Applied Developmental Science

ISSN: 1088-8691 (Print) 1532-480X (Online) Journal homepage: https://www.tandfonline.com/loi/hads20

The long-term benefits of Montessori pre-K for

Latinx children from low-income families

Arya Ansari & Adam Winsler

To cite this article: Arya Ansari & Adam Winsler (2020): The long-term benefits of Montessori
pre-K for Latinx children from low-income families, Applied Developmental Science, DOI:
10.1080/10888691.2020.1781632

To link to this article: https://doi.org/10.1080/10888691.2020.1781632

Published online: 30 Jun 2020.

Submit your article to this journal

Article views: 19

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=hads20
APPLIED DEVELOPMENTAL SCIENCE
https://doi.org/10.1080/10888691.2020.1781632

The long-term benefits of Montessori pre-K for Latinx children from low-
income families
Arya Ansaria and Adam Winslerb
a
The Ohio State University; bGeorge Mason University

ABSTRACT
This study used covariate adjusted regression techniques to compare the third-grade out-
comes of low-income Latinx children who attended Montessori pre-K programs (n ¼ 161)
with those who graduated from more conventional programs (n ¼ 4975) in Miami-Dade
County, Florida. Children who experienced one year of Montessori education demonstrated
stronger pre-academic skills at the end of pre-K and, in turn, performed better on standar-
dized assessments of math and reading in third grade than those who did not. No differen-
ces emerged in students’ identification as gifted and talented nor in third-grade GPA. Taken
together, these findings suggest that the benefits of one year of Montessori at age 4 may
carry forward over time and to the extent that they do, these benefits are attributed to the
fact that Montessori graduates entered kindergarten more ready academically. At the same
time, however, the persisting benefits of Montessori were 60–70% smaller four years after
program exit and were less robust than the end of pre-K outcomes.

Publicly funded early childhood education programs of different types of preschool curricula, and there are
are often viewed as an important means of preparing even fewer studies on the potential longer-term bene-
our nation’s youngest children for school, especially fits. One such model that has received surprisingly lit-
for low-income and ethnic minority children who are tle attention in the educational literature includes the
often at risk for academic and behavioral difficulties Montessori Method, which has been around since the
(Phillips et al., 2017). Despite this promise, recent early 1900s (Montessori, 1994) and has been advo-
findings have brought into question the long-term cated for as a potential remedy for inequality (Lillard,
efficacy of these programs (Hill et al., 2015; Lipsey 2016). To help redress some of these gaps in know-
et al., 2018; Puma et al., 2012); and, therefore, there ledge, the present investigation follows up on the
has been increased interest in understanding the spe- work of Ansari and Winsler (2014) who studied the
cific features of early childhood programs that might short-term experiences of children from low-income
increase opportunities for all children to succeed families in Miami-Dade County, Florida and found
throughout their educational careers (Burchinal, that Montessori pre-K programs were more effective
2018). One such aspect of early schooling that has than conventional programs for developing low-
become a focal point of educational research is the income, Latinx (but not Black) children’s pre-aca-
different types of preschool curricula implemented by demic skills at kindergarten entry. More specifically,
programs, which are intended to help teachers estab- here we push this discussion forward by considering
lish their day-to-day routines and goals for their stu- whether the short-term benefits of Montessori educa-
dents (Jenkins & Duncan, 2017). That is, there are tion observed for Latinx children at age 4 (Ansari &
growing questions concerning whether: (a) certain Winsler, 2014) are sustained into third grade.1 To test
curricula are more effective in facilitating children’s for the long-term benefits of Montessori education,
early learning and development; and (b) these early we examine a broad set of academic outcomes that
benefits of curricula are sustained over time. include children’s standardized test scores, grade point
To date, however, there has been very limited average, and enrollment in gifted and talented classes.
research that has considered the short-term outcomes In doing so, this study responds to recent calls for

CONTACT Arya Ansari ansari.81@osu.edu The Ohio State University, 1787 Neil Avenue, Columbus, OH 43210.
We also explore, as secondary, supplemental analyses, whether late “sleeper” effects emerge in third grade for the Black children in the study sample.
1

ß 2020 Taylor & Francis Group, LLC

2 A. ANSARI AND A. WINSLER

longitudinal studies that focus on the experiences of programs. Reflecting the above, Montessori is also
students from low-income communities and students often seen as more phonetic than traditional pro-
of color in public Montessori classrooms (Debs & grams, and this emphasis on sounds and visuals may
Brown, 2017). also facilitate Latinx children’s engagement in aca-
demic learning (Bodrova, 2008; Denton et al., 2000).
But even with its longstanding history, the evidence
The Montessori Method
base surrounding the effectiveness of Montessori edu-
One of the most well-known curricula in the United cation is surprisingly limited and oftentimes dated,
States is the Montessori Method. As brief background, especially for students of color and low-income chil-
Montessori programs are generally characterized by dren. For example, most of the older studies in this
three-year age groupings and focus on individualized area are often limited in their ability to isolate
learning. Unlike more conventional educational mod- Montessori impacts because study children are not
els, Montessori classrooms are less structured and randomly assigned to programs and many of these
contain less teacher-directed instruction. Instead, studies lack adequate pretest data (Lopata et al., 2005;
Montessori classrooms place greater emphasis on White et al., 1976). And other studies, including
child-centered education, choice, and collaboration newer work, that have tried to derive a more causal
among children, which some developmental and edu- link between Montessori education and children’s
cational scholars argue is more ideal for promoting early learning and development are hampered by
children’s early academic and social-behavioral devel- other factors, such as small sample sizes (Lillard et al.,
opment (Lillard, 2016). Additionally, Montessori class- 2017; Miller & Bizzell, 1984); or lack of economic and
rooms are well known for their large open spaces that racial/ethnic diversity (Lillard, 2012).
allow children to work on activities independently for These limitations in the literature, especially con-
extended periods of time and learn at their own pace cerning demographic and socioeconomic diversity,
and leisure. The classroom materials and activities are may reflect the fact that, historically, Montessori pro-
also arranged in a manner that allows for children to grams were generally located in private schools and
advance from simple to complex and from concrete to served higher-income families in the United States.
abstract materials, and once children master these Recent research, however, has found that there has
tasks, they move on to more challenging activities been an expansion of Montessori programs to the
(Lillard, 2016). In a Montessori classroom, the role of public sector (Debs & Brown, 2017; National Center
the teacher is not to dictate information, but rather, for Montessori in the Public Sector, 2014). In fact, in
to help connect children to different activities that the last several decades, the Montessori Method has
meet their developmental needs and that involve self- expanded to over 500 public schools across the
guided learning. Thus, one central premise of United States and, at present, public Montessori pro-
Montessori is that children learn best when adults do grams serve roughly 125,000 students per year, includ-
not interfere and when children can concentrate with- ing a large share of low-income and ethnic minority
out interruption in the classroom (Lillard, 2016). children (Debs & Brown, 2017; National Center for
Perhaps most germane to the lives of low-income, Montessori in the Public Sector, 2014).
Latinx children is the fact that the Montessori Method
is well recognized for culturally responsive teaching
Recent research on Montessori education
with a focus on individualization, flexibility, and
autonomy, which is particularly relevant for the early Even with these limitations and changes in the demo-
learning and development of young Latinx children, graphic composition of Montessori programs across
many of whom are learning English as a second lan- the country, there have been a number of studies that
guage (Garcia & Jensen, 2007, 2009). Accordingly, have tried to address some of these methodological
Montessori classrooms provide culturally and linguis- concerns outlined above by using matched control
tically diverse Latinx children with the opportunity to groups and lottery designs, or by controlling for
be self-directed learners and to move at their own demographic characteristics of children and families.
pace and master the skills necessary for future school When taken as a whole, what these studies suggest is
success. Within this type of environment, it is likely that children who participate in Montessori programs
that Latinx children with varying English fluency in preschool and kindergarten generally do better in
would be better able to participate fully in Montessori school in the short-term as compared with children
classrooms compared with more conventional pre-K who do not participate in Montessori (Ansari &
 APPLIED DEVELOPMENTAL SCIENCE 3

