Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
© 2014 Save the Children
Literacy Boost
Gampaha District
Sri Lanka Country Office
Endline Report – December 2014
Author: Pubudinie Wickramasekara, Subashini Navaratnam, and Jarret
Guajardo
Thank you to Chandima Liyanagamage, Helmalie Vitharana Charith Weerasinghe
for their excellent support during data collection
With special thanks to the team of assessors: Mr. M.P.R.K. Mapa Pathirana, Mr. K.T.D.S
Isurupala , Ms.R.A.N. Darshika , Ms. G.A. Madhavee Muthumali, Ms.W.L.U.R. Warnasuriya,
Ms. A.D.S.D Ranasinghe, Ms. L.U.E.I Bandara, Ms.R.K. S. Wickramarathne, Ms. M.P.D.S.
Dharmapriya, Mr. A.K.L.A.C Perera, Mr. J.P.A.G. Abekoon, Ms. U.G.S.M Senavirathne,
Ms. K.Y.S Jayathilaka, Ms. Kariyawasam, Ms. L.A.D.K. S. Liyanarachchi, Ms. M.K. Hettiarachchi,
and Ms. K.S. Dilrukshi
1
Table of Contents Executive Summary .................................................................................................................................. 2
1. Introduction, Context, and Implementation History ........................................................................ 3
Introduction ............................................................................................................................................. 3
Context .................................................................................................................................................... 4
Implementation History .......................................................................................................................... 4
11. Methods ................................................................................................................................................ 5
Sampling and Attrition Analysis............................................................................................................... 5
Measurement ........................................................................................................................................... 6
Analysis .................................................................................................................................................... 7
111. Children’s Background ..................................................................................................................... 7
IV. Children’s Home Literacy Environment ........................................................................................... 8
Comparability of Literacy Boost and comparison students at baseline .................................................. 9
V. Impact of Literacy Boost on Reading Skills ..................................................................................... 10
VI. Impact of Literacy Boost for Slow Learners and Children Who Screened Positive for Language
and Learning Challenges......................................................................................................................... 12
VII. Skill Gains of Boys and Girls ............................................................................................................ 14
VIII. Classroom and Home Literacy Environment .............................................................................. 14
IX. Participation in Community Action Activities ............................................................................... 16
X. Children’s Reading Skills .................................................................................................................... 17
Concepts About Print ............................................................................................................................ 17
Letter Identification ............................................................................................................................... 17
Word Recognition: Most Used Words ................................................................................................... 17
Proportion of Readers and Nonreaders ................................................................................................ 18
Fluency and Accuracy ............................................................................................................................ 18
Reading Comprehension........................................................................................................................ 19
Readers with Comprehension ................................................................................................................ 20
XI. Literacy Boost Skill Profile and Skill-Specific Recommendations ................................................ 21
XII. Summary of Findings and General Program Recommendations ............................................... 22
Appendix A: Implementation Timeline ................................................................................................ 24
Appendix B: Inter-rater reliability ......................................................................................................... 27
Appendix C: Full Background and Home Literacy Environment Statistics at Endline .................... 28
Appendix D. Regression Results ............................................................................................................ 30
2
Executive Summary Among 257 Literacy Boost (LB) students and 262 comparison students assessed during a
January/February 2013 baseline and September 2014 endline, impact analysis reveals that the LB
program significantly improved students’ reading fluency. This is especially true for students identified
as slow learners at baseline. LB programming also helped students who screened positive for language
& learning challenges catch up to their non-slow learner peers in almost every reading skill. However,
all students still need help to improve their reading comprehension.
LB programing also appears to have helped teachers better utilize formative assessment to monitor
student achievement and target support for smaller groups of students, and to use songs and games to
develop students’ phonologic skills. For home literacy environment (HLE), LB children reported a
larger gain in the percent of household members reading to them and telling them stories in the past
week. Overall, slow learners were able to participate equally or more so than non-slow learners in
most LB community activities.
As SC’s involvement in the project comes to a close, it will be up to government and
community stakeholders to ensure the continuation of the LB program and to pursue its
enhancement through the set of recommendations in bold type throughout this report.
3
1. Introduction, Context, and Implementation History
Introduction
This report examines the results of a learner background survey and reading assessment conducted in
September 2014 as the endline evaluation of the USAID All Children Reading (ACR) Grand Challenge in
Sri Lanka. The survey and reading assessment was administered to as many of the same children
assessed during the January/February 2013 baseline as could be found. The endline survey and reading
assessment covered 519 grade 3 learners throughout 30 schools in three Education Divisions. The 30
schools are split into 15 primary schools which received the Literacy Boost intervention and 15
comparison primary schools which did not receive the Literacy Boost intervention. The Literacy Boost
program includes teacher training, community reading activities, and age-appropriate local language
material creation to support emergent literacy skills among early-grade children. These skills include
concepts about print, letter awareness, single word reading of most used words, reading fluency,
reading accuracy, and reading comprehension. As part of Literacy Boost, learners are periodically
assessed in each of these skills through an adaptable assessment tool to inform programming and
estimate program impact. The data gathered from these schools is analyzed to present a snapshot of
the emergent literacy skills of grade 3 learners in these schools and to estimate the impact of the
Literacy Boost program.
The key research questions to be explored in this report include:
1. How comparable are learners in Literacy Boost schools versus comparison schools in terms of
reading skills, background characteristics, home literacy environment, and school environment?
2. What can the endline tell us about learners’ reading skills in grade 3? What does this mean for
future Literacy Boost programming should the Sri Lankan government continue the program?
3. What can we estimate about the impact of the Literacy Boost program? If impact exists, how
did impact vary by student background, home literacy environment, and other dimensions of
equity?
4. What role did Literacy Boost play in the reading skills improvement of students considered
‘slow learners’ versus their non-slow peers?
To investigate these questions, this report will first describe the research methods used; including
sampling, measurement, and analysis. Next, in order to see if groups are statistically similar, the
comparability of Literacy Boost and comparison schools will be examined through t-tests. The report
will then compare Literacy Boost and comparison learners’ scores for each of the emergent literacy
skills, exploring learners’ strengths and weaknesses in each skill. Next, the report will examine the
literacy skills that are already present in the sample, and what areas should Literacy Boost focus on.
The report will then investigate student backgrounds examined through t-tests. Finally, the report will
investigate any correlations with student background, school environment, or home literacy practices
& environment variables using multilevel regression analysis.
4
As SC’s involvement in the project comes to a close, it will be up to government and
community stakeholders to ensure the continuation of the LB program and to pursue its
enhancement through the set of recommendations in bold type throughout this report.
