Content uploaded by Ben Richards
Author content
All content in this area was uploaded by Ben Richards on Jun 18, 2020
Content may be subject to copyright.
Early childhood education, language of instruction and child development
in Myanmar: Comparisons among ethnolinguistic groups
Ben Richards,a* Mami Umayahara,a Naing Maw,b and Nirmala Raoa
aFaculty of Education, The University of Hong Kong, Hong Kong;
bDepartment of Educational Psychology, Yangon University of Education, Myanmar.
*Corresponding author: Ben Richards, 507 Meng Wah Complex, The University of Hong
Kong, Pokfulam Road, Hong Kong. (852) 3917 0361. benrich@hku.hk
ORC-ID
Ben Richards: https://orcid.org/0000-0002-8809-9097
Nirmala Rao: https://orcid.org/0000-0002-5695-3156
Twitter
Ben Richards: @richards_bd
Citation for final published paper:
Richards, B., Umayahara, M., Maw, N., & Rao, N. (2020). Early Childhood Education,
Language of Instruction, and Child Development in Myanmar: Comparisons among
Ethnolinguistic Groups. Early Education and Development, 1-22.
https://doi.org/10.1080/10409289.2020.1755179
First author links:
Google Scholar: https://scholar.google.com/citations?user=c4brqikAAAAJ&hl=en
HKU Webpage: https://web.edu.hku.hk/staff/academic/benrich
HKU Scholars Hub: http://hub.hku.hk/cris/rp/rp02400
Twitter: https://twitter.com/richards_bd?lang=en
Running head: EARLY CHILDHOOD EDUCATION IN MYANMAR
Early childhood education, language of instruction, and child development
in Myanmar: Comparisons among ethnolinguistic groups
Research Findings
This study examined the relation between early childhood education (ECE) participation and
early child development in Myanmar. It considered whether this relation varied by ethnic
group, language(s) spoken at home, and language(s) of ECE instruction. Participants were
1,494 children (759 girls) from Myanmar, aged between 36 and 71 months. Children were
directly assessed using the East Asia-Pacific Early Child Development Scales. ECE
participation was associated with higher composite developmental scores and higher scores
across 7 domains, and ethnic group did not moderate ECE-development associations. ECE
dosage of between 12 and 18 months was associated with the highest developmental scores.
For ethnic minority children, minority-language ECE instruction was associated with higher
developmental scores regardless of language(s) spoken at home, but Myanmar-language ECE
instruction was associated with higher developmental scores only for children also exposed to
Myanmar at home.
Practice or Policy
Findings highlight the benefits of ECE participation for the holistic development of both
ethnic majority and minority children. The policy of ECE programme expansion in Myanmar
, but further efforts may be needed to
deliver language support for ethnic minority children not exposed to Myanmar language at
home.
EARLY CHILDHOOD EDUCATION IN MYANMAR
2
Introduction
achievement and well-being is well established in many diverse fields of research (Campbell
et al., 2014; Irwin, Siddiqi, & Hertzman, 2007). However, much of what is known about early
child development and its influence on later life is from high-income countries (Henrich,
Heine, & Norenzayan, 2010). Much less is known about developmental patterns in low- and
middle- income countries (LMICs), as well as possible risk and protective factors that may
born in LMICs (UNESCO, 2015), and that children in LMICs can experience a greater
number of risks to their development than children living in high-income countries. A recent
estimate suggests that as many as 250 million children (43%) below five years of age in
LMICs are at risk of not attaining their developmental potential (Black et al., 2017).
Research from high-income countries and emerging evidence from LMICs suggests
that participation in early childhood education (ECE) programmes can be beneficial to
t (Agirdag, Yazici, & Sierens, 2015; Britto et al., 2017; Larson, Russ,
Nelson, Olson, & Halfon, 2015; Rao, Richards, Sun, Weber, & Sincovich, 2019). ECE
programmes in LMICs have been shown to be associated with improved cognitive, motor,
and psychosocial outcomes (Rao, Yousafzai, & Ip, 2017). ECE participation may protect
against some of the increased risks faced by children in LMICs and may therefore be an
important policy tool for governments seeking to improve developmental outcomes and
mitigate disadvantage. ECE participation rates in LMICs are much lower than in high-income
countries but are increasing rapidly (World Bank, 2018), and as governments in LMICs
continue to implement policies of ECE expansion, it is crucial that up-to-date and
contextually appropriate evidence from LMICs is available to guide policymaking.
EARLY CHILDHOOD EDUCATION IN MYANMAR
3
ECE participation rates can vary markedly by ethnolinguistic group, with ethnicity-
based disadvantages in education present in many countries (UNICEF, 2016). Nevertheless,
when ethnolinguistic minority children do participate, evidence from some countries suggests
an association with improved child outcomes (Gibbs, 2014). Determining the language(s) of
instruction (LOI) used in ECE programmes is a key issue in ensuring equitable outcomes for
children from different ethnolinguistic groups, particularly in countries with large
ethnolinguistic minority populations. Evidence points to the importance of early instruction in
a language with which children are familiar and suggests that bilingual education may be
beneficial (Durán, Roseth, & Hoffman, 2010), but there may be constraints to delivering
instruction in multiple languages, including difficulties in obtaining translated curriculum
materials, and the language competencies of teachers (Ball, 2010). Decisions about LOI may
also be guided by a desire to promote a common language, aiming to ensure that children are
in future economically productive, and able to take part in public affairs and national politics
(Barry, 2002). Despite the importance of determining the most appropriate model of LOI in
ECE, there is a dearth of robust research examining relations between ECE programme
(Rao, Yousafzai, et al., 2017), including
variations in ECE participation and child development by ethnolinguistic background and
LOI. Such research is fundamental to address issues of theoretical importance to
developmental science and of practical importance to educational policy.
In this context, this study examines the relation between ECE participation and early
child development, and whether this relation varies by ethnic group, language(s) spoken at
home, and language(s) of ECE instruction, amongst 3- to 5- year-olds in Myanmar. The
country has a large number of different ethnolinguistic groups, with those who do not speak
the dominant Myanmar (Burmese) language as their first language constituting nearly a
quarter of the population (World Bank, 2014). Further, the Government of Myanmar recently
EARLY CHILDHOOD EDUCATION IN MYANMAR
4
implemented a policy of ECE programme expansion (Republic of the Union of Myanmar,
2014). This study uses data from the East Asia-Pacific Early Child Development Scales
(EAP-ECDS) collected in Myanmar in 2016-17 (Rao, Richards, & Umayahara, 2017), which
provide a unique opportunity to understand whether ECE participation is associated with
early child development for children in a variety of different circumstances.
ECE participation and developmental outcomes in LMICs
Existing evidence has demonstrated ECE to be be
but not all, LMICs. Gove, Dombrowski, King, Pressley, and Weatherholt (2018) found
associations between earlier ECE participation and literacy skills in primary school in 7 out
of 16 LMICs. ECE has been shown to predict outcomes across a number of different domains
including cognition, motor, and social skills although ECE interventions in LMICs have
typically had the largest associations with cognition (Rao, Yousafzai, et al., 2017).
Key elements of structural and process quality of centre-based ECE have been shown
(Slot, 2018). Aspects of process quality,
including staff-child interactions, have been shown to be particularly beneficial
development (Sylva et al., 2006). Indicators of structural quality such as group size, staff to
child ratios, and staff qualifications, have been seen as important preconditions of process
quality (Phillips, Mekos, Scarr, McCartney, & Abbott Shim, 2000). However, whilst high
structural quality has been shown to predict high process quality in some cases (Rao, Koong,
Kwong, & Wong, 2003), other studies have found structural and process quality indicators to
be unrelated (Slot, Leseman, Verhagen, & Mulder, 2015). In addition to quality, overall
(Ou, Arteaga, & Reynolds,
2019). Studies have suggested that increased duration spent in ECE programmes is associated
with improved school readiness, including an expanded vocabulary (Burchinal, Zaslow, &
Tarullo, 2016), and better literacy and numeracy skills for children exposed to two years of
EARLY CHILDHOOD EDUCATION IN MYANMAR
5
ECE rather than just one (Domitrovich et al., 2013). Although most evidence is from high-
income countries, Behrman (2006) found that IQ scores in the Philippines were higher in
children exposed to more than 17 months of ECE compared to less than 17 months, and
Nores and Barnett (2010) reported that interventions lasting longer than one year were more
beneficial than those lasting less than one year in a meta-analysis of 23 non-US countries.
Language exposure at home and school for ethnolinguistic minority children
Changes in global migration patterns have contributed to a large number of children being
language and the dominant language of the country in which they reside (Hoff, 2018).
Evidence suggests that young children can learn several languages nearly simultaneously
(Wagner, 2017), but language and literacy experiences at home can have differential
influences on abilities in each language (Lewis, Sandilos, Hammer, Sawyer, & Méndez,
2016). Studies from Europe and the United States suggest that children exposed to more than
one language often reach school age with lower levels of skills in the majority language than
their monolingual peers, whilst also having low levels of skills in their heritage language
(Garcia, 2016; Hammer et al., 2014; Scheele, Leseman, & Mayo, 2010). These differences
persist even after controlling for socioeconomic differences between children from immigrant
and non-immigrant families (Hoff, 2018). Evidence suggests that children are not confused
by learning two languages (Hammer et al., 2014), so rather than reflecting difficulties in
to acquire two languages instead of just one. Despite lower levels of proficiency in the
dominant language than their monolingual peers, the total vocabulary acquired across both
languages combined by dual language learners has been shown to be greater than that
acquired by monolingual children in one language (Hoff, 2018).
