
80
90
100
Respondent reads with
the child
Respondent looks at
books with the child
Sibling Other adult in the
household
Other family member reads with the child
Percentage
Leer Juntos (A) Leer Juntos-school only (B) Prevailing practice (C)
Source: Leer Juntos, Aprender Juntos Household Survey—Midline 2014 and 2015.
Note: None of the differences between treatment groups presented in this table is statistically significant at the 
0.05 level.
Leer Juntos, Aprender Juntos aimed to increase the time scheduled for reading in the 
home, but the effects were not apparent by the time we conducted the household survey.
For most children in each treatment group (73 percent in Group A, 68 percent in Group B, and 
70 percent in Group C), parents or caregivers schedule a time for their children to dedicate to 
reading activities at home (Table IV.4). On average, the dedicated time for reading at home is 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
41
three hours per week in each treatment group. We found no statistically significant differences 
between treatment qgroups in the scheduled time for reading at home. 
Table IV.4. Dedicated time for reading at home
Blank cell
Leer
Juntos
(A)
Leer 
Juntos￾school only
(B)
Prevailing 
practice
(C)
Impact of 
community 
action
(A–B)
Impact of 
in-school 
compone
nt
(B–C)
Percentage of children who have dedicated time 
for reading at home
72.7 68.3 70.1 4.4
(0.324)
-1.8
(0.699)
Hours per week scheduled for reading at home 
(average)
3.2 3.3 3.0 −0.1
(0.752)
0.3
(0.337)
Number of children 358 338 326
Source: Leer Juntos, Aprender Juntos Household Survey—Midline 2014 and 2015.
Note: Regression-adjusted means. P-values in parentheses.
* Difference in group means is statistically significant at the 0.05 level.
We did not find any statistically significant differences between treatment groups in the 
percentage of children who read with their parents or other family members at home, or in the 
dedicated times at home for reading activities, despite the intention of the reading buddies 
programs and parent workshops offered in Group A communities to encourage more siblings and 
other volunteers to read to children and create spaces at home for reading. The other family 
members who read with the children in all treatment groups are siblings or other adults living in 
the household, such as aunts, uncles, and grandparents (Figure IV.2). 
Children in all three treatment groups engaged in reading activities outside of school 
hours. At home or in the community, most children in the evaluation (at least 81 percent in each 
treatment group) participated in reading activities such as reading books, doing homework, and 
hearing or telling stories. Slightly more children in Group A (99 percent) than in Group B (97 
percent) spent time doing homework (Figure IV.3). In addition, more children in Group A (72 
percent) than in Groups B and C (15 and 18 percent, respectively) played reading games with 
other children in their community. This finding is not surprising, because children in Group A 
were receiving the community reading activities of the Leer Juntos, Aprender Juntos
intervention that children in Groups B and C were not. 
We also asked parents about the amount of time children spent in the week preceding the 
household survey on reading activities.14 Children in the three treatment groups did homework 
on three to four days of the week preceding the household survey and heard or told stories on 
two days of that week. Children in Group A played reading games with other children in the 
communities one day of that week, whereas children in Groups B and C did not spend time on 
that activity in that week. Again, this is consistent with children in the communities in Group A 
having access to the community action component activities of the intervention, whereas 
children in communities in Group B and C did not. 
14 These results are not presented in the report but are available from authors upon request.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
42
Figure IV.3. Reading-related activities at home (percentage of children)
Source: Leer Juntos, Aprender Juntos Household Survey—Midline 2014 and 2015.
* Difference in group means is statistically significant at the 0.05 level.
c. Activities at home other than reading
Children in the group receiving the community action component of the intervention 
spent less time at home helping with household chores. As we show in Table IV.5, in the 
week before administration of the household survey, more children in Group B (95 percent) than 
in Group A (91 percent) helped with housework, and children in Group B did that activity on 
more days (more than four days, on average) than children in Group A (fewer than four days, on 
average). This statistically significant difference suggests that the community component 
intervention activities to which the children in Group A had access might have led them to 
replace time on household chores with time spent participating in the reading activities offered 
by the program in their community. 
We also found another statistically significant difference between treatment groups in the 
activities not related to children reading at home: a smaller percentage of children in Group B 
(92 percent) than in Group C (97 percent) listened to the radio or watched television while at 
home in the week before the household survey was administered, and children in Group B did 
that activity on fewer days (five days), on average, than children in Group C (six days). Finally, 
all children in each treatment group played inside or outside of their homes, and did so almost 
every day of that week. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
43
Table IV.5. Children’s activities at home that are not related to reading
Blank cell
Leer
Juntos
(A)
Leer 
Juntos￾school only
(B)
Prevailing 
practice
(C)
Impact of
community 
action
(A–B)
Impact of in￾school 
component
(B–C)
Percentage of children who in the week before the administration of the household survey:
Played inside or outside the house 99.8 100.0 100.0 −0.2
(0.192)
0.0
(0.983)
Listened to the radio or watched television 94.7 91.5 96.9 3.2
(0.109)
−5.3*
(0.003)
Helped with housework 91.2 94.6 94.5 −3.4*
(0.049)
0.1
(0.949)
Helped with vegetable garden/corn field and/or 
with livestock
79.3 77.1 77.4 2.1
(0.541)
−0.2
(0.953)
Cared for younger siblings 53.5 50.1 52.5 3.4
(0.358)
-2.4
(0.557)
Worked for pay or wage 7.2 7.1 6.5 0.1
(0.961)
0.6
(0.828)
Number of days (average) in the week before the administration of the household survey in which children:
Played inside or outside the house 6.4 6.5 6.5 −0.1
(0.465)
0.0
(0.926)
Listened to the radio or watched television 5.5 5.4 5.8 0.1
(0.577)
−0.4*
(0.024)
Helped with housework 3.6 4.2 4.2 −0.6*
(0.002)
0.1
(0.787)
Helped with vegetable garden and/or with 
livestock
3.1 3.2 3.2 -0.1
(0.540)
0.1
(0.775)
Cared for younger siblings 2.8 2.8 3.0 0.0
(0.930)
-0.2
(0.493)
Worked for pay or wage 0.1 0.1 0.1 0.1
(0.065)
−0.0
(0.212)
Number of children 358 338 326
Source: Leer Juntos, Aprender Juntos Household Survey—Midline 2014 and 2015.
Note: Regression-adjusted means. P-values in parentheses.
* Difference in group means is statistically significant at the 0.05 level.
This page has been left blank for double-sided copying.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
45
V. WHAT ARE THE IMPACTS ON CHILDREN’S EARLY-GRADE READING 
SKILLS? 
In this chapter, we present evidence that the Leer Juntos, Aprender Juntos program in Peru 
accomplished one of its main goals: to have a favorable impact on the reading comprehension of 
the children in the schools that implemented the program. Specifically, we found that for both 
the full sample and the subgroup of girls, the percentage of children in the 3rd grade with a basic 
level of competency in reading comprehension was larger in schools implementing the in-school 
component (Group B) than in the prevailing practice schools (Group C). We also found that 
children in the schools that implemented the in-school component of the intervention had, on 
average, higher decoding and fluency accuracy scores than the children in schools assigned to 
the prevailing practice group.15 The magnitude of the impacts is 0.19 standard deviations for the 
decoding accuracy score and 0.21 standard deviations for the fluency accuracy score. Impacts on 
decoding and fluency were also positive when we measured those constructs by the number of 
words (or invented words) read correctly per minute, but those impacts were not large enough to 
be statistically significant. We did not find impacts of the community component on any literacy 
skills. We discuss these findings in more detail below. 
A. How we measured early-grade reading skills
We assessed the reading skills in Spanish of the 3rd grade children in the evaluation with 
three tasks: (1) pseudo-word reading (decoding), (2) reading fluency, and (3) reading 
comprehension. These three tasks were adapted for Peru following the guidelines of the EGRA 
Toolkit (RTI International 2015). Passages for the fluency and reading comprehension tasks 
were drawn from the Young Lives study (Niños del Milenio; Guerrero et al. 2012) in Peru and 
from the Evaluaciones Censales de Estudiantes (ECE; Ministerio de Educación 2009). All 
children were assessed in Spanish, as every school in the evaluation used Spanish as the 
language of instruction. We measured decoding and fluency skills because they are precursors to 
reading comprehension; that is, children must have the abilities to translate a printed word into 
sound (decoding skills) and to read connected text with speed, accuracy, and expression (fluency 
skills) before they can achieve reading comprehension (Kim et al. 2016; National Reading Panel 
2000). 
In the pseudo-word decoding task, children read aloud as many pseudo-words as they can 
within 60 seconds, with a maximum of 50 pseudo-words. Pseudo-words are vowel-consonant 
combinations that follow Spanish phonological and spelling rules but are not actual meaningful 
words (for example, garo and tabra). To administer this task, examiners were instructed to stop 
(discontinue) the task and move onto the next task (the fluency task) if the child answered the 
first five items in the task incorrectly or could not complete the first five items in 60 seconds. We 
used the number of pseudo-words read or decoded correctly to examine program effects. 
15 Although not the main focus of our evaluation, we did examine the impacts of the full intervention. That is, we 
examined the contrast between Group A (receiving both the in-school and the community action components) and 
Group C (implementing the prevailing practice in Peru). From that analysis, we found a similar pattern of 
differences in children’s reading skills to those we found when we compared Group B with Group C. See Table G.4 
in the Appendix for more details on the impacts of the full intervention on children’s reading skills.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
46
In the reading fluency task, children read aloud as many words as they can within 60 
seconds from a short story with a total of 154 words (11 sentences). This task has a time limit so 
that we can obtain a measure of children’s ability to read text quickly and accurately. We used 
the number of words read correctly in 60 seconds to examine the effects of the program. 
Examiners were instructed to stop the task if the child could not read correctly any of the 20 
words included in the first two lines of text within 60 seconds. In addition, examiners did not 
administer the subsequent reading comprehension task to the children whose fluency task was 
stopped in order to avoid causing them excessive stress. 
From the decoding and reading fluency tasks, we also constructed “accuracy” scores. We 
calculated these accuracy scores by dividing the number of correctly read words by the total 
number of words that the child read (correctly or incorrectly). We also used the decoding and 
fluency accuracy scores to examine program effects. Higher decoding and fluency accuracy 
scores suggest that children are more deliberate in translating print into sound (decoding or 
reading pseudo-words) and in reading words with fluency with a time limit, respectively.
The reading comprehension task in the evaluation’s 3rd grade assessment consists of three 
passages that children are asked to read quietly to themselves. Children received up to four 
minutes to finish reading a passage. Lookbacks were not allowed. After a child finished reading 
each passage, the examiner asked him or her to answer five questions about that passage. We 
constructed two variables from this task to examine program impacts on reading comprehension 
achievement: (1) the total number of correct answers in the task (up to 15), and (2) an indicator 
for whether the child correctly answered at least one of the questions in the task. The second 
construct is a dichotomous measure indicating whether the child demonstrates any emergent 
reading comprehension skills. 
We present standard deviations of all reading measures in Table F.1 in Appendix F. 
B. Impacts on reading skills
The in-school component of the intervention had positive, statistically significant impacts on 
some of the literacy outcomes of the evaluation’s children in the 3rd grade. However, we did not 
find statistically significant effects of the community component on the literacy outcomes of the 
children in the evaluation sample. The positive impact finding is consistent with the reports from 
teachers described in Chapter III. In focus groups, teachers from the schools that received the in￾school component of the intervention reported feeling motivated by the training and coaching 
because they established good relationships with the program specialists from Kallpa and 
because they received concrete guidance that they could incorporate into their regular practice. 
The finding of no impacts of the community action component of the intervention is also 
consistent with what we learned from the focus groups conducted in the analysis of qualitative 
information: implementing the community component activities was challenging because 
recruiting and retaining volunteers for the community activities was difficult in Peru. In addition, 
attendance at intervention activities such as parent workshops was lower than expected, and 
some parents reported difficulty creating spaces dedicated to reading activities at the home. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
47
1. Impacts on decoding and fluency
The impacts of Leer Juntos, Aprender Juntos on decoding and reading fluency skills 
measured in terms of correctly read words per minute were not large enough to be 
statistically significant. As mentioned earlier, we assessed the decoding and reading fluency 
skills of the children in the evaluation as they completed the 3rd grade to examine whether the 
intervention had an impact on those foundational reading skills. Decoding and reading fluency 
skills are essential elements in achieving reading comprehension (Fuchs et al. 2001). Decoding 
skills—the ability to translate a printed word into sound—encompass alphabet knowledge 
(ability to recognize and link written letters to their sound) and phonological awareness (ability 
to focus on, manipulate, break apart, and put together sounds orally). Reading fluency is the 
ability to read with speed, accuracy, and expression. 
For each of the treatment groups, Figure V.1 presents the average number of pseudo-words 
read correctly in the decoding task and Figure V.2 presents the average number of words read 
correctly in the fluency task. On average, the children in the schools that implemented the Leer 
Juntos, Aprender Juntos program (Groups A and B) read correctly in one minute more pseudo￾words in the decoding task (about 46 percent of the 50 pseudo-words included in the task) and 
more words in the fluency task (about 27 percent of the words included in the task) than children 
in the schools assigned to the prevailing practice group (about 42 percent of the words in the 
decoding task and 25 percent of the words in the fluency task). However, we did not find any 
statistically significant differences between treatment groups in the sum of correct answers in the 
decoding and reading fluency tasks that the children in the evaluation sample completed when 
they attended the 3rd grade (Figures V.1 and V.2). 
Figure V.1. Similar number of pseudo-words read correctly in the decoding 
task (of 50) for each treatment group
23 23 21
0
5
10
15
20
25
30
35
40
45
50
Number of pseudo-words
Leer Juntos (A) Leer Juntos-school only (B) Prevailing practice (C)
Source: Leer Juntos, Aprender Juntos Students’ Reading Skills Assessment—Final Follow-ups 2015 and 2016.
Note: None of the differences between treatment groups presented in this figure is statistically significant at the 
0.05 level.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
48
Figure V.2. Similar number of words read correctly in the fluency task (of 
154) for each treatment group
Source: Leer Juntos, Aprender Juntos Students’ Reading Skills Assessment—Final Follow-ups 2015 and 2016.
Note: None of the differences between treatment groups presented in this figure is statistically significant at the 
0.05 level.
The in-school component of Leer Juntos, Aprender Juntos had positive impacts on some 
of the indicators of decoding and fluency. Figure V.3 presents children’s decoding and fluency 
accuracy scores on a scale of 0 to 100. As shown in that figure, we found statistically significant 
differences between Groups B and C in the accuracy scores of the children participating in the 
evaluation: the children in the schools implementing the in-school component (Group B) 
obtained, on average, higher decoding and fluency accuracy scores than the children in schools 
implementing the prevailing practice approach (Group C). 
Figure V.3. Positive program impacts on decoding and fluency accuracy 
scores
Source: Leer Juntos, Aprender Juntos Students’ Reading Skills Assessment—Final Follow-ups 2015 and 2016.
* Difference in group means is statistically significant at the 0.05 level.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
49
This finding suggests that the children who had teachers who were trained and coached by 
the intervention, when compared with children who did not, were more deliberate in translating 
print into sound (decode or read pseudo-words) and in reading words with fluency with a time 
limit. This result is not surprising, because teachers in the schools that implemented the in-school 
component of the program were trained to provide instruction focused on improving 
foundational skills like alphabet knowledge and phonemic awareness that in turn help build 
decoding skills; these teachers were also trained on improving children’s reading fluency. 
We found no impacts of the community action component on fluency accuracy score.
The difference on fluency accuracy score between Groups A and B, which would reflect any 
additional effects of the community action component was not statistically significant at the 0.05 
level. However, the difference in the mean fluency accuracy scores between children in Groups 
A (74 percent) and B (78 percent), shown in Figure V.3, is statistically significant at the 0.10 
level (p = 0.058). 
2. Impacts on reading comprehension
The Leer Juntos, Aprender Juntos program had favorable impacts on the reading 
comprehension outcomes of children in the evaluation in Peru. The program targeted 
foundational reading skills with the objective of helping children in a multilingual context 
achieve reading comprehension in Spanish by the time they attended the 3rd grade. Children 
need to understand what they read so they can learn. Children who do not learn to read with 
comprehension in the earlier grades face greater difficulties in learning in later grades, as they 
have limited ability to understand information presented in print form, do not develop proper 
writing skills, and cannot evaluation materials in other subject areas by themselves (Gove and 
Cvelich 2011). 
We found that more children in schools that implemented the in-school component 
compared to children in schools in the prevailing practice group achieved basic reading 
comprehension. As presented in Figure V.4, the percentage of children who achieved basic 
reading comprehension skills (children who were able to complete both the decoding and fluency 
tasks and answer at least one question correctly in the reading comprehension task) is larger by a 
statistically significant margin in schools that implemented the in-school component of the 
intervention (Group B) than in schools that did not (schools in the prevailing practice group, or 
Group C). The children in the schools in Group B also answered correctly more questions in the 
reading comprehension task of the evaluation’s assessment, on average, than children in the 
schools in the prevailing practice group (see Figure V.5). However, the difference between 
Groups B and C in the number of questions answered correctly in the reading comprehension 
task is statistically significant only at the 0.10 level (p = 0.077). We did not find any statistically 
significant difference between Groups A and B in the reading comprehension outcomes of 
students. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
50
Figure V.4. Positive program impacts on the percentage of children who 
achieved a basic level of reading comprehension skills
Source: Leer Juntos, Aprender Juntos Students’ Reading Skills Assessment—Final Follow-ups 2015 and 2016.
* Difference in group means is statistically significant at the 0.05 level.
Figure V.5. Similar number of questions answered correctly in the reading 
comprehension task (of 15) for each treatment group
Source: Leer Juntos, Aprender Juntos Students’ Reading Skills Assessment—Final Follow-ups 2015 and 2016.
Note: None of the differences between treatment groups presented in this figure is statistically significant at the 
0.05 level.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
51
3. Program impacts in effect sizes
The size of the impacts we found in Peru is comparable to the impacts of other interventions 
targeting reading outcomes of children in the early grades. In our evaluation of Leer Juntos, 
Aprender Juntos, the statistically significant impacts on the in-school component of the 
intervention in terms of effect sizes16 (or standardized mean difference between Groups A and B) 
range from 0.19 to 0.32 standard deviations, as shown in Table V.1. A recent systematic review 
of evaluations studies on reading interventions in the developing world (Kim et al. 2016) found 
that the impacts in effect sizes of reading interventions in the studies they reviewed ranged from 
0.14 to 0.73 for reading fluency outcome measures, and from 0.05 to 0.58 for reading 
comprehension outcome measures.
Table V.1. In-school component effects on literacy outcomes in standard 
deviation units
Literacy outcome Effect size
Decoding
Number of pseudo-words read correctly in one minute 0.13
Accuracy score 0.19*
Fluency
Number of words read correctly in one minute 0.09
Accuracy score 0.21*
Reading comprehension
Number of questions answered correctly 0.13
Percentage of children who achieved basic reading comprehension skills 0.32*
Source: Leer Juntos, Aprender Juntos Students’ Literacy Skills Assessment—Final Follow-ups 2015 and 2016.
* Effect is statistically significant at the 0.05 level.
4. Sensitivity analysis 
The findings are not sensitive to changes in assumptions for the impact estimation model. 
We conducted robustness checks on two important assumptions we made in the estimation of 
program impacts. First, we reweighted the data to account for children’s nonresponse to the 3rd 
grade assessment. Second, we estimated program impacts based on a regression model that 
included control variables without imputations, which reduced the size of the analytic sample in 
about 15 percent. The results from these alternative specifications are the same as those 
presented in this chapter. In particular, based on the two alternative specifications, we found that 
the in-school component of the program had a favorable effect on the percentage of children who 
have basic reading comprehension skills in the 3rd grade. We present the findings from these 
sensitivity analyses in more detail in Appendix G (in Tables G.1 and G.2). 