Context
Gampaha district is located approximately 35km north of the greater Colombo area and Minuwangoda
is one of the 4 education zones that belong to Gampaha district. Though there exist some small Tamil
and Muslim communities in the project area, the vast majority are Sinhala speaking children. Because of
the nature of semi-urban setting, the socio-economic status of the people residing in this district is
generally better than those residing in rural areas. According to the Primary Director of Ministry of
Education, the district’s average score of the Grade 5 scholarship examination, an annual nationwide
examination carried out at the end of the primary education, is higher than the average for districts in
the entire country. Nonetheless, the results in Minuwangoda area show that there is some room for
improvement.
In Sri Lanka, primary education lasts 5 years beginning from Grade 1 to Grade 5, which are further
divided into 3 key stages; Key Stage 1 consisting of Grade 1 and 2, Key Stage 2 consisting of Grade 3
and 4, and Key Stage 3 solely consisting of Grade 5. Throughout each key stage, students are taught by
the same teacher. For instance, a teacher teaching Grade 1 students in a given year will be teaching
Grade 2 in the following year as the students go on to Grade 2; likewise, a teacher teaching Grade 2 in
a given year will be teaching Grade 1 in the subsequent year. At the end of Grade 5, students may elect
to take the national scholarship exam. Since the students who mark exceptional scores can seek
advance study at the recognized schools, most of the parents are eager to send their children to after-
school tuition classes.
Implementation History
The project has two major components: a teacher training component at school level and a community
level component with activities targeting 40 villages surrounding the 15 intervention schools. The
project was implemented in collaboration with the Provincial Ministry of Education - Western
Province. The National Ministry of Education provided guidance to SC for designing and conduction
teacher trainings with In-Service Advisors (ISAs). Teacher training modules were adapted to the Sri
Lankan context through a series of adaptation workshops led by ISAs. Subsequently, teacher training
programmes were conducted for selected primary school teachers. This was a very successful
programme except for some delays in conducting the adaptation workshops due to the unavailability of
ISAs during the particular period. It was recommended to incorporate the modules to regular pre-
service training programme for primary teachers.
An implementing partner, a local NGO, Siddhartha Foundation was selected to deliver community
level activities. Forty community volunteers were appointed to lead reading activities at the community
5
level. In this way, the relevant education authorities focused on the school-level component only and
Siddhartha Foundation focused on the community-level component only.
One implementation challenge encountered was a delay in conducting the vision and hearing screening
portion of the learning challenges screening assessment, as the required process and entry points were
initially not cleared to SC. This was due to lack of proper coordination mechanisms between the
Ministry of Health and Ministry of Education. A detailed timeline of intervention activities can be found
in Appendix A
11. Methods
Sampling and Attrition Analysis
The sample for this endline assessment encompasses 519 grade 3 learners, divided between 15 schools
which received the Literacy Boost intervention (n of learners =257) and 15 comparison schools (n of
learners = 262). Schools were selected based on meeting following criteria:
Similar socio-economic status
Similar level of the educational achievement
More than 50 children attending the school (Government of Sri Lanka is providing special
scheme to the schools where there are fewer than 50 students)
As Minuwangoda Educational Zone consists of three educational divisions, namely Divulapitiya,
Meerigama and, Minuwangoda, ten schools that meet the aforementioned criteria were randomly
selected from each division. Once selected, two clusters were formulated based on the proximity. If it
was found that any particular schools is close enough from any one of the schools in the another
cluster, the school was omitted from the cluster and another randomly selected school was added.
This process continued until all schools in one cluster are far enough from the other cluster. After the
completion of this process, one cluster was assigned as the LB schools and the other as comparison
schools. Literacy Boost and comparison schools are not receiving additional interventions from SC or
other organizations that may have affected the implementation of Literacy Boost activities.
In each of the 30 schools, as many of the students sampled at baseline as could be found were
assessed. The baseline sampling guidance had called for a random sample of 20 learners and a
purposive sample of up to 6 slow learners. However due to class size constraints, at baseline the total
number of randomly sampled students ranged from 6 to 20, and the total number of slow learners
assessed at each school ranged from 1 to 6.
Of the 640 children assessed at baseline, 519 were found and assessed at endline. The attrition rate
of 19% did not differ significantly between LB and comparison schools. Of the students who could
6
not be found, two-thirds had transferred to another school and one-third were absent from school
on the day of assessment. No children had dropped out of school.
Measurement
Table 1 offers examples of background and home literacy indicators and offers a detailed description of
reading indicators.
Table 1: Data Collected
Student background Examples
General Sex, age, language spoken at home, work
School-related Distance to walk to school, repetition history
Socioeconomic status
Type of home, household size, household
amenities/possessions Home Literacy Environment
Access to print Materials present in home, types of materials
Reading at home
Presence and percentage of family members who children see read,
and who read to children Teacher Practices and Student Participation in Community Activities
Teacher practices How teacher reads to class, how teacher teaches letters, words,
etc., help offered to students
Student participation in
community activities
Frequency of reading camp attendance, frequency of book bank
borrowing, frequency of reading with reading buddy
Reading Outcome Description
Concepts About Print N of concepts demonstrated correctly of 10
Alphabet knowledge N of letters/sounds known of 61
Vocabulary/Decoding N of single words read correctly of 20
Fluency N of words in a connected text read correctly in a minute
Accuracy Percentage of words in a connected text read correctly
Listening
Comprehension
N of 9 comprehension questions answered correctly after
listening to a text read aloud by the assessor (only for non- readers)
Reading
Comprehension
Number of comprehension questions answered correctly of 9 after
reading a text read aloud (only for readers)
These assessments were developed and pilot tested prior to baseline data collection
using the Literacy Boost Toolkit Assessment Component. All questions were asked to students in
Sinhala. For information on inter-rater reliability, please see Appendix B.
In addition to the data specified above, teacher interview information was collected, including which
trainings the teacher had attended, their feedback on the usefulness of the trainings, how often they
use certain active learning strategies in their class, and how strongly they believe in statements such as
7
‘With the right support, students are capable of reading, understanding, and summarizing a short story
by the end of grade 2.’ Due to teacher absence, this information was collected from 28 of the 30 grade
3 teachers.
Analysis
The critical purpose of this analysis is twofold: first, to test whether the students in the Literacy Boost
intervention schools and their comparison counterparts are equal in terms of background and skills.
That is, at baseline do these students possess the same resources and capabilities? Second, given the
findings about comparability between intervention and comparison students, this analysis will estimate
the impact of the Literacy Boost intervention controlling for any significant differences between groups.