Winsler, 2014; Bhatia et al., 2015; Kayılı, 2018; Kayili to four years later (Bailey et al., 2017). Thus, whether
& Ari, 2011; Lillard, 2012; Lillard & Else-Quest, 2006). the benefits of one year of Montessori education per-
In particular, Montessori participants demonstrate sist throughout children’s educational careers (as
stronger academic skills, social-behavior, and executive opposed to extended exposure) is unclear.
functioning than children who attend more conven- Even with these promising findings regarding both
tional educational programs. Likewise, students who the short-term benefits of Montessori education as
experience Montessori education in the older grades well as extended exposure, it is important to acknow-
also demonstrate stronger school performance ledge that a few evaluations of the Montessori model
(Culclasure et al., 2018; Dohrmann et al., 2007; Lillard appear to suggest that programs that implement this
& Else-Quest, 2006; Rathunde & Csikszentmihalyi, curriculum might not necessarily support school-aged
2005); and these benefits extend to low-income chil- children any more than conventional educational pro-
dren and children of color (Ansari & Winsler, 2014; grams (Cox & Rowlands, 2000; Kirkham & Kidd,
Brown & Lewis, 2017; Culclasure et al., 2018; 2017; Lopata et al., 2005). In particular, these studies
Dohrmann et al., 2007; Mallett & Schroeder, 2015). have found that children in Montessori programs
In addition to these short-term evaluations of demonstrate similar levels of creativity (Cox &
Montessori education, two studies also provide Rowlands, 2000; Kirkham & Kidd, 2017); and, in
important insight into the potential longer-term bene- some cases, perform significantly worse on assess-
fits of these programs. The first study by Dohrmann ments of language skills (but not math; Lopata et al.,
and colleagues (2007) found that middle class children 2005) as compared with children in conventional pro-
with 8–9 years of experience in Montessori schools grams. Additionally, Laski, Vasilyeva, and Shiffman
performed significantly better on high school assess- (2016) found that despite the initial academic benefits
ments of math and science as compared with a demo- of Montessori participation in kindergarten, these
graphically similar group of classmates who attended benefits conferred by Montessori did not persist over
regular non-Montessori schools during the earlier time such that children in Montessori and non-
years. And the second study conducted by Lillard and Montessori programs demonstrated similar academic
colleagues (2017), which was based on a lottery skills by the end of third grade.
design, suggested that ethnically diverse children from
middle-class families who experienced three years of
The persisting benefits of Montessori and
Montessori education demonstrated elevated levels of
models of skill building
academic achievement, improvements in social-behav-
ior, and stronger executive function skills. Even Notwithstanding the mixed-evidence regarding the
though both of these studies certainly provide import- long-term efficacy of Montessori education, there has
ant insight into the efficacy of Montessori education, been limited research on why Montessori programs
it is important to note that both focused on middle may (or may not) have sustained benefits for children.
class children who experienced Montessori education And even though studies aimed at understanding why
for extended periods of time. early childhood programs more generally have down-
The above is of note because the reality for many stream benefits for children are scarce, one of the
children from low-income homes is that they may most basic developmental arguments behind investing
attend these programs for one or two years before in the early childhood years is that skills beget skills
transferring to other (and perhaps regular) public (Cunha et al., 2006). That is, all capabilities are built
schools. And, in general, we know from other work in on a foundation of capacities developed early in the
the education literature that the short-term impacts of life course and early mastery of various academic and
other types of targeted curricula (Clements et al., socioemotional skills can reinforce motivation to learn
2013) and pre-K programs more generally (Bailey more in the future.
et al., 2017; Duncan & Magnuson, 2013; Phillips et al., In the present study, we test this conceptual model
2017; Yoshikawa et al., 2013); do not always persist of skill building and consider whether the benefits of
three to four years later. For example, a meta-analysis Montessori can be sustained over time because higher
of over 67 early childhood education and intervention level skills are based on lower level ones developed
programs showed that the benefits of one year of pro- during the early years. In other words, if Montessori
gram participation corresponds to roughly 25% of a programs promote the foundational skills needed for
standard deviation at program exit, but these benefits future academic learning, these programs, in turn,
often diminish to only 5% of a standard deviation two may be able to influence third-grade outcomes. As
4 A. ANSARI AND A. WINSLER