16 Effect sizes are calculated as the difference in adjusted means between two treatment groups (A and B, or B and 
C) divided by the pooled and weighted standard deviation of the two groups (Hedges’ g) for continuous outcome 
measures, and as the log odds ratio divided by 1.65 (Cox index) for dichotomous outcome measures. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
52
We also examined program impacts on literacy outcomes for subgroups of students defined 
by phase (cohort). We found that the impacts of the in-school component for the overall sample 
are due to impacts for students in Phase II schools only (see Table G.3 in Appendix G). That is, 
in Phase II, children in schools in Group B had significantly better decoding and fluency 
accuracy scores than their counterparts in Group C. Also, the percentage of children in Phase II 
schools who have a basic level of competency in reading comprehension was larger in Group B 
than in Group C. We did not find any pattern of statistically significant impacts of the 
community action component for children in schools in either Phase I or Phase II. These findings 
are consistent with programmatic reports from the implementers (Save the Children and Kallpa 
2013, 2014, 2015). The challenges of delivering the in-school component of the intervention—
such as acquiring buy-in from local stakeholders to ensure participation in the teachers’ training, 
providing the training early enough in the school year, and meeting the demand for coaching 
visits with teachers—were felt more strongly during the first year of implementation of the 
program, when the evaluation included Phase I schools only (Phase II schools entered the 
evaluation in the second year of program implementation). 
Finally, we examined the contrast between Group A and Group C to assess the effect of the 
full intervention (the community action component and the teacher training and coaching 
component) on children’s reading outcomes (see Table G.4 in Appendix G). We found that the 
full intervention had positive, statistically significant impacts at the 0.05 level on children’s 
decoding skills. In particular, we found that children in Group A read more pseudo-words per 
minute and had higher decoding accuracy scores than children in Group C (Table G.4 in 
Appendix G). We also found that children in Group A read correctly more words in the fluency 
task and answered correctly more questions in the reading comprehension task than children in 
Group C, and those differences were statistically significant at the 0.10 level. These findings are 
consistent with the findings we presented earlier in this chapter regarding the effects of the in￾school component of the program. 
C. Impacts by gender
The effects of the in-school component of the program on literacy outcomes were due 
to impacts for girls and not boys. We examined the program effects for subgroups defined by 
sex to explore whether the components of the program work for girls and boys in different ways. 
As shown in Table V.2, we found statistically significant effects of the in-school component, as 
contrasted with the prevailing practice approach, for the girls in the sample. In particular, the in￾school component had a statistically significant effect on girls’ decoding skills (number of 
pseudo-words read correctly in one minute and decoding accuracy score) and fluency accuracy 
scores, as well as on the percentage of girls who have achieved basic reading comprehension 
skills in the 3rd grade. 
We did not find statistically significant differences between treatment groups for the 
boys in the sample. Furthermore, we found that the difference between girls and boys on the 
effect of the in-school component on the indicator of basic reading comprehension skills is 
statistically significant (the p-value of this difference is 0.025; this result is not presented in 
Table V.2), providing more evidence that the in-school component might have been more 
successful in creating fluent readers for girls than for boys. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
53
Table V.2. Impacts on the literacy outcomes of girls and boys in the 
evaluation
 
 
Literacy outcome
Leer
Juntos
(A)
Leer 
Juntos￾school 
only
(B)
Prevailing 
practice
(C)
Impact of
community 
action
(A–B)
Impact of 
in-school 
component
(B–C)
Girls
Decoding
Number of pseudo-words read correctly in 
one minute (of 50)
23.3 24.0 20.8 -0.7
(0.564)
3.2*
(0.007)
Accuracy score (0 to 100) 70.8 73.6 64.3 -2.8
(0.379)
9.3*
(0.001)
Fluency
Number of words read correctly in one 
minute (of 154)
42.6 43.4 38.2 -0.8
(0.800)
5.2*
(0.040)
Accuracy score (0 to 100) 76.6 82.1 71.4 -5.5
(0.082)
10.7*
(0.001)
Reading comprehension
Number of questions answered correctly 
(of 15)
4.9 4.8 4.3 0.1
(0.766)
0.5
(0.099)
Percentage of girls who achieved basic 
reading comprehension skills
78.6 87.3 72.0 -8.7
(0.080)
15.3*
(0.001)
Boys
Decoding
Number of pseudo-words read correctly in 
one minute (of 50)
23.2 21.3 21.0 1.9
(0.188)
0.3
(0.832)
Accuracy score (0 to 100) 68.0 66.2 64.0 1.8
(0.599)
2.2
(0.544)
Fluency
Number of words read correctly in one 
minute (of 154)
41.2 38.7 39.0 2.5
(0.393)
-0.3
(0.917)
Accuracy score (0 to 100) 70.9 74.2 72.0 -3.3
(0.352)
2.2
(0.581)
Reading comprehension
Number of questions answered correctly 
(of 15)
4.7 4.7 4.3 0.0
(0.938)
0.4
(0.244)
Percentage of boys who achieved basic 
reading comprehension skills
79.3 78.7 76.7 0.6
(0.910)
2.0
(0.680)
Number of girls 178 175 167
Number of boys 180 163 159
Source: Leer Juntos, Aprender Juntos Students’ Reading Skills Assessment—Final Follow-ups 2015 and 2016.
Note: Regression-adjusted means. P-values in parentheses.
* Difference in group means is statistically significant at the 0.05 level.
Impacts on reading outcomes for girls were considerable. Table V.3 show the effects of 
the in-school component for girls and boys in terms of standard deviation units (effect sizes). 
They range from 0.15 to 0.60 standard deviations. In particular, the effect of the in-school 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
54
component on the percentage of girls who have basic skills in reading comprehension in 3rd 
grade is larger than half of a standard deviation (60 percent). We present the effects of the 
community action component for girls and boys in terms of standard deviation units in Table H.1 
in Appendix H. 
Table V.3. In-school component effects on literacy outcomes for girls and 
boys, in standard deviation units
Literacy outcome Effect size for girls Effect size for boys
Decoding
Number of pseudo-words read correctly in one minute 0.24* 0.02
Accuracy score (0 to 1) 0.32* 0.07
Fluency
Number of words read correctly in one minute 0.20 -0.01
Accuracy score (0 to 1) 0.36* 0.06
Reading comprehension
Number of questions answered correctly 0.15 0.11
Percentage of children achieved basic reading 
comprehension skills
0.60* 0.07
Source: Leer Juntos, Aprender Juntos Students’ Literacy Skills Assessment—Final Follow-ups 2015 and 2016.
* Effect is statistically significant at the 0.05 level.
The finding that the program had impacts for girls but not for boys does not mean that 
the boys did not learn. Figures V.6 and V.7 illustrate the change in the literacy skills over the 
evaluation period for girls (Figure V.6) and separately for boys (Figure V.7). In all treatment 
groups, girls and boys, essentially had no decoding skills at baseline (when they were in 1st 
grade), but at the endline assessment (at the end of the school year when they were supposed to 
be in 3rd grade) they had an average of 21 to 24 words per minute. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
55
Figure V.6. Growth in girls decoding skills
Source: Leer Juntos, Aprender Juntos Students’ Reading Skills Assessment—Baselines 2013 and 2014, and Final 
Follow-ups 2015 and 2016.
Figure V.7. Growth in boys decoding skills 
Source: Leer Juntos, Aprender Juntos Students’ Reading Skills Assessment—Baselines 2013 and 2014, and Final 
Follow-ups 2015 and 2016.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
56
At baseline, we found differences among the randomly assigned groups for boys and for 
girls (Figures V.6 and V.7 and Table J.1 in Appendix J), which were likely due to pure chance. 
Girls in the two treatment groups (Groups A and B) read about one fewer word per minute at 
baseline than the girls in the prevailing practice group (Group C), but the Leer Juntos, Aprender 
Juntos program helped them close the gap by the endline (Figure V.6). For boys, there were 
small baseline differences among all three groups, and the program helped them close the gap 
with the students in Group C, but not all the way and not enough to be statistically significant 
(Figure V.7).
To better understand the finding that the program had impacts for girls but not for boys, we 
explored whether certain aspects of the learning environment in the classroom or some family 
characteristics could explain these results. For instance, we looked at whether the seating 
arrangements in the 3rd grade classrooms could be favoring girls in Group B, but we did not find 
any statistically significant differences in classroom seating arrangements for girls or boys 
between Groups B and C.17 We also examined, for the subgroups of boys and girls, differences 
between treatment groups in baseline characteristics such as age, preschool attendance, and 
socioeconomic characteristics (parents’ education, household income, and household assets); and 
characteristics of the literacy environment at home (number of books available in the home, 
reading activities with adults and siblings, and reading activities in the community).19 
We did not find any statistically significant differences in preschool attendance, 
socioeconomic characteristics, or characteristics of the literacy environment at home for girls or 
boys between Groups B and C, with two exceptions. First, at baseline, the girls in Group B were 
about one month older, on average, than the girls in Group C. There were no differences in age at 
baseline between Groups B and C for boys. Second, fewer girls in Group B (87 percent) than in 
Group C (97 percent) spent time at home watching television or listening to the radio. We found 
no differences between boys in Group B and boys in Group C in the amount of time spent 
watching television or listening to the radio. These differences provide context to the finding of 
favorable impacts of the in-school component for girls. However, they do not explain why there 
were no statistically significant program impacts for boys. 
Because these differences do not explain the results, we looked at differences between girls 
and boys in literacy skills at baseline. At baseline (see Table J.1 in Appendix J), the girls in the 
treatment groups (Groups A and B) had slightly higher scores in assessments of emergent 
literacy skills (letter identification, emergent reading, emergent writing, phonemic awareness, 
pseudo-word decoding, and listening comprehension skills), on average, than the boys in the 
treatment groups. It is possible that the program enhanced these differences between girls and 
boys, resulting in statistically significant program impacts at the end of 3rd grade for girls but not 
for boys. Our finding of different levels of literacy skills in the early grades between girls and 
boys at baseline is consistent with the results for 2016 of the Ministry of Education’s annual 
reading and mathematics skills national evaluations (Evaluación Censal de Estudiantes [ECE]) 
for students in 2nd and 4th grade. The Ministry of Education (Ministerio de Educación 2017) 
reported that in 2016, a higher percentage of girls in 2nd (49 percent) and 4th grade (33 percent) 
than boys (44 percent in 2nd grade and 30 percent in 4th grade) performed at the level expected 
17 These findings are not presented in this report but are available from the authors upon request.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
57
for their grade in a reading test. In addition, scores on the reading test were higher, on average, 
for girls than for boys (Ministerio de Educación 2017). Our findings on the differences between 
boys and girls in reading skills are also consistent with the findings from other recent studies 
(Deasley et al. 2016; Friedlander and Goldenberg 2016, Loveless 2015; Loveless 2015; Price￾Mohr and Price 2017). In particular, a recent evaluation in Rwanda of Literacy Boost, the 
program model on which Leer Juntos, Aprender Juntos is based (Friedlander and Goldenberg 
2016), found significantly larger effects of the program on reading fluency and comprehension 
scores for girls than for boys. 
Finally, we examined in more detail the data from the focus groups and interviews with 
stakeholders conducted during our qualitative evaluation (Chapter III) and from the quarterly 
reports conducted by the program implementation team (Save the Children and Kallpa 2013, 
2014, 2015). Those examinations did not reveal any particular aspect of the program in its design 
or implementation that might have favored girls. However, during a site visit conducted in 
November 2017, program implementation staff commented that in small, rural communities in 
Peru, there is a tendency for teachers to engage boys more than girls during their classroom 
practice. A program like Leer Juntos, Aprender Juntos, which emphasizes training teachers to 
engage and encourage participation of all students, could have opened opportunities for girls to 
engage more with their teacher and other students during the reading/literacy lessons, which in 
turn could have resulted in more benefits from the program for girls than boys. 
D. Cost-effectiveness
The in-school component of Leer Juntos, Aprender Juntos generated impacts of effect sizes 
that ranged from 0.09 to 0.32 standard deviations (as shown in Table V.1), depending on which 
reading skill is being measured. This pattern of positive impacts in Peru suggests the in-school 
component of the program could be a cost-effective way to improve early grade reading, 
depending on the best alternative uses of the funding for the program. 
We estimated the cost per child of Leer Juntos, Aprender Juntos in Peru to be approximately 
$527 for the full Leer Juntos, Aprender Juntos program and $434 for the in-school component 
only. The exact cost of each of the program’s components may be lower, as we assumed that 
there were no savings from reduced use of services offered by the national and regional 
education authorities to all schools. We had to make several other assumptions, documented in 
the Appendix I. 
Despite the uncertainty about the costs of the program and the likelihood of there being 
unmeasured benefits of the program, the cost of generating the impacts presented in Table V.1 
lie in between what Evans and Ghosh (2008) refer to as the high cost interventions and what they 
call low cost interventions, described below. Their global review of education interventions of all 
kinds, where the authors arrayed the interventions by the cost of generating a 0.10 standard 
deviation impact, provide several points of comparison. We can divide the cost per child of the 
in-school component by its impact per child in standard deviation units to obtain a cost 
effectiveness of $136 per 0.10 standard deviation increase in the basic reading comprehension 
rate. This cost of a positive impact equal to an improvement of 0.10 standard deviations in the 
basic reading comprehension rate is higher than some of the cost estimates for the same size 
impact reported by Evans and Ghosh (2008), which included $19 for preschool programs in the 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
58
Philippines, $23 for class size reduction in Honduras, and $47 for school vouchers in Colombia. 
It is not surprising that to improve reading comprehension in multilingual populations that are 
geographically harder to reach would be at the higher end of the range of cost-effective 
interventions. A few other programs examined by Evans and Ghosh (2008) had higher costs than 
what we calculated for the in-school component of Leer Juntos, Aprender Juntos. Those 
programs included a psychosocial and nutrition program for stunted children in Jamaica ($296 
for an impact of 0.10 standard deviations of test scores) and student subsidies in Chile ($465 for 
the same size impact on test scores).
Finally, based on the evidence presented in this report, the community action component of 
the program is not likely a good investment—at least in the context of this particular 
implementation and evaluation effort—given that it did not generate impacts beyond those of the 
in-school component. It is difficult to pinpoint why a program with many literacy-supporting 
activities, such as reading fairs, reading buddies, and book banks, would have no measurable 
impacts on children beyond the in-school component. However, the challenges in implementing 
the community action component we documented in Chapter III can provide additional 
information to interpret that finding. For instance, it was challenging to recruit and retain 
volunteers, which had resulted in delayed implementation of the community action component 
relative to the implementation of the in-school component. In addition, the reading buddies 
activity was eventually folded into the in-school component, with teachers delivering the activity 
in their classrooms during school hours once a week instead of volunteers conducting the activity 
in the community outside of school hours. Another challenge was to fully involve parents in the 
community action component activities. For example, it was difficult to get parents to attend the 
workshops aimed at supporting them in providing an enriched print environment for their 
children at home. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
59
VI. CONCLUSIONS, LIMITATIONS, LESSONS LEARNED, AND 
RECOMMENDATIONS 
This evaluation sought to answer three questions. In this chapter, we present conclusions 
corresponding to each of those questions and discuss lessons learned and recommendations. 
A. Conclusions
1. Were the program components implemented as intended? 
Teachers received the training and individualized coaching largely as intended, but 
with some challenges. More teachers in the Leer Juntos, Aprender Juntos intervention groups 
reported receiving training and coaching than did teachers in the prevailing practice group 
(Group C). However, the differences in training rates (53 percent in Groups A and B versus 39 
percent in Group C) were not large enough to be statistically significant, but the differences in 
coaching between Groups B and C (71 versus 47 percent, respectively) were. Challenges in the 
implementation of this component included difficulty retaining qualified program staff for 
training and coaching in the in-school component, difficulty recruiting and retaining staff to lead 
the activities of the community action component, and logistical challenges for teachers, 
coaches, trainers, and volunteer coordinators due to the geography of the region and the limited 
access to public transportation. 
2. What is the impact of the teacher training and support component of Leer Juntos, 
Aprender Juntos on early-grade reading and other intermediate outcomes relative to 
prevailing practice?
Schools implementing only the in-school component of Leer Juntos, Aprender Juntos 
showed favorable changes in the classroom literacy environment. For example, teachers in 
these schools participated in the training and coaching activities offered by the program. In 
addition, more classrooms in these schools than in prevailing practice schools had a complete 
alphabet in Spanish and Quechua that was visible to students and more books with text. Our 
findings also show that teachers in schools implementing only the in-school component of Leer 
Juntos, Aprender Juntos established positive and supportive relationships with the program 
specialists who conducted training workshops and coaching sessions with them. In addition, 
teachers in schools implementing Leer Juntos, Aprender Juntos did use the reading instruction 
practices they learned from the program, and teachers reported observing increased student 
engagement.
In Peru, we also found that the program’s in-school component (teacher training and 
coaching) had favorable impacts on some, but not all, of the children’s reading skills we 
measured, and the size of those impacts is comparable to the size of the impacts other 
evaluations of reading interventions in developing countries have found. For example, three 
years after the program started, about 83 percent of the children in schools assigned to the in￾school intervention component only could read a simple passage and answer at least one question 
correctly—an indication of basic reading comprehension—compared with only 74 percent of 
children in schools using prevailing practices. This was a statistically significant positive impact. 
We also found positive, statistically significant impacts of the in-school component on decoding 
and fluency accuracy. The sizes of some of the impacts were substantial, equivalent to effect 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
60
sizes of 0.19, 0.21, and 0.13 for decoding accuracy, fluency accuracy, and reading 
comprehension, respectively. However, we found no impacts on non-accuracy reading measures, 
such as reading fluency (and decoding) measured by number of correct words read per minute 
and reading comprehension measured by number of questions answered correctly. The evidence 
of favorable impacts of the in-school component is consistent with our findings from classroom 
observations and teacher survey data: as discussed earlier in this report, the schools that 
implemented only the in-school component provided a more print-rich environment in their 
classrooms than did schools that did not implement any of the components of the intervention. 
The positive effects of the in-school component on reading skills were larger for girls 
than for boys. This does not mean that boys in Peru did not advance their reading skills or that 
they did not benefit from the program. A good example of how all the children in the evaluation 
acquired literacy skills is the change in decoding skills from the baseline to the final follow-up: 
In all three treatment groups, boys and girls went from having essentially no decoding skills at 
baseline (when children in the evaluation started the 1st grade) to decoding an average of more 
than 22 words per minute by the final follow-up assessment (at the end of the school year when 
they were in 3rd grade). At baseline, we found differences in emergent literacy skills among the 
randomly assigned groups, which were likely due to pure chance: girls in the two Leer Juntos 
groups were behind, reading about one fewer word per minute than the prevailing practice group. 
However, the program helped them close the gap by the final follow-up. For boys, there were 
small differences among all three groups at baseline, and the full Leer Juntos, Aprender Juntos
program helped them close the gap with the students in prevailing practice schools, but not all 
the way and not enough to be statistically significant. 
One possible explanation for the finding that the positive effects of the in-school component 
were larger for girls than for boys could be that the learning environment in the prevailing 
practice schools might not be gender-neutral; girls might not be encouraged to engage actively in 
reading activities to the same extent as boys. Introducing the in-school component of Leer 
Juntos, Aprender Juntos in Group B schools might have eliminated such barriers for girls, 
allowing them to surpass the performance of girls in the prevailing practice group (Group C). 
Another possible explanation could be that girls in Group B had more room to improve their 
reading skills than girls in Group C, and introducing the Leer Juntos, Aprender Juntos program 
(Groups A and B) helped the girls in the two treatment groups catch up with the girls in Group C. 
Finally, the finding of positive impacts in Peru suggests that the in-school component of 
the program could be a cost-effective way to improve early-grade reading, depending on 
the best alternative uses of the funding for the program. We estimated the cost of the in￾school component of Leer Juntos, Aprender Juntos in Peru to be approximately $434 per child. 
The exact cost of the in-school component might be lower, as we assumed there were no savings 
from reduced use of services offered by the national and regional education authorities to all 
schools. As described in Chapter V, we can also calculate the cost of the program’s in-school 
component in terms of units of improvement in children’s reading outcomes, which would reflect 
the cost effectiveness of the component. For example, we can divide the cost per child of the in￾school component ($434) by its impact per child in standard deviation units (0.32) to obtain a 
cost effectiveness of $136 per 0.10 standard deviation increase in the basic reading 
comprehension rate. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
61
3. What is the impact of the community action component of Leer Juntos, Aprender 
Juntos on early-grade reading and other intermediate outcomes relative to an 
intervention that does not have the community action component?