EARLY CHILDHOOD EDUCATION IN MYANMAR
6
A key factor that can affe
LOI used in ECE programmes. Ethnolinguistic minority children participating in ECE
programmes taught in a majority language face the dual challenges of acquiring the new
language and learning the curriculum in that new language (Ball, 2010), which may plausibly
affect the ability of ethnolinguistic minority children to benefit from the advantages
associated with ECE. Nonetheless, evidence from high-income countries has demonstrated
associations between ECE and improved child outcomes for ethnic minority children
(Biedinger, Becker, & Rohling, 2008; Gibbs, 2014; Williams, Uchikoshi, Bunge, & Zhou,
2019). However, findings for ethnic minority compared to majority children vary across
countries, with a lower rate of ECE participation and weaker school readiness documented in
Germany (Biedinger et al., 2008), whilst in the United States ECE had a greater positive
impact on literacy skills for Hispanic than non-Hispanic children (Gibbs, 2014).
Models of ECE language instruction for ethnolinguistic minorities
Educational programmes for bilingual or multilingual children can differ by the extent to
which chil
LOI (Ball, 2010) (see Table 1 for a summary of bilingual and multilingual education
approaches). Programme models fall on a continuum, ranging from a mother tongue-based
programme in which instruction is delivered entirely in L1, bilingual approaches in which
both L1 and L2 are used to some degree, multilingual approaches which use more than two
languages of instruction, programmes where all instruction is delivered in a
language initially unfamiliar to the child, and approaches which ignore
(Ball, 2010). Some programmes may employ both
languages systematically by teaching in one language on specific days of the week or by
having two teachers with each teaching in a different language.
EARLY CHILDHOOD EDUCATION IN MYANMAR
7
Findings from high-income countries on the outcomes of L1-based bilingual
education compared to L2-based monolingual education suggest that L1-based bilingual
education can lead to better academic performance (Durán et al., 2010; Marian, Shook, &
Schroeder, 2013; Rolstad, Mahoney, & Glass, 2005), more efficient acquisition of L2, and
better early literacy skills in L2 (Durán et al., 2010; Lindholm-Leary & Block, 2010; Lindsey,
Manis, & Bailey, 2003). Evidence also suggests that L1-based bilingual education in the early
years helps children develop cognitive flexibility and metalinguistic awareness in a way that
is more advanced than monolingual children (Bialystok, 2001; Cummins, 2000). However,
there is considerable debate over the most appropriate age to introduce L2 exposure for
ethnolinguistic minority children (Hammer et al., 2014).
There are many L1-based bilingual education programmes in LMICs. Despite
research suggesting L1-based education is effective in high-income countries, the success of
these programmes for child outcomes is mixed (Ball, 2010; Benson, 2002). Possible
explanations include a shortage of trained teachers and material resources in LMICs, and that
in some programmes the transition to the exclusive use of L2 has been implemented too soon
(Benson, 2002). Systematic investigation of the effectiveness of L1-based programmes in
LMICs is limited, yet evidence is accumulating. Graham (2010) has highlighted the potential
benefits of L1-based ECE programmes in Kenya. Lee, Watt, and Frawley (2015) found that
Grade 1 ethnic minority children attending bilingual schools in Cambodia performed better in
mathematics than their peers in L2-based monolingual schools, while the differences in
Khmer literacy and oral Khmer performance were not statistically significant (Lee et al.,
2015). These findings are consistent with those from high-income countries suggesting that
L1-based bilingual education is beneficial for the academic development of ethnic minority
children, and also does not hamper their second language acquisition.
EARLY CHILDHOOD EDUCATION IN MYANMAR
8
Given these debates internationally, a greater understanding of whether ECE
participation in Myanmar is associated with improved child outcomes for both ethnic
majority and minority children, and whether outcomes for ethnic minority children vary by
the language(s) exposed to at home and the language(s) of ECE instruction, would inform not
only ECE policy in Myanmar but also the emerging academic knowledge base on ECE policy
in LMICs more broadly.
[Table 1 here].
ECE policy in Myanmar
Policy context
(Department of Population, 2015) is
ethnically diverse with 135 officially recognised ethnic groups speaking over 100 different
languages. The Bamar (Burman) ethnic group accounts for approximately 69% of the
population (Central Statistical Organization, 2015) and speaks the Myanmar language as
mother tongue, which is the official language and the main medium of education, government
and the justice system (Bradley, 2016). Seventy percent of the population resides in rural
areas (Department of Population, 2015), and many ethnic minority groups reside in border
areas with little access to early childhood services (Republic of the Union of Myanmar,
2014).
Since 2011 there have been substantial political and economic reforms in Myanmar.
The Myanmar Government has indicated a commitment to expanding access to preschool (for
3- to 5- year olds) and kindergarten (for 5-year olds) (Republic of the Union of Myanmar,
2014). A commitment to improving access to quality preschool and kindergarten education
has focused on access for children living in rural and remote areas and ethnic minority
communities (Ministry of Education Myanmar, 2016). The pre-primary gross enrolment ratio
increased from 7.6% in 2010 to 24.2% in 2014 (UNESCO Institute for Statistics, 2017).
EARLY CHILDHOOD EDUCATION IN MYANMAR
9
However, the implementation of pre-primary expansion policies is particularly
challenging in Myanmar due to the ethnic and linguistic diversity of the population and the
literacy rate is 89.5%, but disparities exist between the urban (95.2%) and rural (87%)
populations, and among Regions and States, with a literacy rate of 64.6% in Shan compared
to 96.6% in Yangon (Department of Population, 2015). Large inequities in ECE participation
rates also exist between children living in urban and rural areas, between the richest and
poorest wealth quintiles, and across States and Regions (Ministry of National Planning and
Economic Development, 2011).
Languages of instruction
The Myanmar language is the dominant, only official language in Myanmar and the main
medium of instruction used by the government education system. However, the use of ethnic
minority languages as a supplement to Myanmar in the justice system and education is
permitted (Bradley, 2016), and there are possibilities for alternative LOIs in ECE institutions.
Nevertheless, there has been little progress towards making ethnic minority languages a
medium of instruction in government schools, and some government schools have begun
teaching ethnic minority languages as a school subject rather than using them as a medium of
instruction (South & Lall, 2016). Language and education issues are deeply implicated in
ethnic conflicts in Myanmar, with some ethnic communities perceiving and resisting forced
of strategies, including armed conflict and
the development of minority ethnic education regimes (South & Lall, 2016).
The current study
This study presents analyses of associations between ECE programme participation and early
child development of 3- to 5- year-olds in Myanmar, as measured by the EAP-ECDS (Rao,
Sun, et al., 2019), and assesses whether associations are moderated by ethnic group,
EARLY CHILDHOOD EDUCATION IN MYANMAR
10
language(s) spoken at home, and language(s) of ECE instruction. It has three main research
questions (RQs).
RQ1. First, it asks whether ECE participation and programme dosage are associated
with early child development measured using a composite score and seven individual EAP-
ECDS domains, and whether there are differences by domain. Our hypothesis is that it is
likely we will find associations between ECE participation and EAP-ECDS scores across at
least some domains but, in keeping with findings elsewhere (Rao, Yousafzai, et al., 2017),
that associations may be larger for cognitive than socio-emotional or motor skills. Following
findings on dosage elsewhere (Domitrovich et al., 2013) we hypothesise that exposure to
ECE for two years compared to just one will be associated with higher scores. We have data
on ECE participation and the duration of ECE exposure but not on ECE programme quality,
so we acknowledge as a limitation of this study that we cannot provide insights into how
ECE-development associations vary by quality.
RQ2. Second, we ask whether ethnicity moderates any association between ECE and
development. Children are split into ethnic majority and minority groups based on whether
from high-
income countries suggest that ethnic minority as well as majority children benefit from ECE
(Biedinger et al., 2008). However, variations by ethnic group have been found in both ECE
participation rates and preschool quality, with black children more likely to receive poorer
quality care than white children in the United States (Magnuson & Waldfogel, 2005). In some
cases, larger benefits from ECE have been found for ethnic minority compared to majority
children. These may in part be due to minority children being more likely to be disadvantaged
in other ways, such as having lower literacy skills upon entry to ECE, and being more likely
to exhibit indicators of poverty such as eligibility for means-tested free lunches, with ECE
EARLY CHILDHOOD EDUCATION IN MYANMAR
11
(Gibbs, 2014). In Myanmar, however, there may be substantial challenges in providing
language-appropriate ECE. Difficulties include the very large number of different
ethnolinguistic groups (World Bank, 2014), challenges in translating curricula into multiple
languages and recruiting teachers with relevant language skills (Ball, 2010), and the fact that
the Myanmar language has historically been the dominant language of education provision
(South & Lall, 2016). Given these difficulties, we hypothesise that whilst ECE may be
positively associated with early child development for ethnic minority children, it is possible
that associations will be larger for ethnic majority children.
RQ3. Third, we examine whether any associations between ECE and EAP-ECDS
scores for ethnic minority children are moderated by the
and the language(s) of ECE instruction. We compare ECE-development associations for L1-
only instruction, L2-only instruction, and bilingual instruction. Based on findings from high-
income countries (Durán et al., 2010; Marian et al., 2013), we hypothesise that associations
will be larger for L1-only and bilingual instruction than for L2-only instruction. We compare
scores between ethnic minority children exposed or not exposed to Myanmar-language
speaking at home, to understand whether ECE LOI and home language(s) interact in any
association with EAP-ECDS scores. Findings from qualitative research in Myanmar suggest
that mother tongue-based teaching is especially valuable at younger ages (South & Lall,
2016), so we therefore anticipate that we will find differences in ECE-development
associations by LOI and home language(s).