In addition, this report will present a profile of grade 3 student’s reading skills, as well as an in-depth
analysis of each skill. Summary statistics will be used to analyze students’ performance in each of the
reading sub-tests. The purpose of this is to provide useful information for government and community
stakeholders to continue and expand the Literacy Boost program, given the learning gaps that exist
among grade 3 students in this area.
To test the comparability of learners in the Literacy Boost and comparison samples, this report will use
comparison of means through t-tests assuming unequal variance between the two samples. Summary
statistics, accompanied by t-tests, will be used to analyze learners’ performance in each of the reading
sub-tests. The impact analysis will be conducted through a ‘difference in difference’ multilevel
regression analysis. This means that the learning gains over time among comparison students will be
statistically compared to the learning gains over time among Literacy Boost students, controlling for
any variables what were found to statistically differ between the two groups during the comparability
analysis. Finally, this report will look to multilevel regression models to explore relationships between
literacy skill gains and student background characteristics, school environment, and home literacy
environment.
111. Children’s Background
The students are about eight years old on average, and all students speak Sinhala as their native tongue.
On average, students live with four other family members and have three of five common amenities in
the region (bike, electricity, refrigerator, toilet with water facility, and computer). Finally, almost all
children (96 percent) do chores, 23 percent do chores less than 3 days per week and 36 percent do
chores more than 3 days per week. Another 36 percent have to do chores every day. Girls and boys
report doing chores with equal frequency.
On average, students spend more than two hours studying out of school each day. Seventy-seven
percent of the students attend tuition classes. Most students attend tuition classes 2 days per week.
8
However only 30 percent spend time studying at home. More than 95 percent from both Literacy
Boost and comparison schools have attended early-childhood development (ECD) programs.
IV. Children’s Home Literacy Environment
An important aspect of reading development concerns the home literacy environment (HLE). How are
children exposed to the printed word in the home? How much access do they have to books and print
to practice their nascent reading skills? Many Literacy Boost activities are centered on helping parents
and communities to enhance the HLE. As such, it is important to measure where learners' HLEs begin,
and how they change over the course of time. Figure 1 displays the different types of printed materials
that learners may or may not have at home.
Nearly all learners have some type of reading materials at home. The most common reading material in
students’ homes is text books. The least common reading materials are comics and booklet. No
statistically significant differences exist between Literacy Boost and comparison groups. Further analysis
discovered more than 80 percent of students have at least 4 types of books at home. The
prevalence of print, especially storybooks, is a strong foundation for home support of
literacy learning. However, the situation could be improved even further by diversifying
the types of reading materials available to students – especially child-friendly types of
reading materials such as comics.
The HLE is not only about materials in the home, but how those materials are used to engage the child
in reading and learning. Hess and Halloway (1984) identified five dimensions of the home literacy
environment that are theoretically related to reading achievement in children. The first is value placed
on literacy, which we operationalize by asking the learners whether they see anyone reading at home.
97%
78%
91% 94%
44% 48%
0%
20%
40%
60%
80%
100%
% o
f st
ud
en
ts r
ep
ort
ing
pri
nt
in h
om
e
Figure 1: Presence of Reading Materials in the Household
9
The second is press for achievement, which we operationalize as individuals telling the student to study.
The third is the availability of reading and print materials, which we operationalize as the amount of
printed materials at home (see Figure 1). The fourth dimension is reading with children, which we
operationalize by asking the learners whether anyone reads to them at home. The last is opportunities
for verbal interaction, which we operationalize as family members telling stories to learners. Figure 2
shows how students measure up in terms of engagement in these four home literacy environment
activities.
Students report seeing almost 90 percent of their household members read in the past week. Further,
almost all student report that their household members have encouraged them to study in past week.
On average, 82 percent of their household have read to them, or told them a story in past week. No
statistically significant differences exist between Literacy Boost and comparison groups.
Comparability of Literacy Boost and comparison students at baseline
Literacy Boost and comparison students differ significantly on 8 of 33 baseline attributes. Some
differences favor the children in comparison schools: at baseline, they had more types of books in the
home, a greater proportion of household members read or told stories to them, more read to others
in the community, and their teachers had a few more years of experience on average. However, other
differences favor the LB group: LB students reported lower chore frequency and more study time,
their class size is slightly smaller, and their schools are closer to the district center on average. Despite
these background dissimilarities, there are no significant differences in children’s baseline reading skills
between the two groups. Thus, of the remaining sample of children assessed at endline, the
two groups were similar at baseline but differ in some important ways. Therefore, all of
these variables are controlled for during impact analysis.
91% 99%
81% 84%
0%
20%
40%
60%
80%
100%
seen reading Encouraging child tostudy
Reading to child Telling story to child
% o
f h
ou
seh
old
mem
ber
s
Figure 2: In the past week, % of household members reported to be...
10
V. Impact of Literacy Boost on Reading Skills
To estimate the impact of the Literacy Boost program, a series of difference-in-difference regressions
were run to compare the learning gains of LB students with the learning gains of comparison students,
controlling for all baseline characteristics found to differ between the two groups. Standard errors
were clustered to represent the fact that students are clustered within schools. The output of these
regressions can be seen in Appendix D.
As shown in Figure 3 below, LB students gained significantly more words correct per minute (a
measure of fluency) between baseline and endline than did comparison students. The additional six
words correct per minute gain over time is equivalent to an additional 1.6 months of learning (0.35
effect size). However, the endline scores and gains in all other skills were equal between LB and
comparison students. Thus, from the data collected at baseline and endline assessments we
are able to conclude that LB programming accelerated the development of students’
fluency skills.
Denotes significant difference at * p<0.05, **p<0.01, ***p<0.001
Figure 4 below displays the proportion of students who qualify as ‘emergent,’ ‘beginner,’ or ‘readers
with comprehension.’ Emergent students either cannot read or can read but can answer less than
three of the five literal comprehension questions asked after reading the story. Beginner students can
answer three of the five literal comprehension questions, and readers with comprehension can answer
four or five literal questions correctly. In addition, beginners and readers with comprehension must
also have a minimum level of fluency or accuracy.
92 90 86 87 82 82
16 18
63 66 47 49
5 7 10 9 11 11
41 35
27 23
21 15
0
10
20
30
40
50
60
70
80
90
100
LB Control LB Control LB Control LB Control LB Control LB Control
Concepts about
Letters Familiar Words Fluency** Accuracy Reading
Comprehension% ite
ms
co
rrect
(or
wo
rds
co
rrect
per
min
ute
fo
r fl
uen
cy)
Figure 3: Reading Skills Baseline Scores and Gain Scores by Sample
Group
Gain
Baseline
11
In addition to the literal comprehension questions, students were also asked one summary, two
inferential, and one evaluative comprehension question. Figure 5 shows the relative baseline scores and
gain scores of LB and comparison students in terms of their advanced reading comprehension skills.