The community action component had some of the intermediate impacts predicted by 
the program logic model on the literacy environment in the classrooms and at home, but 
these intermediate results did not translate into impacts on students’ reading outcomes. We 
found that teachers in schools implementing the community action and the in-school components 
of Leer Juntos, Aprender Juntos participated in the training and coaching activities offered by 
the program, and reported they used what they learned from the program in their teaching 
practices. We also found that most children in the evaluation communities read at home with a 
parent or other family members. However, these positive aspects of the classroom and home 
environments did not translate into impacts of the community action component on students’ 
reading outcomes. That is, we did not find evidence that students in schools implementing both 
Leer Juntos, Aprender Juntos components had better reading skills or spent more time on 
reading activities at home than students in schools that implemented only the Leer Juntos, 
Aprender Juntos in-school component. 
Two possible explanations for the lack of impacts relate to challenges with the 
implementation of the community action activities and to the local context. First, the 
implementation of the community action component was delayed and modified in response to 
challenges. Program staff reported recruiting and retaining volunteers was challenging, which 
resulted in delayed implementation of the community action component relative to the 
implementation of the in-school component in Peru. In addition, the reading buddies activity was 
eventually folded into the in-school component, with teachers delivering the activity in their 
classrooms during regular school hours once a week instead of volunteers conducting the activity 
in the community outside of school hours. Second, the program was operating in a context where 
most parents in all treatment groups reported being literate (about three-quarters of the household 
survey respondents in each treatment group) and reading and looking at books with the children. 
Therefore, it probably was more challenging for the community action component to create an 
effect in a context where parents w ere already engaging in reading-related activities with their 
children. 
Finding no impacts of the community action component does not mean that children in 
schools offering both components of Leer Juntos, Aprender Juntos did not advance their 
reading skills. Children in schools offering both components of the program did acquire literacy 
skills: on average, children in these schools read correctly 23 pseudo-words (out of 50) in the 
decoding task and 42 words (out of 154) in the fluency task, and answered correctly 5 questions 
(out of 15) in the reading comprehension task. However, the progress of children in schools 
implementing both components of Leer Juntos, Aprender Juntos was not significantly different 
from the progress of children in schools implementing only the in-school component.
B. Limitations
This evaluation of Leer Juntos, Aprender Juntos has some limitations to bear in mind. The 
first refer to the evaluation design, specifically the timing of the baseline data collection and the 
use of a three-armed randomized design, and the last is about generalizability of the findings. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
62
The baseline of the evaluation was not a “true” baseline. The evaluation team 
administered the baseline assessment of students’ early literacy skills about three months after 
random assignment, which was about two months after the rollout of the program’s teacher 
training activities. It was not feasible to measure baseline student outcomes before random 
assignment because several evaluation activities needed to be completed before data collectors 
could begin baseline data collection; these included identifying a local partner that could assist in 
recruiting and training field workers, implementing the data collection plan, and supervising data 
collection activities. As a result of the late baseline, the children’s skills that we captured in the 
baseline assessment (when students in the evaluation were in first grade) reflect the ability of the 
students after up to three months of potential exposure to different conditions caused by 
assignment of schools to different intervention groups. 
Despite this limitation, it is still appropriate to consider the evaluation’s baseline assessment 
as the reference point for the evaluation. Under most circumstances, a late baseline will still be 
useful for estimating the impact of that intervention as long as the impact on skill development is 
slow in the early period of program implementation (Schochet 2010). That is the case in the 
current evaluation because the teachers in the groups receiving the program (Groups A and B) 
had only been practicing their newly acquired skills for at most two months when the baseline 
data collection took place, and literacy acquisition in children is a process that happens gradually 
over many months or even years. 
The evaluation design assumes that the impacts of in-school and community action 
components of the intervention are additive. Specifically, the evaluation design assumes that 
the effect of the community action component on its own relative to prevailing practice is equal 
to the sum of the impact of the combined in-school and community action components relative to 
the in-school component on its own. This assumption could be violated if implementing the in￾school component makes it easier or harder to simultaneously do community activities. We did 
not find clear evidence of such a violation, but it should be taken into account by any 
policymakers who wish to apply the lessons of this evaluation to a future implementation of 
community action on its own. 
Findings from this evaluation may not necessarily apply to other regions. It is always 
tempting to generalize from the experience of one or two evaluations to other contexts, but 
caution is warranted. The evaluation included schools from just 3 of 7 provinces Apurímac, 
which is just one of 24 departments in Peru. While we also conducted the evaluation in the El 
Quiche region of Guatemala (Lugo-Gil et al. 2021b; Lugo-Gil et al. 2021), these communities 
were selected with the following criteria in mind: the locations had to be within reasonable 
driving distance from each other to facilitate intervention and evaluation activities. They had to 
contain a high percentage of families that spoke one language besides Spanish, in this case 
K’iche’ (in Guatemala) or Quechua (in Peru). Thus, results could differ if an approach like Leer 
Juntos, Aprender Juntos was taken in other communities with different characteristics and 
challenges than those communities included in this evaluation. 
C. Lessons learned
Teacher training and coaching on how to teach literacy skills and guidance for 
applying it in their classrooms can be effective when it is distinct from other teaching 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
63
approaches already in use. In Peru, we found positive impacts of the in-school component on 
some student reading skills. Furthermore, during the qualitative evaluation, teachers mentioned 
they valued receiving training on how to provide reading instruction and prepare their own 
reading instruction materials. Teachers also expressed that the demonstration and step-by-step 
layout of Leer Juntos, Aprender Juntos strategies for reading instruction helped them apply in 
the classroom what they had learned during the program’s trainings. As discussed earlier, the 
prevailing practice approach in Peru advocates a communicative and textual approach which is 
very different from the approach of the Leer Juntos, Aprender Juntos program. In Guatemala, the 
prevailing practice approach to early reading instruction emphasizes children’s alphabet 
knowledge, phonological awareness, vocabulary, and decoding skills, and seeks to gradually 
improve children’s reading fluency and reading comprehension skills, similarly to the Leer 
Juntos, Aprender Juntos approach. Therefore, the stronger contrast between the Leer Juntos, 
Aprender Juntos approach and the prevailing practice in Peru than in Guatemala can explain why 
the in-school component of Leer Juntos, Aprender Juntos program had some impacts in Peru but 
not in Guatemala. 
Relying on volunteer work has substantial risks for fidelity of implementation. We 
learned that implementing and evaluating a program component that relies on volunteers has 
substantial risks. In the Apurímac region where the program and evaluation took place, doing 
volunteer work related to children’s schooling or learning is not common practice. In the 
communities where the program was implemented, people expect payment for work; therefore, 
recruiting and retaining volunteers was challenging. The discrepancy between the program 
requirements for volunteer participation and the cultural expectations in the communities caused 
delays and disruptions in implementing the community action component and required some 
course corrections. 
D. Recommendations
Program implementers, regional and national education authorities, and donors should 
continue implementing the in-school component of Leer Juntos, Aprender Juntos and 
possibly expand to other schools in the Apurímac region if they wish to improve literacy 
skills and do not have an alternative intervention that can deliver these impacts for this 
cost. The in-school component showed evidence of positive impacts on some reading outcomes, 
so sustaining and possibly expanding this component of the program might help further the 
region’s progress in reading achievement. In addition, the region already has the capacity and 
experience to support the sustainability of the in-school component, and teachers who 
participated in the program and the evaluation found the Leer Juntos, Aprender Juntos
methodology and training approach to be practical and feasible to apply in Peruvian classrooms. 
Nevertheless, the costs of the program and the best alternative uses of available funds are 
important to consider when thinking about continuing and expanding the program. As discussed 
earlier in this report, we estimated that the cost of the in-school component of the program in 
Peru was approximately $434 per child. In addition, to estimate the cost of the component in 
terms of improvement in children’s reading outcomes, we divided the cost per child of the 
component by its impact per child in standard deviation units to obtain a cost effectiveness of 
$136 per child per 0.10 standard deviation increase in the basic reading comprehension rate. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
64
Program implementers, regional and national education authorities, and donors should 
consider alternative ways to conduct activities that increase the time children spend 
reading outside of school hours. If there is a desire to continue testing activities that aim to 
increase time-on-task (reading) outside of school hours—such as the book banks, reading camps, 
story hour, and reading festivals that were part of the community component of Leer Juntos, 
Aprender Juntos—we recommend exploring an alternative way to implement the activities that 
does not rely primarily on volunteer work. We also recommend assessing the feasibility and 
potential take-up of the alternative approach to ensure it suits the context of the communities for 
which it is intended. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
65
REFERENCES
Banerjee, Abhijit, Shawn Cole, Esther Duflo, and Leigh Linden. “Remedying Education: 
Evidence from Two Randomized Experiments in India.” Quarterly Journal of Economics, 
vol. 122, no. 3, 2007, pp. 1235–1264. 
Banerjee, Abhijit, Rukmini Banerji, Esther Duflo, Rachel Glennerster, and Stuti Khemani. 
“Pitfalls of Participatory Programs: Evidence from a Randomized Evaluation in Education 
in India.” American Economic Journal: Economic Policy, vol. 2, 2010, pp. 1–30. 
Benavides, Martín, Magrith Mena, and Carmen Ponce. “Estado de la Niñez Indígena en el Perú.” 
Lima, Perú: Instituto Nacional de Estadística e Informática (INEI) y Fondo de las Naciones 
Unidas para la Infancia (UNICEF), 2010. 
Bruns, B., and J. Luque. “Great Teachers: How to Raise Student Learning in Latin America and 
the Caribbean.” Washington, DC: World Bank, 2014. 
Chesterfield, Ray, and Adriana Abreu-Combs. “Centers for Excellence in Teacher Training 
(CETT): Two-Year Impact Study Report (2008–2010).” Washington, DC: U.S. Agency for 
International Development, January 2011. 
Chinen, M., and J. Bonilla. “Evaluación de Impacto del Programa de Soporte Pedagógico del 
Ministerio de Educación del Perú.” Documento Técnico GRADE-FORGE. Lima, Perú: 
2017. Available at http://www.grade.org.pe/forge/descargas/Evaluación%20de%20impacto
%20Soporte%20pedagogico.pdf. Accessed August 17, 2017. 
Comings, John. “Evidence-Based Reading Projects.” Version I. Washington, DC: U.S. Agency 
for International Development, December 2012. 
Consejo Nacional de Educación. “Balance y Panorama de Cuatro Programas de Educación 
Básica: Acompañamiento Pedagógico, Colegios de Alto Rendimiento, Educación 
Intercultural Bilingüe y Jornada Escolar Completa.” Boletín del Consejo Nacional de 
Educación, Number 40. Lima: Perú: Consejo Nacional de Educación, November 2016. 
Crouch, Luis, Medina Korda, and David Mumo. “Improvements in Reading Skills in Kenya: An 
Experiment in the Malindi District.” Washington, DC: U.S. Agency for International 
Development, Bureau for Economic Growth, Agriculture and Trade, March 2009. 
Cueto, S., G. Guerrero, J. León, E. Seguín, and I. Muñoz. “Explaining and Overcoming 
Marginalization in Education: Ethnic and Language Minorities in Peru.” In Childhood 
Poverty. Multidisciplinary Approaches, edited by J. Boyden and M. Bourdillon. 
Basingstoke, UK: Palgrave Macmillan, 2012, pp. 261–282. 
Deasley, Shanna, Mary Ann Evans, Sarah Nowak, and David Willoughby. “Sex Differences in 
Emergent Literacy and Reading Behaviour in Junior Kindergarten.” Canadian Journal id 
School Psychology, 1-18, 2016. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
66
Evans, David, and Arkadipta Ghosh. “Prioritizing Educational Investments in Children in the 
Developing World.” RAND Working Paper No. WR-587, June 2008. 
Friedlander, Elliott, and Claude Goldenberg (eds.). “Literacy Boost in Rwanda: Impact 
Evaluation of a 2-year Randomized Control Trial.” Stanford, CA: Stanford University, 2016. 
Fuchs, L., D. Fuchs, M. K. Hosp, and J. Jenkins. “Oral Reading Fluency as an Indicator of 
Reading Competence: A Theoretical, Empirical, and Historical Analysis.” Scientific Studies 
of Reading, vol. 5, no. 3, 2001, pp. 239–256. 
Glazerman, Steven, Larissa Campuzano, Nancy Murray, and Camila Fernandez. “Final Leer 
Juntos, Aprender Juntos Evaluation Plan.” Washington, DC: Mathematica Policy Research, 
March 22, 2013. 
Gove, A., and P. Cvelich. “Early Reading: Igniting Education for All: A Report by the Early 
Grade Learning Community of Practice, Revised Edition.” Research Triangle Park, NC: 
Research Triangle Institute, 2011.
Guerrero, Gabriela, Juan Leon, Elizabeth Rosales, Mayli Zapata, Silvana Freire, Víctor 
Saldarriaga, and Santiago Cueto. “Young Lives: School Survey in Peru.” Young Lives 
Working Paper No. 92. Oxford, UK: Young Lives, December 2012. 
Hernandez-Zavala, Martha, Harry Anthony Patrinos, Chris Sakellariou, and Joseph Shapiro. 
“Quality of Schooling and Quality of Schools for Indigenous Students in Guatemala, 
Mexico, and Peru.” World Bank Policy Research Working Paper 3982. Washington, DC: 
World Bank, August 2006. 
Ikeda, Miyako, and Emma García. “Grade Repetition: A Comparative Study of Academic and 
Non-Academic Consequences.” OECD Journal: Economic Studies, vol. 2013, no. 1, 2014. 
Instituto Nacional de Estadística e Informática. “Idioma o Lengua con el que Aprendió Hablar.” 
Lima, Perú: Censo de Población y Vivienda, 2007. 
Kim, Y.-S. G., H. N. Boyle, S. S. Zuikowski, and P. Nakamura. “Landscape Report on Early 
Grade Literacy.” Washington, DC: U.S. Agency for International Development, 2016. 
Loveless, Tom. “How Well are American Students Learning? With sections on the gender gap in 
reading, effects of the Common Core, and student engagement.” The 2015 Brown Center 
Report on American Education, vol. 3, no.4, Washington, DC: Brookings, March 2015. 
Lugo-Gil, Julieta, Kara Contreary, Nancy Murray, Steven Glazerman, Larissa Campuzano, 
Camila Fernandez, and Kate Place. “Evaluation of Leer Juntos, Aprender Juntos Early 
Grade Intervention in Guatemala: Final baseline report.” Washington, DC: U.S. Agency for 
International Development, 2016a. Available at 
https://pdf.usaid.gov/pdf_docs/PA00MMTD.pdf. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
67
Lugo-Gil, Julieta, Kate Place, Nancy Murray, Steven Glazerman, Larissa Campuzano, Camila 
Fernandez. “Evaluation of Leer Juntos, Aprender Juntos Early Grade Intervention in Peru: 
Final baseline report.” Washington, DC: U.S. Agency for International Development, 
2016b. Available at https://pdf.usaid.gov/pdf_docs/PA00MMTF.pdf. 
Lugo-Gil, Julieta, Nancy Murray, Camila Fernandez, Steven Glazerman, Ivonne Padilla, and 
Larissa Campuzano, “Evaluation of Leer Juntos, Aprender Juntos Early Grade Intervention 
in Peru: Midline report.” Washington, DC: U.S. Agency for International Development, 
2017a. Available at https://pdf.usaid.gov/pdf_docs/PA00X21P.pdf. 
Lugo-Gil, Julieta, Nancy Murray, Camila Fernandez, Steven Glazerman, and Larissa 
Campuzano, “Evaluation of Leer Juntos, Aprender Juntos Early Grade Intervention in 
Guatemala: Midline report.” Washington, DC: U.S. Agency for International Development, 
2017b. Available at https://pdf.usaid.gov/pdf_docs/PA00X21Q.pdf. 
Lugo-Gil, Julieta, Nancy Murray, Camila Fernandez, Steven Glazerman, and Larissa 
Campuzano, “Evaluation of Leer Juntos, Aprender Juntos Early Grade Intervention in 
Guatemala: Final report.” Washington, DC: U.S. Agency for International Development, 
2021b. 
Lugo-Gil, Julieta, Nancy Murray, and Steven Glazerman, “Evaluation of Leer Juntos, Aprender 
Juntos Early Grade Intervention in Peru and Guatemala: Summary report.” Washington, 
DC: U.S. Agency for International Development, 2021. 
Ministerio de Educación. “Diseño Curricular Nacional de Educación Básica Regular.” Lima, 
Perú: Dirección General de Educación Básica Regular, 2008. 
Ministerio de Educación. “Evaluación Censal de Estudiantes (ECE). Segundo Grado de Primaria 
y Cuarto Grado de Primaria de IE EIB: Marco de Trabajo.” Lima, Perú: Secretaría de 
Planificación Estratégica, Unidad de Medición de la Calidad Educativa, Ministerio de 
Educación, 2009. 
Ministerio de Educación. “Programa Nacional de Redes Educativas Rurales.” Lima, Perú: 
Ministerio de Educación, 2012. 
Ministerio de Educación. “Hacia una Educación Intercultural Bilingüe de Calidad. Propuesta 
Pedagógica.” Lima, Perú: Dirección General de Educación Intercultural Bilingüe y Rural, 
2013. 
Ministerio de Educación. “Presentación de Resultados de la Evaluación Censal de Estudiantes: 
ECE 2016.” Lima, Perú: Oficina de Medición de la Calidad de los Aprendizajes (UMC)-
Ministerio de Educación, 2017. Available at http://umc.minedu.gob.pe/resultadosece2016/. 
Accessed December 22, 2017. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
68
National Reading Panel. “Teaching Children to Read: An Evidence-Based Assessment of the 
Scientific Research Literature on Reading and Its Implications for Reading Instruction.” 
National Institute of Child Health and Human Development, NIH Publication No. 00-4769. 
Washington, DC: U.S. Government Printing Office, 2000. 
Patrinos, Harry Anthony, and Eduardo Velez. “Costs and Benefits of Bilingual Education in 
Guatemala: A Partial Analysis.” International Journal of Education Development, vol. 29, 
2009, pp. 594–598. 
Piper, Benjamin, Stephanie S. Zuikowski, and Salome Ong’ele. “Implementing Mother Tongue 
Instruction in the Real World: Results from a Medium Scale Randomized Controlled Trial in 
Kenya.” Comparative Education Review, vol. 60, no. 4, November 2016, pp. 776–807. 
Price-Mohr, Ruth, and Colin Price. “Gender Differences in Early Reading Strategies: A 
Comparison of Synthetic Phonics Only with a Mixed Approach to Teaching Reading to 4-5 
Year-Old Children.” Early Childhood Education Journal, 45, 2017, pp. 613-620. 
RTI International. Early Grade Reading Assessment (EGRA) Toolkit, Second Edition. 
Washington, DC: U.S. Agency for International Development, 2015. 
Save the Children. “Literacy Boost Toolkit: Teacher Training.” Washington, DC: Save the 
Children, 2012a. 
Save the Children. “Literacy Boost Toolkit: Community Action.” Washington, DC: Save the 
Children, 2012b. 
Save the Children. “Beyond the Classroom Walls: A Boost for Readers.” Fairfield, CT: Save the 
Children, 2013a. 
Save the Children. “Leer Juntos, Aprender Juntos Performance Management Plan” Washington, 
DC: Save the Children, 2013b. 
Save the Children, and Kallpa. “Leer Juntos, Aprender Juntos Quarterly Report, September 
2012–December 2012 (Quarter 1).” Guatemala City, Guatemala and Lima, Peru: Save the 
Children, 2012. 
Save the Children, and Kallpa. “Leer Juntos, Aprender Juntos Quarterly Reports, January 2013–
December 2013 (Quarters 2–5).” Guatemala City, Guatemala and Lima, Peru: Save the 
Children, 2013. 
Save the Children, and Kallpa. “Leer Juntos, Aprender Juntos Quarterly Reports, January 2014–
December 2014 (Quarters 6–9).” Guatemala City, Guatemala and Lima, Peru: Save the 
Children, 2014. 
Save the Children, and Kallpa. “Leer Juntos, Aprender Juntos Quarterly Reports, January 2015–
December 2015 (Quarters 10–13).” Guatemala City, Guatemala and Lima, Peru: Save the 
Children, 2015. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
69
Schochet, Peter. “The Late Pretest Problem in Randomized Control Trials of Education 
Interventions.” Journal of Educational and Behavioral Statistics, vol. 35, no. 4, 2010, pp. 
379–406. 