Materials and Methods
Sample
Data are from the Short Form of the East Asia-Pacific Early Child Development Scales
(EAP-ECDS), conducted in three predominantly ethnic minority States (Kachin, Mon and
EARLY CHILDHOOD EDUCATION IN MYANMAR
12
Rakhine) and two predominantly Bamar Regions (Sagaing and Yangon) (see Supplementary
Table 1),1 and administered in Myanmar between November 2016 and January 2017. Prior to
data collection, a nationwide survey on the development of young children in Myanmar did
not exist (Republic of the Union of Myanmar, 2014), and so the EAP-ECDS in Myanmar
present a unique opportunity to guide ECE policy. The original EAP-ECDS of 85 items
across seven domains of early childhood development and learning (Rao et al., 2014) were
developed and validated between 2010 and 2014, and a Short Form of the EAP-ECDS
(33 items of the same seven domains) was later developed (Rao, Sun, Bacon-Shone, Ip, &
Becher, 2016) and administered for the first time in Myanmar (Rao, Richards, et al., 2017).
A total of 1,494 (759 or 51% girls) aged between 36 and 71 months were assessed in
seven different domains of development, and a caregiver survey was used to collect
demographic and socioeconomic information. Informed consent was obtained from all
caregivers prior to data collection, and the final dataset was anonymised by removing all
names or other identifiable information. The sampling plan was developed using government
census data from 2014. The five States or Regions were selected to represent a diverse range
of economic and demographic characteristics. The sample was stratified by urbanicity, with
wards (urban) or villages (rural) randomly sampled separately from within each state or
region, with the probability of selection being in proportion to an estimate (based on census
data) of the number of 3- to 5- year olds residing in that ward or village. A total of 87 wards
and villages were selected. Within each ward or village, a list of all children was used to
randomly select approximately equal numbers of 3-, 4-, and 5- year olds, and approximately
equal numbers of girls and boys. The sample was therefore stratified by urbanicity, age (in
years), and gender. Assessments were conducted by trained assessors in the Myanmar
1 All Supplementary Tables can be freely obtained from the authors upon request (please contact benrich@hku.hk
or nrao@hku.hk).
EARLY CHILDHOOD EDUCATION IN MYANMAR
13
language and in four ethnic minority languages: Sakaw Kayin, Mon, Jingphaw, and Tai Nai
Shan. Back translation procedures were used to ensure equivalence between different
languages.
Measures
Early child development
The EAP-ECDS Short Form assesses children using 33 items in seven different domains of
development: Cognitive Development; Socio-Emotional Development; Motor Development;
Language and Emergent Literacy; Health, Hygiene, and Safety; Cultural Knowledge and
Participation; and Approaches to Learning
composite score representing overall early child development was also created based on a
was
excluded from the composite mean score because factor analysis revealed that the other six
domains were represented by one common factor (loadings 0.47 to 0.71), whereas Motor
Development was not well represented (loading 0.27). Month-of-age adjusted z-scores were
created for the composite development score and for each individual domain to ensure that
interpretation of differences in scores was consistent across different ages. We refer to this
-ECDS score. Scores
were recentred to have a mean of 100 and a standard deviation (SD) of 10, such that a score
of 100 indicates a mean score for children of that age (in months) and a score of 110 indicates
a score of 1 SD above the mean for children of that age (in months).
ECE participation and dosage
ECE participation and dosage were determined using questions in the caregiver survey. Two
measures were used. A binary variable representing ECE participation was based on a
question about having ever participated in ECE. A variable representing ECE dosage was
created based on 6-month categories of the number of months spent in ECE, and had 5
EARLY CHILDHOOD EDUCATION IN MYANMAR
14
categories: (i) no participation; (ii) between 0 and 6 months; (iii) between 6 and 12 months;
(iv) between 12 and 18 months; and (v) more than 18 months in ECE.
Ethnic and linguistic status
Ethnic minority status was determined from a question in the caregiver survey asking for the
belonging to
belonging to an ethnic minority if another description was given (including split or multiple
ethnic identifications). Of those categorised as ethnic minority, linguistic status was split
between those whose caregiver reported speaking Myanmar at home (either as the sole home
language or as one of several languages), and those whose caregiver reported no Myanmar
being spoken at home. Two binary variables were therefore created, with one representing
ethnic majority or minority status, and the other representing linguistic status at home for
ethnic minority children.
Language of instruction
A categorical variable was created based on a question from the caregiver survey asking
created: L1-based (minority-language) instruction; bilingual instruction; and L2-based
(Myanmar-language) instruction. Supplementary Table 3 shows the sample composition by
ECE participation status, language of ECE instruction, ethnicity, and language at home.
Control variables
The following control variables were used: age, sex, urban or rural residence, household
wealth index, and parental socioeconomic status (SES). Different measures of wealth and
SES were used because of findings from LMICs demonstrating that a variety of
socioeconomic measures can be associated with child outcomes (Richards, Bacon-Shone, &
Rao, 2018). The household wealth index was created based on questions on asset ownership
EARLY CHILDHOOD EDUCATION IN MYANMAR
15
(UNICEF, 2019). The variable was created by using weights determined by the first
component of a Principal Components Analysis, following a technique used extensively to
create indicators of household wealth (Schady et al., 2015) and as outlined by Filmer and
Pritchett (2001). Following Weber, Darmstadt, and Rao (2017), the variable representing
parental SES was created using a latent factor measurement model using maximum
likelihood estimation and allowing for missing values. This was based on four variables from
the caregiver survey: maternal and paternal education (coded over 8 levels, assumed to be
continuous), and maternal and paternal occupation (coded over 9 levels, assumed to be
continuous). All control variables were centred at their grand mean.
Analytic plan
All analyses were conducted using Stata 13.1. Descriptive statistics were created showing
means and standard deviations of composite and domain-specific EAP-ECDS scores across
different demographic groups. Reliability of domain scores was assessed by calculating
The factor structure of the EAP-ECDS was assessed using
exploratory factor analysis (principal factors). A model building procedure was followed to
assess whether the control variables improved model fit (Supplementary Table 4). Using OLS
regression with the overall EAP-ECDS development score as the dependent variable, sex,
urbanicity, wealth index, and parental SES index all improved model fit (measured by an
increased adjusted R-squared value). These 4 variables were then used as control variables in
all regression models. Including age in the OLS regression did not change model fit because
the EAP-ECDS score was calculated using month-of-age adjusted z-scores. However, age
was also included as a control variable in all models as an additional check against
confounding due to age being a very strong predictor of developmental scores and also being
correlated with ECE participation.
EARLY CHILDHOOD EDUCATION IN MYANMAR
16
Statistical power analysis was conducted by first computing the effective sample size
(ESS) given clustering within wards or villages. The effective sample size was computed
based on the intraclass correlation coefficient (ICC) for the main dependent variable (overall
developmental score) of .16, and the formula ESS = mk / 1 + ICC (m-1) (Killip, Mahfoud, &
Pearce, 2004), where m is the number of subjects in a cluster (average 17.17), and k is the
number of clusters (87), giving an ESS of 416. GPower 3.1 (Faul, Erdfelder, Buchner, &
Lang, 2009) was used to compute power for a linear multiple regression with two tails, a
sample size of 4
size of .02, and power of approximately equal to 1 for a medium effect size of .15.
RQ1. A series of hierarchical linear modelling (HLM) regressions was used to analyse
associations between EAP-ECDS scores (overall and in each of the 7 individual domains)
and (i) ECE participation and (ii) ECE dosage. Associations between ECE participation and
EAP-ECDS scores, and between ECE dosage and EAP-ECDS scores, were tested for
statistical significance in each model. To account for potential similarities between children
living in the same geographical area, ward or village was used as the level 2 variable and
province was used as the level 3 variable. HLM is a useful strategy when independence of
cases may be a concern (Bartholomew, Steele, Galbraith, & Moustaki, 2008) although the
focus of the analyses was primarily on results from the level 1 variables. All HLM
regressions used robust standard errors.
Sensitivity analysis was conducted using inverse probability weighting (IPW) to
create a synthetic sample to test associations between ECE participation and developmental
scores. Inverse probability weights can adjust for biases created where those exposed to a
treatment (in this case ECE participation) and those not exposed to a treatment are not
assigned at random, and makes it possible to design and analyse an observational study in
such a way that mimics some of the characteristics of the design and analysis of a randomised
EARLY CHILDHOOD EDUCATION IN MYANMAR
17
control trial (Austin, 2011). In this study, IPW adjusts for the fact that those participating and
not participating in ECE were not allocated at random but instead had different characteristics
with, for instance, children living in urban rather than rural areas being more likely to
participate. This adjustment attempts to increase confidence in concluding that any positive
ics in the likelihood of ECE
participation. The same set of HLM regressions as before was run, but this time using
province-specific inverse probability weights and province as the level 2 variable.
RQ2. Next, interaction terms between (i) ethnic minority status and ECE
participation, and (ii) ethnic minority status and ECE dosage, were used to assess whether
ethnic minority status moderated associations between ECE and EAP-ECDS scores (overall
and for each individual domain). Associations between ECE and EAP-ECDS scores were
tested for significance for ethnic majority and minority children separately, and differences
between coefficients were examined to test for moderation. A series of HLM regressions was
used as before. Sensitivity analysis using inverse probability weighting was also conducted
once more.