This is likely due to the fact that the baseline inferential questions tested basic inference (all clues
contained within the story yield only one plausible answer) while the endline tested advanced inference
(student must draw on their own knowledge to infer one of multiple plausible answers.)
Denotes significant difference at * p<0.05
LB students had a statistically significant higher gain in correctly answering the evaluative
question (0.13 effect size). Thus, LB may be helping students improve their evaluation
reading comprehension skills. However, children’s advanced comprehension skills still lag
their literal comprehension skills, and more work remains to be done to improve all types
of comprehension.
14%
11%
42%
46%
12%
11%
19%
12%
74%
78%
39%
42%
0% 20% 40% 60% 80% 100%
Control
LB
Control
LB
End
line
Bas
elin
e
Figure 4: Reader with Comprehension Tiers by Sample Group
Emergent Beginner Reader with Comprehension
33% 31% -10% -18% 25% 13%
-40%
-20%
0%
20%
40%
60%
80%
100%
LB Control LB Control LB Control
Summary Inferential Evaluative*
Fre
qu
en
cy o
f co
rre
ct r
esp
on
ses
Figure 5: Advanced Comprehension Baseline Scores and Gain Scores
Gain Baseline
12
15 17
46 35
010203040506070
LB Control
Wo
rds
co
rrect
per
min
ute
Significant at p<0.05
Figure 6: Fluency Baseline Scores
and Gain Scores for Slow Learners
Gain
Baseline
VI. Impact of Literacy Boost for Slow Learners and Children Who Screened
Positive for Language and Learning Challenges
The classification system of ‘slow learner’ is one already used by teachers in assessing/evaluating
students. Save the Children provided teachers with a narrower definition by which to identify up to 6
slow learners per classroom at baseline: "a child who has trouble identifying letters and/or who cannot
count up to 20.” Students designated as slow learners at baseline appear to have benefitted even more
than the average LB learner in terms of fluency. Figure 6 shows baseline scores and gains for LB and
comparison slow learners.
Relative to comparison slow learners, LB
slow learners improved their fluency by
an additional 11 words correct per
minute. This improvement is the
equivalent of nearly an additional three
months of schooling for slow learners
(0.53 effect size). Other than fluency, within
both LB and comparison schools slow learners
improved their skills at the same rate as non-
slow learners. Figure 7 below displays the gains
of non-slow and slow learners in LB schools.
This finding may be explained in part by the
fact that slow learners with lower baseline reading scores were much more likely to have
transferred schools and thus could not be found at endline. For this reason it is difficult to
estimate the impact of LB on the neediest of the slow learners.
7 8 9 12 10 15
40 43
27 26
22 18
0102030405060708090
100
No
n-S
low
Slo
w
No
n-S
low
Slo
w
No
n-S
low
Slo
w
No
n-S
low
Slo
w
No
n-S
low
Slo
w
No
n-S
low
Slo
w
Concepts aboutPrint
Letters Familiar Workds Fluency Accuracy ReadingComprehension%
of
corr
ect
ite
ms
(or
wo
rds
corr
ect
pe
r m
inu
te f
or
flu
en
cy)
Figure 7: Reading Skills Baseline Scores and Gain Scores by LB Non-Slow and LB Slow Learners
Gain
Baseline
13
At endline, teachers were again asked to identify the slow learners among the assessed students.
Interestingly, the correlation between baseline and endline designation is almost zero, meaning that
children who were considered slow learners at baseline were not considered slow learners at endline,
and those considered slow learners at endline had been considered non-slow at baseline. No
statistically significant difference was found in the change in number of slow learners over time
between LB and comparison schools. Students identified as slow learners at endline show significantly
lower baseline and endline scores than students without this endline designation, and the learning gap
between them and their non-slow peers actually widened over time. Among this group of endline slow
learners, LB slow learners did not learn significantly more than comparison slow learners over time.
All of the baseline slow learners were screened for various types of learning challenges. The largest
number of students screened positive for background-related challenges (76 students, equal to 15% of
the sample) and language & learning challenges (15 students, equal to 3% of the sample).1 Both types of
challenges were strongly associated with much lower literacy skills at baseline. No evidence could be
found that LB helped students with background-related challenges. In both LB and comparison schools,
students with background-related challenges continued to learn less over time than students without
these challenges.
However, LB appears to have considerably supported the learning of students with possible language
and learning challenges. Figure 8 below displays the much larger gains in Concepts about Print, Letters,
Fluency, and Accuracy among LB students with language & learning challenges versus comparison
students with these challenges. Except in the case of fluency, these gains were enough for LB
students to make notable progress closing the learning gap with their non-slow peers.
Strikingly, 66% of LB students with language & learning challenges became readers by
endline versus only 22% of comparison students with the same challenges. While these are
promising indications, it should however be kept in mind that the sample size was extremely small –
only six LB students and nine comparison students screened positive for language & learning challenges.
1 It should be noted that this screening is not equivalent to a medical diagnosis.
14
14 20 45 52 34
42 21 14
0
20
40
60
80
100
Boys Girls Boys Girls
Fluency* ReadingComprehension~
Wo
rds
corr
ect
pe
r m
inu
te
(flu
en
cy)
or
% o
f w
ord
s co
rre
ct
(acc
ura
cy)
Figure 9: Fluency and Accuracy Baseline Scores and Gain Scores by Sex
Gain Baseline
Denotes significant difference at ~p<0.1, * p<0.05
VII. Skill Gains of Boys and Girls
When examining student skill gains between boys
and girls, girls tended to score higher in many
skills at baseline (especially in the LB group). By
endline, boys had improved their skills slightly
more than girls in all skills except fluency, and
thus the two groups are equal in most skills at
endline. This trend occurred in both LB and
comparison schools, and is represented by the
Reading Comprehension scores in Figure 9.
Denotes significant difference at ~p<0.1, * p<0.05
Fluency did not fit this trend. Although girls again scored higher than boys at baseline, girls also
improved their fluency skills more than boys over time and thus girls have much higher average fluency
at endline. This occurred equally in LB and comparison schools.