Vargas Dávila, José Luis. “El Programa Presupuestal Estratégico Logros de Aprendizaje: 
Avances, Tensiones, Perspectivas.” Lima, Perú: Tarea Asociación de Publicaciones 
Educativas, 2009. 
APPENDIX A:
TIMELINE OF EVALUATION ACTIVITIES IN PERU, 2013–2016 
This page has been left blank for double-sided printing.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
A.1 
Timeline of evaluation activities in Peru
Planning for the start of implementation of the Leer Juntos, Aprender Juntos program 
activities in Groups A and B in Peru began in fall 2012 with the development of a work plan and 
project strategy, hiring of program implementation staff, a training workshop for key 
implementation staff from Save the Children, and meetings of key stakeholders, such as U.S. 
Agency for International Development, local education authorities, and relevant local 
government entities (Save the Children and Kallpa 2012, 2013). Preparing for rolling out the 
program in Peru continued in winter and early spring 2013 with activities that included recruiting 
and orienting technical and project administration staff, developing an implementation 
monitoring and evaluation system, mapping elements of the teacher training with the national 
curriculum, translating project materials to Quechua, producing reading materials for local book 
banks, and recruiting schools to participate in the evaluation. 
Figure A.1 shows the timeline of Leer Juntos, Aprender Juntos activities in Peru from 2013 
to 2016. The school year in Peru begins in March and ends in December. Random assignment for 
the Phase I cohort took place in April 2013. The Leer Juntos, Aprender Juntos intervention 
rolled out in Peru in May 2013 with the first teacher training sessions for Phase I teachers in 
Groups A and B. 
Figure A.1. Timeline of Leer Juntos, Aprender Juntos activities 2012–2016 in 
Peru
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
A.2 
The teacher training in Phase I schools continued in 2013 with sessions conducted in July, 
October, and December. Data collection for the baseline of the Phase I cohort occurred in 
September and October 2013, about two to three months after teacher training began.18
Recruitment and training of the community volunteers took place in August 2013 in Group A 
Phase I communities, and implementation of the community-based intervention activities such as 
reading banks, story hour, reading buddies, and reading camps began in September 2013. 
For the Phase II cohort, random assignment occurred in March 2014. The first teacher 
training sessions for Phase II teachers in the treatment groups also took place in March 2014 
(Phase I teachers also participated in this training session), and teacher training for Phase II 
teachers continued with sessions in April, July, and December. Data collection for the baseline of 
the Phase II cohort took place in April and May 2014.The first training of community volunteers 
for Phase II took place in May 2014, as did the start of implementation of the community action 
components in Group A. Data collection for the midline of the Phase I cohort took place in 
November and December 2014. The midline data collection for the Phase II cohort occurred 
from August through October 2015. 
The teacher training and support (coaching) activities, as well as training of teacher leaders 
and principals, continued throughout 2014 and 2015 for the Phase I and II cohorts in the 
treatment groups (A and B). Recruitment and training of community volunteers and community 
reading activities also continued in 2014 and 2015 for the Phase I and II communities in Group 
A. The implementation of the teacher training and support and community components of the 
intervention (Groups A and B) ended in December 2015 for the Phase I and II cohorts. The final 
data collection (endline) took place in September and October 2015 for the Phase I cohort and in 
August and September 2016 for the Phase II cohort. 
18 Although it would have been ideal to have the baseline data collected before the rollout of teacher training in 
Phase I schools, it was not possible to procure a local data collection firm in the time between reaching an agreement 
with stakeholders on the evaluation design (April 2013) and the rollout of the teacher training (May 2013). We
submitted and agreed upon the evaluation design with stakeholders in April 2013, and only after that we were able 
to share the evaluation design in the terms of reference in the request for proposals from prospective data collectors. 
We also had to allow time to prepare and review proposals and, after securing a data collection partner, GRADE, 
time for them to recruit and train interviewers. For Phase II schools we were able to accomplish baseline data 
collection closer to the time of teacher training rollout.
APPENDIX B:
QUALITATIVE DATA COLLECTION
This page has been left blank for double-sided printing
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
B.1 
Qualitative data collection
Table B.1 provides detailed information on the respondent types, methods, topics of inquiry, 
and location for all qualitative data collection activities.
Table B.1. Respondents, methods, topics of inquiry, and location in the 
qualitative data collection
Respondents Number of focus 
groups / interviews 
(number of 
participants)
Topics of inquiry Location
Provincial education 
unit (UGEL) 
directors and 
education 
specialists
3 interviews with 
directors and 4 
interviews with 
specialists
● Challenges in reading instruction
● Knowledge about Leer Juntos, Aprender Juntos
and other education interventions in the region
Lima
Save the Children 
staff
2 interviews
(Project manager; 
Monitoring and 
Evaluation 
Coordinator) 
● Program leadership and management 
● Literacy Boost adaptation to Peruvian context
● Program implementation strategy
● Facilitators and barriers to implementation
Lima
Kallpa program 
implementation 
specialists 
1 focus group 
(4 female; 4 male 
participants)
● The role and activities of specialists
● Rollout of program components
● Facilitators and barriers to implementation
● Perception of program participation and take-up 
Andahuaylas
Community 
volunteers
4 focus groups and 
3 interviews
(12 female; 2 male 
participants)
● The role and activities of volunteers
● Motivation and barriers to volunteer
● Perception of program participation and take-up
Andahuaylas
Program teachers 
(in schools in 
Groups A and B)
4 focus groups
(23 female; 13 male 
participants)
● Program knowledge and expectations
● Program participation and take-up
● Barriers to implementation of program activities 
● Changes in reading instruction practices
● Perception of program activities 
● Perception of program outcomes
Andahuaylas, 
Chincheros, 
Huancarama
Prevailing practice 
teachers (in schools 
in Group C)
1 focus group
(5 female 
participants)
● Reading instruction practices and challenges
● In-service training 
● Participation in reading instruction programs
Chincheros
Parents of children 
in schools 
implementing the 
program 
8 focus groups
(38 female; 9 male 
participants)
● Program knowledge and expectations
● Program participation and take-up
● Barriers to participation
● Perceived benefits of program services
Andahuaylas, 
Chincheros, 
Huancarama
Observations of 
community activities
6 observations ● Context of community activities (location, 
leaders, volunteers, and participants)
● Types of activities
● Participant engagement
Andahuaylas, 
Huancarama
This page has been left blank for double-sided printing.
APPENDIX C:
CONTROL VARIABLES IN CHILD-LEVEL REGRESSION ANALYSES
This page has been left blank for double-sided printing.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
C.1 
Control variables in child-level regression analyses
Table C.1 lists the control variables included in the regression models we used to estimate 
program impacts on the reading skills in the 3rd grade of the children in the evaluation. The table 
also indicates whether the variable is dichotomous or continuous, as well as the source of the 
variable. All the variables in this list are unlikely to be affected by the intervention, and they 
were measured at baseline or at the midline data collections in the evaluation. 
Table C.1. Control variables included in the analyses of the reading skills of 
3rd grade children in the evaluation
Level Variables Type Source 
Student ● Child age at final follow-up assessment (in months)
● Child sex
Continuous Classroom rosters
Student ● Baseline letter identification score
● Baseline emergent reading score
● Baseline emergent writing score
● Baseline phonemic awareness score
● Baseline number of pseudo-words read in one minute
● Baseline passage comprehension score
Continuous Leer Juntos, 
Aprender Juntos
(LJAJ) Emergent 
Literacy Skills 
Assessment, 
Baseline 2013 and 
2014
Student ● Child was assessed in Spanish only at baseline
● Child assessed in both Spanish and in mother tongue at 
baseline
Binary LJAJ Emergent 
Literacy Skills 
Assessment, 
Baseline 2013 and 
2014
Student ● Child attended preschool or kindergarten
● Child’s mother completed at least one year of formal 
schooling
● Child’s parents are able to read
● Language spoken at home is mother tongue only
Binary LJAJ Household 
Survey, Midline 2014 
and 2015
Student ● Number of people living in the household
● Number of rooms in the household
● Family monthly income in dollars
Continuous LJAJ Household 
Survey, Midline 2014 
and 2015
Student ● Family owns land
● Household has floor built with finished materials
● Household has roof built with finished materials
● Household has electricity
● Household has phone service
● Household has a radio
● Household has a television
● Household has a refrigerator
● Household has a computer
● Household has a bicycle
● Household has a motorcycle
● Household has a car or truck
Binary LJAJ Household 
Survey, Midline 2014 
and 2015
School ● At baseline, school participated in the One Laptop per Child 
program 
● At baseline, school participated in the National Strategic 
Learning Achievement Program (PELA)
● At baseline, school participated in the National Rural 
Education Networks Program (REDES)
● At baseline, school received education materials from the 
Ministry of Education
Binary LJAJ School 
Infrastructure 
Observation, Baseline 
2013 and 2014
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
C.2 
Level Variables Type Source 
● At baseline, school participated in the National School 
Nutrition Program (Qali Warma)
● School has multi-grade classrooms
● School has library or resource room
● School has computers for students
● School has computers for teachers
● School has Internet connectivity
● School has phone service
● School has piped water supply
● School has unpiped water supply
● School has potable drinking water
● School has electricity
● School is a permanent building
● School has painted interior walls
● School has painted exterior walls
● School classrooms have ceilings
● School has perimeter fence
● School has working restroom facilities for children
● School has latrines or urinals
● School has functional toilets
● School has separate facilities for boys and girls
● School has hand-washing facilities
● School has working waste disposal
● School has plumbing waste disposal
● School has septic or similar disposal
● School has other waste disposal
● School has kitchen
● School has vegetable garden
● School has teachers’ lounge
● School has outdoors recreational space
● School has gymnasium or sports facilities
● School has infirmary
School ● Number of physical hazards observed in school
● Number of health hazards observed in school
Continuous LJAJ School 
Infrastructure 
Observation, Baseline 
2013 and 2014
For details on the construction of variables from baseline and midline data sources, refer to 
the Leer Juntos, Aprender Juntos baseline (Lugo-Gil et al. 2016a) and midline (Lugo-Gil et al. 
2017a) reports, respectively. All the variables listed in Table C.1 except age at baseline and sex, 
had missing values. One hundred and fifty six observations were missing data in control 
variables, which reduces the size of the analytic sample from 1,022 children to 866 children. To 
retain in the analysis all the observations that have reading outcome data (1,022 children), we 
imputed the missing values in control variables. To do imputations, we conducted regressions 
using each of the variables in the list (nonmissing data) as the dependent variable. In the 
regressions for the baseline literacy skills and child socioeconomic characteristics variables, we 
used baseline age and baseline age squared as the independent variables. In the regressions for 
the rest of the variables (household assets and school characteristics and resources), we used as 
independent variables the subgroup of other variables from that list that contained the least 
amount of missing data. All of the regression models we used to estimate program impacts on 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
C.3 
children’s reading outcomes included a binary variable indicating, for each child in the sample, 
whether imputations were done on any of the control variables. 
This page has been left blank for double-sided printing.
APPENDIX D:
SCHOOL INFRASTRUCTURE CHARACTERISTICS
This page has been left blank for double-sided printing.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
D.1 
School infrastructure characteristics
The utilities and facilities in evaluation schools are basic (Tables D.1), but most schools 
have running water, electricity, and painted interior walls. For example, although all of the 
schools have classrooms covered with roofs (result not in tables), only about half of the schools 
in each treatment group are located in a permanent building. 
Table D.1. Available utilities in evaluation schools (percentages of schools)
Utilities
Leer
Juntos
(A)
Leer 
Juntos￾school only
(B)
Prevailing 
practice
(C) (A–B) (B–C)
Permanent building 46.9 46.9 49.0 0.0
(0.997)
-2.2
(0.826)
Painted interior walls 96.6 99.8 95.5 -3.2
(0.323)
4.3
(0.189)
Painted exterior walls 38.6 47.0 48.0 -8.3
(0.383)
-1.1
(0.911)
Perimeter fence 63.4 57.5 68.2 5.9
(0.524)
-10.8
(0.252)
Finished (cement, wood, etc.) classroom floors 95.9 100.0 100.0 -4.2
(0.086)
0.0
(1.000)
Electricity 96.2 97.8 97.8 -1.6
(0.658)
0.0
(1.000)
Piped water supply 79.3 79.2 85.6 0.1
(0.989)
-6.5
(0.386)
Potable drinking water supply 57.2 57.3 56.2 -0.0
(0.999)
1.1
(0.914)
Unpiped water supply 22.8 23.0 12.2 -0.2
(0.983)
10.8
(0.179)
Plumbing waste disposal 26.6 31.4 24.9 -4.8
(0.566)
6.5
(0.445)
Septic (or similar) disposal 71.3 62.4 66.7 8.9
(0.324)
-4.3
(0.637)
Number of schools 49 47 48
Source: Leer Juntos, Aprender Juntos School Infrastructure Observation Form—Final Follow-ups 2015 and 2016.
Note: Regression-adjusted means. P-values in parentheses. None of the differences between treatment groups 
presented in this table is statistically significant at the 0.05 level.
Almost all of the schools in each treatment group have working restroom facilities for 
children (Table D.2). However, few schools have amenities such as sports facilities, a music 
room, infirmary, or teachers’ lounge. Importantly, we found one difference between treatment 
groups in the schools’ available utilities and facilities: more schools in Group B than in Group A 
have a vegetable garden. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
D.2 
Table D.2. Other facilities available in schools
Facilities
Leer
Juntos
(A)
Leer 
Juntos￾school 
only
(B)
Prevailing 
practice
(C) (A–B) (B–C)
Working restroom facilities for children 97.9 97.9 100.0 0.0
(0.989)
-2.2
(0.386)
Separate restroom facilities for boys and girls 67.3 71.4 70.4 -4.2
(0.669)
1.1
(0.914)
Latrines or urinals 83.7 74.7 83.3 9.0
(0.222)
-8.6
(0.248)
Functional toilets 73.1 83.1 79.9 -9.9
(0.209)
3.2
(0.686)
Hand-washing facilities 87.6 91.8 89.7 -4.2
(0.533)
2.2
(0.752)
Kitchen 97.9 91.5 98.0 6.4
(0.132)
-6.5
(0.131)
Vegetable garden 59.4 80.3 76.0 -20.9*
(0.021)
4.3
(0.634)
Teachers' lounge 6.2 10.4 10.4 -4.2
(0.488)
0.0
(1.000)
Outdoor recreational space 94.2 97.9 91.5 -3.7
(0.416)
6.5
(0.168)
Gymnasium or sports facilities 0.0 2.2 0.0 -2.1
(0.221)
2.2
(0.219)
Infirmary 0.0 2.2 0.0 -2.1
(0.221)
2.2
(0.219)
Computers for students 80.1 87.5 79.9 -7.4
(0.336)
7.5
(0.334)
Computers for teachers 65.5 68.1 69.2 -2.6
(0.782)
-1.1
(0.909)
Internet connectivity 20.4 28.3 20.8 -7.9
(0.265)
7.5
(0.295)
Library or resource room 15.1 8.7 19.4 6.4
(0.366)
-10.8
(0.136)
Music room 2.1 2.1 2.1 -0.0
(1.000)
0.0
(1.000)
Number of schools 49 47 48
Source: Leer Juntos, Aprender Juntos School Infrastructure Observation Form—Final Follow-ups 2015 and 2016.
Note: Regression-adjusted means. P-values in parentheses.
* Difference in group means is statistically significant at the 0.05 level.
As presented in Table D.3, on average, there were no statistically significant differences 
between treatment groups in the number of hazards—related to either physical (such as broken 
glass or furniture, dangerous materials, and unsafe building structures) or health (such as 
stagnant water reservoirs) factors—observed in the schools. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
D.3 
Table D.3. Physical and health hazards observed in schools
Hazards
Leer
Juntos
(A)
Leer 
Juntos￾school 
only
(B)
Prevailing 
practice
(C) (A–B) (B–C)
Number of physical hazards observed 
(average)
3.3 2.8 3.0 0.5
(0.169)
-0.2
(0.514)
Number of health hazards observed (average) 2.2 2.0 2.0 0.2
(0.414)
-0.1
(0.774)
Number of schools 49 47 48
Source: Leer Juntos, Aprender Juntos School Infrastructure Observation Form—Final Follow-ups 2015 and 2016.
Note: Regression-adjusted means. P-values in parentheses. None of the differences between treatment groups 
presented in this table is statistically significant at the 0.05 level.
This page has been left blank for double-sided printing.
APPENDIX E:
CLASSROOM AND HOUSEHOLD ENVIRONMENTS
This page has been left blank for double-sided printing.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
E.1 
Classroom and household environments
In this section, we present additional information on school, classrooms, and households in 
Peru that supports the findings presented in Chapter IV. 
1. Schools’ participation in programs different to Leer Juntos, Aprender Juntos
We measured school participation in other programs besides Leer Juntos, Aprender Juntos
to help explore the setting in which these schools operate and gather information on what the 
prevailing practice offers. Schools in the evaluation participate in a number of education and 
nutrition programs. In particular, we gathered information about school participation in the 
following programs:
● Una Laptop por Niño (One Laptop per Child), an initiative of the Ministry of Education in 
Peru that aims to integrate the use of technological resources in the education process by 
providing a laptop to every student. 
● Programa Estratégico Logros de Aprendizaje al finalizar el III ciclo de la Educación Básica 
Regular, or PELA (Strategic Learning Outcomes Program), a national program that aims to 
improve reading and mathematics achievement of students in preschool and grades 1 and 2. 
The program provides resources to improve school governance efficacy, increase teachers’ 
participation in professional development activities, improve schools’ infrastructure, and 
increase access to high quality early education programs for all children. 
● Programa Nacional de Redes Educativas Rurales, or REDES (National Rural Education 
Network Program), a national program that focuses on improving the academic achievement 
of students in rural schools by incorporating computer use in the classroom and by providing 
resources to enhance teachers’ performance.
● Educational materials provided by the Ministry of Education in Peru. The materials include 
any type of educational or instructional material, and they may or may not be provided in 
relation to any of the three programs described earlier in this list.
● Programa Nacional de Alimentación Escolar, also known as Qali Warma (National 
School Nutrition Program), an initiative of the Ministry of Social Development 
(Ministerio de Desarrollo e Inclusión Social) in Peru. This program provides a free 
breakfast to students in preschool and elementary public schools. In addition, students 
attending schools in the most disadvantaged districts receive breakfast and lunch. This 
program is not included in the programs listed in Table III.1 in Chapter III, but we 
included participation in this program at baseline to control for the reading outcomes 
impact analysis. 
● Soporte Pedagógico (Pedagogical Support), one of two initiatives that the Ministry of 
Education launched in 2015. Soporte Pedagógico targets urban areas, and the other 
initiative, Soporte Pedagógico Intercultural (Intercultural Pedagogical Support), aims to 
serve bilingual schools located primarily in rural areas. The goal of these two initiatives 
is to improve student learning in elementary schools by (1) providing supports to 
teachers; (2) developing students’ communication, math, and social studies skills that 
require additional support; (3) improving allocation and use of appropriate teaching 
materials and educational resources; and (4) building capacity of principals, teachers, and 
parents to ensure that students learn. These initiatives also employ teachers who work 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
E.2 
after regular school hours (docentes fortaleza) to provide remedial support to students 
who are falling behind. 
2. Participation in professional development activities 
Table E.1. Teacher-reported participation in reading instruction professional 
development (PD) activities in the endline year (percentages of teachers)
Blank cell
Leer
Juntos
(A)
Leer Juntos
school-only
(B)
Prevailing 
practice
(C)
Impact of 
community 
action
(A–B)
Impact of 
in-school 
component
(B–C)
Teachers who participated in reading instruction 
professional development activitiesa
52.8 53.0 39.0 -0.2
(0.981)
14.0
(0.128)
Teachers who reported they received reading 
instruction professional development from:
Ministry of Education or other institution 27.9 36.2 39.5 -8.3
(0.371)
-3.2
(0.730)
Save the Children through Leer Juntos, 
Aprender Juntos
38.4 36.4 0.0 2.0
(0.759)
36.4*
(0.000)
Teachers who reported they received coaching 63.2 70.9 47.2 -7.7
(0.384)
23.7*
(0.009)
Teachers who reported they received in-class 
coaching from:
Ministry of Education or other institution 46.6 54.9 47.5 -8.3
(0.383)
7.4
(0.443)
Save the Children through Leer Juntos, 
Aprender Juntos
38.4 40.6 0.0 -2.2
(0.742)
40.6
(0.000)
*
Total number of teachers 49 47 48 . . 