part of this effort, we focus on two potential media- Method

tors, namely children’s pre-academic and social-behav-
The current investigation uses a subset of data from
ioral skills, which the extant literature has found are
the Miami School Readiness Project (MSRP; Winsler
malleable to early intervention (Phillips et al., 2017;
et al., 2008), which was a university-community
Weiland & Yoshikawa, 2014); and shape children’s
applied partnership and program evaluation. As part
long-term educational success (Duncan et al., 2007;
of the sampling frame, the MSRP recruited most of
McClelland et al., 2007; Raver et al., 2011). We con- the low-income population of 4-year-olds (n ¼ 41,339;
nect these mechanisms to a constellation of outcomes, 58% Latinx, 35% Black; 7% White/other) in Miami-
including third grade test scores and course grades. Dade County between 2002 and 2006 who were
And given the substantial links between Montessori enrolled in either a public school pre-K program
education and end of pre-K outcomes (Ansari & (58%) or receiving subsidies for low-income families
Winsler, 2014), we also examined whether participa- to attend child care in the community (42%).
tion in Montessori programs shifted children’s educa- Approximately 90% of the population of children in
tional experiences in other fundamental ways (i.e. these programs consented to participate in the MSRP.
placement in gifted and talented classes; Phillips et al., It is important to note that the sampling frame of the
2016; Bailey et al., 2017; Winsler et al., 2013). MSRP did not include children at age four who were
(a) attending Head Start, (b) being cared for exclu-
The current study sively by a parent, or (c) attending private child care
without subsidies (e.g. more advantaged families). For
When taken as a whole, the small quantity of research our purposes, we include only Latinx children from
on Montessori education as compared with other edu- all five years/cohorts of the MSRP if they attended a
cational models demonstrates the need for more Title-I public school pre-K program at age four. We
research in this area to better assess the extent to restricted our sample to Title-I programs because all
which Montessori education results in short- and Montessori classrooms were housed in Title-I schools
long-term benefits for students, especially for ethnic (i.e. high poverty schools), whereas conventional pro-
minority children and children from low-income grams were located in both Title-I and non-Title-I
homes. To help fill in these gaps in knowledge, we schools. Additionally, we only included children who
address two research questions: remained in the county through the end of third
grade and entered third grade on time. The latter
1. Do Montessori pre-K programs have sustained restriction criteria was implemented to isolate com-
academic benefits for Latinx children from low- parisons among typically developing children without
income homes four years later as they finish social and academic confounds associated with reten-
third grade? tion. These exclusions resulted in a final sample of
2. Do the sustained benefits of Montessori preschool 5136 children (end of pre-K sample ¼ 7045).
programs operate through Latinx children’s pre- Although the Latinx children in the MSRP were
academic and social-behavioral skills? not more or less likely to leave the district as a func-
tion of their pre-K type, those who left the district
Given the conflicting evidence regarding the lon- were slightly younger (65.99 months of age vs.
ger-term benefits of early childhood programs more 66.42 months of age, p < 001), more likely to be male
generally (Bailey et al., 2017; Phillips et al., 2017; (54% vs. 48%, p < .001), and slightly less likely to
Yoshikawa et al., 2013), and for pre-K curricula in speak Spanish at home (87% vs. 89%, p < .05). On
particular (Clements et al., 2013; Laski et al., 2016), average, however, children in our final analytic sample
we do not make directional hypotheses and leave our were 66.42 months of age (SD ¼ 3.55) at kindergarten
study objectives as exploratory. However, given con- entry, 11% of children were born outside of the
cerns in the literature regarding selection effects, in United States, and 89% spoke Spanish at home.
this study, we control for preexisting differences in Reflecting the sampling frame, roughly 70% of chil-
children’s early learning upon pre-K entry, which is dren received free lunch in kindergarten and another
recognized as one of the strongest adjustments for 18% received reduced price lunch (the remainder of
omitted variable bias in the context of correlational children did not apply or were denied). Additionally,
research (Duncan & National Institute of Child 48% of study participants were boys. It is important
Health & Human Development Early Child Care to acknowledge that children who attended
Research Network, 2003). Montessori pre-K programs were more likely to speak
 APPLIED DEVELOPMENTAL SCIENCE 5

Table 1. Demographic characteristics and outcomes of children in conventional and Montessori pre-
K classrooms.
Conventional Montessori Effect size difference
Child characteristics and experiences
Age at kindergarten entry 66.42 (3.54) 66.39 (3.69) 0.01
Male 0.48 0.53 0.11
Home language Spanish 0.89 0.95 0.21 
Preschool assessment in Spanish 0.25 0.50 0.57 
Immigrant 0.11 0.15 0.11
Free/reduced lunch
Did not apply 0.06 0.03 0.11
Denied 0.07 0.05 0.06
Reduced 0.18 0.10 0.21 
Free 0.69 0.81 0.27 
Attended pre-K in 2005/2006 0.41 0.49 0.16
Pre-K school quality 3.13 (0.99) 2.83 (0.79) 0.31 
Kindergarten school quality 3.28 (0.95) 2.98 (0.85) 0.31 
Child outcomes
Preschool entry
Academic skills 43.90 (24.64) 41.54 (23.37) 0.09
Social behavior 56.18 (24.11) 50.02 (21.49) 0.25 
Preschool exit
Academic skills 60.07 (23.60) 66.23 (22.07) 0.26 
Social behavior 63.82 (24.23) 61.75 (20.44) 0.08
Third grade outcomes
Grade point average 3.14 (0.58) 3.10 (0.61) 0.08
Reading test score 320.81 (57.36) 320.64 (60.36) 0.00
Math test score 340.71 (60.54) 344.29 (53.63) 0.06
Identified as gifted 0.16 0.16 0.02
Sample size 4,975 161
Notes. Estimates correspond to means or proportions. Estimates in brackets correspond to standard deviations. Proportions
might not sum to 1.00 due to rounding.
p < .001;
p < .01;
p < .05.

Spanish at home and were more likely to be assessed policy change in the county starting in 2005, which
in Spanish in pre-K compared with children who we account for in our models (discussed in more
attended conventional programs (see Table 1 for effect depth below). Even though it is certainly true that
sizes for these differences). Additionally, Montessori only a small share of the overall sample experienced
graduates were also more likely to receive free lunch Montessori pre-K programs in this community, our
in kindergarten and attended lower quality public estimates closely match national estimates of pre-K
schools in pre-K and kindergarten than their peers programs across the country that implement the
who attended conventional pre-K programs at age 4. Montessori Method (2%; Jenkins & Duncan, 2017).
Thus, in general, Montessori graduates were some- Because the Montessori programs in Miami-Dade
what more disadvantaged as compared with their County were located in magnet/charter schools, fami-
peers who received a more conventional public school lies were allowed to apply to enroll their children
pre-K education. from other areas in the district. Unfortunately, the
schools in the district were not able to accommodate
all children and, thus, families who wanted to enroll
Measures
their children in Montessori pre-K programs had to
Age 4 pre-K experience apply through a lottery. Regardless of curricula, the
At the time, eight of the Title-I public schools used a public pre-K programs in Miami-Dade operated for
Montessori pre-K curriculum (n ¼ 161 children) while 3–4 h per day and were required to be staffed by certi-
the remainder of the pre-K programs in the district fied teachers with a child–adult ratio of no more than
used a more conventional High/Scope curriculum 20:2. However, unlike regular pre-K programs that
with a literacy supplement (Building Early Language required teachers to have a bachelor’s degree,
and Literacy – 2002–2004) or the Houghton Mifflin Montessori classrooms required that teachers have a
curriculum within the High/Scope framework master’s degree and to have received special teacher
(2005–2006; more details on High/Scope provided training and certification to become official
below). This change of curriculum among the conven- Montessori teachers. To meet the needs of the cur-
tional programs in the district corresponded to a riculum, complete Montessori materials and activities
6 A. ANSARI AND A. WINSLER