VIII. Classroom and Home Literacy Environment
Differences between LB and comparison students in terms of student-reported classroom and home
literacy environment were tested. In terms of classroom environment, while some insignificant
student-reported indications exist that LB teachers may be more likely to ask students questions while
20 18
45 33 45
45
17 8
58
28 37 16
0
20
40
60
80
100
LB Control LB Control LB Control LB Control LB Control LB Control
Concepts aboutPrint~
Letters* Familiar Words Fluency* Accuracy* ReadingComprehension%
co
rre
ct it
em
s (o
r w
ord
s co
rre
ct p
er
min
ute
fo
r fl
ue
ncy
) Figure 8: Baseline and Gain Scores for Slow Learners who Screened Positive for
Language and Learning Challenge
Gain
Baseline
15
in the middle of reading a story and read newspapers to children, these require additional research to
verify as the sample size of this study may be too small to confirm these sample differences.
Grade 3 teachers from the classroom of assessed students were also given a survey about their
practices and beliefs. LB teachers are more likely to use formative assessment to monitor student
achievement and target support to smaller groups of students, but comparison teachers are more likely
to use it to revisit topics the class is struggling. Regarding the responsibility for helping slow learners,
LB teachers are less likely to state that it is the responsibility of the slow student him/herself or the
student’s parents to help, but LB teachers are more likely to state that it is the responsibility of other
students to help. Nearly all LB and comparison teachers stated that the responsibility also falls on the
teacher.
LB teachers report using “activities such as singing the alphabet song, playing letter games and reciting
rhymes to develop my students’ phonologic skill” with much greater frequency than comparison
teachers. LB teachers are less likely to agree that “It is more important to treat students as individual
learners rather than as a group.” There are indications that LB teachers more frequently ask specific
questions about the story when the teacher reads to students and more frequently schedule time for
students to read their own material, although these differences are not statistically significant and
require a larger sample size to confirm or refute.
Thus, the clearest impact of LB is on how teachers use formative assessment to monitor
student achievement and target support for smaller groups of students, and how teachers
use songs and games to develop students’ phonologic skills.
However, more work needs to be done to enhance teachers’ beliefs and practices where there is
room for improvement among both LB and comparison teachers. In terms of beliefs, LB programming
should help impress upon teachers the importance of treating students as individuals, of reading to
students on a daily basis, of differential instruction, of the proper way to correct incorrect student
answers, and of believing that all students are equally capable of learning although some need more
help than others. In terms of practices, LB programming should help teachers more frequently talk with
parents about how they can support their child’s learning and decorate their classroom with students’
work.
For home literacy environment (HLE), LB children reported a larger gain in the percent
of household members reading to them (marginally insignificant) and telling them stories
(significant) in the past week. Figure 10 below illustrates this finding. Despite these gains, 13% of
LB students come from households in which at least one person was seen reading in the past week,
but in which the student was not read to in the past week. Parents need to be continuously
encouraged to read to their children on a regular basis.
16
Denotes significant difference at ~p<0.1, * p<0.05
IX. Participation in Community Action Activities
Table 3 displays the self-reported rates of participation in LB Community Action activities by LB
students. Non-slow learners are more likely than slow learners to have met with a Reading Buddy in
the past week. However, slow learners are more likely to have attended a Reading Camp in the past
two weeks, to have used Make and Take materials at home, and to have participated in a Read-a-thon.
The LB program seems to be reaching slow learners in most activities.
Table 2: Participation in Community Action Activities
among LB Students by Learning Status
Community Action Activity Non-Slow Learner Slow Learner Overall
Borrowed book from Book Bank in past two weeks 63% 57% 62%
Borrowed book from Book Bank in past two weeks
and can name Book Bank 34% 30% 33%
Has had a problem borrowing from Book Bank 19% 20% 19%
Has a Reading Buddy and met with him/her last week 53% 43% 50%
Has attended a Reading Camp and did so in past two
weeks 24% 32% 26%
Used Make and Take materials at home 16% 24% 18%
Participated in a Read-a-thon 26% 34% 28%
In an attempt to verify student self-reporting of visiting the Book Bank, students were asked to name
their local Book Bank. Once this is factored in, the proportion of students verified to be visiting the
Book Bank drops by nearly half. However, it is possible that not all students know the name of their
Book Bank. In any case, efforts should be made to increase access and participation in LB
40 47
40 47
17 8 19 8
0
10
20
30
40
50
60
LB Control LB Control
Reading to child~ Telling story to child*% o
f h
ou
seh
old
mem
bers
Figure 10: In the Past Week, Percent of
Household Members ...
Gain Baseline
17
community activities among children, and reports of problems using the Book Bank more
closely investigated.
X. Children’s Reading Skills
This section presents a profile of grade 3 students’ reading skills, as well as an in-depth analysis of each
skill. The purpose of this is to provide useful information for government and community stakeholders
to continue and expand the Literacy Boost program, given the learning gaps that exist among grade 3
students in this area.
Concepts About Print
This report will not analyze the Concepts about Print subtest, as children have mastered this skill (96
percent correct).
Letter Identification
The letter identification subtest examined learners’ letter awareness. Learners were shown a chart of
basic & compound letters and asked to name the letter or pronounce the letter sound. On average,
Literacy Boost and comparison learners correctly identified 95 percent (58 letters) out of a combined
total of 37 basic and 24 compound letters. There was no statistically significant difference between LB
and comparison students.
Except for one letter which was difficult for many students, all the other basic letters were correctly
identified by more than 90 percent of students. The difficult basic letter was ‘ළු’, which 79 percent of
students were able to identify.
In terms of compound letters, students had difficulty reading (ද්) and (ව ෝ) compound letters. Eighty-
one percent of students were able to identify those difficult letters on average. Other compound
letters were identified by above 90 percent of students.
It is clear that the letters that were most commonly known are some of the most
commonly used in Sinhala. They are found in names and many other words commonly
used and related to everyday life.
Word Recognition: Most Used Words
The most used words (MUW) sub-test consists of a chart of 20 words that the student is asked to
read. These 20 words were identified as ‘most used’ by tabulating the number of times a word
appeared in learners’ language arts textbooks.
On average, learners in Literacy Boost and comparison schools were able to read 93 percent of MUW.
There is no statistically significant difference between the two groups, meaning that we cannot
determine whether learners in Literacy Boost schools as a whole would be able to read more MUW
than learners in comparison schools.
18
Learners had the easiest time reading words that were two letters long and words that refer to family
members. These words have been correctly identified by more than 95 percent students. Only three
words out of 20 were correctly read by only 83 percent students. Two out of these three words were
confused by the learners due to the availability of other words with same sound which are commonly
used. The other word which on which students scored low contained six consonants in a row.
As Figure 11 below shows, the distribution of most used word scores is clustered around the average
(93 percent correct).