Source: Leer Juntos, Aprender Juntos Teacher Survey—Final Follow-ups 2015 and 2016.
Note: Regression-adjusted means. P-values in parentheses.
a The percentages shown refers to the participation during the year when the endline survey was administered (2015 
for Phase I teachers and 2016 for Phase II teachers).
* Difference in group means is statistically significant at the 0.05 level.
Considering the entire period of program implementation, the participation of intervention 
teachers (Groups A and B) in PD activities was considerable. For instance, in the midline data 
collection (November–December 2014 in Phase I and September–October 2015 in Phase II), 
more than 80 percent of the intervention teachers (89 percent in Group A and 84 percent in 
Group B) reported they participated in PD activities in the 12 months before the midline survey 
administration (Lugo-Gil et al. 2017a). Also, during the year before the endline survey 
administration (2014 for Phase I and 2015 for Phase II teachers), the percentage of teachers in 
the intervention groups who reported participating in PD activities was close to 80 percent, and 
the difference in the participation rate with teachers in Group C was 51 percentage points, which 
is statistically significant. We present the findings on teachers’ reports of participation in PD 
activities in the year before the endline teacher survey in Table E.2. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
E.3 
Table E.2. Teacher-reported participation in reading instruction professional 
development (PD) activities in the year before endline survey administration 
(percentages of teachers)
Blank cell
Leer
Juntos
(A)
Leer 
Juntos￾school 
only
(B)
Prevailing 
practice
(C)
Impact of 
community 
action
(A–B)
Impact of in￾school 
component
(B–C)
Teachers who participated in reading 
instruction PD activities
79.2 84.6 34.1 -5.5
(0.522)
50.5*
(0.000)
Teachers who reported they received coaching 76.0 82.6 55.8 -6.7
(0.439)
26.9*
(0.003)
Number of teachers 49 47 48
Source: Leer Juntos, Aprender Juntos Teacher Survey—Final Follow-ups 2015 and 2016.
Note: Regression-adjusted means. P-values in parentheses.
* Difference in group means is statistically significant at the 0.05 level.
At the endline or final follow-up, teachers in the intervention groups (Groups A and B) also 
reported higher rates of participation in reading instruction PD activities focused on topics that 
are specifically addressed in the Leer Juntos, Aprender Juntos training. As we show in Figure 
E.1, more than 40 percent of the teachers in Groups A and B reported receiving training on 
teaching letter knowledge, reading fluency, reading comprehension, vocabulary, and 
phonological awareness. In contrast, only 19 to 37 percent of the teachers in the prevailing 
practice group reported participating in PD focused on teaching specific literacy skills. The 
difference between Groups B and C is statistically significant for all types of PD focused on 
teaching specific literacy skills, except for reading comprehension for which the differences 
between groups range from 2 to 16 percentage points but were not statistically significant. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
E.4 
Figure E.1. Teachers in intervention groups reported higher rates of 
participation in professional development (PD) activities focused on specific 
literacy skills
Source: Leer Juntos, Aprender Juntos Teacher Survey— Final Follow-ups 2015 and 2016.
* Difference in group means is statistically significant at the 0.05 level.
Teachers also reported participation in PD activities focused on other topics related to 
reading instruction in the endline or final follow-up year: as shown in Table E.3, approximately 
one-third of teachers in each of the intervention groups reported participation in PD focused on 
teaching reading in Quechua or teaching bilingual students. In contrast, only 13 percent of 
teachers in the prevailing practice group reported participating in training focused on reading in 
Quechua or teaching bilingual children. At least 40 percent of teachers in each of the intervention 
groups reported participating in PD focused on making or adapting materials to teach reading, 
and on creating a suitable classroom environment for learning to read. None of these differences 
between treatment groups was statistically significant (see Table E.3). 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
E.5 
Table E.3. Teacher-reported participation in reading instruction professional 
development (PD) activities focused on specific topics in the endline year
Blank cell
Leer
Juntos
(A)
Leer Juntos￾school only
(B)
Prevailing 
practice
(C)
Impact of 
community 
action
(A–B)
Impact of 
in-school 
component
(B–C)
Percentage of teachers who participated in other PD focused on:
Teaching reading in Quechua or 
teaching bilingual students
35.5 28.5 13.4 7.1
(0.406)
15.1
(0.081)
Support for literacy activities at home 38.7 35.3 20.2 3.4
(0.707)
15.1
(0.104)
Classroom management 31.4 28.9 23.5 2.5
(0.797)
5.4
(0.588)
Making or adapting materials to teach 
reading
40.4 40.7 24.5 -0.3
(0.979)
16.1
(0.094)
Creating a suitable classroom 
environment for learning reading
40.3 33.1 27.8 7.2
(0.448)
5.4
(0.575)
Use of regular evaluations 42.7 27.3 16.5 15.5
(0.084)
10.8
(0.233)
Parent/community participation in 
reading activities
32.8 23.0 12.3 9.7
(0.262)
10.8
(0.220)
Other curricular areas 30.7 30.0 36.5 0.6
(0.947)
-6.5
(0.496)
Management of multiple-grade 
classrooms
11.4 8.2 9.3 3.1
(0.606)
-1.1
(0.861)
Other related to literacy instruction 5.5 10.2 11.3 -4.7
(0.434)
-1.1
(0.859)
Total number of teachers 49 47 48 . . 
Source: Leer Juntos, Aprender Juntos Teacher Survey—Final Follow-ups 2015 and 2016.
Note: Regression-adjusted means. P-values in parentheses.
* Difference in group means is statistically significant at the 0.05 level.
3. Teacher-reported practices
At the endline data collection in Peru, we surveyed teachers about the methods, strategies, 
materials, and allocation of instruction time they use to teach reading, and about the needs and 
challenges they face in teaching their students to read. 
In that teacher survey, we asked open-ended questions to teachers about the methods they 
used more frequently to teach reading to their students in 3rd grade. In the survey, teachers also 
reported on the strategies they usually use when they teach reading fluency, new words, and 
reading comprehension. Table E.4 shows the methods and strategies that teachers mentioned 
most frequently in their answers to the survey. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
E.6 
Table E.4. Teacher-reported methods and strategies to teach reading
Blank cell
Leer
Juntos
(A)
Leer 
Juntos￾school only
(B)
Prevailing 
practice
(C)
Impact of 
community 
action
(A–B)
Impact of 
in-school 
compone
nt
(B–C)
Percentage of teachers who reported using the following methods to teach reading:
Syllabic 14.4 9.2 14.5 5.3
(0.411)
-5.4
(0.409)
Communicative-textual 36.9 35.3 37.4 1.6
(0.877)
-2.2
(0.837)
Phonological/global 16.5 12.3 14.5 4.2
(0.569)
-2.2
(0.773)
Guided reading 32.4 17.2 19.4 15.1
(0.064)
-2.2
(0.793)
Strategies for reading comprehension 16.3 22.1 14.6 -5.8
(0.471)
7.5
(0.359)
Percentage of teachers who reported using the following strategies to teach
Reading fluency
Reading with chronometer 40.7 30.4 22.9 10.3
(0.230)
7.5
(0.385)
Reading aloud in groups or pairs 44.9 56.4 52.1 -11.5
(0.251)
4.3
(0.671)
Guided oral reading 33.1 26.3 28.4 6.8
(0.472)
-2.2
(0.822)
Reading in chain 30.4 35.5 37.7 -5.2
(0.589)
-2.2
(0.824)
Silent reading 18.7 23.0 14.4 -4.3
(0.612)
8.6
(0.317)
Strategies for reading comprehension 12.1 7.9 4.7 4.1
(0.445)
3.2
(0.554)
New words
Discover the name of things using the dictionary 81.0 85.2 88.4 -4.1
(0.573)
-3.2
(0.662)
Synonyms and antonyms 28.3 32.1 25.6 -3.7
(0.697)
6.5
(0.507)
Build sentences and phrases with word lists 1.7 1.7 2.8 0.0
(0.995)
-1.1
(0.712)
Reading comprehension
Ask questions about the reading 67.6 68.5 77.1 -0.9
(0.926)
-8.6
(0.382)
Identify the sequence of events 22.0 1.6 7.0 20.4*
(0.001)
-5.4
(0.384)
Predictions about history or pre-reading 
reflections
6.3 16.7 14.6 -10.5
(0.118)
2.2
(0.749)
Abstracts, paraphrasing, or storytelling 22.5 39.0 18.6 -16.5
(0.072)
20.4*
(0.028)
Identify characters, problems, and solutions 7.6 13.3 17.6 -5.7
(0.406)
-4.3
(0.532)
Number of teachers 49 47 48 . . 
Source: Leer Juntos, Aprender Juntos Teacher Survey—Final Follow-ups 2015 and 2016.
Note: Regression-adjusted means. P-values in parentheses.
* Difference in group means is statistically significant at the 0.05 level.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
E.7 
More than 30 percent teachers in all treatment groups reported using the communicative￾textual method to teach reading, whereas less than 20 percent of teachers in all treatment groups 
reported using the phonological/global approach. We found no statistically differences between 
treatment groups in the methods teachers reported using to teach reading. However, we found 
statistically significant differences between treatment groups in two of the strategies teachers use 
to teach reading comprehension. As shown in Table E.4, more teachers in Group A than in 
Group B reported they identify the sequence of events while teaching reading comprehension (22 
and 2 percent, respectively). Additionally, more teachers in Group B than teachers in Group C 
reported using abstracts, paraphrasing, or storytelling to teach reading comprehension. 
We also asked teachers about other literacy-related practices they use in their classrooms, 
such as the types of reading materials students use and the frequency with which they assess their 
students’ literacy skills. More teachers in Group C than in Group B reported their students use 
picture books every day (Table E.5). That difference (22 percentage points) is statistically 
significant. In contrast, a higher percentage of teachers in the intervention groups than in the 
prevailing practice group reported their students use books that provide reading opportunities, 
such as simple reading books and workbooks or exercises guides. Although these differences are 
not statistically significant, these results are encouraging because the use of reading books 
strengthens the development of decoding, fluency, and reading comprehension skills of 3rd grade 
students. 
Table E.5. Students’ use of different types of text, as reported by teachers 
Blank cell
Leer
Juntos
(A)
Leer Juntos￾school only
(B)
Prevailing 
practice
(C)
Impact of 
community 
action
(A–B)
Impact of in￾school 
component
(B–C)
Percentage of teachers who reported their students use the following types of text every day:
School textbooks 45.8 43.2 44.3 2.6
(0.799)
-1.1
(0.918)
Workbooks or exercise guides 28.7 37.8 22.7 -9.1
(0.344)
15.1
(0.123)
Picture or artwork books 24.4 10.0 31.5 14.4
(0.097)
-21.5
(0.015)
*
Books to name objects or letters of the 
alphabet
14.5 23.8 28.1 -9.3
(0.243)
-4.3
(0.593)
Simple reading books 24.2 23.2 16.8 0.9
(0.916)
6.5
(0.475)
Cards with syllables or words 13.1 13.2 8.9 -0.1
(0.991)
4.3
(0.497)
Instructional materials issued by the 
Ministry of Education
3.8 16.8 19.0 -13.0
(0.058)
-2.2
(0.754)
Books with chapters 2.0 0 4.2 2.0
(0.375)
-4.3
(0.082)
Number of teachers 49 47 48
Source: Leer Juntos, Aprender Juntos Teacher Survey—Final Follow-ups 2015 and 2016.
Note: Regression-adjusted means. P-values in parentheses.
* Difference in group means is statistically significant at the 0.05 level.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
E.8 
Teachers also reported on the time their students spend on reading-related activities during a 
typical language class (Table E.6). Those in Group B reported their students spent 8 more 
minutes (on average) narrating familiar stories than students in Group C, and that they spent 
more time, on average, on motivational activities than students in Group A. These differences are 
statistically significant. According to their teachers, students in the intervention groups (Groups 
A and B) spent a similar amount of time (on average) in other activities related to reading during 
communication class such as learning to identify the letters of the alphabet, phonemic awareness, 
reading aloud, building vocabulary, and taking dictation. 
Table E.6. Number of minutes students spend on the following activities 
during a typical language and reading class, as reported by teachers
Activity 
Leer
Juntos
(A)
Leer 
Juntos￾school 
only
(B)
Prevailing 
practice
(C)
Impact of 
community 
action
(A–B)
Impact of in￾school 
component
(B–C)
Learning to identify the letters of the alphabet 16.7 17.2 17.5 -0.5
(0.848)
-0.2
(0.937)
Practicing phonemic awareness activities 17.1 19.3 16.6 -2.3
(0.348)
2.7
(0.275)
Practicing reading fluency 26.6 30.1 28.7 -3.5
(0.201)
1.4
(0.614)
Practicing reading comprehension 35.7 37.9 36.6 -2.2
(0.527)
1.3
(0.714)
Building vocabulary and learning new words 25.1 24.8 24.8 0.3
(0.920)
0.0
(1.000)
Narrating familiar stories without reference to 
text
29.0 30.2 22.4 -1.2
(0.721)
7.8*
(0.024)
Copying letters or words 22.4 25.8 23.3 -3.4
(0.173)
2.5
(0.330)
Taking dictation 20.9 21.7 20.7 -0.8
(0.694)
1.0
(0.638)
Motivational activities 10.9 14.9 11.9 -4.1*
(0.038)
3.0
(0.123)
Number of teachers 49 47 48
Source: Leer Juntos, Aprender Juntos Teacher Survey—Final Follow-ups 2015 and 2016.
Note: Regression-adjusted means. P-values in parentheses.
* Difference in group means is statistically significant at the 0.05 level.
Interviewers also asked teachers about the needs and challenges they face in teaching 
reading. The most mentioned barrier by teachers in intervention groups (Groups A and B) was 
lack of support from parents. However, there were no statistically significant differences between 
treatment groups in that report. In contrast, between teachers in Groups B and C, the difference 
in the reports of lack of training in methods to teach literacy as a barrier to teach (a difference of 
19 percentage points) was statistically significant. Table E.7 shows the teachers’ reports on the 
resources that they perceive as lacking when they teach reading to their students and the issues 
they reported as challenges to teach reading.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
E.9 
Table E.7. Teacher-reported needs and challenges (percentages of teachers)
Needs and challenges
Leer
Juntos
(A)
Leer 
Juntos￾school 
only
(B)
Prevailing 
practice
(C)
Impact of 
community 
action
(A–B)
Impact of in￾school 
component
(B–C)
Lack of supports and resources for teaching reading and writing
Lack of books 41.4 56.4 66.0 -15.0
(0.152)
-9.7
(0.358)
Lack of teaching resources 67.2 62.3 76.2 4.9
(0.596)
-14.0
(0.137)
Lack of sufficient instructional or class time 32.4 31.5 27.1 1.0
(0.919)
4.3
(0.658)
Lack of training in methods to teach literacy 71.3 64.2 83.6 7.1
(0.424)
-19.4*
(0.033)
Lack of other types of training 64.1 62.5 63.6 1.6
(0.876)
-1.1
(0.916)
Lack of support from parents 81.7 76.5 74.4 5.2
(0.570)
2.2
(0.815)
Lack of training in bilingual intercultural 
education
76.8 59.1 61.2 17.7
(0.079)
-2.2
(0.832)
Lack of classroom equipment or material 67.2 58.7 67.3 8.5
(0.429)
-8.6
(0.427)
Lack of Quechua language teaching skills 20.7 18.6 16.5 2.0
(0.792)
2.2
(0.784)
Other -0.0 -0.1 6.3 0.1
(0.972)
-6.5*
(0.031)
Challenges to teaching reading
Students’ absenteeism 41.5 56.6 27.5 -15.0
(0.113)
29.0
(0.003)
*
Students' lack of motivation 49.9 40.8 54.8 9.1
(0.400)
-14.0
(0.201)
Students’ lack of Spanish-language 
knowledge
24.8 16.3 22.8 8.4
(0.321)
-6.5
(0.453)
Lack of parents’ involvement in school 
activities
81.8 90.3 80.6 -8.5
(0.244)
9.7
(0.189)
Students’ malnutrition or poor health of 
students
81.8 88.2 78.5 -6.4
(0.405)
9.7
(0.214)
Students’ vision problems 41.8 50.9 33.7 -9.1
(0.356)
17.2
(0.087)
Number of teachers 49 47 48
Source: Leer Juntos, Aprender Juntos Teacher Survey—Final Follow-ups 2015 and 2016.
Note: Regression-adjusted means. P-values in parentheses.
* Difference in group means is statistically significant at the 0.05 level.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
E.10
4. Observed practices
To measure observed teaching practice, the evaluation team observed the instruction in the 
3rd grade classrooms during a typical language or reading class. On average, teachers in the 
three treatment groups were observed for 39 minutes (there were no statistically significant 
differences between treatment groups in the average time of the classroom observations). 
Across all treatment groups, teachers spent a majority of the observed literacy 
instruction time on two activities: reading texts and reading comprehension. Figure E.2 
shows that teachers in all treatment groups spent about 35 percent of the observed time in 
activities that involved reading text such as reading aloud or in groups or individually in a silent 
manner. The observed teachers also spent 24 to 29 percent of the observed time identifying 
characters and event sequences and posing questions about the text contents to students with the 
goal of promoting reading comprehension. We did not find statistically significant differences 
between treatment groups on how teachers spent the time during the language or reading lesson 
that we observed, with one exception: teachers in classrooms in Group B spent a larger 
proportion of the observed time (10 percent) than teachers from Group A (3 percent) on other 
activities to practice reading such as drawing, playing, and showing visual representations of a 
text.
Figure E.2. Observed allocation of instruction time in a typical language or 
reading class
Source: Leer Juntos, Aprender Juntos Classroom Observation Form—Final Follow-ups 2015 and 2016.
* Difference in group means between Groups A and B is statistically significant at the 0.05 level.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
E.11
Teachers in the three treatment groups were similar in terms of their use of 
instructional practices focused on teaching the five foundational literacy skills. In the 
evaluation’s classroom observations, observers recorded whether teachers used instructional 
practices focused on teaching the five foundational literacy skills that were the focus of the 
teacher training provided by Leer Juntos, Aprender Juntos: (1) alphabet/letter knowledge, (2) 
vocabulary, (3) phonemics and phonological knowledge, (4) fluency, and (5) reading 
comprehension. As expected, most teachers in all treatment groups were observed providing 
instruction on reading fluency and activities to guide students in using reading comprehension 
strategies, such as helping students understand the text by making predictions and asking 
questions about events and encouraging predictions or inferences. In contrast, the percentage of 
teachers observed providing instruction on letter knowledge, phonemic awareness, 
pronunciation, and segment of words was less than 15 percent in all treatment groups. 
We found only one statistically significant difference between treatment groups in teachers’ 
use of vocabulary-building activities, as shown in Table E.8: the percentage of classrooms in 
which teachers introduced or rehearsed vocabulary words and explained their meaning was 
higher in Group A than in Group B (68 and 38 percent, respectively). 