RQ3. Analysis then focused specifically on ethnic minority children to assess whether
associations between ECE and EAP-
language(s) at home. A variable was created that split ethnic minority children into 7
categories: (1) ECE non-participants; (2) L2-based ECE participants with minority-only
language(s) at home; (3) L2-based ECE participants with exposure to Myanmar language at
home; (4) bilingual ECE participants with minority-only language(s) at home; (5) bilingual
ECE participants with exposure to Myanmar language at home; (6) L1-based ECE
participants with minority-only language(s) at home; and (7) L1-based ECE participants with
exposure to Myanmar at home. HLM regressions were run to examine associations between
EARLY CHILDHOOD EDUCATION IN MYANMAR
18
this variable and EAP-ECDS scores (overall and each individual domain). Differences
between children who did not participate in ECE and each of the 6 ECE participant groups
were tested for statistical significance.
Next, a binary variable was used to compare ethnic minority children who spoke only
a minority language at home with those who were exposed to Myanmar at home. An
interaction term between this variable and ECE dosage was used to examine associations
between ECE dosage and EAP-ECDS scores, broken down by home language exposure.
Differences in coefficients between those exposed and not exposed to Myanmar at home
were tested for statistical significance.
Missing values
The original sample included 1,502 children. 8 children were outside the target age range of
36 to 71 months and so were excluded from the final analytic sample. Of the 1,494 (759 girls)
in the analytic sample, no missing values were found for: age; sex; province; ward; ethnic
minority status; household wealth; ECE participation; and ECE duration. Missing values
language(s) of ECE instruction (n=26); maternal education (n=4); paternal education (n=27);
maternal occupation (n=7); and paternal occupation (n=31).
There were also missing values for the early child development domain scores.
Missing values were found for Cognitive Development (n=27), Socio-emotional
Development (n=20), Motor Development (n=5), Language and Emergent Literacy (n=37),
Health, Hygiene and Safety (n=14), Cultural Knowledge and Participation (n=5), and
Approaches to Learning (n=2). The composite child development score, based on the means
of 6 domains, had 87 missing values (this variable was coded as missing where any 1 of the 6
domain scores was missing).
EARLY CHILDHOOD EDUCATION IN MYANMAR
19
Missing values for maternal and paternal education and occupation were estimated
using maximum likelihood estimation as part of the creation of a composite SES variable, as
described above. Missing values for all other variables were imputed using the multiple
imputation function in Stata 13.1. Sequential imputation was used with chained equations to
calculate 25 imputations for each missing value, using linear, logistic, ordered logistic, and
multinomial logistic regressions.
Results
Table 2 shows descriptive statistics for the EAP-ECDS sample, including composite and
domain specific scores, broken down by demographic group. Girls, children living in urban
areas, and ethnic majority children had higher month-of-age-adjusted composite development
scores than other children. Older children had higher raw development scores than younger
children (not shown). Supplementary Table 5 shows intraclass correlation coefficients (ICCs)
between provinces and wards for all variables, with the largest ICC by both province and
ward for the variable indicating whether Myanmar was spoken at home.
[Table 2 here].
RQ1. ECE participation (binary) was significantly associated with higher composite
EAP-ECDS scores (b = 4.5, R-squared = .14), and model fit was significantly better than the
intercept-only model (F(6, 1838) = 2784.1, p < .001). ECE participation was also associated
with EAP-ECDS scores in all 7 individual domains (b = 1.5 to 4.8), the association was
largest for the Language and Emergent Literacy domain (b = 4.8), and random effects
estimates showed that the composite score intercept differed between provinces (SD = 2.2)
and wards (SD = 2.9) (Table 3, Model 1). Sensitivity analysis using IPW also showed that
ECE participation was significantly associated with higher composite scores and scores in all
7 domains (Supplementary Table 6). ECE duration was significantly associated with
composite EAP-ECDS scores (b = 1.7) and scores in all individual domains (b = 0.8 to 1.7);
EARLY CHILDHOOD EDUCATION IN MYANMAR
20
the association was largest for the Language and Emergent Literacy domain (b = 1.7); and the
composite score intercept varied between provinces (SD = 2.2) and wards (SD = 2.8) (Table
4, Model 1). Marginal mean composite EAP-ECDS scores broken down by ECE duration (5
categories) showed the highest score at 12-
between no ECE participation and all duration categories of ECE dosage were statistically
significant (ps < .05). For ECE participants, 12-
significantly higher scores than a duration of 0-6 months (b = 3.2, p < .05).
[Tables 3 and 4, and Figure 1, here].
RQ2. ECE participation (binary) was significantly associated with EAP-ECDS scores
for both ethnic majority and ethnic minority children based on the composite score (b = 5.2
and 3.8, respectively), Cognitive Development (b = 3.5 and 2.6), Language and Emergent
Literacy (b = 5.5 and 4.0), and Approaches to Learning (b = 2.9 and 2.5) (Table 3, Model 2).
Ethnicity did not significantly moderate ECE-development associations for the composite
score or any individual domain. Sensitivity analysis using IPW showed that ECE was
significantly associated with EAP-ECDS scores for both majority and minority children
based on the composite score and in 5 individual domains. When using IPW, ethnicity did not
moderate ECE-development associations for the composite score, but minority children had
significantly larger ECE-development associations than majority children in the Motor
Development and Approaches to Learning domains (ps < .05, Supplementary Table 6).
Significant associations were found between ECE duration and EAP-ECDS scores for both
ethnic majority and minority children for the composite score and 5 individual domains (b =
0.8 to 2.0, ps < .05), and ethnicity did not significantly moderate associations between ECE
duration and EAP-ECDS scores (ps > .05, Table 4, Model 2). Marginal mean composite
EAP-ECDS scores broken down by ECE duration (5 categories) showed the highest score for
both ethnic majority and minority children at 12-18 months (Figure 2). Marginal mean scores
EARLY CHILDHOOD EDUCATION IN MYANMAR
21
at 6-
(p < .001, not shown).
[Figure 2 here].
RQ3. Figure 3 shows marginal mean composite EAP-ECDS scores for ethnic
minority children, split into 7 categories based on whether or not they participated in ECE, if
they had L1-based, bilingual, or L2-based instruction, and if they were exposed to Myanmar
language at home. Ethnic minority children who participated in L1-based ECE and who were
exposed to Myanmar at home had the highest scores of the 7 groups. Table 5 shows that,
compared to no ECE participation, L1-based ECE was associated with significantly higher
composite scores both for children who were exposed (b = 8.6) and not exposed (b = 4.5) to
Myanmar at home. L1-based ECE was associated with higher scores in 6 individual domains
for those who were exposed to Myanmar at home, and 5 individual domains for those who
were not. Bilingual-based ECE was associated with higher scores in Language and Emergent
Literacy (b = 5.3) and Cultural Knowledge and Participation (b = 4.3) for those exposed to
Myanmar at home, and with higher scores in Motor Development (b = 2.8) for those who
were not. Ethnic minority children with L2-based ECE who were exposed to Myanmar at
home had significantly higher scores than non-participants on the composite score (b = 3.8)
and in 4 individual domains (b = 1.7 to 3.8) (all ps < .05). However, compared to non-
participants, L2-based ECE was not associated with higher scores for children not exposed to
Myanmar at home (ps > .05). Supplementary Table 7 shows that ECE duration was
associated with higher scores both for ethnic minority children who were exposed and who
were not exposed to Myanmar at home (ps < .05).
[Table 5 and Figure 3 here].
Discussion
EARLY CHILDHOOD EDUCATION IN MYANMAR
22
This study analysed the relation between ECE participation and early child development in
Myanmar using data from the East Asia-Pacific Early Child Development Scales, a
pioneering survey measuring the holistic development of children in the East Asia-Pacific
region. It asked: whether ECE participation and dosage were associated with early child
development across seven domains (RQ1); whether associations were moderated by ethnic
group (RQ2); and whether ECE-development associations for ethnic minority children varied
depending on the language(s) spoken at home and the language(s) of ECE instruction (RQ3).
ECE programme participation and dosage and early child development
Children who had participated in ECE had higher composite development scores and higher
scores in all individual domains compared to children who had not participated, and increased
ECE programme duration also predicted higher composite development scores and higher
scores in all individual domains. This suggests that ECE programme participation in
Myanmar is associated with improved overall child development, and not only with improved
-development association was with the
Language and Emergent Literacy domain, suggesting that ECE participation may be
especially beneficial for language skills. Children participating in ECE had language scores
that were almost half a standard deviation (.48 SD) higher than those of their same-age peers
who were not participating, even after controlling for covariates. These findings are
consistent with those from high-income countries which demonstrate that ECE is important
-cognitive skills (Sylva et al., 2006; Weiland &
Yoshikawa, 2013), and contribute to the accumulating evidence that ECE can also be
important for child development in some LMICs (Gove et al., 2018). The finding that the
largest ECE-development associations were with language is also consistent with evidence
suggesting that ECE programmes typically have most impact on aspects of cognition
compared to other domains of development (Rao, Yousafzai, et al., 2017), supporting our
EARLY CHILDHOOD EDUCATION IN MYANMAR
23
hypothesis. ECE dosage of between 12 and 18 months was associated with the highest
composite development scores, which tentatively supports other research suggesting that
dosage of two years rather than only one year of ECE is more
development (Domitrovich et al., 2013).