were also provided, but like many other studies in this determined the language of assessment (26% Spanish)
area (Jenkins et al., 2018), information on fidelity was based on the child’s strongest language. Given the
not available. Accordingly, Montessori programs in strong inter-correlations among the three subscales
this study, at least with regard to training and materi- ranging from 0.48 to 0.62, we created an overall com-
als, were as rigorous as those considered classic posite of pre-academic skills (a ¼ .78). This approach
Montessori programs (Lillard, 2012). With that said, it corresponds to the total score presented in the LAP-D
is important to note that in Miami-Dade County, the manual (Nehring et al., 1992).
Montessori Method was only used during the pre-kin-
dergarten year; after children transitioned to kinder- Social-behavior
garten, the Montessori Method was no longer used in Children’s social skills and behavior problems were
any of the public school classrooms. reported on by teachers at the beginning
In contrast to the Montessori Method, the High/ (September-October) and end (April-May) of the
Scope curriculum balances child-initiated and teacher- preschool year with the Devereux Early Childhood
directed activities, and is one of the most frequently Assessment (DECA; LeBuffe & Naglieri, 1999). As
used curricula in pre-K (and in MDCPS). The four part of the DECA, preschool teachers were asked to
principles of High/Scope include active learning, posi- rate children’s social-behavior from the prior four
tive child–adult interactions, a child-friendly environ- weeks (0 ¼ never to 4 ¼ very frequently). Questions
ment, and consistent routines. High/Scope teachers from the social skills subscale include: “starts or
organize the environment, materials, and routines to organizes play with other children,” whereas an
facilitate active learning and, more importantly, to fos- example of a behavior concerns item is “fights with
ter a positive social environment through small- and other children.” Prior studies have shown that the
large-group activities. Throughout the school day, the teacher-rated DECA has strong internal consistency
role of teachers is to observe, extend, and facilitate and has been validated with Latinx children (as ¼
opportunities for active learning and scaffold actions .90–.94; Crane et al., 2011). Given the strong correl-
to best support children’s development (Hohmann & ation between the social skills and behavior problems
Weikart, 2002). Although smaller experimental trials sub-scales of this instrument (absolute value of corre-
that were carried out in the 1960s (i.e. Perry lations ranged from 0.62 to 0.64), we created a com-
Preschool) suggest that children exposed to High/ posite of the two for social-behavior (a ¼ .77). For
Scope demonstrate better outcomes (Schweinhart this measure, higher scores are indicative of more
et al., 2005), in larger and more recent studies, the positive social behavior.
benefits of the curriculum have been questioned
(Preschool Curriculum Evaluation Research Third-grade math and reading test scores
Consortium, 2008). During the third grade year, the state of Florida
required that all children take the Florida
Pre-academic skills Comprehensive Assessment Test (FCAT; Human
Children’s cognitive, language, and fine motor skills Resources Research Organization & Harcourt
were directly assessed at the beginning (September- Assessment, 2007), which includes standardized
October) and end (April-May) of the preschool year assessments of children’s math and reading skills
using the Learning Accomplishment Profile (range of 100–500, a ¼ 0.98). These math and reading
Diagnostic (LAP-D; Nehring et al., 1992). The LAP-D assessments are criterion-based and are designed to
is a nationally normed instrument that has strong test for the skills students must achieve at each grade
internal consistency, both nationally (a ¼ .76–.92; level in order to be promoted. To be promoted to
Nehring et al., 1992) and within the MSRP (a ¼ fourth grade in particular, children must earn a Level
.93–.95; Winsler et al., 2008) Prior validation studies 2 or higher on the reading FCAT (range 1–5); how-
have also found that this assessment battery is highly ever, to be considered at grade level, children must
correlated with other assessments of school readiness score at a Level 3 or higher.
(e.g. Woodcock Johnson, Peabody Picture Vocabulary
Test; Hardin et al., 2005). Spanish and English ver- Third-grade point average
sions of the LAP-D were available, both of which At the end of the third grade, children also received
have demonstrated strong test–retest reliability marks (A ¼ 4, B ¼ 3, C ¼ 2, D ¼ 1, F ¼ 0) from teach-
(.93–.97; Hardin et al., 2005). The LAP-D was admin- ers in multiple subject areas (i.e. reading, writing, lan-
istered by children’s pre-K teachers who also guage arts, math, science, social studies, art, music,
 APPLIED DEVELOPMENTAL SCIENCE 7