Proportion of Readers and Nonreaders
After the most used words sub-test, students were then asked to read aloud a passage of connected
text of 112 words in length. This passage was created with the support of In – Service Advisors from
Zonal Education Department in age appropriate basis. At this point in the assessment, assessors classify
students as either ‘readers’ or ‘nonreaders.’ Readers are defined as students who were able to read at
least 5 words correctly in the first 30 seconds of reading. At endline, 94 percent of students read at least 5 words correctly in 30 seconds and thus were classified as readers. All other students were
classified as nonreaders, and were read the passage by assessors. There were no significant differences
between the Readers and Nonreaders of Literacy Boost and comparison students.
Fluency and Accuracy
Fluency (words per minute read correctly) and accuracy (percent of the passage read correctly) are
presented together here because they are measured together in a single sub-test in which learners
read a passage aloud. The number of words learners read correctly in a minute is tracked for fluency.
As the student continues to read after the first minute, the total number of words read correctly from
the passage as a whole, no matter how long it takes the student, is computed for accuracy. This section
0%
10%
20%
30%
40%
50%
60%
0 5 10 15 20 25
% o
f sa
mp
le
# of most used words correct
Figure 11: Distribution of Most Used Word Score
19
presents this data for readers only2 in order to better understand the strengths and weaknesses of
students who can read a degree of connected text.
The average endline fluency rate for Literacy Boost and comparison readers was 58 words correct per
minute and their accuracy was 94 percent. Although the learning gain in fluency was significantly larger
for LB student overall, among endline readers there were no significant differences between the
fluency and accuracy skills of LB and comparison readers. As the Figure 12 histogram below shows, the
distribution of accuracy scores was clustered at or above the average. The highest frequency of
readers read with 100 percent accuracy / read all 112 words correctly. The distribution of fluency
scores was also clustered close to the average of 94 percent correct (106 out of 112 words).
Figure 12
Reading Comprehension
The final sub-test quizzed students who qualified as readers on a series of nine comprehension
questions related to the reading passage. This section presents this data for readers only3 in order to
better understand the strengths and weaknesses of students who can read a degree of connected text.
On average, both Literacy Boost and comparison students who could read correctly answered 70
percent of reading comprehension questions. There was no significant difference between the scores
of Literacy Boost and comparison students.
2 As opposed to including the zero scores of the nonreaders, which would lower the average scores for this data.
3 As opposed to including the zero scores of the nonreaders, which would lower the average scores for this data.
20
The easiest type of comprehension question to answer for all students were the factual questions,
where students were asked to recall facts that were presented in the story. This type of question
asked the students questions like "Whose is that monkey?" and "Did which fruit Udara show to the
monkey?” The hardest questions were the inferential questions. Inferential questions ask students to
use the information from the text to make inferences. The summary question, which asked students to
retell the story in their own words, was not much harder to the students when compared to the
factual questions. Students received 1 point on this question if they were able to accurately recount
four of the six most important parts of the story. Students scored similarly well on the evaluative
question, which asks student to state an opinion on a feature of the story and support that opinion
with reasons from the story. Figure 13 shows how students who qualified as readers performed on
each comprehension question.
Readers with Comprehension
Students reading with comprehension is the ultimate goal of Literacy Boost. As such, a new composite
measure to focus attention on this goal as well as to track progress in terms of equity, in terms of all
children reading with comprehension, is displayed in Figure 14 below. The Literacy Boost program
classifies students into Emergent, Beginning, and Reading with Comprehension tiers based on their
comprehension skills. Emergent students either cannot read or can read but can answer less than three
of the five literal comprehension questions asked after reading the story. Beginner students can answer
three of the five literal comprehension questions, and readers with comprehension can answer four or
five literal questions correctly. In addition, beginners and readers with comprehension must also have a
minimum level of fluency or accuracy.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
% o
f sa
mp
le
Figure 13: Average Scores on Each Comprehension Question among Readers
LB Comp
21
In this sample, 13 percent of LB and comparison students are classified as being Emergent readers.
Eleven percent of students fall under the Beginner category, and most of the students qualify as
Readers with Comprehension (76 percent). While three-fourths of Literacy Boost and comparison
students have achieved the ultimate goal of Reading with Comprehension, one-fourth of children still
have not. There were no significant differences between Literacy Boost and comparison students on
this measure.
XI. Literacy Boost Skill Profile and Skill-Specific Recommendations
14%
11%
42%
46%
12%
11%
19%
12%
74%
78%
39%
42%
0% 20% 40% 60% 80% 100%
Control
LB
Control
LB
End
line
Bas
elin
e
Figure 14: Reader with Comprehension Tiers by Sample Group
Emergent Beginner Reader with Comprehension
84.2
77.9 72.8
14.7
55.0
75.6
90.9
85.1 79.1
35.0
90.0
60.5
89.8 86.8 83.1
16.3
64.8 68.6
97.6 97.0 96.0
64.1
95.4
73.1
0.0
20.0
40.0
60.0
80.0
100.0
CAP ( % correct ) Letter ID ( % correct ) Familiar words ( %correct )
Fluency ( Words perminute)
Accuracy ( % Readcorrect )
ReadingComprehension
Literacy Skill Profile, Literacy Boost Students Only
Slow learnerBaseline
Slow learnerEndline
Non slow learnerBaseline
Non slow learnerEndline
22
Table 3. Program Recommendations by Skill
Skill Program implications
Letter identification Letter knowledge is the ability to identify the names of letters of the Sinhala
alphabet. Familiarity with the letters of the alphabet is a powerful predictor of
early reading success. When children can quickly recognize letters, they can
begin to appreciate that all words are made of sequences and patterns of
letters. Letter games using low cost materials, letters songs and group
activities can be used to increased letter knowledge.
Reading Single
Words- familiar
words
More than half of the assessed students at the endline have been able to read
the familiar words. Fifty six percent (56%) of children can read 20 words
correctly. It is recommended to create more opportunities for reading
through book banks, daily reading activities in schools and continued
encouragement of children.
Fluency (words per
minute)
The students have significantly improved their reading fluency. This project
was implemented in the semi urban area but we cannot expect the same level
of fluency among students in the remote area. Fluency develops over time and
through substantial practice. Teachers can support students to improve their
fluency by giving opportunities to read aloud in the class room and providing
dedicated time to read story books. In the community level, children should
be encouraged more to practice their reading with reading buddies and
participate in other community level reading activities such as read-a-thons and
readings camps.
Accuracy At the school level, reading with children – e.g. choral reading and echo
reading – would improve accuracy. At the community level, reading camp
leaders and parents should continue to support children on reading with
accuracy.