Teachers’ use of general teaching practices such as providing verbal reinforcement and 
feedback to students was similar across treatment groups. We found two statistically 
significant differences between treatment groups in the teachers’ use of general instructional 
practices: more teachers in Group A and Group C than teachers in Group B asked open-ended 
questions to students with the goal of stimulating oral language, as presented in Table E.9. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
E.12
Table E.8. Observed instructional practices to teach foundational literacy skills 
Practice
Leer
Juntos
(A)
Leer 
Juntos￾school 
only
(B)
Prevailing 
practice
(C)
Impact of 
community 
action
(A–B)
Impact of 
in-school 
component
(B–C)
Percentage of classrooms where teachers used the following practices to teach:
Vocabulary
Introduced or rehearsed vocabulary words and 
explained word meaning or elaborated on new 
concepts
67.4 37.6 48.3 29.9*
(0.002)
-10.8
(0.269)
Fluency 96.2 87.7 95.2 8.5
(0.097)
-7.5
(0.144)
Letter knowledge 
Encouraged students to recognize letters, 
identify the names of letters, or distinguish 
upper and lower case
8.3 15.7 12.4 -7.3
(0.235)
3.2
(0.605)
Phonemics and phonological awareness 
Provided instruction on phonemic awareness 10.0 15.3 13.1 -5.2
(0.443)
2.2
(0.756)
Provided instruction on the correct pronunciation 
of words in Quechua or Spanish
11.0 8.9 8.9 2.1
(0.729)
0.0
(1.000)
Prompted students to segment words into 
specific units (syllables or phonemes)
1.8 10.1 6.9 -8.4
(0.081)
3.2
(0.504)
Provided instruction on grammar, mechanics, or 
spelling
45.9 38.2 29.6 7.7
(0.423)
8.6
(0.375)
Provided instruction on word writing 64.9 78.7 63.7 -13.8
(0.161)
15.1
(0.130)
Book reading and reading comprehension
Pre-reading or book-preview 64.9 69.6 67.4 -4.7
(0.658)
2.2
(0.842)
Encouraged predictions 36.5 31.2 35.5 5.3
(0.605)
-4.3
(0.677)
Explored children’s interest in the story and 
facilitated connections with their lives
56.9 51.7 48.5 5.2
(0.600)
3.2
(0.746)
Explained the uses of different types of texts 
and written materials
10.0 17.9 12.6 -7.9
(0.281)
5.4
(0.468)
Encouraged students to use context clues for 
reading comprehension
36.2 33.2 27.8 3.0
(0.750)
5.4
(0.577)
Discussed text structure 12.4 14.5 14.5 -2.1
(0.765)
0.0
(1.000)
Discussed the characters in the text, who they 
are, their motivation and/or goals
54.9 49.8 38.0 5.0
(0.641)
11.8
(0.279)
Asked questions about events in history, 
encourages predictions or inferences
87.2 76.6 86.3 10.6
(0.169)
-9.7
(0.212)
Helped students understand the text by making 
predictions, applying prior knowledge
85.5 77.6 81.9 7.9
(0.293)
-4.3
(0.569)
Used other tools in addition to books for the 
practice of reading skills
34.5 29.4 20.8 5.1
(0.478)
8.6
(0.235)
Reinforced the pedagogical objective with 
playful or motivational activities
30.7 25.0 20.7 5.7
(0.514)
4.3
(0.624)
Number of classrooms 49 47 48
Source: Leer Juntos, Aprender Juntos Classroom Observation Form—Final Follow-ups 2015 and 2016.
Note: Regression-adjusted means. P-values in parentheses.
* Difference in group means is statistically significant at the 0.05 level.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
E.13
Table E.9. Observed general instructional practices
Blank cell
Leer
Juntos
(A)
Leer 
Juntos￾school 
only
(B)
Prevailing 
practice
(C)
Impact of 
community 
action
(A–B)
Impact of in￾school 
component
(B–C)
Percentage of classrooms in which teachers used the following practices to:
Stimulate oral language
Gave students opportunities to respond 
to the print/text materials
53.1 50.0 50.0 3.1
(0.766)
0.0
(1.000)
Asked open-ended questions 71.9 48.1 70.7 23.8*
(0.010)
-22.6*
(0.016)
Provide verbal reinforcement to students 94.1 85.2 89.5 8.9
(0.139)
-4.3
(0.478)
Provide appropriate feedback to students 97.9 95.8 95.8 2.1
(0.549)
0.0
(1.000)
Number of classrooms 49 47 48
Source: Leer Juntos, Aprender Juntos Classroom Observation Form—Final Follow-ups 2015 and 2016.
Note: Regression-adjusted means. P-values in parentheses.
* Difference in group means is statistically significant at the 0.05 level.
5. Language use in the classroom 
Spanish was the spoken language that teachers used most often to communicate with their 
students, but we did not find statistically significant differences in teachers’ language use 
between treatment groups. As shown in Table E.10, teachers in at least 85 percent of the 
classrooms in each treatment group were observed using only Spanish or mostly Spanish to 
discipline and manage students’ behavior, and most of the observed teachers in all treatment 
groups used Spanish to provide literacy instruction. In contrast, teachers were observed using 
only Quechua or mostly Quechua to communicate with their students in only 2 to 6 percent of 
the classrooms in each treatment group. 
Table E.10. Observed teachers’ use of language in the classroom
Blank cell
Leer
Juntos
(A)
Leer 
Juntos￾school 
only
(B)
Prevailing 
practice
(C)
Impact of 
community 
action
(A–B)
Impact of in￾school 
component
(B–C)
Classrooms in which teachers were observed using only Spanish or mostly Spanish to:
Discipline and manage students’ behavior 91.7 85.1 93.9 6.5
(0.285)
-8.8
(0.156)
Provide literacy instruction 93.8 91.7 93.8 2.1
(0.685)
-2.2
(0.684)
Classrooms in which teachers were observed using only Quechua or mostly Quechua to:
Discipline and manage students’ behavior 4.1 2.0 2.0 2.1
(0.489)
0.0
(1.000)
Provide literacy instruction 6.2 2.0 2.0 4.2
(0.228)
0.0
(1.000)
Number of classrooms 49 47 48
Source: Leer Juntos, Aprender Juntos Classroom Observation Form—Final Follow-ups 2015 and 2016.
Note: Regression-adjusted means. P-values in parentheses. None of the differences between evaluation means 
presented in this table is statistically significant.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
E.14
In most of the observed classrooms, at least half of the students used Spanish to 
communicate with their teacher, but we found no statistically significant differences between 
treatment groups in the language use of students in the classroom (Table E.11). 
Table E.11. Observed students’ use of language in the classroom
Blank cell
Leer
Juntos
(A)
Leer 
Juntos￾school 
only
(B)
Prevailing 
practice
(C)
Impact of 
community 
action
(A–B)
Impact of in￾school 
component
(B–C)
Classrooms in which half or more of the students speak to the teacher in:
Spanish 93.8 93.7 98.0 0.1
(0.987)
-4.3
(0.316)
Quechua 12.4 16.6 14.5 -4.2
(0.579)
2.2
(0.779)
Number of classrooms 49 47 48
Source: Leer Juntos, Aprender Juntos Classroom Observation Form—Final Follow-ups 2015 and 2016.
Note: Regression-adjusted means. P-values in parentheses. None of the differences between evaluation means 
presented in this table is statistically significant.
6. Classroom environment 
Table E.12 shows the books and materials in Spanish and Quechua observed in the 3rd 
grade classrooms in the evaluation. 
Table E.12. Books and materials observed in the classroom
Books and materials
Leer
Juntos
(A)
Leer 
Juntos￾school 
only
(B)
Prevailing 
practice
(C)
Impact of 
community 
action
(A–B)
Impact of in￾school 
component
(B–C)
Percentage of classrooms in which:
More than 25 books in Spanish were 
observed
92.0 80.5 82.7 11.5
(0.122)
-2.2
(0.773)
More than 25 books in Quechua were 
observed
79.6 72.4 68.1 7.2
(0.449)
4.3
(0.656)
10 or more of the following were observed:
In Spanish 
Printed materials 30.0 19.0 21.1 11.0
(0.167)
-2.2
(0.788)
Handmade materials prepared by the 
teacher
79.6 65.4 70.8 14.1
(0.134)
-5.4
(0.572)
Handmade materials prepared by the 
students
45.2 50.0 43.5 -4.8
(0.621)
6.5
(0.511)
In Quechua 
Printed materials 0.0 2.1 2.1 -2.1
(0.396)
0.0
(1.000)
Handmade materials prepared by the 
teacher
10.3 6.2 4.1 4.1
(0.425)
2.2
(0.682)
Handmade materials prepared by the 
students
6.2 4.2 0.0 2.0
(0.599)
4.2
(0.269)
Number of classrooms 49 47 48
Source: Leer Juntos, Aprender Juntos Classroom Observation Form—Final Follow-ups 2015 and 2016.
Note: Regression-adjusted means. P-values in parentheses. None of the differences between evaluation means 
presented in this table is statistically significant. 
APPENDIX F:
STANDARD DEVIATIONS OF CHILDREN’S READING SKILLS MEASURES
This page has been left blank for double-sided printing.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
F.1 
 Standard deviations of children’s reading skills measures
Table F.1 presents the standard deviations of the reading skills measures (Chapter V) we 
used to examine the impacts of the components of Leer Juntos, Aprender Juntos. 
Table F.1. Standard deviations of children’s reading skills outcome measures
Outcome
Leer
Juntos
(A)
Leer Juntos￾school only
(B)
Prevailing 
practice
(C)
Decoding 
Number of pseudo-words read correctly in one minute (of 
50)
13.49 13.63 13.70
Accuracy score (0 to 1) 0.30 0.31 0.31
Fluency
Number of words read correctly in one minute (of 154) 28.36 26.15 27.71
Accuracy score (0 to 1) 0.31 0.31 0.32
Reading comprehension
Number of questions answered correctly (of 15) 3.71 3.60 3.71
Percentage of children who answered at least one 
question in the comprehension task correctly (achieved 
basic reading comprehension capabilities)
0.40 0.41 0.41
Number of children 358 338 326
Source: Leer Juntos, Aprender Juntos Students’ Literacy Skills Assessment—Final Follow-ups 2015 and 2016.
This page has been left blank for double-sided printing.
APPENDIX G:
SENSITIVITY ANALYSES
This page has been left blank for double-sided printing.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
G.1 
Sensitivity analyses
As mentioned in Chapter V, we conducted robustness checks on two important assumptions 
we made in the estimation of program impacts. First, we re-weighted the data to account for 
differences in the number of children who completed the 3rd grade literacy skills assessment 
(also called “respondents”) and avoid underweighting students whose classmates were absent. 
We reweighted the data of each child i in the 3rd grade in school s using the following formula: 
𝑤𝑤𝑒𝑒𝑖𝑖𝑖𝑖ℎ𝑡𝑡𝑖𝑖𝑖𝑖 = 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑒𝑒𝑁𝑁 𝑜𝑜𝑜𝑜 𝑠𝑠𝑡𝑡𝑁𝑁𝑠𝑠𝑒𝑒𝑠𝑠𝑡𝑡𝑠𝑠 𝑖𝑖𝑠𝑠 𝑡𝑡ℎ𝑒𝑒 𝑠𝑠𝑠𝑠𝑁𝑁𝑠𝑠𝑠𝑠𝑒𝑒 𝑖𝑖𝑠𝑠 𝑠𝑠𝑠𝑠ℎ𝑜𝑜𝑜𝑜𝑠𝑠 𝑠𝑠
𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑒𝑒𝑁𝑁 𝑜𝑜𝑜𝑜 𝑁𝑁𝑒𝑒𝑠𝑠𝑠𝑠𝑜𝑜𝑠𝑠𝑠𝑠𝑒𝑒𝑠𝑠𝑡𝑡𝑠𝑠 𝑡𝑡𝑜𝑜 𝑡𝑡ℎ𝑒𝑒 𝑠𝑠𝑠𝑠𝑠𝑠𝑒𝑒𝑠𝑠𝑠𝑠𝑁𝑁𝑒𝑒𝑠𝑠𝑡𝑡 𝑖𝑖𝑠𝑠 𝑠𝑠𝑠𝑠ℎ𝑜𝑜𝑜𝑜𝑠𝑠 𝑠𝑠
By reweighting data that way, our results on regression-adjusted mean differences between 
treatment groups will correspond to the mean reading outcomes of the average respondent in the 
schools in the sample. The results do not vary when we estimate program effects based on the 
reweighted data. Table G.1 presents the regression-adjusted means of the literacy assessment 
skills we assessed in 3rd grade, by treatment group. As the table shows, we found that the main 
result presented in Chapter V still holds when we account for students’ nonresponse: the 
percentage of children who have basic reading comprehension skills in the 3rd grade is larger in 
Group B (the schools that implemented the in-school component of the Leer Juntos, Aprender 
Juntos program) than in Group C (the schools in the prevailing practice group that did not 
implement any of the program components). 
Table G.1. Differences between treatment groups in children’s literacy 
outcomes, accounting for nonresponse to the 3rd grade assessment
Outcome
Leer
Juntos
(A)
Leer 
Juntos￾school 
only
(B)
Prevailing 
practice
(C)
Impact of 
community 
action
(A–B)
Impact of in￾school 
component
(B–C)
Decoding 
Number of pseudo-words read correctly in one 
minute (of 50)
23.1 22.9 21.0 0.2
(0.836)
1.9
(0.067)
Accuracy score (0 to 100) 69.3 70.3 64.3 -1.0
(0.665)
6.0*
(0.020)
Fluency
Number of words read correctly in one minute 
(of 154)
41.6 41.4 38.7 0.2
(0.915)
2.7
(0.225)
Accuracy score (0 to 100) 73.6 78.5 71.7 -4.9*
(0.035)
6.8*
(0.014)
Reading comprehension
Number of questions answered correctly (of 
15)
4.7 4.8 4.3 0.1
(0.874)
0.5
(0.088)
Percentage of children who answered at least 
one question in the comprehension task 
correctly (achieved basic reading 
comprehension capabilities)
78.6 83.4 74.3 -4.8
(0.213)
9.1*
(0.014)
Number of children 358 338 326
Source: Leer Juntos, Aprender Juntos Students’ Literacy Skills Assessment—Final Follow-ups 2015 and 2016.
Note: Regression-adjusted means. P-values in parentheses.
* Difference in group means is statistically significant at the 0.05 level.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
G.2 
For a second check on the robustness of our assumptions, we also calculated differences 
between treatment groups in mean literacy outcomes in the 3rd grade based on the sample of 
children that has complete data for all the control variables included in the regression model. The 
sample with complete data in literacy outcome variables and control variables includes 866 
children. In comparison, the sample that has complete data on literacy outcome variables but has 
incomplete data in control variables (for which we imputed in the analyses presented in Chapter 
V) includes 1,022 children. Using complete data reduces the sample size by about 15 percent. As 
Table G.2 shows, the findings do not change when we use a sample with complete data. In 
particular, under this alternative assumption we also found that the percentage of children who 
acquired basic reading comprehension skills in the 3rd grade is larger in the schools that 
implemented the in-school component of the program than in the schools that did not receive the 
program. 
Table G.2. Differences between treatment groups in children’s literacy 
outcomes, based on sample with complete data
Outcome
Leer
Juntos
(A)
Leer 
Juntos￾school 
only
(B)
Prevailing 
practice
(C)
Impact of 
community 
action
(A–B)
Impact of in￾school 
component
(B–C)
Decoding 
Number of pseudo-words read correctly in one 
minute (of 50)
23.3 23.7 20.8 -0.4
(0.692)
2.9*
(0.002)
Accuracy score (0 to 100) 68.8 72.5 63.5 -3.7
(0.097)
9.0*
(0.002)
Fluency
Number of words read correctly in one minute 
(of 154)
42.3 41.9 38.5 0.4
(0.837)
3.4
(0.101)
Accuracy score (0 to 100) 73.9 79.4 71.0 -5.5*
(0.030)
8.4*
(0.002)
Reading comprehension
Number of questions answered correctly (of 
15)
4.7 4.7 4.4 0.0
(0.996)
0.3
(0.248)
Percentage of children who answered at least 
one question in the comprehension task 
correctly (achieved basic reading 
comprehension capabilities)
79.8 84.2 72.5 -4.4
(0.275)
11.7*
(0.002)
Number of children 301 283 282
Source: Leer Juntos, Aprender Juntos Students’ Literacy Skills Assessment—Final Follow-ups 2015 and 2016.
Note: Regression-adjusted means. P-values in parentheses.
* Difference in group means is statistically significant at the 0.05 level.
We also examined program impacts on literacy outcomes for subgroups of students defined 
by phase (cohort). Table G.3 presents the findings from that analysis. In Phase II, children in 
Group B had better decoding and fluency accuracy scores than their counterparts in Group C. 
Also, in Phase II, the percentage of children who have a basic level of competency in reading 
comprehension was larger than in Group C. We did not find any pattern of statistically 
significant impacts of the community action component for either Phase I or Phase II children. 
Finally, we did not find any statistically significant differences in the program impacts on 
children’s reading outcomes between Phase I and Phase II.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
G.3 
Table G.3. Impacts on the literacy outcomes of children in Phase I and Phase 
II schools
Outcome
Leer
Juntos
(A)
Leer 
Juntos￾school 
only
(B)
Prevailing 
practice
(C)
Impact of 
community 
actiona 
(A–B)
Impact of in￾school 
componentb 
(B–C)
Children in Phase I schools
Decoding
Number of pseudo-words read correctly 
in one minute (of 50)
24.5 22.1 21.9 2.46
(0.119)
0.2
(0.896)
Accuracy score (0 to 100) 71.0 68.8 65.5 2.2
(0.501)
3.3
(0.346)
Fluency
Number of words read correctly in one 
minute (of 154)
45.2 38.8 39.3 6.4
(0.051)
-0.5
(0.869)
Accuracy score (0 to 1) 76.0 75.9 71.5 0.1
(0.971)
4.4
(0.269)
Reading comprehension
Number of questions answered correctly 
(of 15)
5.1 4.6 4.4 0.4
(0.239)
0.3
(0.453)
Percentage of girls who achieved basic 
reading comprehension skills
79.6 78.9 74.4 0.7
(0.905)
4.5
(0.412)
Children in Phase II schools
Decoding
Number of pseudo-words read correctly 
in one minute (of 50)
22.2 22.9 19.8 -0.7
(0.580)
3.1*
(0.027)
Accuracy score (0 to 100) 68.3 70.6 62.7 -2.3
(0.474)
7.9*
(0.010)
Fluency
Number of words read correctly in one 
minute (of 154)
39.5 42.4 37.8 -2.9
(0.264)
4.6
(0.123)
Accuracy score (0 to 1) 72.3 79.8 72.0 -7.6*
(0.012)
7.8*
(0.021)
Reading comprehension
Number of questions answered correctly 
(of 15)
4.6 4.8 4.2 -0.2
(0.518)
0.6
(0.120)
Percentage of boys who achieved basic 
reading comprehension skills
79.3 86.6 74.3 -7.3
(0.105)
12.3*
(0.009)
Number of children in Phase I schools 185 178 183
Number of children in Phase II schools 173 160 143
Source: Leer Juntos, Aprender Juntos Students’ Reading Skills Assessment—Final Follow-ups 2015 and 2016.
Note: Regression-adjusted means. P-values in parentheses.
aThe differences between Phase I and Phase II in impacts of the community of action component were not 
statistically significant for any reading outcomes.
bThe differences between Phase I and Phase II in impacts of the in-school component were not statistically significant 
for any reading outcomes.
* Difference in group means is statistically significant at the 0.05 level.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
G.4 
Finally, Table G.4 presents our findings on the effects of the full intervention (the 
community action component and the teacher training and coaching component) on children’s 
reading outcomes. We found statistically significant differences between children in schools 
implementing the full intervention (Group A) and children in schools implementing the 
prevailing practice (Group C): children in Group A read more pseudo-words per minute and had 
higher decoding accuracy scores than children in Group C. The percentage of children who 
achieved a basic level of reading comprehension was higher in Group A than in Group C, but 
this difference is not statistically significant. Also, children in Group A read correctly more 
words in the fluency task and answered correctly more questions in the reading comprehension 
task than children in Group C, and those differences are statistically significant at the 0.10 level 
(p = 0.086 for the difference in the fluency task and p = 0.055 for the difference in the reading 
comprehension task). 
Table G.4. Impacts of the full intervention
Outcome
Leer
Juntos
(A)
Prevailing 
practice
(C)
Impact of full 
intervention
(A–C)
Decoding 
Number of pseudo-words read correctly in one minute 
(of 50)
23.2 20.9 2.3*
(0.019)
Accuracy score (0 to 100) 69.5 64.1 5.4*
(0.016)
Fluency
Number of words read correctly in one minute (of 154) 42.0 38.6 3.4
(0.086)
Accuracy score (0 to 100) 73.8 71.7 2.1
(0.362)
Reading comprehension
Number of questions answered correctly (of 15) 4.8 4.3 0.5
(0.055)
Percentage of children who answered at least one 
question in the comprehension task correctly 
(achieved basic reading comprehension capabilities)
79.0 74.3 4.8
(0.162)
Number of children 358 326
Source: Leer Juntos, Aprender Juntos Students’ Literacy Skills Assessment—Final Follow-ups 2015 and 2016.
Note: Regression-adjusted means. P-values in parentheses.
* Difference in group means is statistically significant at the 0.05 level.