and physical education). In light of the strong correla- (3) these end of preschool academic and social out-
tions in grades between subject areas (typically comes serve as mediators in our longitudinal analyses.
.35–.73, lowest being .22 involving physical educa- After establishing the end of preschool associations
tion), we created an overall composite of children’s between Montessori pre-K education and children’s
GPA in third grade (a ¼ .87). school readiness, we then turn to the third grade out-
comes. For these analyses, we estimated two sets of
Gifted and talented classification models. In the first set of models, we included the
Eligibility for gifted and talented classes in Miami covariates—including children’s preschool entry
Dade County was determined by state criteria and skills—and children’s pre-K type to examine the asso-
there were two pathways to enrollment (for more ciations between Montessori education and children’s
details see: Winsler et al., 2013). First, children who third grade outcomes (i.e. Model 1). Then, in Model
were referred by parents or teachers were adminis- 2, we incorporate measures of children’s end of pre-
tered intelligence testing by a school psychologist and school academic and social-behavioral skills to test for
those who exceeded a score of 130 were considered mediation, which was confirmed with the INDIRECT
eligible for gifted and talented services. Because the command in Mplus.
school district had pushed to increase the representa- All continuous variables were standardized to have
tion of minority students, children with limited a mean of 0 and standard deviation of 1 and, there-
English skills, and low-income children, a second fore, all estimates can be interpreted as effect sizes.
pathway was established for enrollment in gifted serv- Although gifted status was dichotomous (0 ¼ no, 1¼
yes), we estimated linear probability models for this
ices. Specifically, children who scored 112–129 on the
outcome to facilitate interpretation of indirect effects.
IQ test were also eligible for gifted services if they: (a)
As a precaution, parallel logistic regression models
scored at or above the 89th percentile on the FCAT;
were estimated for this outcome and all findings were
(b) received grades of As or Bs in their coursework;
the same as reported below (results available from
or (c) were nominated by their teachers or parents as
authors). All analyses employed full information max-
being above average in creativity, very rapid learners,
imum likelihood estimation to address missing data.
or as demonstrating strong leadership abilities. Once
Although missing data averaged out to roughly 15%,
children were classified as gifted and talented, it
there was variability as a function of the indicator of
remained in their school records; however, families
interest. For example, in terms of third grade out-
could (and some likely did) opt-out of these services. comes and demographic covariates (discussed in more
Unfortunately, we do not have access to information detail below), missing data ranged from 0–2% and
about gifted services used. Our indicator of gifted and 0–10%, respectively. The highest level of missing data
talented services is pulled from the school records was for the pre-K entry and exit outcomes, which
and, therefore, any child who had ‘gifted’ listed as were approximately 55% for pre-academic and 15%
their primary exceptionality between kindergarten and for social behavior. Moreover, we accounted for
third grade was considered to be gifted in dependence in child outcomes with school-level clus-
our analyses. tered standard errors in third grade (for similar meth-
ods see: Jenkins et al., 2018; Weiland & Yoshikawa,
Analysis plan 2014). It is important to note that our findings were
not sensitive to the year of clustering (i.e. pre-K ver-
All analyses were estimated in Mplus (Muthen & sus third grade).
Muthen, 1998-2013). First, to replicate the work of Finally, to reduce the possibility of spurious associ-
Ansari and Winsler (2014), we estimated multivariate ations and account for differential selection into the
regression models that considered the benefits of different pre-K programs, all models control for a set
Montessori education for children’s end of preschool of covariates that were drawn from school records.
academic and social-behavior development. We esti- These covariates include: children’s age at kindergar-
mated these models for three reasons: (1) the current ten entry, children’s gender, children’s primary home
study sample was slightly different given our require- language (English or Spanish), children’s country of
ment that children remain in Miami Dade County birth (United States or overseas), and an indicator of
Public Schools through the end of third grade; (2) as socioeconomic status as captured by receipt of free or
a means of reducing spurious associations, we include reduced lunch in kindergarten (did not apply, denied,
a broader set of covariates in the present study; and received reduced lunch, received free lunch). We also
8 A. ANSARI AND A. WINSLER

Table 2. Results of regression models predicting children’s lowest 25% who made learning gains on the
end of preschool and third grade outcomes as a function of FCAT (25%).
Montessori enrollment.
Model 1 Model 2
B (se) R 2
B (se) R2 Results
End of pre-K outcomes As discussed above, the first aim of the present inves-
Pre-academic skills 0.433  0.496 – –
(0.107) tigation was to examine whether there were differen-
Social-behavior 0.016 0.433 – – tial school readiness outcomes associated with
(0.066)
Third grade outcomes Montessori pre-K participation for low-income, Latinx
Grade point average 0.046 0.199 – – children. Confirming the results reported by Ansari
(0.058)
Math test 0.141  0.196 0.037 0.241 and Winsler (2014), we found that net of socio-demo-
(0.069) (0.069) graphic characteristics of children and their preschool
Reading test 0.158  0.177 0.075 0.213
(0.076) (0.077) entry skills, Latinx children who experienced
Identified as gifted 0.030 0.113 – – Montessori education in this community demon-
(0.026)
strated greater pre-academic skill gains at the end of
Notes. Coefficients correspond to effect sizes and the “effect” of
Montessori education (vs. conventional programs). Both models 1 and 2 the pre-K year as compared with their Latinx peers
controlled for child age at kindergarten entry, child gender, home lan- who attended more conventional public school pre-K
guage, assessment language, immigrant status, receipt of free/reduced
lunch, an indicator that captured the roll out of voluntary pre-K in the programs. As can be seen in Model 1 of Table 2, the
community, and school quality (for end of pre-K outcomes only pre-K effect sizes for the documented benefits of Montessori
quality was controlled for whereas for end of third grade outcomes
both pre-K and kindergarten quality was included). Model 2 included education corresponded to roughly 43% of a standard
children’s end of pre-K academic and social behavioral skills and was deviation (SD; p < .001). In contrast, no significant
only estimated for the third-grade outcomes that differed as a function
of Montessori education in Model 1. All R2 estimates were significant at differences emerged for the end of pre-K social-behav-
p < .001. ior skills of children in Montessori and conventional
p < .001;
p < .01; programs (see Table 2).
p < .05. Having established the short-term academic bene-
fits of Montessori pre-K education, our next set of
analyses considered children’s third grade school per-
formance. Results from these longitudinal analyses
controlled for a dummy indicator that captured the
revealed that the low-income, Latinx children who
roll out of voluntary pre-K in the community in 2005
(i.e. 0 ¼ child attended pre-K prior to 2005; 1¼ child participated in Montessori programs during the pre-K
attended pre-K or during/after 2005) along with child- year scored higher on standardized assessments of
ren’s language of assessment on the LAP-D (English math and reading in third grade, with effect sizes of
or Spanish). All models also controlled for children’s approximately 14–16% of a SD (p’s < .05). However,
preschool entry pre-academic and social-behavioral no significant differences emerged in their GPA, and
skills, which as discussed above, is one of the stron- Latinx children who experienced Montessori educa-
gest adjustments for omitted variable bias (Duncan & tion at age 4 were no more (or less) likely to be
National Institute of Child Health & Human enrolled in gifted and talented classes by the end of
Development Early Child Care Research Network, third grade (see Model 1 of Table 2).
2003). Moreover, to account for the fact that children Having established the short- and long-term bene-
who experienced Montessori education were some- fits of Montessori education, our next set of models
what more disadvantaged than their peers who explored the possibility of indirect effects for the out-
attended conventional pre-K programs in Title-I comes with observable and significant differences in
schools (Ansari & Winsler, 2014), we controlled for third grade. After including measures of children’s
school quality in both preschool and kindergarten, end of preschool pre-academic and social-behavioral
which was determined by the Florida Department of skills in our models, we found that both dimensions
Education. As discussed in more depth elsewhere of early learning were independently associated with
(Hines & Winsler, 2016), every year, schools were children’s math (pre-academic skills ¼ .26 SD, p <
graded on a five point scale (A ¼ 4, B ¼ 3, C ¼ 2, .001; social-behavioral skills ¼ .08 SD, p < .001) and
D ¼ 1, F ¼ 0), which was determined by: (a) school- reading test scores (pre-academic skills ¼ .21 SD, p <
wide achievement on the FCAT (50%); the percent of .001; social-behavioral skills ¼ .10 SD, p < .001).
students who made learning gains in each domain of Results from the mediation models also revealed sig-
the FCAT (25%), and the percent of students in the nificant indirect effects via children’s pre-academic
 APPLIED DEVELOPMENTAL SCIENCE 9