Reading
Comprehension
Reading comprehension, or being able to understand, interpret and use
information derived from text, is the main objective of reading. Even though
activities were conducted by focusing on comprehension, the end line value
for reading compression is 70%, which is quite low compared to other reading
skills. Teachers, reading camp leaders and parents need to be encouraged on
asking questions focusing on literal and inferential comprehension questions.
Story time is an opportunity for improving comprehension by asking questions.
XII. Summary of Findings and General Program Recommendations
LB programming accelerated the development of reading fluency skills, especially among slow
learners.
LB programming may have enhanced students’ advanced comprehension skills, however much
more work is needed to build students’ comprehension by integrating this with other skills at all
grade levels.
23
LB programming helped students who screened positive for language & learning challenges catch up
to their peers in almost every skill.
LB programming helped teachers improve their use of formative assessment, songs, and games.
However, more work is needed on teachers’ beliefs and practices where there is room for
improvement.
LB is making a positive impact on household literacy practices. Furthermore, slow learners are
reporting equal or greater access to LB community activities. However, community action
participation can be increased.
This promising data should be used to advocate for the LB model of supporting school and
communities to help children to read.
24
Appendix A: Implementation Timeline
Intervention Dates
I. Literacy Boost Teacher Training
Orientation of the LB programme for Education authorities December 2012
First Training of trainer (ToT) workshop to Primary ISAs (In Service advisors) (9
ISAs) and adopted to the 1st training module according to the Sri Lankan context.
May 2013
Pilot workshop session conducted by 9 ISAS to primary teachers represent from
another school and ISAs got confident to do the session in future - 28 primary
teachers and 04 directors in the Zone
31-5-2013
First Teacher Training: Introduction to Development and Instruction for
Children and Letter Knowledge
Three parallel trainings were conducted in Minuwangoda, Diulapitiya and
Meerigama division
68 (55 females & 13 males) grade two and four -primary teachers and ISAS
10,11/ 6/ 2013
Second ToT and adaptation session with Primary ISAs (In Service advisors ) ( 9
ISAs) : Struggling learners - SNAP (Special Need Action Pack ),
Formative assessment
27/30 /7/2013
Second Teacher Training on Struggling learners - SNAP (Special Need
Action Pack ), Formative assessment
Three parallel trainings were conducted in Minuwangoda, Diulapitiya and
Meerigama division - 43 grade two and four -primary teachers and 9 ISAs
07,09/10/13
Third TOT and adaptation session with Primary ISAs (In Service advisors ) (9
ISAs and 2 directors) : Phonemic Awareness and : Reading Fluency
05/11/2013
Third Teacher Training on Phonemic Awareness and Reading Fluency
30 (26 females & 4males) grade two teachers conduced as three parallel sessions
conducted in Meerigama, Minuwangoda and Diulapitiya.
04, 06/11/2013
Refresher training programmer for 44 Primary teachers on letter knowledge,
SNAP, Phonemic awareness and fluency.
01/02/ 2013
Fourth ToT and adaptation session with Primary ISAs (In Service advisors) (9 ISAS
and 1 director): Vocabulary.
05,07/02/2013
Fourth Teacher Training: Vocabulary
27 grade two and three primary teachers attended conducted as three parallel
workshops in Meerigama, Minuwangoda and Diulapitiya.
19/2/2013
Fifth ToT and adaptation session with Primary ISAs (In Service advisors ) (9 ISAs
and 1 director) : Comprehension
09/05/2013
25
1.1 SNAP (Special Need Action Pack )
Peers support meeting with teachers to discuss the progress of catch up classes and
application of the SNAP in the class rooms – 43 primary teachers, 9 ISAs and 2
directors.
Feb, March May, July
2013
Three Screening tests were conducted by doctors for providing spectacles and
hearing aids to children in Minuwangoda, Diulapitiya and Meerigama
2/14 , 01/14 , /2/14
Provision of hearing and visual aids for identified children - came form 15 schools 27/3/14
2.Community activities
2.1 Enhancing the Literacy Environment
Introduce Literacy boost programme to government officials-Discussion with Non
formal project officers to identify the catchment areas of the project, identifying the
community leaders. (13 participants)
20 /3/14
Discussion with the ISAs on activities in the community action module (14
participants)
17/5/13
Aware the community action module for 38 community leaders – conducted by
three ISAs in Marapola Maha Vidyalaya
29/30 /6 13
Training programmer for 40 community leaders, and 28 camp leaders to discuss
the community action module – conducted by four ISAS
15, 16/8/14
Distributed books to book banks – 74 books were given at the beginning totally
148 books were distributed to each book bank
29/8/13
Training on 40 community leaders – functioning of the book bank, leveling the
book, lending and borrowing the books.
7,8//9/13
Work shop for camp leaders (54) ,community volunteer (40) and reading buddies
(266) on preparation of low cost materials for book banks facilitated by 3 ISAs and
10 primary teachers
27/28/12/13
A total of 2869 children (Male) 1370 (Female) 1499 attending. Book banks and
facilitated by 40 community leaders- 2 days per month
From August 13 to date
2.2 Community reading activities
Workshop for Reading buddies , camp leaders, community leaders to discuss the
role of reading buddies , camp leaders –facilitated by the non-formal project officer
and 2 ISAs
16/17/11/13
Training for 291 reading buddies– role of the reading buddies, effective
communication related to reading stories, how to guide younger children to
inculcate reading habits. Facilitated by staff in room to read and three Non formal
officers
28,29/ 6/14
Training for the 40 camp leaders and 40 community leaders – how to conduct
reading camp activities, leadership skills – facilitated by partner staff
28,29/ 6/14
Workshop for 40 community leaders, 270 reading buddies, 40 camp leaders
focusing in reading methods , enjoining with reading activities
Resource person – Ramya Nawarathne
15/16/2/14
26
Reading buddies
pairs form across 40 reading camps 333 elder children read for 950 younger
children to books - twice a month ( first day on 3rd week Second day on 4th
week )
From November 2013 up
to date
Reading camps
78 Reading camps facilitated by 40 Reading Camp Volunteers and supervised by
community leaders , social mobilizers of the partner organization .Total number of
1979 children 913 (Male) 1066 (Female )attending to the reading camps
Typical activities: Open play, Song, Story including Reading for Children, Activity
based learning, Make and Take, Journaling (Drawing, Writing), conducting - 2 days
per month
From October 2013 up
to date
Community- Read -A- Thon
Reading festival celebrated in 40 book banks. Reading completion and reading
festival were conducted involving -1492 children, 161 parents.
From September 2013 up
to date
Story time
Two days per month - the elderly members from the community were invited to
tell traditional oral stories to 1463 children in 40 Reading Camps.