APPENDIX H: 
EFFECTS OF THE COMMUNITY COMPONENT FOR BOYS AND GIRLS
This page has been left blank for double-sided printing.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
H.1 
Effects of the community component for boys and girls
As described in Chapter V, we did not find any statistically significant differences between 
Groups A and B (effect of the community component) in the literacy outcomes of the 
evaluation’s girls and boys in the 3rd grade. Table H.1 presents the differences between Groups 
A and B in the means of the literacy outcomes of the evaluation’s boys and girls, expressed in 
standard deviation units (effect sizes). 
Table H.1. Community component effects on literacy outcomes for girls and 
boys, in standard deviation units
Outcome Effect size for girls Effect size for boys
Decoding
Number of pseudo-words read correctly in one minute -0.06 0.13
Accuracy score (0 to 1) -0.09 0.06
Fluency
Number of words read correctly in one minute -0.03 0.09
Accuracy score (0 to 1) -0.19 -0.10
Reading comprehension
Number of questions answered correctly 0.03 -0.01
Percentage of children achieved basic reading 
comprehension skills
-0.38 0.02
Source: Leer Juntos, Aprender Juntos Students’ Literacy Skills Assessment—Final Follow-ups 2015 and 2016.
Note: Effect sizes are calculated as the difference in adjusted means between two treatment groups (in this table, 
Groups A and B) divided by the pooled and weighted standard deviation of the two groups (Hedges’ g) for 
continuous outcome measures, and as the log odds ratio divided by 1.65 (Cox index) for dichotomous 
outcome measures. None of the effect sizes presented in this table is statistically significant at the 0.05 
level.
This page has been left blank for double-sided printing. 
APPENDIX I: 
THE COSTS OF LEER JUNTOS, APRENDER JUNTOS
This page has been left blank for double-sided printing.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
I.1 
The costs of Leer Juntos, Aprender Juntos
We estimated the cost per child of Leer Juntos, Aprender Juntos in Peru to be approximately 
$527 for the full Leer Juntos, Aprender Juntos program and $434 for the in-school component 
only. We estimated these values using a combination of expenditure data and program 
implementation records supplied by Save the Children, the implementer of the program. The 
approach was to begin with the total cost of the program based on the size of the grant awarded 
to Save the Children and allocate those costs to the two countries in which the program operated 
(Peru and Guatemala), the two types of activities (in-school and community), and finally, to the 
correct number of children who were served. We tried to account for fixed costs that were 
specific to operating the program at this particular scale of 197 schools across the two countries. 
The goal was to estimate a marginal cost per child that would be comparable to the impact per 
child that we estimated using the evaluation sample of children who were in the 1st grade at the 
point of random assignment. Marginal cost is the cost to provide services to one additional child 
if the program is operational. This contrasts with average cost, which spreads the cost of central 
office staff and program start-up over the number of children being served. We had to make 
several assumptions, as follows.
● The only difference in costs between Group C schools and the intervention schools 
(Groups A and B) were those associated with the Leer Juntos, Aprender Juntos program. 
For example, any routine teacher training provided by the Ministry of Education was 
offered to all three groups equally and participation in Leer Juntos did not offset any of 
these costs. This assumption may overstate the costs of Leer Juntos if there were such 
offsets.
● All of the costs of Leer Juntos, Aprender Juntos were accounted for in the Save the 
Children budget. Specifically, no donated goods or services enhanced the program 
offerings. In fact, the program as implemented in Group A (which included the 
community component) relied on community volunteers. When we attempted to estimate 
the dollar value of time for program volunteers, even generous estimates resulted in 
trivial amounts when divided by the number of children. Many of the volunteers were 
parents or older siblings, whose time was difficult to quantify using a shadow price. To 
the extent that such volunteer hours were in fact taken away from productive activities, 
this assumption may understate the full cost of Leer Junto, Aprender Juntos. 
● The ratio of costs of the program as offered to Group A versus Group B (with or without 
the community component) was the same for Guatemala as it was for Peru. We had to 
make this assumption because the expenditure data were sufficiently detailed for 
Guatemala to make distinctions between the in-school and community components, 
whereas for Peru they were not. In Peru, many services were subcontracted with an 
organization whose costs appeared as a line item for which we could not determine the 
nature of the activities. The ratio of Group A to Group B costs that we estimated using 
Guatemala expenditure data was 1.87. We applied this ratio to the total costs in Peru. 
● The duration of treatment for all children is the same as the duration of treatment for 
children in the evaluation’s impact analysis sample.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
I.2 
● To calculate the number of participants for the program in its steady state, we first 
calculated the maximum enrollment in each year and then averaged across the two years 
of observation. We did this for Phase I and II of the program and then calculated the final 
number by summing across both phases. It was important to calculate the number of 
children participating in each component of the program in order to generate a cost per 
child. Some children in Group A were also counted as enrolled even if they were not in 
grades 1 to 3 because the reading buddies community activity took place in schools. The 
number of participants calculated for Group A only includes students enrolled in grades 1 
to 3. 
● We converted costs to constant 2014 dollars using the Consumer Price Index – Urban. 
● The marginal cost calculation assumed that 9 percent of the total costs were fixed and the 
rest were variable (by child). This number came from the amount spent by Save the 
Children as a fraction of the total grant. In reality, the costs per child of implementing a 
program like Leer Juntos, Aprender Juntos will depend on the number of children per 
school, the distance of the schools from the program office, the number of schools, and 
other factors. Readers interested in using our estimate of costs to plan future 
implementation should examine the conditions described in Chapter I and consider the 
local context before extrapolating from the Peru experience. 
Costs were mostly made up of personnel salaries and benefits, followed by travel costs 
associated with sending technical staff into the field to conduct trainings and coaching. Other 
costs included educational materials and supplies.
APPENDIX J: 
EMERGENT LITERACY SKILLS OF GIRLS AND BOYS IN THE EVALUATION AT 
BASELINE
This page has been left blank for double-sided printing.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
J.1 
Emergent literacy skills of girls and boys in the evaluation at baseline
To provide more context to the impact findings for subgroups of girls and boys, we looked 
at the results for girls and boys on the emergent literacy skills assessments we conducted during 
the baseline data collection (in 2013 for Phase I and 2014 for Phase II). In terms of levels, the 
girls in the schools in the treatment groups (Groups A and B) had, on average, slightly higher 
scores in the six assessments than the boys in the treatment groups (Table J.1). For example, girls 
in Groups A and B scored, on average, 14.9 and 14.3 points, respectively, in the letter 
identification assessment, while boys in Groups A and B scored, on average, 14.4 and 14.1 
points, respectively, in the same assessment. 
Table J.1. Baseline emergent literacy skills of girls and boys in the 
evaluation
Literacy outcome
Leer
Juntos
(A)
Leer 
Juntos￾school 
only
(B)
Prevailing 
practice
(C) (A–B) (B–C)
Girls
Letter identification (average score out of 
24 possible points)
14.9 14.3 14.5 0.6
(0.364)
-0.2
(0.805)
Emergent writing (average score out of 22 
possible points)
10.6 9.7 12.0 0.8
(0.282)
-2.3*
(0.004)
Emergent reading (average score out of 9 
possible points)
2.0 1.8 2.7 0.2
(0.588)
-1.0*
(0.007)
Phonemic awareness (average score out 
of 10 possible points)
3.2 3.0 3.1 0.2
(0.449)
-0.1
(0.630)
Pseudo-word decoding (number correct in 
one minute, up to 50)
1.9 1.9 3.1 0.0
(0.955)
-1.2
(0.050)
Listening comprehension (average score 
out of 6 possible points)
2.1 2.2 2.2 0.0
(0.951)
0.0
(0.832)
Boys
Letter identification (average score out of 
24 possible points)
14.4 14.1 14.1 0.3
(0.758)
0.1
(0.943
Emergent writing (average score out of 22 
possible points)
9.7 9.3 10.2 0.4
(0.637)
-0.9
(0.254)
Emergent reading (average score out of 9 
possible points)
1.8 1.7 2.3 0.1
(0.786)
-0.6
(0.119)
Phonemic awareness (average score out 
of 10 possible points)
2.8 2.9 3.1 -0.1
(0.774)
-0.2
(0.411)
Pseudo-word decoding (number correct in 
one minute, up to 50)
1.4 1.1 3.1 0.3
(0.593)
-1.9*
(0.010)
Listening comprehension (average score 
out of 6 possible points)
1.9 2.0 2.2 -0.1
(0.623)
-0.2
(0.482)
Number of children 349 324 316
Source: Leer Juntos, Aprender Juntos Students’ Emergent Literacy Skills Assessment—Baseline 2013 and 2014.
Note: Regression-adjusted means. P-values in parentheses.
* Difference in group means is statistically significant at the 0.05 level.
This page has been left blank for double-sided printing.
APPENDIX K: 
DATA COLLECTION INSTRUMENTS
This page has been left blank for double-sided printing.
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
K.1 
K.1. Survey instruments
The baseline, midline, and endline survey instruments are available online upon request at 
USAID’s Development Data Library website (https://data.usaid.gov/)
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
K.2 
K.2. Qualitative data collection instruments
This section includes the following qualitative data collection instruments
1. Save the Children staff interview instrument
2. Save the Children specialist interview instrument
3. Community volunteers interview instrument 
4. Teachers in treatment Groups A and B focus group protocol 
5. Teachers in Group C focus group protocol 
6. Parents in treatment Groups A and B focus group protocol 
7. Community action component activities observation instrument 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
K.3 
1. LEER JUNTOS, APRENDER JUNTOS: SAVE THE CHILDREN STAFF
A. ROLES AND RESPONSIBILITIES
1. What is your role and what are your responsibilities in the implementation of LJAJ? What were 
some of your primary activities in the LJAJ implementation? 
2. How/when did you learn the program model and intervention goals? What type of support did you 
receive to understand the program goals?
● Probe: hires dedicated exclusively for LJAJ are more committed or steeped into the program? 
How much of the original intervention model gets lost in translation as it is adapted and 
implemented by different organizations?
3. How is the implementation team organized? 
● Probe: communication line, decision-making, changes to the organization during the project
4. What challenges in the LJAJ implementation have you encountered in your role? How have you 
handled them?
B. ADAPTATION OF THE LITERACY BOOST MODEL TO PERU
5. How was the Literacy Boost model adapted to Guatemala, particularly to the sociocultural and 
linguistic context of Quiché?
● Probe: who, when, how was it done; explore adaptations to the teacher training and the 
community action components 
6. How is the K’iche’ language and culture incorporated into the program? 
7. What challenges came up during the cultural adaptation? How were those challenges addressed? 
Has the model or the teaching strategies evolved over the course of the project? How?
C. FIDELITY
8. How was the teacher training component organized and how did it take place? What have been 
the main challenges in training and coaching teachers? How were those addressed?
● Probe: explore specific challenges by activity type: book banks, book fairs, reading buddies, 
reading camps
9. How was the reading assessment component of the program implemented? What were the 
challenges in administering or using assessment data? How was the reading assessment data 
used? 
● Probe: were teachers able to use assessment data to inform instruction? How did you provide 
assessment data to teachers, how they received it? Did they use it?
10. How has LJAJ been implemented in phase I and phase II schools? What are the main differences 
between phases? In what ways has implementation of LJAJ varied from school to school? 
11. What aspects of the program did not take place according to plan or to SC guidelines?
D. SC KEY STAFF’S PERCEPTIONS ON PROGRAM TAKE UP AND IMPACT
12. How have school principals, teachers, parents and children received the LJAJ activities? 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
K.4 
13. What challenges have come up with parents’ and children’s participation in program activities? 
How were those challenges addressed?
14. In what ways do you think LJAJ can be improved to better serve teachers and students in that 
region? 
15. How can program take up be improved? What changes would be needed?
E. PROGRAM GOALS AND PERCEIVED IMPACTS
16. Is the program reaching the intended target population, and, if not, why not?
17. How do you track and measure the program’s goals?
● Probes: how are benchmarks set and tracked? How have you adjusted to meet program 
goals?
18. Do you think students in the program are actually reading better? How has the program benefited 
children and families?
● Do you think students in the program are actually spending more time reading at home and at 
school?
D. WRAP UP
19. In your view, what aspects of the program have been the most successful ones, and which ones 
the least?
20. What recommendation would you make to future implementers if this programs were to be scaled 
up or implemented in other regions in Guatemala or other countries with indigenous populations?
Thanks for your collaboration!
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
K.5 
2. LEER JUNTOS, APRENDER JUNTOS: VOLUNTARIOS
Presentación
Gracias por aceptar reunirse con nosotros el día de hoy. Grade y Mathematica Policy Research, una 
firma de investigación y evaluación independiente con sede en los EE.UU., está trabajando con Save the 
Children para llevar a cabo una evaluación del Programa “Leer Juntos, Aprender Juntos (LJAJ)” y para 
entender los programas educativos de la región de Apurímac. El propósito de nuestra conversación de 
hoy es conocer más acerca de sus experiencias como voluntarios. Nuestra conversación durará unos 60 
minutos aproximadamente. Nos gustaría grabar y tomar notas durante la discusión para que podamos 
recordar con precisión las experiencias que ustedes compartan con nosotros.
Un grupo focal es una conversación grupal guiada, en este caso con voluntarios como ustedes, para 
aprender acerca de sus experiencias. El moderador del grupo hará preguntas acerca de la enseñanza de 
la lectura en primer y segundo grado. Por favor, tenga en cuenta que no hay respuestas correctas o 
incorrectas en la conversación. Las preguntas del moderador sirven para organizar la conversación, pero 
esperamos que los participantes quieran compartir sus opiniones y experiencias relacionadas al tema. La 
participación en la discusión de grupo es voluntaria y no es necesario responder a preguntas que no 
quiera contestar. 
Les pediremos a todos los participantes mantener en reserva lo que otras personas compartan durante el 
grupo y no compartir lo conversado fuera del grupo. Lo que ustedes compartan aquí se mantendrá en 
privado. Sus opiniones no se compartirán con personas fuera del equipo de investigación, y los 
participantes no se identificarán con información personal. En las transcripciones y notas de los grupos 
focales se omitirán los nombres de las personas que participan y su nombre no aparecerá en informes o
presentaciones. 
Vamos a grabar la discusión en grupo para facilitar la toma de notas, sin embargo solamente el equipo 
de investigación tendrá acceso a la grabación. La grabación de audio se borrará después de que termine 
el estudio. 
Si tienen alguna pregunta o duda, por favor, contactar a María Balarin mbalarin@grade.org.pe, a cargo 
del estudio, o a Kristell Benavides kbenavides@grade.org.pe / telf. 01 247 9988. También puede 
comunicarse con Camila Fernandez en cfernandez@mathematica-mpr.com. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
K.6 
A. ROL Y MOTOVACIÓN DE LOS VOLUNTARIOS
Quisiera empezar preguntándoles sobre su rol como voluntarios en el programa LJAJ. ¿Que los llevó 
a convertirse en voluntarios de LJAJ?
1. ¿Cuáles han sido sus funciones y responsabilidades? 
2. ¿En qué consisten las actividades en las que ustedes colaboran?
3. ¿Cuál es su opinión sobre las actividades de la comunidad?
4. ¿De qué manera ayudan estas actividades a los estudiantes?
5. ¿De qué manera ayudan estas actividades a los docentes?
6. ¿De qué formas han contribuido a las actividades de LJAJ? (Tiempo, materiales)
7. ¿Qué actividades que realizaban en su vida diaria han dejado de hacer por participar como 
voluntarios de LJAJ?
8. ¿Qué desafíos han tenido que afrontar como voluntarios del programa LJAJ? ¿Cómo los han 
afrontado?
9. ¿Qué les ha impedido participar en más actividades de su comunidad?
10. ¿Recomendarían el voluntariado a otras personas en su comunidad? ¿Por qué sí/Por qué no?
11. ¿Qué les ha gustado más de ser voluntarios para LJAJ?
B. PERCEPCIONES DE LOS VOLUNTARIOS SOBRE EL PROGRAMA
12. ¿Cómo describirían la percepción de los miembros de la comunidad sobre las actividades 
comunitarias de LJAJ?
13. ¿Cómo se incorporan en las actividades comunitarias de LJAJ el lenguaje y la cultura quechua?
14. ¿Cuáles han sido los principales desafíos que han encontrado con respecto a la participación de 
padres y niños?
15. ¿Ha habido alguna dificultad específica en la implementación de las actividades comunitarias del 
programa LJAJ?
16. ¿Cómo sienten que ha sido la recepción del Programa por parte de la comunidad y las familias 
con las que participan?
17. ¿De qué manera creen que LJAJ puede mejorar para adaptarse a las necesidades de los 
miembros de la comunidad? ¿Qué aspectos del programa dirían que están funcionando bien y 
cuáles no? ¿Por qué?
D. CIERRE
Para finalizar,
18. ¿Tienen alguna idea o información adicional sobre el programa, las actividades del programa, o la 
implementación del programa que les gustaría compartir?
¡Gracias por su participación!
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
K.7 
3. LEER JUNTOS, APRENDER JUNTOS : GRUPOS FOCALES CON
ESPECIALISTAS DEL PROGRAMA
Presentación
Gracias por aceptar reunirse con nosotros el día de hoy. Como ustedes saben, GRADE y Mathematica 
Policy Research, una firma de investigación y evaluación independiente con sede en los EE.UU., están 
trabajando con Save the Children para llevar a cabo una evaluación del Programa “Leer Juntos, 
Aprender Juntos (LJAJ)” y para entender los programas educativos de la región de Apurímac. 
El propósito de esta conversación es conocer sus experiencias en la implementación de las actividades 
del Programa Leer Juntos, Aprender Juntos (LJAJ). Nuestra conversación durará alrededor de 90 
minutos. 
Por favor, tenga en cuenta que no hay respuestas correctas o incorrectas en la conversación. La 
participación en la discusión de grupo es voluntaria y no es necesario responder a preguntas que no 
quiera contestar. Les pediremos a todos los participantes mantener en reserva lo que otras personas 
compartan durante el grupo y no compartir lo conversado fuera del grupo. Lo que ustedes compartan 
aquí se mantendrá en privado. Sus opiniones no se compartirán con personas fuera del equipo de 
investigación, y los participantes no se identificarán con información personal. En las transcripciones y 
notas de los grupos focales se omitirán los nombres de las personas que participan y su nombre no 
aparecerá en informes o presentaciones. 
Vamos a grabar la discusión en grupo para facilitar la toma de notas, sin embargo solamente el equipo 
de investigación tendrá acceso a la grabación. La grabación de audio se borrará después de que termine 
el estudio. 
Si tienen alguna pregunta o duda, por favor, contactar a María Balarin mbalarin@grade.org.pe, a cargo 
del estudio, o a Kristell Benavides kbenavides@grade.org.pe / telf. 01 247 9988. También puede 
comunicarse con Camila Fernandez en cfernandez@mathematica-mpr.com. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
K.8 
A. ROL DE LOS ESPECIALISTAS 
Quisiera comenzar preguntándoles ¿cómo le enseñan ustedes a leer a los estudiantes de primero y 
segundo grado? 
1. ¿Cuáles han sido sus responsabilidades?
2. ¿Cuáles han sido sus actividades principales en la implementación del programa?
3. ¿Qué tipo de preparación y soporte han recibido durante la implementación del programa? ¿De 
parte de quién?
4. ¿Cómo está organizado el equipo de implementación? ¿Con quienes se comunican para planear 
el trabajo y conversar sobre sus responsabilidades y actividades en el campo?
5. ¿En qué forma sienten ustedes que han contribuido con la implementación del Programa? 
(tiempo, habilidades, materiales)
6. ¿Qué desafíos ha enfrentado en su rol? ¿Cómo han manejado esos desafíos hasta el momento?
B. FIDELIDAD DE LA IMPLEMENTACIÓN DEL PROGRAMA
7. ¿Cómo se ha implementado LJAJ en las escuelas de la fase I y la fase II? ¿Qué diferencias ha 
habido en la implementación del programa en las escuelas de la fase 1 las de la fase 2? 
8. ¿Cómo se organizó el componente de formación a los docentes y cómo se realizó? ¿Cuáles han 
sido los principales desafíos? ¿Cómo fueron asumidos por los docentes?
9. ¿Cómo se organizó el componente de acompañamiento a los docentes y cómo se realizó? 
¿Cuáles han sido los principales desafíos? ¿Cómo fueron asumidos por los docentes?