skills. Specifically, children who experienced possible that the educational benefits of Montessori
Montessori education at age 4 scored higher on stand- accrue later in the life course (Clarke, 1981). Given
ardized assessments of math and reading (indirect this possibility, we replicated the aforementioned
effect sizes ¼ 0.08–0.10, p’s < .001) in part because models among the subgroup of Black children
they entered kindergarten more ready academically (n ¼ 4573) who remained in the county through third
than their peers who attended conventional pre-K grade. Results from these analyses revealed that Black
programs. On the other hand, because Montessori Montessori graduates did not perform significantly
pre-K enrollment was not linked with children’s end better (or worse) on any third-grade benchmark as
of preschool social-behavior, no indirect effects compared with Black children who attended conven-
emerged through this pathway (see Model 2 of tional programs. Although there were no significant
Table 2). differences on any study outcome, for three of the
four benchmarks the direction of associations, though
negligible, were negative, suggesting that Black
Robustness checks
Montessori graduates were doing slightly, but not sig-
One common issue in the developmental and educa- nificantly, worse than those who graduated from con-
tional sciences has to do with concerns of omitted ventional pre-K programs The differences were as
variable bias that may invalidate findings within the follows: In third grade, Black Montessori graduates
context of correlational research. To test these issues, performed 1% of a SD higher on the math test (p >
we employed a replacement of cases framework, .25), 3% of a SD worse on the reading test (p > .25),
which quantifies how many children in the and had a lower GPA by approximately 5% of a SD
Montessori condition would have to be replaced with (p > .25) and were 2% less likely to be identified as
counterfactual cases with a zero effect to invalidate an gifted and talented than Black children who graduated
inference (Frank et al., 2013). In doing so, this frame- from more conventional pre-K programs (p > .25).
work enables us to quantify bias within the context of
nonrandom assignment. Results from this effort
Discussion
revealed a few different patterns. First, the benefits of
Montessori education upon pre-K exit would become Despite the growth of Montessori education across
non-significant if a little over 50% of Montessori cases the country, there has been limited research on the
were replaced with cases that had a zero effect. These benefits of this educational model in the public and
findings suggest that the initial benefits of Montessori private sector for children’s early learning and devel-
education were fairly robust and, importantly, opment (Debs & Brown, 2017; National Center for
exceeded the estimates documented in other observa- Montessori in the Public Sector, 2014). Hence, as part
tional educational studies (32%; Frank et al., 2013). of the current investigation, we sought to push the
Despite the fairly robust patterns at pre-K entry, the field forward and provide new insight into whether
direct “effects” reported for children’s math and read- attending public Montessori pre-K programs has sus-
ing test scores in third grade were far less robust. To tained benefits for low-income, Latinx children as
invalidate the inference regarding the persisting bene- they progressed through their early educational
fits of Montessori, only 5% of cases would have to be careers. As noted earlier, previous research with these
replaced with cases for which there was a zero effect. data had considered the short-term benefits of one
At the same time, however, the documented indirect year of Montessori education and found that 4-year-
effects were more robust as they would require old Latinx Montessori attendees in Miami-Dade
approximately 45% of cases to be replaced. Thus, County demonstrated greater school readiness gains
when taken together, these findings suggest that the as compared with their peers in more conventional
immediate benefits of Montessori education were programs through the end of pre-K (Ansari &
more robust than the longer-term associations. Winsler, 2014). Our study examines the longer-term
Moreover, and as noted earlier, Ansari and Winsler (third grade) benefits of Montessori pre-K education
(2014) found that Montessori pre-K programs were and, thus, extends this evidence-base in two meaning-
more effective than conventional programs for devel- ful ways, which we discuss in more detail below.
oping low-income, Latinx (but not Black) children’s First, the results from the present investigation
pre-academic skills in this community, hence the reveal that one year of Montessori public school pre-
focus of the present investigation on Latinx children. K education in Miami-Dade County had some sus-
Despite the lack of initial program benefits, it is tained benefits for children four years down the line.
10 A. ANSARI AND A. WINSLER