Variation in ECE-development associations by ethnic group
Ethnic group (comparing the majority with minority groups) did not moderate ECE-
development associations overall, and associations were found between ECE and both
composite development and domain-specific development for ethnic minority children,
suggesting that these children were able to benefit from ECE participation to an extent that is
comparable with that of majority children in Myanmar. We hypothesised that some of the
barriers that have been documented elsewhere that may prevent ethnolinguistic minority
children from sharing in the benefits of ECE (Ball, 2010) could mean that ECE-development
associations were moderated by ethnolinguistic group in Myanmar. Encouragingly, we did
not find this moderating effect, suggesting that at least some ECE programmes in Myanmar
may have been able to overcome these barriers. Indeed, sensitivity analysis even suggested
that ethnic minority children may receive significantly greater benefits from ECE in two
individual domains compared to ethnic majority children. However, overall developmental
ethnic minority compared to majority children, suggesting that ethnic minority children may
be at a disadvantage relative to their majority peers at this level of dosage.
Language at home and language of instruction for ethnic minority children
For ethnic minority children, developmental scores varied by language spoken at home and
LOI. Compared to ethnic minority non-participants in ECE, children with L1 instruction had
significantly higher overall scores regardless of whether they had been exposed to Myanmar
language at home. Where children had L2 instruction, however, scores were significantly
EARLY CHILDHOOD EDUCATION IN MYANMAR
24
higher for ECE participants, but only when the participants had been exposed to Myanmar
language at home. No significant differences in scores between non-participants and
participants with L2 instruction were found overall or in any individual domain where
children spoke only a minority language at home. Minority-language children participating in
bilingual ECE programmes only had higher scores than non-participants in the Motor
Development domain. Longer ECE duration was associated with higher overall
developmental scores both for ethnic minority children who did speak Myanmar at home and
those who did not, and language at home did not moderate associations between ECE
duration and overall development.
Implications
Overall, these results are encouraging as they suggest there are benefits to ECE participation
and longer ECE duration for children from a wide range of different backgrounds in
Myanmar. Positive associations were found between ECE and child development for both
ethnic majority and minority children, and for ethnic minority children both exposed and not
exposed to the dominant national language at home. We found little evidence to support
research elsewhere that suggests ECE may even be more beneficial for children from ethnic
or linguistic minority backgrounds (Gibbs, 2014), although sensitivity analysis did show
greater benefits for ethnic minority compared to majority children in two individual domains.
However, given the substantial barriers in providing ECE for ethnolinguistic minority
children in Myanmar, including the translation of materials into multiple languages,
recruitment of teachers with appropriate language skills (Ball, 2010), difficulties in providing
services in remote areas (Ministry of Education Myanmar, 2016), and overcoming ethnic
conflict (South & Lall, 2016), it is nevertheless reassuring that ECE-development
associations were found for ethnic and linguistic minority children.
EARLY CHILDHOOD EDUCATION IN MYANMAR
25
Despite these positive findings, results also demonstrated differences in ECE-
development associations by LOI. L1-based ECE was associated with higher scores both for
children who were exposed to the dominant Myanmar language at home and those who were
not, but L2-based ECE was only associated with higher scores where children were exposed
to Myanmar at home, and bilingual ECE was only associated with higher Motor
Development scores for minority-language children. This suggests that the use of the mother
tongue as a medium of ECE provision may be of great importance in achieving benefits from
ECE participation for ethnic minority children who do not speak Myanmar. These findings
are consistent with those elsewhere suggesting that mother tongue-based education has
(Baker, 2001; Cummins,
2000). Despite mixed evidence in other LMICs on the effectiveness of minority-language
programmes (Ball, 2010; Benson, 2002) it also suggests that, in Myanmar, where ECE
programmes are implemented in a minority language they are associated with improved child
outcomes. ECE programmes in Myanmar may therefore either have been able to overcome
some of the obstacles to effective mother tongue instruction documented in other LMICs,
such as a shortage of trained teachers and lack of teaching and learning materials (Ball,
2010), or alternatively may be effective despite such potential difficulties.
Limitations
This study has several limitations. First, the data are cross-sectional, so it was not possible to
analyse changes in ECE-development associations over time for a cohort of children as they
mature. Future research using longitudinal data could expand on the findings reported here
and give insight into the processes by which ECE-development associations for ethnic
minority children are moderated by language at home and LOI. Second, the indicator of LOI
was based on parent report. Data from teacher report or classroom observations may have
been more reliable, and may have allowed more nuanced and detailed analysis of variations
EARLY CHILDHOOD EDUCATION IN MYANMAR
26
in developmental scores by LOI, but these data were unfortunately unavailable. There was
languages, so it was not possible to understand whether this influenced ECE-development
associations. Further, data were not available on ECE programme quality, so we
acknowledge as a limitation that we cannot provide insights into how ECE-development
associations vary by quality.
Conclusions
ECE programme participation in Myanmar was associated with higher child development
scores overall and across all individual domains, and associations were largest in the
Language and Emergent Literacy domain (scores were .48 SD higher for ECE participants
than non-participants). Longer ECE duration was also associated with higher scores overall
highest scores. Ethnic group did not moderate ECE-development associations, suggesting that
both ethnic majority and minority children have been able to take advantage of the benefits
associated with ECE. For ethnic minority children, ECE was associated with higher
development scores both for those who were exposed to the dominant Myanmar language at
home and those who were not. However, ethnic minority ECE participants who did not speak
the dominant language at home only had higher overall scores where ECE instruction was
L1-based. Findings highlight the benefits of ECE for both ethnolinguistic majority and
minority children in Myanmar, suggesting that the policy of ECE programme expansion may
speaking the dominant language at home, these benefits may vary by the LOI used in ECE
programmes. Further investigation could shed light on the processes underlying this variation
and seek to understand how language-appropriate instruction could be provided so that all
children are able to realise the benefits of ECE.
EARLY CHILDHOOD EDUCATION IN MYANMAR
27
Acknowledgements
This study is based on a grant awarded to Nirmala Rao by UNICEF Myanmar and was
funded by UNICEF Myanmar, Denmark, and the European Union as part of the Building on
Quality Basic Education Programme. We are very grateful to the Myanmar team for
overseeing and implementing the survey on which this study is based. We thank the Rector of
the Yangon University of Education, the professors and students from the Education
Psychology Department of the Yangon University of Education, the education officials from
the Department of Basic Education and the Department of Education, Research, Planning and
Training of the Ministry of Education, and the team at UNICEF Myanmar. We also express
our gratitude to the children and parents who participated in this study.
Declaration of Interest Statement
The authors declare no conflict of interest.
EARLY CHILDHOOD EDUCATION IN MYANMAR
28
References
Agirdag, O., Yazici, Z., & Sierens, S. (2015). Trends in pre-school enrolment in Turkey:
unequal access and differential consequences. Comparative Education, 51(4), 1-18.
Austin, P. C. (2011). An Introduction to Propensity Score Methods for Reducing the Effects
of Confounding in Observational Studies. Multivariate Behavioral Research, 46(3),
399-424.
Baker, C. (2001). Foundations of Bilingual Education and Bilingualism (Vol. 3rd ed).
Clevedon, England: Multilingual Matters.
Ball, J. (2010). Enhancing Learning of Children from Diverse Language Backgrounds:
Mother Tongue-based Bilingual or Multilingual Education Early Childhood and
Early Primary School Years. Paris: UNESCO.
Barry, B. (2002). Culture and equality: An egalitarian critique of multiculturalism.
Cambridge: Polity Press.
Bartholomew, D. J., Steele, F., Galbraith, J., & Moustaki, I. (2008). Analysis of multivariate
social science data. Chapman and Hall/CRC.
Behrman, J. R. K., Elizabeth M. Armecin, Graeme Duazo, Paulita Ghuman, Sharon Gultiano,
Socorro Lee, Nannette. (2006). Early Childhood Development Through An Integrated
Program: Evidence From The Philippines. Washington, D.C.: World Bank.
Benson, C. J. (2002). Real and Potential Benefits of Bilingual Programmes in Developing
Countries. International Journal of Bilingual Education and Bilingualism, 5(6), 303-
317.
Bialystok, E. (2001). Bilingualism in Development: Language, Literacy, and Cognition.
Cambridge: Cambridge University Press.
EARLY CHILDHOOD EDUCATION IN MYANMAR
29
Biedinger, N., Becker, B., & Rohling, I. (2008). Early Ethnic Educational Inequality: The
Influence of Duration of Preschool Attendance and Social Composition. European
Sociological Review, 24(2), 243-256.
Black, M. M., Walker, S. P., Fernald, L. C. H., Andersen, C. T., Digirolamo, A. M., Lu,
C., . . . Grantham-Mcgregor, S. (2017). Early childhood development coming of age:
science through the life course. The Lancet, 389(10064), 77-90.
Bradley, D. (2016). The Languages of Myanmar. Report to UNICEF Myanmar 2016.
Retrieved from http://www.academia.edu/32976865/The_Languages_of_Myanmar
Britto, P. R., Lye, S. J., Proulx, K., Yousafzai, A. K., Matthews, S. G., Vaivada, T., . . .
Bhutta, Z. A. (2017). Nurturing care: promoting early childhood development. The
Lancet, 389(10064), 91-102.
Burchinal, M., Zaslow, M. J., & Tarullo, L. (2016). Quality thresholds, features and dosage in
early care and education: secondary analyses of child outcomes. In P. J. Bauer (Ed.),
Monographs of the Society for Research in Child Development (Vol. 81, 2).
Campbell, F., Conti, G., Heckman, J. J., Moon, S. H., Pinto, R., Pungello, E., & Pan, Y.