10. ¿Cómo se han organizado los talleres con padres y cómo se han realizado? ¿Cuáles han sido los 
principales desafíos? ¿Cómo fueron asumidos por los padres?
11. ¿Cómo se organizó el banco de libros y cómo ha sido usado en la comunidad? 
● Explorar el contenido de la actividad y los roles de los diferentes actores. ¿Cuáles han sido 
los principales desafíos con respecto a esta actividad? 
● ¿Cómo se respondió a dichos desafíos?
12. ¿Cómo se organizó el préstamo de libros y cómo se ha llevado a cabo en la comunidad? 
● Explorar el contenido de la actividad y los roles de los diferentes actores. ¿Cuáles han sido 
los principales desafíos con respecto a esta actividad? 
● ¿Cómo se respondió a dichos desafíos?
13. ¿Cómo se organizó la actividad amigos de la lectura y cómo se ha llevado a cabo en la 
comunidad? ¿Cuáles han sido los principales desafíos con respecto a esta actividad?
● Explorar el contenido de la actividad y los roles de los diferentes actores. ¿Cómo se 
respondió a dichos desafíos?
14. ¿Cómo se organizaron los campamentos de lectura y cómo se ha llevado a cabo en la 
comunidad? ¿Cuáles han sido los principales desafíos? 
● Explorar el contenido de la actividad y los roles de los diferentes actores. ¿Cómo se 
respondió a dichos desafíos?
15. ¿Qué aspectos del programa no se han desarrollado de acuerdo al planeamiento o a las 
indicaciones de Kallpa?
16. ¿Qué tipo de adaptaciones han tenido que hacer para satisfacer las necesidades de la población 
objetivo?
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
K.9 
17. ¿Qué tipo de diferencias se han producido en la implementación del programa en las escuelas 
según su tamaño, ubicación geográfica, tipo o cualquier otra característica relevante?
C. PERCEPCIONES DE LOS ESPECIALISTAS
18. ¿Cómo han recibido los profesores, padres y estudiantes las actividades de LJAJ?
19. ¿Cómo describirían las características de los estudiantes/padres que participan de las actividades 
en la comunidad?
20. ¿Cómo se incorporan en las actividades de la comunidad el lenguaje y la cultura quechua?
21. ¿Cuáles han sido los principales desafíos que han encontrado con respecto a la participación de 
padres y niños?
22. ¿De qué manera creen que LJAJ puede mejorar para adaptarse a las necesidades de los 
miembros de la comunidad?
D. CIERRE
23. ¿Tienen alguna idea o información adicional sobre el programa, las actividades del programa, o la 
implementación del programa que les gustaría compartir?
24. ¿Cuáles prácticas pedagógicas utiliza con mayor frecuencia, y por qué? ¿Podrían dar algunos 
ejemplos?
25. ¿Qué aspectos de la lectura y la escritura son más difíciles para los estudiantes?
26. ¿Qué aspectos de la lectura y la escritura aprenden con facilidad los estudiantes?
¡Gracias por su participación!
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
K.10
4. LEER JUNTOS, APRENDER JUNTOS : GRUPOS FOCALES DE MAESTROS
(GRUPO A & B)
Presentación
Gracias por aceptar reunirse con nosotros el día de hoy. Como ustedes saben, GRADE y Mathematica 
Policy Research, una firma de investigación y evaluación independiente con sede en los EE.UU., están 
trabajando con Save the Children para llevar a cabo una evaluación del Programa “Leer Juntos, 
Aprender Juntos (LJAJ)” y para entender los programas educativos de la región de Apurímac. 
El propósito de nuestra conversación de hoy es conocer más acerca de sus experiencias enseñándole a 
leer a sus alumnos de primer y segundo grado. Nuestra conversación durará unos 60 minutos 
aproximadamente. Nos gustaría grabar y tomar notas durante la discusión para que podamos recordar 
con precisión las experiencias que ustedes compartan con nosotros.
Un grupo focal es una conversación grupal guiada, en este caso con maestros como ustedes, para 
aprender acerca de sus experiencias. El moderador del grupo hará preguntas acerca de la enseñanza de 
la lectura en primer y segundo grado. Por favor, tenga en cuenta que no hay respuestas correctas o 
incorrectas en la conversación. Las preguntas del moderador sirven para organizar la conversación, pero 
esperamos que los participantes quieran compartir sus opiniones y experiencias relacionadas al tema. La 
participación en la discusión de grupo es voluntaria y no es necesario responder a preguntas que no 
quiera contestar. 
Les pediremos a todos los participantes mantener en reserva lo que otras personas compartan durante el 
grupo y no compartir lo conversado fuera del grupo. Lo que ustedes compartan aquí se mantendrá en 
privado. Sus opiniones no se compartirán con personas fuera del equipo de investigación, y los 
participantes no se identificarán con información personal. En las transcripciones y notas de los grupos 
focales se omitirán los nombres de las personas que participan y su nombre no aparecerá en informes o 
presentaciones. 
Vamos a grabar la discusión en grupo para facilitar la toma de notas, sin embargo solamente el equipo 
de investigación tendrá acceso a la grabación. La grabación de audio se borrará después de que termine 
el estudio. 
Si tienen alguna pregunta o duda, por favor, contactar a María Balarin mbalarin@grade.org.pe, a cargo 
del estudio, o a Kristell Benavides kbenavides@grade.org.pe / telf. 01 247 9988. También puede 
comunicarse con Camila Fernandez en cfernandez@mathematica-mpr.com. 
A. PEDAGOGÍA DE LA LECTURA A LOS ALUMNOS DE PRIMER Y SEGUNDO 
GRADO
Quisiera comenzar preguntándoles ¿cómo le enseñan ustedes a leer a los estudiantes de primero y 
segundo grado? 
1. ¿Cuáles son las prácticas pedagógicas para la enseñanza de la lectura que funcionan mejor con 
sus alumnos? ¿Qué métodos de instrucción se utilizan? ¿Podrían dar algunos ejemplos?
2. ¿Cuáles prácticas pedagógicas utiliza con mayor frecuencia, y por qué? ¿Podrían dar algunos 
ejemplos?
3. ¿Qué aspectos de la lectura y la escritura son más difíciles para los estudiantes?
4. ¿Qué aspectos de la lectura y la escritura aprenden con facilidad los estudiantes?
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
K.11
B. PARTICIPACIÓN EN EL PROGRAMA, APRENDIZAJE Y CAMBIOS EN LAS 
PRÁCTICAS
Ahora hablemos del programa Leer Junto, Aprender Juntos (LJAJ).
5. ¿Qué conocen sobre el programa LJAJ? ¿Cuáles son los objetivos del programa? 
6. ¿En qué tipo de formación o actividades de entrenamiento de LJAJ han participado? 
7. ¿Qué cosas has aprendido en el entrenamiento LJAJ? ¿Podrían dar algunos ejemplos?
8. ¿Qué tipo de cosas aprendió a través del coaching entrenamiento (acompañamiento) LJAJ? 
¿Podrían dar algunos ejemplos?
9. ¿De qué manera LJAJ ha cambiado su manera de enseñarles a sus alumnos a leer? ¿Podrían 
dar algunos ejemplos?
10. ¿De qué manera LJAJ ha cambiado sus rutinas en la clase o en la forma de interactuar con sus 
estudiantes? ¿Podrían dar algunos ejemplos?
11. ¿Cuáles son las diferencias y similitudes entre la pedagogía de la lectura del Currículo Nacional 
Base o del Programa Nacional de Lectura y la de LJAJ?
12. ¿Enfrentó algún desafío para estar disponible para participar en actividades de capacitación o 
entrenamiento de LJAJ? ¿Cuáles fueron esos desafíos?
13. ¿Qué aspectos de LJAJ les han gustado más? ¿Por qué? 
14. ¿Qué aspectos de LJAJ les han gustado menos gustó menos? ¿Por qué?
C. ASPECTOS SOCIOCULTURALES DEL PROGRAMA
15. ¿Qué aspectos del programa toman en cuenta la cultura, el idioma y las costumbres de su 
comunidad?
16. ¿Qué aspectos del programa son difíciles de integrar en la comunidad?
D. CIERRE
17. ¿En qué medida creen que LJAJ le ha ayudado a mejorar su capacidad de enseñar a los niños a 
leer?
18. ¿Qué retos o dificultades tienen enfrentan aún al enseñarle a leer a los alumnos de primer y 
segundo grado? ¿Qué les podría ayudar a superar esos desafíos?
19. ¿Cuáles son los aspectos del programa que se pueden mejorar?
20. ¿Recomendaría el entrenamiento LJAJ a otros en la comunidad? ¿Por qué / por qué no?
21. ¿Tiene usted alguna idea o información adicional sobre el programa, las actividades del programa 
o de la implementación que le gustaría compartir?
¡Gracias por su participación!
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
K.12
5. LEER JUNTOS, APRENDER JUNTOS : GRUPOS FOCALES DE MAESTROS
(GRUPO C)
Presentación
Gracias por aceptar reunirse con nosotros el día de hoy. Como ustedes saben, GRADE y Mathematica 
Policy Research, una firma de investigación y evaluación independiente con sede en los EE.UU., están 
trabajando con Save the Children para llevar a cabo una evaluación del Programa “Leer Juntos, 
Aprender Juntos (LJAJ)” y para entender los programas educativos de la región de Apurímac. 
El propósito de nuestra conversación de hoy es conocer más acerca de sus experiencias enseñándole a 
leer a sus alumnos de primer y segundo grado. Nuestra conversación durará unos 60 minutos 
aproximadamente. Nos gustaría grabar y tomar notas durante la discusión para que podamos recordar 
con precisión las experiencias que ustedes compartan con nosotros.
Un grupo focal es una conversación grupal guiada, en este caso con maestros como ustedes, para 
aprender acerca de sus experiencias. El moderador del grupo hará preguntas acerca de la enseñanza de 
la lectura en primer y segundo grado. Por favor, tenga en cuenta que no hay respuestas correctas o 
incorrectas en la conversación. Las preguntas del moderador sirven para organizar la conversación, pero 
esperamos que los participantes quieran compartir sus opiniones y experiencias relacionadas al tema. La 
participación en la discusión de grupo es voluntaria y no es necesario responder a preguntas que no 
quiera contestar. 
Les pediremos a todos los participantes mantener en reserva lo que otras personas compartan durante el 
grupo y no compartir lo conversado fuera del grupo. Lo que ustedes compartan aquí se mantendrá en 
privado. Sus opiniones no se compartirán con personas fuera del equipo de investigación, y los 
participantes no se identificarán con información personal. En las transcripciones y notas de los grupos 
focales se omitirán los nombres de las personas que participan y su nombre no aparecerá en informes o 
presentaciones. 
Vamos a grabar la discusión en grupo para facilitar la toma de notas, sin embargo solamente el equipo 
de investigación tendrá acceso a la grabación. La grabación de audio se borrará después de que termine 
el estudio. 
Si tienen alguna pregunta o duda, por favor, contactar a María Balarin mbalarin@grade.org.pe, a cargo 
del estudio, o a Kristell Benavides kbenavides@grade.org.pe / telf. 01 247 9988. También puede 
comunicarse con Camila Fernandez en cfernandez@mathematica-mpr.com. 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
K.13
A. PEDAGOGÍA DE LA LECTURA A LOS ALUMNOS DE PRIMER Y SEGUNDO 
GRADO
Quisiera comenzar preguntándoles ¿cómo le enseñan ustedes a leer a los estudiantes de primero y 
segundo? ¿Qué métodos de instrucción se utilizan?
1. ¿Qué tipo de actividades de lectura y escritura hacen los niños en su salón de clases? ¿Puede 
darnos algunos ejemplos?
2. ¿Cuáles son las prácticas pedagógicas para la enseñanza de la lectura que funcionan mejor con 
sus alumnos?
3. ¿Cuáles prácticas pedagógicas utilizan con mayor frecuencia, y por qué? 
4. ¿Qué aspectos de la lectura y la escritura son más difíciles para los estudiantes? ¿Puede darnos 
algunos ejemplos?
5. ¿Qué aspectos de la lectura y la escritura aprenden con facilidad los estudiantes? ¿Puede darnos 
algunos ejemplos?
6. ¿Qué retos o dificultades enfrentan al enseñarle a leer a los alumnos en primer y segundo grado?
B. EXPERIENCIAS DE FORMACIÓN Y APOYO DOCENTE
Ahora hablemos del programa de sus experiencias formándose como maestro, y del tipo de apoyo 
pedagógico que reciben en esta escuela. 
7. ¿Han recibido alguna capacitación o entrenamiento enfocado específicamente en enseñar a leer 
a los niños en la escuela primaria?
8. ¿Cuál fue ese entrenamiento?¿Quién lo impartió? ¿Cuándo? ¿Puede darnos algunos ejemplos?
9. ¿Qué cosas aprendieron en el entrenamiento? ¿Puede darnos algunos ejemplos?
10. ¿En qué medida han podido aplicar lo que aprendieron en esas capacitaciones al enseñar la 
lectura? ¿Puede darnos algunos ejemplos?
11. ¿En qué medida lo que aprendieron ha mejorado su capacidad para enseñar a los niños a leer? 
¿Puede darnos algunos ejemplos?
D. CIERRE
Para finalizar,
12. ¿Hay algo que crean que puede ayudarles a ustedes a mejorar la manera le enseñan a aprenden 
a leer a los alumnos? 
13. ¿Qué tipo de apoyo piensan que podría ayudar a superar los desafíos que enfrentan al enseñarle 
a leer a los estudiantes de primer y segundo grado?
¡Gracias por su participación!
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
K.14
6. LEER JUNTOS, APRENDER JUNTOS: PADRES DE FAMILIA (GRUPOS A&B)
- Presentación
Muchas gracias por aceptar reunirse con nosotros en este día. Nosotros hacemos parte de un equipo 
que está realizando un estudio sobre programas educativos en la región de Apurímac. En el estudio 
están colaborando varias organizaciones, entre ellas GRADE, Mathematica Policy Research (centro de 
investigación basada en EEUU) y Save The Children. El propósito de la conversación de hoy es 
aprender más acerca de de sus hijos y miembros de la familia en el proceso de aprendizaje de la 
lectoescritura y conocer sobre su participación en actividades dentro y fuera de la escuela para ayudarle 
a los niños a aprender a leer. Nuestra conversación durara alrededor de 60 minutos.
Un grupo focal es una conversación grupal guiada, en este caso con padres de familia como ustedes, 
para aprender acerca de sus experiencias. El moderador del grupo hará preguntas acerca de la lectura 
de los niños de primer y segundo grado. Por favor, tenga en cuenta que no hay respuestas correctas o 
incorrectas en la conversación. Las preguntas del moderador sirven para organizar la conversación, pero 
esperamos que los participantes quieran compartir sus opiniones y experiencias relacionadas al tema. La 
participación en la discusión de grupo es voluntaria y no es necesario responder a preguntas que no 
quieran contestar. 
Les pediremos a todos los participantes mantener en reserva lo que otras personas compartan durante el 
grupo y no compartir lo conversado fuera del grupo. Lo que ustedes compartan aquí se mantendrá en 
privado. Sus opiniones no se compartirán con personas fuera del equipo de investigación, y los 
participantes no se identificarán con información personal. En las transcripciones y notas de los grupos 
focales se omitirán los nombres de las personas que participan y su nombre no aparecerá en informes o 
presentaciones. 
Vamos a grabar la discusión en grupo para facilitar la toma de notas, sin embargo solamente el equipo 
de investigación tendrá acceso a la grabación. La grabación de audio se borrará después de que termine 
el estudio. 
Si tienen alguna pregunta o duda, por favor, contactar a María Balarin mbalarin@grade.org.pe, a cargo 
del estudio, o a Kristell Benavides kbenavides@grade.org.pe / telf. 01 247 9988. También puede 
comunicarse con Camila Fernandez en cfernandez@mathematica-mpr.com. 
 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
K.15
A. CONOCIMIENTOS Y EXPECTATIVAS DE LOS PADRES SOBRE EL 
PROGRAMA LJAJ
-Quisiera empezar preguntándoles: ¿Cómo han aprendido a leer sus hijos? 
1. ¿Cómo participan ustedes en el aprendizaje de la lectura? ¿Cuánto tiempo le dedican a la 
práctica de la lectura en su hogar?
2. ¿Qué saben ustedes sobre el programa LJAJ?
3. ¿Cuáles son los objetivos de las actividades de LJAJ? ¿Para qué sirven? ¿A quiénes están 
dirigidas?
4. ¿Por qué son importantes las actividades de LJAJ para ustedes?
5. ¿Cómo participan ustedes en las actividades propuestas por LJAJ? 
6. ¿Cuáles eran sus expectativas con respecto LJAJ? ¿Esas expectativas han cambiado? ¿De qué 
manera? ¿Por qué? 
B. PARTICIPACIÓN, APRENDIZAJE Y CAMBIOS EN LAS PRÁCTICAS
7. ¿En qué tipos de actividades de LJAJ han participados ustedes? ¿Con qué frecuencia?
8. ¿Qué han aprendido ustedes en esas actividades? 
9. ¿En qué tipos de actividades de LJAJ han participados sus hijos? ¿Con cuánta frecuencia?
10. ¿Qué han aprendido sus hijos en esas actividades?
11. ¿Han afrontado algún desafío para poder participar en las actividades de LJAJ? ¿Cuáles fueron 
esos desafíos? (tiempo, relevancia percibida de la actividad, etc.) ¿Les gustaría participar más de 
las actividades de LJAJ?
12. ¿Cuáles son los aspectos de las actividades de LJAJ que más les gustaron? ¿Por qué?
13. ¿Cuáles son los aspectos de LJAJ que menos les gustaron? ¿Por qué? ¿De qué formas creen 
que LJAJ los ha beneficiado a ustedes y a sus familias?¿De qué forma LJAJ ha cambiado sus 
rutinas en casa o el modo en que interactúan con sus hijos? 
14. ¿Las actividades de LJAJ con los padres les parecen apropiadas para esta región 
(Andahuaylas/Chincheros etc.) y para su cultura? ¿Por qué?
15. ¿Las actividades de LJAJ con los niños les parecen apropiadas para esta región 
(Andahuaylas/Chincheros etc.) y para su cultura? ¿Por qué?
C. CIERRE
-Para finalizar,
16. ¿Creen que podrían haber algunos cambios para mejorar las actividades de LJAJ? ¿Cuáles 
serían sus sugerencias?
17. ¿Recomendaría participar de LJAJ a otras personas en la comunidad? ¿Por qué sí/Por qué no?
18. ¿Tienen alguna idea o información adicional sobre el programa, las actividades del programa, o la 
implementación del programa que les gustaría compartir?
¡Gracias por su participación!
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
K.16
7. LEER JUNTOS, APRENDER JUNTOS 
FORMATO DE OBSERVACIONES DE ACTIVIDES COMUNITARIAS
ESTUDIO CUALITATIVO DE IMPLEMENTACIÓN
Fecha de la observación: | | | / | 1 1 | | / | 2 | 0 | 1 | 4 |
Nombre del observador:
Municipio: 
Tipo de Actividad: 
Descripción del lugar y del contexto donde se lleva a cabo la actividad:
_________________________________________________________________________________
____________
 Número de Participantes: | , 
Número de Lideres/Voluntarios en la actividad: | | | | 
Descripción de Participantes: 
 | | | | | 
EVALUATION OF LEER JUNTOS, APRENDER JUNTOS EARLY-GRADE
READING INTERVENTION IN PERU: FINAL REPORT MATHEMATICA POLICY RESEARCH
K.17
Descripción de Actividad de Lideres/Voluntarios:
Descripción de la Actividad (lenguaje utilizado, objetivos de la actividad, materiales, 
instrucciones, etc.): 
_________________________________________________________________________________
___________
Involucramiento de participantes’ en la Actividad (ej. qué tan involucrados estaban los 
participantes en la tarea, entendieron y siguieron las instrucciones dadas, parecían disfrutar la 
actividad): 
 
_________________________________________________________________________________
__________
Otras notas (ej. qué tan involucrados estuvieron los voluntarios y otros líderes en la tarea, 
dieron ellos instrucciones claras): 
U.S. Agency for International Development 
1300 Pennsylvania Avenue, NW 
Washington, DC 20523 
Telephone: (202) 712-0000 
Fax: (202) 216-3524 
www.usaid.gov