From September 2013 to
date
40 community leaders participated to cluster meeting conducted in three places
to discuss the progress of the community activities – (Minuwangoda, Meerigama
and Diulapitiya )
Once a month
2.3.Reading awareness workshops
Local resource group(15) formed ( selected 15 out of 40 community leaders ) and
trained on parents awareness sessions
17/12/13
Awareness programme for 32 community leaders about the content of the
reading awareness workshops – Minuwandoga Pradesheya Shaba
26/1/14
36 parents awareness workshop for Mothers (684) Fathers (72) conducted by 15
local resource group.
Feb –May -2014
Discussion with 28 community leaders in terms of the application of the parents
awareness session - Venue Udugampolaa Pradeshiya Shaba
15/3/14
Phasing out of the community action component with over 1500 participants
representing of children, parents, community leaders, Provincial and Zonal
education authorities, representatives of the Chief Minister, Siddhartha foundation
staff and SCI. Certificates of appreciation were awarded to Reading buddies,
Community facilitators, reading camp leaders and other key active participants in
the community
27/9/14
27
Appendix B: Inter-rater reliability To test inter-rater reliability, 12 percent of learners (60 out of 519) were assessed by two
enumerators simultaneously. Long one-way ANOVA techniques were used to calculate the intra-class
correlation within pairs of assessors for a measure of reliability. Table A presents the results below.
Using Fleiss’ benchmarks for excellent (ICC>0.75), good or fair (0.75>=ICCA>0.4), and poor
(0.4>=ICC).
Table A. Interrater Accuracy and Reliability
Literacy Skill Sub-Test Inter-rater Reliability Rating
Concepts about Print 0.98 Excellent
Letter Knowledge 0.99 Excellent
Most Used Words 0.99 Excellent
Reader or Nonreader? 1.0 Excellent
Fluency 0.99 Excellent
Accuracy 0.97 Excellent
Reading Comprehension 0.98 Excellent
There was excellent inter-rater reliability on every measure. This means that assessors had near-
perfect agreement on the scoring of every subtest. Because inter-rater reliability was very high,
we can be confident that the internal validity of the scores is good.
28
Appendix C: Full Background and Home Literacy Environment Statistics at
Endline
Table B1: Background Characteristics at Endline, by Sample Group
Child Background Characteristics Comparison
(N=262) Literacy Boost
(N=257) Significant difference
Age 7.9 7.9
% female 48% 45%
% have attended ECD program 97% 98%
Number of family members in household 5 5
Number of amenities (of 5) 3 3
% who do chores at home 96% 95%
time spent studying (in minutes) 145 136 0.046
Height in centimeters 125 124
Weight in kilograms 23 22
Body Mass Index 14.40 14.23
29
Table B2. Home Literacy Environment at Endline, by Sample Group
Home Literacy Environment
Comparison (N=262)
Literacy Boost (N=257)
Significant
Difference
Reading Materials in Home 100% 100%
Religious books 78% 77%
Newspaper 92% 91%
Textbooks 97% 97%
Comics 43% 46%
Story books 95% 92%
Booklet 47% 48%
See people read at home 93% 89%
People at home read to you 81% 81%
Total people in home 856 861
Total people seen reading at home 573 590
Total people read to you at home 461 495
% of Family Seen Reading 93% 95%
% of Family that Helps Student to Study 99% 99%
% of Family that Reads to the Student 81% 81%
% of Family that Tells the Student Stories 84% 84%
30
Appendix D. Regression Results
(1) (2) (3) (4) (5) (6) (7) (8)
Gain Scores # CAP Qs #
Letters #
MUWs %
Readers #
WCPM Accuracy
(%) # Reading
Comprehension Qs % RWC
LB student screened positive for language challenge (interaction) 1.324~ 9.355* 2.965 0.479~ 7.580 0.431* 2.084 0.0763 Student screened positive for language challenge -1.169~ -5.537 -2.221 -0.519** -27.8*** -0.490*** -2.998** -0.56***
Student attends LB school -0.0626 -0.204 -0.140 -0.00699 6.043** -0.00394 0.320 0.0161
Female 0.0446 0.395 0.256 0.0159 7.978*** 0.0274~ -0.139 -0.062~
SES quintiles 0.0266 0.0963 0.0658 0.0129* 1.162 0.0132* 0.145* 0.032**
Log total reading material 0.0682 -0.231 -0.139 0.0192 1.445 0.00446 0.0378 -0.0116
% of household reading to child -0.211 0.140 -0.195 -0.0189 -0.535 -0.00235 0.813~ 0.0224
% of household telling story to child 0.127 0.666 0.152 0.0413 -0.497 0.0402 -0.0231 0.0858
Quintiles of chore intensity -0.0215 -0.0075 -0.0267 -0.0144* 0.775 -0.0121* 0.0613 0.0254*
Log of chore frequencies 0.135 1.092 0.132 -0.00557 -1.332 0.00285 0.0407 -0.0267
Attended nursery school 0.190 2.539 0.468 0.136 1.414 0.124 1.063 0.213
Age in years -0.0694 -3.825** -1.97*** -0.092*** -3.432 -0.0816** -0.430* -0.0392
Quintiles of study intensity 0.0331~ -0.0261 0.0274 0.00736~ 0.105 0.00430 0.0926 0.0116
Reads books to others -0.0410 0.694 0.639~ 0.0415 0.397 0.0335 0.0907 0.0459
Lends books to others 0.0283 -0.187 0.112 -0.00946 -0.487 -0.00763 -0.205 -0.0525
Family size including child -0.0336 0.0925 0.143~ 0.00346 -0.190 0.00226 0.00543 -0.0061
BMI -0.00536 -0.008 -0.0710 -0.00736 -1.000~ -0.00925* -0.147** -0.025*
Distance to district center in km -0.00170 -0.0342 -0.0022 -0.00079 -0.0194 -0.00116 -0.00543 -0.0017
# of students in assessed classroom -0.00303 -0.035~ -0.0156 -0.00043 -0.0575 -0.000588 -0.00893 -7.7e-05
Baseline teacher yrs of experience -0.000662 -0.006 0.00033 -0.00023 0.118 -0.000195 0.00935 0.00361~
Baseline score -0.854*** -0.72*** -0.73*** -0.882*** -0.0112 -0.813*** -0.737*** -0.733***
Constant 8.639*** 63.56*** 25.78*** 1.314*** 59.21** 1.232*** 7.249*** 0.875**
Observations 491 491 491 491 490 491 491 491
R-squared 0.747 0.812 0.751 0.806 0.152 0.782 0.638 0.497
Robust standard errors in parentheses, *** p<0.001, ** p<0.01, * p<0.05, ~ p<0.1