Specifically, the Latinx children who attended they are noteworthy given that Montessori graduates
Montessori programs at age 4 and subsequently tran- experienced a greater degree of inequality as evi-
sitioned to regular public schools for kindergarten and denced by the higher rates of free lunch and lower
beyond performed somewhat better on standardized school quality. The fact that Montessori graduates
tests for math and reading. At the same time, how- were doing somewhat better in light of their circum-
ever, there were no significant direct effects of stances is impressive and is aligned with findings
Montessori education for children’s grade point aver- from Lillard and colleagues (2017) who found that
age or placement in the gifted and talented program. three-years of Montessori education reduced income-
One potential explanation for the lack of direct effects based differences in children’s academic achievement.
on children’s school grades and giftedness is the sub- Moreover, the fact that Montessori programs were
jective nature of teachers’ judgments. Indeed, prior housed in lower quality elementary schools would
studies have shown that teachers have a general ten- appear to indicate that these programs may have ino-
dency to judge the academic achievement of low- culated Latinx children who came from low-income
income children as low (Alvidrez and Weinstein, communities somewhat from broader struc-
1999; Muller et al., 1999), and this might be the case tural inequality.
regardless of children’s pre-K history. But when taken The second way in which our study adds to the
as a whole, these results echo a number of existing existing literature is the focus on the mechanisms for
studies that have found that participation in the long-term benefits of Montessori education. In
Montessori has short-term benefits for children’s early particular, our work supports models of skill building
learning and development (Ansari & Winsler, 2014; (Cunha et al., 2006) and reveals that the reason
Bhatia et al., 2015 Kayili & Ari, 2011; Kayılı, 2016; Montessori programs have longer-term benefits for
Lillard, 2012; Lillard & Else-Quest, 2006); and illus- children through the end of third grade is largely
trate that some of these short-term benefits can be because Montessori graduates—at least in this com-
sustained over time. These findings also mirror some munity—enter kindergarten with stronger pre-aca-
of the work done by Dohrmann and colleagues (2007) demic skills. Similar to other work in the Miami
and Lillard and colleagues (2017) who documented community by Ansari and colleagues (2017), we found
long-term benefits of Montessori enrollment. that these early improvements in children’s early aca-
To our knowledge, however, our study is one of demic knowledge, in turn, meant that children per-
the first to demonstrate that the early academic bene- formed better on standardized assessments of math
fits of attending one year of Montessori at the age of and reading. Put another way, participation in
4 might have persisting benefits for students’ school Montessori at age 4 allowed children to seize the
performance, for low-income, Latinx children. opportunities provided to them in the early elemen-
Nonetheless, our findings also illustrate a more com- tary school years, which in turn, had long-term cas-
mon pattern documented in the early childhood lit- cading benefits. At the same time, however, it is
erature more broadly: The benefits of early childhood important to acknowledge that Latinx children’s
education diminish over time (Bailey et al., 2017; social-behavior also played an important role for their
Duncan & Magnuson, 2013; Phillips et al., 2017; long-term school success, but we found that
Yoshikawa et al., 2013). Indeed, the benefits of Montessori graduates did not demonstrate stronger
Montessori education for Latinx children diminished social-behavioral skills through the transition to
by roughly 60–70% by third grade. At program exit, school than their peers in more conventional pro-
Montessori graduates scored roughly 45% of a SD grams. Accordingly, the primary mechanism for the
higher on academic test scores than non-participants, sustained benefits of Montessori pre-K education was
but only 15% of a SD higher at the end of third grade. children’s earlier academic achievement.
This degree of convergence is similar to recent meta Moreover, even though the results of our investiga-
analyses that found that the benefits of early child- tion are suggestive of long-term benefits of
hood programs more generally diminish by roughly Montessori education and touch on potential mecha-
70–75% in the 4–5 years after program completion. nisms for these long-term associations, one question
Just as importantly, we found that the persisting bene- that remains is what specific aspects of the Montessori
fits of Montessori education for Latinx children were Method matter most for promoting Latinx children’s
less robust than those documented at program exit. short- and long-term school success. Although the
With that said, and even though the persisting bene- Montessori Method is often thought of as a collection
fits of Montessori are small by traditional standards, of themes and practices, there are a variety of
 APPLIED DEVELOPMENTAL SCIENCE 11

potential explanations for the stronger academic out- (Lillard, 2016), which may mean that programs that
comes of Montessori classrooms, including the materi- implement this model are better equipped to take
als used in these programs and the focus on an asset-based view of children’s background and
individualization, flexibility, and autonomy (Lillard, competencies than more conventional pre-
2012, 2016), which plays a critical role in the develop- K programs.
ment of young Latinx children (Garcia & Jensen, Despite these contributions to the literature, there
2007, 2009). Beyond the differences in programs as a are a number of limitations and future directions that
result of curriculum, it is also possible that another need to be taken under consideration when interpret-
reason children do better in Montessori—both in the ing our study findings. First, the programs that par-
short- and long-term—is because their teachers are ticipated in this study only enrolled 4-year-old
more educated as compared with teachers in regular children and, therefore, were not mixed-age, which is
pre-K programs, which is beyond the scope of the one of the components of the Montessori Method
data available. Indeed, prior studies have shown that (Lillard, 2016). These findings require attention in the
more educated teachers display higher quality interac- future as they might mean that mixing ages, a charac-
tions with students and are more engaged in the class- teristic of Montessori education, is not necessary for
room (Pianta et al., 2005, 2018). Even with this having positive long-term effects. It was also unfortu-
possibility, other studies of Montessori programs have nate that we did not have information on fidelity or
found that the materials and curriculum itself play an how well the curriculum was implemented in the pre-
important role (Lillard, 2012). Put another way, while K classrooms, which has important ramifications for
teachers can (and do) play a central role in children’s early learning (Lillard, 2012). Although the
Montessori classrooms, it is unlikely that they are the lack of information regarding implementation fidelity
sole reason for the findings reported herein given that is certainly a limitation of our work, the same holds
other studies suggest that changes in the materials true for many other studies that examine early educa-
matter for children’s early learning and development. tion models (Dohrmann et al., 2007; Jenkins et al.,
And even though some scholars argue that the 2018; Kayili & Ari, 2011; Lillard & Else-Quest, 2006;
Montessori model is effective because it is bundled Miller & Bizzell, 1984). For these reasons, future
into one package (Culclasure et al., 2018), it would researchers should more carefully consider implemen-
seem critical in moving forward that future studies tation fidelity. Nevertheless, in light of prior work
more carefully consider what aspects of the done on Montessori fidelity, it is likely that our docu-
Montessori model are most effective in facilitating mented findings are biased downward given that not
children’s learning. One way in which this can be all aspects of the Montessori curriculum were imple-
done is leveraging variation in implementation fidel- mented and, as a result, our results likely represent a
ity, which may allow researchers to tease apart what conservative estimate of the benefits of Montessori
aspects of this program yield the greatest impact. And education (see, Lillard, 2012).
on a related note, it was also unfortunate that, beyond Related to implementation fidelity, the study chil-
school-level quality, we did not have other indicators dren only experienced one year of Montessori educa-
available to isolate other potential school level tion, which might reflect the experiences of many
differences. children, but limits the type of questions that we
One might also wonder why Latinx children in par- could address. For example, as a result of data limita-
ticular benefit more from pre-K programs that imple- tions, we could not consider the potential benefits of
ment Montessori as compared with conventional multiple years of Montessori enrollment, which is an
programs. One possibility is that the Montessori area that requires continued attention (Laski et al.,
Method is better aligned with Latinx children’s needs. 2016; Lillard et al., 2017). Therefore, similar study
Indeed, many educational and developmental scholars designs in fully implemented Montessori programs are
have argued that Latinx children, especially those who necessary to determine if the results reported herein
are learning English as a second language, benefit are consistent after the full three-year cycle. Caution
most from educational environments that are cultur- is also warranted when interpreting our findings
ally responsive and provide children with the oppor- because: (a) children were not randomly assigned to
tunity to progress at their own leisure (Garcia & the different pre-K programs; and (b) our results are
Jensen, 2007, 2009). The Montessori Method not only only applicable to programs in Miami-Dade County
emphasizes individualized and child-driven instruc- serving low-income Latinx children, some of which
tion, but is also committed to cultural openness are no longer in operation.
12 A. ANSARI AND A. WINSLER