(2014). Early Childhood Investments Substantially Boost Adult Health. Science,
343(6178), 1478-1485.
Central Statistical Organization. (2015). Myanmar Statistical Information System. Retrieved
from http://mmsis.gov.mm/
Cummins, J. (2000). Language, Power and Pedagogy: Bilingual Children in the Crossfire
(Vol. 23). Clevedon, England: Multilingual Matters.
Department of Population. (2015). The 2014 Myanmar Population and Housing Census: The
Union Report. Census Report Volume 2. Nay Pyi Taw.
Domitrovich, C. E., Morgan, N. R., Moore, J. E., Cooper, B. R., Shah, H. K., Jacobson, L., &
Greenberg, M. T. (2013). One versus two years: Does length of exposure to an
EARLY CHILDHOOD EDUCATION IN MYANMAR
30
enhanced preschool program impact the academic functioning of disadvantaged
children in kindergarten? Early Childhood Research Quarterly, 28(4), 704-713.
Durán, L. K., Roseth, C. J., & Hoffman, P. (2010). An experimental study comparing
English-only and Transitional Bilingual Education on Spanish-
early literacy development. Early Childhood Research Quarterly, 25(2), 207-217.
Faul, F., Erdfelder, E., Buchner, A., & Lang, A.-G. (2009). Statistical power analyses using
G*Power 3.1: Tests for correlation and regression analyses. Behavior Research
Methods, 41(4), 1149-1160.
Filmer, D., & Pritchett, L. (2001). Estimating Wealth Effects Without Expenditure Data Or
Tears: An Application To Educational Enrollments In States Of India*. Demography,
38(1), 115-132.
Garcia, E. E. (2016). Dual Language Learners. In A. Farrell, S. L. Kagan, & E. K. M. Tisdall
(Eds.), The SAGE Handbook of Early Childhood Research (pp. 363-379). London:
SAGE Publications Ltd.
Gibbs, C. R. (2014). Experimental evidence on early intervention: The impact of full-day
kindergarten. Batten School of Leadership and Public Policy. University of Virginia.
Gove, A., Dombrowski, E., King, S. J., Pressley, J., & Weatherholt, T. (2018). Persistence
and Fadeout of Preschool Participation Effects on Early Reading Skills in Low- and
Middle-Income Countries. Global Education Review, 5(2), 85-109.
Graham, B. E. (2010). Mother tongue education: necessary? Possible? Sustainable?
Language and Education, 24(4), 309-321.
Hammer, C. S., Hoff, E., Uchikoshi, Y., Gillanders, C., Castro, D. C., & Sandilos, L. E.
(2014). The language and literacy development of young dual language learners: A
critical review. Early Childhood Research Quarterly, 29(4), 715-733.
EARLY CHILDHOOD EDUCATION IN MYANMAR
31
Henrich, J., Heine, S. J., & Norenzayan, A. (2010). Most people are not WEIRD. Nature,
466(7302), 29-29.
Hoff, E. (2018). Bilingual Development in Children of Immigrant Families. Child
Development Perspectives, 12(2), 80-86.
Irwin, L. G., Siddiqi, A., & Hertzman, G. (2007). Early child development: a powerful
equalizer. Vancouver, BC: Human Early Learning Partnership (HELP)
Killip, S., Mahfoud, Z., & Pearce, K. (2004). What Is an Intracluster Correlation Coefficient?
Crucial Concepts for Primary Care Researchers. The Annals of Family Medicine, 2(3),
204-208.
Larson, K., Russ, S. A., Nelson, B. B., Olson, L. M., & Halfon, N. (2015). Cognitive ability
at kindergarten entry and socioeconomic status. Pediatrics, 135(2), e440-e448.
Lee, S., Watt, R., & Frawley, J. (2015). Effectiveness of bilingual education in Cambodia: a
longitudinal comparative case study of ethnic minority children in bilingual and
monolingual schools. Compare: A Journal of Comparative and International
Education, 45(4), 526-544.
Lewis, K., Sandilos, L. E., Hammer, C. S., Sawyer, B. E., & Méndez, L. I. (2016). Relations
Abilities: A Study of Head Start Dual Language Learners and Their Mothers. Early
Education and Development, 27(4), 478-494.
Lindholm-Leary, K., & Block, N. (2010). Achievement in Predominantly Low SES/Hispanic
Dual Language Schools. International Journal of Bilingual Education and
Bilingualism, 13(1), 43-60.
Lindsey, K. A., Manis, F. R., & Bailey, C. E. (2003). Prediction of First-Grade Reading in
Spanish-Speaking English-Language Learners. Journal of Educational Psychology,
95(3), 482-494.
EARLY CHILDHOOD EDUCATION IN MYANMAR
32
Magnuson, K. A., & Waldfogel, J. (2005). Early Childhood Care and Education: Effects on
Ethnic and Racial Gaps in School Readiness. The Future of Children, 15(1), 169-196.
Marian, V., Shook, A., & Schroeder, S. R. (2013). Bilingual Two-Way Immersion Programs
Benefit Academic Achievement. Bilingual Research Journal, 36(2), 167-186.
Ministry of Education Myanmar. (2016). National Education Strategic Plan 2016 21. The
Republic of the Union of Myanmar. Retrieved from
http://planipolis.iiep.unesco.org/sites/planipolis/files/ressources/myanmar_nesp-
english.pdf
Ministry of National Planning and Economic Development. (2011). Monitoring the situation
of children and women: Multiple Indicator Cluster Survey 2009 2010. Nay Pyi Taw:
UNICEF.
Nores, M., & Barnett, W. S. (2010). Benefits of early childhood interventions across the
world: (Under) Investing in the very young. Economics of Education Review, 29(2),
271-282.
Ou, S.-R., Arteaga, I., & Reynolds, A. J. (2019). Dosage Effects in the Child-Parent Center
PreK-to-3rd Grade Program: A Re-analysis in the Chicago Longitudinal Study.
Children and Youth Services Review, 101, 285-298.
Phillips, D., Mekos, D., Scarr, S., McCartney, K., & Abbott Shim, M. (2000). Within and
beyond the classroom door: assessing quality in child care centers. Early childhood
research quarterly, 15(4), 475-496.
Rao, N., Koong, M., Kwong, M., & Wong, M. (2003). Predictors of preschool process quality
in a Chinese context. Early Childhood Research Quarterly, 18(3), 331-350.
Rao, N., Richards, B., Sun, J., Weber, A., & Sincovich, A. (2019). Early childhood education
and child development in four countries in East Asia and the Pacific. Early Childhood
Research Quarterly, 47, 169-181.
EARLY CHILDHOOD EDUCATION IN MYANMAR
33
Rao, N., Richards, B., & Umayahara, M. (2017). Myanmar East Asia- Pacific Early Child
Development Scales (EAP-ECDS) Project. Hong Kong: HKU Early Childhood
Development and Education Research Team.
Rao, N., Sun, J., Bacon-Shone, J., Ip, P., & Becher, Y. (2016). East Asia-Pacific Early Child
Development Scales - Short Form. Final Report. Singapore: ARNEC.
Rao, N., Sun, J., Ng, M., Becher, Y., Lee, D., Ip, P., & Bacon-Shone, J. (2014). Validation,
Finalization and Adoption of the East Asia-Pacific Early Child Development Scales
(EAP-ECDS). UNICEF. Retrieved from https://arnec.net/static/uploads/EAP-ECDS-
Final-Report1.pdf
Rao, N., Sun, J., Richards, B., Weber, A. M., Sincovich, A., Darmstadt, G. L., & Ip, P.
(2019). Assessing Diversity in Early Childhood Development in the East Asia-Pacific.
Child Indicators Research, 12(1), 235-254.
Rao, N., Yousafzai, A., & Ip, P. (2017). Systematic review and meta-analyses of parenting
and early childhood educational interventions. Web Appendix 2 (pp. 10 -15). The
Lancet, 389(10064), 91-102.
Republic of the Union of Myanmar. (2014). Myanmar Policy for Early Childhood Care and
Development. The Republic of The Union of Myanmar. Retrieved from
http://www.themimu.info/sites/themimu.info/files/documents/Ref_Doc_Early_Childh
ood_Care_Development_Policy_2014.pdf
Richards, B., Bacon-Shone, J., & Rao, N. (2018). Socioeconomic correlates of early child
development: Gradients from six countries in the East Asia-Pacific region.
International Journal of Behavioral Development, 42(6), 581-587.
Rolstad, K., Mahoney, K., & Glass, G. V. (2005). The Big Picture: A Meta-Analysis of
Program Effectiveness Research on English Language Learners. Educational Policy,
19(4), 572-594.
EARLY CHILDHOOD EDUCATION IN MYANMAR
34
Schady, N. R., Behrman, J. R., Araujo, M. C., Bernal S, R., Bravo, D., Lopez Boo, F., . . .
Vakis, R. N. (2015). Wealth Gradients in Early Childhood Cognitive Development in
Five Latin American Countries. Journal of Human Resources, 50(2), 446-463.
Scheele, A. F., Leseman, P. P. M., & Mayo, A. Y. (2010). The Home Language Environment
of Monolingual and Bilingual Children and Their Language Proficiency. Applied
Psycholinguistics, 31(1), 117-140.
Slot, P. L. (2018). Structural characteristics and process quality in early childhood education
and care: A literature review. OECD Directorate for Education and Skills, Education
Working Paper No. 176.
Slot, P. L., Leseman, P. P. M., Verhagen, J., & Mulder, H. (2015). Associations between
structural quality aspects and process quality in Dutch early childhood education and
care settings. Early Childhood Research Quarterly, 33, 64-76.
South, A., & Lall, M. (2016). Schooling and Conflict: Ethnic Education and Mother Tongue-
based Teaching in Myanmar. The Asia Foundation and USAID. Retrieved from
https://asiafoundation.org/resources/pdfs/SchoolingConflictENG.pdf
Sylva, K., Siraj-Blatchford, I., Taggart, B., Sammons, P., Melhuish, E., Elliot, K., & Totsika,
V. (2006). Capturing quality in early childhood through environmental rating scales.
Early Childhood Research Quarterly, 21(1), 76-92.
UNESCO. (2015). EFA Global Monitoring Report. Education for All 2000-2015:
Achievements and Challenges. Paris: UNESCO.
UNESCO Institute for Statistics. (2017). UIS.Stat. UNESCO. Retrieved from
http://data.uis.unesco.org/#
UNICEF. (2016). . New
York: UNICEF. Retrieved from
https://www.unicef.org/publications/files/UNICEF_SOWC_2016.pdf
EARLY CHILDHOOD EDUCATION IN MYANMAR
35
UNICEF. (2019). MICS6 TOOLS. Multiple Indicator Cluster Surveys (MICS). UNICEF.
Retrieved from http://mics.unicef.org/tools
Wagner, D. A. (2017). Learning as Development: Rethinking International Education in a
Changing World. New York and London: Routledge.
Weber, A., Darmstadt, G. L., & Rao, N. (2017). Gender disparities in child development in
the east Asia-Pacific region: a cross-sectional, population-based, multicountry
observational study. The Lancet Child & Adolescent Health, 1(3), 213-224.
Weiland, C., & Yoshikawa, H. (2013). Impacts of a Prekindergarten Program on Children's
Mathematics, Language, Literacy, Executive Function, and Emotional Skills. Child
Development, 84(6), 2112-2130.
Williams, A. I., Uchikoshi, Y., Bunge, S. A., & Zhou, Q. (2019). Relations of English and
Heritage Language Proficiency to Response Inhibition and Attention Shifting in Dual
Language Learners in Head Start. Early Education and Development, 30(3), 357-374.
World Bank. (2014). Myanmar: Ending poverty and boosting shared prosperity in a time of
transition. World Bank Group Report No. 93050-MM. Retrieved from
http://www.worldbank.org/content/dam/Worldbank/document/EAP/Myanmar/WBG_
SCD_Full_Report_English.pdf
World Bank. (2018). School enrollment, preprimary (% gross). World Bank Group.
Retrieved from https://data.worldbank.org/indicator/SE.PRE.ENRR
EARLY CHILDHOOD EDUCATION IN MYANMAR
36
Tables and Figures
Table 1. Summary of bilingual and multilingual education approaches.
Mother tongue-based
instruction
Bilingual education (or
-way bilingual
Minority and majority language children are taught in both
minority and majority languages.
Mother tongue-based
bilingual education (or
bilingualism )
L1 is used as the primary medium of instruction throughout
primary education while L2 is introduced as a subject of study
to prepare students for eventual transition to some academic
subjects in L2.
Multilingual education Formal use of more than two languages in the curriculum.
Transitional
bi/multilingual education
The objective is a planned transition from one language of
instruction to another. r
to programmes that involve an abrupt transition to L2
instruction after only 2 or 3 years in school, w ate
o a switch to L2 instruction after
a child has become fully fluent academically in L1.
Maintenance
bi/multilingual education
(or
)
After L2 has been introduced, both (or all) chosen languages
are media of instruction. L1 instruction continues, often as a
subject of study, to ensure ongoing support for children to
become academically proficient in L1.
Immersion or foreign
language instruction
The entire education programme is provided in a language that
is new to the child.
Submersion ( Sink or
Swim )
Speakers of non-dominant languages have no choice but to
receive education in languages they do not understand.
Note: Source: Ball (2010).
EARLY CHILDHOOD EDUCATION IN MYANMAR
37
Table 2. Descriptive statistics for the EAP-ECDS sample by demographic group.
3yo 4yo 5yo Girls Boys Urban Rural
Ethnic
majority
Ethnic
minority
Children in sample (n) 487 504 503 759 735 725 763 729 765
Overall development 100.00 100.00 100.00 101.02 98.95 100.91 99.12 100.82 99.25
9.88 9.88 9.88 9.88 9.76 9.78 9.88 9.41 10.23
Cognitive Development 100.00 100.00 100.00 100.79 99.18 100.83 99.19 100.23 99.78
10.00 10.01 10.01 9.89 10.05 10.15 9.72 9.70 10.27
Socio-emotional Development 100.00 100.00 100.00 100.49 99.49 100.59 99.38 100.52 99.51
10.00 10.01 10.01 9.86 10.12 9.63 10.31 9.23 10.66
Motor Development 100.00 100.00 100.00 98.65 101.40 99.70 100.28 99.98 100.02
10.00 10.01 10.01 9.84 9.98 10.14 9.90 9.79 10.21
Language & Emergent Literacy 100.00 100.00 100.00 101.30 98.68 101.04 99.02 100.25 99.77
10.01 10.01 10.01 9.91 9.92 10.23 9.71 9.74 10.24
Health, Hygiene & Safety 100.00 100.00 100.00 100.29 99.70 100.38 99.60 101.00 99.05
10.00 10.01 10.01 9.81 10.19 9.57 10.40 9.11 10.69
Cultural Knowledge & Participation 100.00 100.00 100.00 100.69 99.29 100.06 99.98 102.07 98.02
10.00 10.01 10.01 9.94 10.02 9.72 10.26 9.15 10.37
Approaches to Learning 100.00 100.00 100.00 100.48 99.50 100.91 99.14 100.21 99.80
10.00 10.01 10.01 9.91 10.08 10.00 9.95 9.96 10.04
Wealth index -0.05 0.02 0.02 0.01 -0.02 0.21 -0.21 0.16 -0.16
1.01 0.97 1.02 1.01 0.99 0.98 0.97 0.99 0.99
Parental SES Index -0.02 0.04 -0.03 0.03 -0.04 0.26 -0.25 0.13 -0.12
0.96 1.02 1.02 1.00 0.99 1.06 0.86 0.99 0.99
Preschool participation (%) 43% 61% 72% 58% 59% 63% 54% 54% 63%
Preschool duration (months) 6.49 10.14 11.34 10.04 9.50 10.10 9.45 9.54 9.97
4.32 7.18 7.19 7.03 6.72 7.24 6.48 6.91 6.86
Any Myanmar spoken at home (%) 73% 74% 74% 74% 73% 75% 72% 100% 48%
EARLY CHILDHOOD EDUCATION IN MYANMAR
38
Note: Descriptive statistics are shown before imputation of missing values. Overall development and domain-specific development scores are
shown after standardisation by month-of-age. Means (or %) are shown, with SDs in italics underneath.
EARLY CHILDHOOD EDUCATION IN MYANMAR
39
Table 3. Associations between ECE participation (binary) and EAP-ECDS scores (SD=10;
Model 1 = all children; Model 2 = interaction term with ethnic majority/minority group).
Model 1 Model 2
Estimate SE Estimate SE
Overall development
Fixed part
ECE overall 4.46 0.89
ECE ethnic maj 5.24 0.55
ECE ethnic min 3.82 1.47
ECE maj vs min 1.42 1.23
Random part
Province (SD) 2.24 0.59 2.26 0.84
Ward (SD) 2.87 0.31 2.78 0.33
Residual (SD) 8.35 0.21 8.33 0.21
Cognitive development
Fixed part
ECE overall 2.96 0.92
ECE ethnic maj 3.54 0.70
ECE ethnic min 2.65 1.25
ECE maj vs min 0.89 0.74
Random part
Province (SD) 1.71 0.82 2.02 1.07
Ward (SD) 2.84 0.56 2.88 0.60
Residual (SD) 8.88 0.40 8.85 0.40
Socio-emotional development
Fixed part
ECE overall 2.76 1.23
ECE ethnic maj 3.71 0.88
ECE ethnic min 1.76 1.93
ECE maj vs min 1.94 1.85
Random part
Province (SD) 0.00 . 0.00 .
Ward (SD) 3.14 365.47 3.15 2.18
Residual (SD) 9.04 193.69 9.03 1.13
Motor development
Fixed part
ECE overall 1.47 0.64
ECE ethnic maj 1.05 0.86
ECE ethnic min 2.00 0.67
ECE maj vs min -0.96 0.95
Random part
Province (SD) 0.00 . 0.00 .
Ward (SD) 2.36 0.92 2.37 15.01
Residual (SD) 9.59 0.37 9.58 7.94
EARLY CHILDHOOD EDUCATION IN MYANMAR
40
Model 1 Model 2
Estimate SE Estimate SE
Language and emergent literacy
Fixed part
ECE overall 4.80 0.46
ECE ethnic maj 5.52 0.58
ECE ethnic min 3.99 1.14
ECE maj vs min 1.51 1.62
Random part
Province (SD) 2.46 0.70 2.37 0.67
Ward (SD) 2.93 0.50 2.91 0.49
Residual (SD) 8.75 0.16 8.74 0.15
Health, hygiene and safety
Fixed part
ECE overall 3.22 1.02
ECE ethnic maj 3.54 0.47
ECE ethnic min 3.42 1.82
ECE maj vs min 0.12 1.62
Random part
Province (SD) 1.38 0.55 1.89 0.75
Ward (SD) 2.81 0.74 2.79 0.61
Residual (SD) 8.92 0.50 8.85 0.46
Cultural knowledge and participation
Fixed part
ECE overall 2.37 0.66
ECE ethnic maj 3.07 0.39
ECE ethnic min 2.05 1.47
ECE maj vs min 1.02 1.41
Random part
Province (SD) 2.64 0.39 2.18 0.50
Ward (SD) 2.74 0.33 2.45 0.16
Residual (SD) 9.03 0.31 8.99 0.31
Approaches to learning
Fixed part
ECE overall 2.71 0.61
ECE ethnic maj 2.88 0.80
ECE ethnic min 2.52 0.56
ECE maj vs min 0.36 0.71
Random part
Province (SD) 0.41 1.26 0.41 1.34
Ward (SD) 2.18 0.49 2.18 0.50
Residual (SD) 9.51 0.29 9.51 0.29
EARLY CHILDHOOD EDUCATION IN MYANMAR
41
Note: The table shows fixed (bold indicates p < .05) and random estimates from a series of
random
estimates being very close to 0.
EARLY CHILDHOOD EDUCATION IN MYANMAR
42
Table 4. Associations between ECE dosage (5 categories) and EAP-ECDS scores (SD=10;
Model 1 = all children; Model 2 = interaction term with ethnic majority/minority group).
Model 1 Model 2
Estimate SE Estimate SE
Overall development
Fixed part
ECE overall 1.72 0.19
ECE ethnic maj 2.03 0.27
ECE ethnic min 1.55 0.38
ECE maj vs min 0.48 0.40
Random part
Province (SD) 2.22 0.52 2.25 0.81
Ward (SD) 2.80 0.34 2.69 0.36
Residual (SD) 8.38 0.17 8.37 0.17
Cognitive development
Fixed part
ECE overall 1.40 0.24
ECE ethnic maj 1.67 0.41
ECE ethnic min 1.30 0.27
ECE maj vs min 0.37 0.28
Random part
Province (SD) 1.76 0.78 2.06 1.06
Ward (SD) 2.82 0.51 2.85 0.55
Residual (SD) 8.86 0.39 8.83 0.39
Socio-emotional development
Fixed part
ECE overall 1.09 0.21
ECE ethnic maj 1.12 0.21
ECE ethnic min 1.08 0.38
ECE maj vs min 0.04 0.40
Random part
Province (SD) 0.00 . 0.00 .
Ward (SD) 3.12 9.50 3.11 3.02
Residual (SD) 9.05 60.65 9.05 2.99
Motor development
Fixed part
ECE overall 0.80 0.24
ECE ethnic maj 0.76 0.30
ECE ethnic min 0.88 0.27
ECE maj vs min -0.11 0.36
Random part
Province (SD) 0.00 . 0.00 .
Ward (SD) 2.37 2.85 2.38 0.46
Residual (SD) 9.57 1.05 9.57 0.32
EARLY CHILDHOOD EDUCATION IN MYANMAR
43
Model 1 Model 2
Estimate SE Estimate SE
Language and emergent literacy
Fixed part
ECE overall 1.68 0.35
ECE ethnic maj 2.28 0.25
ECE ethnic min 1.14 0.65
ECE maj vs min 1.13 0.82
Random part
Province (SD) 2.39 0.65 2.25 0.62
Ward (SD) 2.83 0.49 2.82 0.47
Residual (SD) 8.82 0.17 8.80 0.16
Health, hygiene and safety
Fixed part
ECE overall 1.22 0.38
ECE ethnic maj 1.14 0.27
ECE ethnic min 1.52 0.50
ECE maj vs min -0.39 0.43
Random part
Province (SD) 1.30 0.48 1.84 0.80
Ward (SD) 2.78 0.77 2.75 0.63
Residual (SD) 8.94 0.48 8.87 0.44
Cultural knowledge and participation
Fixed part
ECE overall 0.80 0.24
ECE ethnic maj 1.14 0.20
ECE ethnic min 0.71 0.55
ECE maj vs min 0.43 0.53
Random part
Province (SD) 2.63 0.40 2.17 0.48
Ward (SD) 2.71 0.31 2.40 0.13
Residual (SD) 9.05 0.32 9.01 0.31
Approaches to learning
Fixed part
ECE overall 1.22 0.28
ECE ethnic maj 1.28 0.58
ECE ethnic min 1.19 0.08
ECE maj vs min 0.09 0.61
Random part
Province (SD) 0.21 2.01 0.00 2.88
Ward (SD) 2.15 0.46 2.16 1.96
Residual (SD) 9.50 0.28 9.50 2.17
EARLY CHILDHOOD EDUCATION IN MYANMAR
44
Note: The table shows fixed (bold indicates p < .05) and random estimates from a series of
random
estimates being very close to 0.
EARLY CHILDHOOD EDUCATION IN MYANMAR
45
Table 5. Associations between ECE participation (compared to non-participation) and EAP-ECDS scores, by language of ECE instruction and
whether or not Myanmar is spoken at home (ethnic minority children only).
Overall
Cognitive
Development
Socio-
emotional
Development
Motor
Development
Language &
Emergent
Literacy
Health,
Hygiene &
Safety
Cultural
Knowledge &
Participation
Approaches to
Learning
Fixed part
L2-based, minority-only at home
2.17 (2.25) -0.61 (1.59) 0.79 (3.35) 1.29 (2.02) 3.34 (2.21) 2.69 (3.29) 1.77 (2.02) -0.60 (1.41)
L2-based, Myanmar at home
3.80 (1.66) 2.17 (1.84) 2.16 (2.19) 0.49 (1.26) 3.51 (0.83) 3.82 (1.64) 2.43 (1.03) 1.75 (0.86)
Bilingual, minority-only at home
2.36 (2.30) 1.56 (1.94) -0.08 (1.72) 2.80 (1.13) 1.92 (1.53) 2.45 (1.93) 2.39 (1.76) 2.46 (1.66)
Bilingual, Myanmar at home
4.12 (2.66) 2.01 (2.12) 1.13 (4.18) 1.03 (0.92) 5.31 (1.59) 2.96 (2.72) 4.35 (1.75) 1.92 (1.57)
L1-based, minority-only at home
4.46 (1.97) 3.62 (1.61) 3.27 (2.26) 2.85 (0.94) 4.81 (1.20) 3.78 (1.70) -0.23 (1.70) 3.14 (1.08)
L1-based, Myanmar at home
8.63 (2.75) 6.79 (2.16) 6.91 (2.35) 4.56 (1.96) 8.67 (3.41) 7.27 (2.12) 4.33 (0.88) 3.67 (2.59)
Random part
Province
2.78 (0.43) 2.32 (0.41) 0.00 (.) 0.00 (.) 2.90 (0.48) 2.20 (0.64) 3.19 (0.29) 0.81 (0.96)
Ward
2.92 (0.49) 2.32 (0.45) 4.03 (3.76) 2.77 (12.21) 2.77 (1.15) 3.42 (1.07) 2.17 (0.30) 1.47 (0.61)
Residual
8.64 (0.24) 9.27 (0.51) 9.37 (1.78) 9.66 (8.02) 8.87 (0.28) 9.22 (0.65) 9.38 (0.27) 9.60 (0.30)
Note. The table shows fixed (bold indicates p < .05) and random (SD units) estimates from a series of HLM regressions, and SEs are shown in
parentheses. being very close to 0.
EARLY CHILDHOOD EDUCATION IN MYANMAR
46
Figure 1. Associations between ECE dosage (5 categories) and marginal mean overall
month-of-age-adjusted EAP-ECDS scores (mean=100; SD=10). Coefficient estimates are
based on an HLM regression with overall development as the dependent variable and ECE
duration as the independent variable, controlling for age, sex, urban-rural residence,
household wealth, and parental SES. Error bars show 95% confidence intervals.
94
96
98
100
102
104
106
108
No ECE 0-6 months 6-12 months 12-18 months >18 months
Marginal mean overall EAP-ECDS score
ECE dosage
EARLY CHILDHOOD EDUCATION IN MYANMAR
47
Figure 2. Associations between ECE dosage (5 categories) and marginal mean overall
month-of-age-adjusted EAP-ECDS scores (mean=100; SD=10), by ethnic group. Coefficient
estimates are based on an HLM regression with overall development as the dependent
variable and an interaction term between ECE duration and ethnic minority status as the
independent variable, controlling for age, sex, urban-rural residence, household wealth, and
parental SES. Error bars show 95% confidence intervals.
92
94
96
98
100
102
104
106
108
No ECE 0-6 months 6-12 months 12-18 months >18 months
Marginal mean overall EAP-ECDS score
ECE dosage
Majority Minority
EARLY CHILDHOOD EDUCATION IN MYANMAR
48
Figure 3. Marginal mean overall month-of-age-adjusted EAP-ECDS scores (mean=100,
SD=10) for ethnic minority children, by ECE participation, language of ECE instruction, and
whether or not Myanmar is spoken at home. Marginal mean estimates are calculated from an
HLM regression including ethnic minority children only, with overall development as the
dependent variable and 7 categories of ECE participation status as the independent variable,
controlling for age, sex, urban-rural residence, household wealth, and parental SES. Error
bars show 95% confidence intervals.
90
92
94
96
98
100
102
104
106
108
110
Minority-only
at home
Myanmar at
home
Minority-only
at home
Myanmar at
home
Minority-only
at home
Myanmar at
home
Does not attend
ECE
L2-based instruction Bilingual instruction L1-based instruction
Marginal mean overall EAP-ECDS score
ECE and language status