It is also important to note that sample distribution low-income, Latinx children. What these results also
of students in Montessori and conventional programs make clear is that early Montessori participation had
provided us adequate power to detect main effect sizes benefits that extend upwards because Montessori
of a little over 0.20, but unfortunately, we could not graduates entered school more ready to learn. Thus,
detect smaller main effects that are often times the when taken with other studies in the field (Bhatia
norm in this area of work nor could we consider vari- et al., 2015; Culclasure et al., 2018; Dohrmann et al.,
ation in program effects within the Latinx population. 2007; Kayili & Ari, 2011; Kayılı, 2016; Lillard, 2012;
Additionally, even though we had access to adminis- Lillard & Else-Quest, 2006; Lillard et al., 2017), our
trative records of third grade school performance, results provide some support for the implementation
some of our measures were somewhat limited. For of Montessori in public school pre-K programs but
example, our findings regarding children’s school also highlight the importance of replication in mov-
grades could be biased if teachers across schools did ing forward.
not use the same criteria for assigning grades (e.g. The project on which this article is based was
percentage of final grade determined by in class funded by the Early Learning Coalition of Miami-
assignments and homework). And even though we Date/Monroe and the Children’s Trust. The Trust is a
used state of the art missing data procedures, it is dedicated source of revenue established by voter refer-
important to acknowledge that a limitation of our endum to improve the lives of children and families
study is the high levels of missing data for the pre-K in Miami-Dade County. Opinions reflect those of the
entry and exit outcomes. authors and do not necessarily reflect those of the
Finally, although it is certainly true that there have granting agencies.
been advances made in statistical methods, such as
propensity score matching, these methods are only as
References
good as the observables available. In fact, recent
research suggests that the results of propensity scores Alvidrez, J., & Weinstein, R. S. (1999). Early teacher percep-
tions and later student academic achievement. Journal of
are not substantively different from models with cova-
Educational Psychology, 91(4), 731–746. https://doi.org/
riates (Elze et al., 2017) and, under certain conditions, 10.1037/0022-0663.91.4.731
matching methods can amplify bias (King and Ansari, A., Lόpez, M., Manfra, L., Bleiker, C., Dinehart,
Nielsen, 2019). That is, propensity scores serve the L. H. B., Hartman, S. C., & Winsler, A. (2017).
same purpose as covariate adjustment because they do Differential third grade outcomes associated with attend-
not change the causal identification strategy (Rohrer, ing publicly funded preschool programs for low-income,
Latino children. Child Development, 88(5), 1743–1756.
2018). Thus, in the absence of an experimental design,
https://doi.org/10.1111/cdev.12663
causality is in question because the same unobserved Ansari, A., & Winsler, A. (2014). Montessori public school
variables that caused children to select into the pre-K programs and the school readiness of low-income
Montessori programs (e.g. parent characteristics) may Black and Latino children. Journal of Educational
have influenced their third grade school performance Psychology, 106(4), 1066–1079. https://doi.org/10.1037/
(e.g. parents more likely to enter a Montessori lottery a0036799
Bailey, D., Duncan, G. J., Odgers, C. L., & Yu, W. (2017).
may also be more likely to stimulate their children
Persistence and fadeout in the impacts of child and ado-
cognitively at home). With that said, we did estimate lescent interventions. Journal of Research on Educational
the robustness of our findings and found that the Effectiveness, 10(1), 7–33. https://doi.org/10.1080/
immediate benefits of Montessori were fairly robust, 19345747.2016.1232459
but some of the longer-term associations were less so. Bhatia, P., Davis, A., & Shamas-Brandt, E. (2015).
Accordingly, before firm conclusions can be drawn Educational gymnastics: The effectiveness of Montessori
practical life activities in developing fine motor skills in
about the efficacy of Montessori education, replication
kindergartners. Early Education and Development, 26(4),
with different samples across different communities 594–607. https://doi.org/10.1080/10409289.2015.995454
with different methods is necessary. Replication across Bodrova, E. (2008). Make-believe play versus academic
different communities is particularly important given skills: a Vygotskian approach to today’s dilemma of early
the fact that our study was not diverse in some childhood education. European Early Childhood
key ways. Education Research Journal, 16(3), 357–369. https://doi.
org/10.1080/13502930802291777
With these limitations and future directions in
Brown, K., & Lewis, C. W. (2017). A comparison of reading
mind, the results of the current investigation build on and math achievement for African American third grade
the prior literature and illustrate the potential long- students in Montessori and other magnet schools. The
term benefits of Montessori pre-K programs that serve Journal of Negro Education, 86, 439–448.
 APPLIED DEVELOPMENTAL SCIENCE 13

Burchinal, M. (2018). Measuring early care and education Covariate Adjustment. Journal of the American College of
quality. Child Development Perspectives, 12(1), 3–9. Cardiology, 69(3), 345–357. https://doi.org/10.1016/j.jacc.
https://doi.org/10.1111/cdep.12260 2016.10.060
Clarke, A. (1981). "Sleeper effects" in development: Fact or Frank, K. A., Maroulis, S. J., Duong, M. Q., & Kelcey, B. M.
artifact?. Developmental Review, 1(4), 344–360. https:// (2013). What would it take to change an inference?
doi.org/10.1016/0273-2297(81)90030-7 Using Rubin’s causal model to interpret the robustness of
Clements, D. H., Sarama, J., Wolfe, C. B., & Spitler, M. E. causal inferences. Educational Evaluation and Policy
(2013). Longitudinal evaluation of a scale-up model for Analysis, 35(4), 437–460. https://doi.org/10.3102/016237
teaching mathematics with trajectories and technologies: 3713493129
Persistence of effects in the third year. American Garcia, E., & Jensen, B. (2007). Language development and
Educational Research Journal, 50(4), 812–850. https://doi. early education of young Hispanic children in the United
org/10.3102/0002831212469270 States. National Task Force on Early Childhood
Cox, M. V., & Rowlands, A. (2000). The effect of three dif- Education for Hispanics.
ferent educational approaches on children’s drawing abil- Garcia, E., & Jensen, B. (2009). Early educational opportuni-
ity: Steiner, Montessori and traditional. British Journal of ties for children of Hispanic origins. Social Policy Report,
Educational Psychology, 70(4), 485–503. https://doi.org/ 23(2), 1–19. https://doi.org/10.1002/j.2379-3988.2009.
10.1348/000709900158263 tb00059.x
Crane, J., Mincic, M. S., & Winsler, A. (2011). Hardin, B. J., Peisner-Feinberg, E. S., & Weeks, S. W.
Parent–teacher agreement and reliability on the Devereux (2005). The Learning Accomplishment Profile–Diagnostic
Early Childhood Assessment (DECA) in English and (LAP-D) Third Edition: Examiner’s manual & technical
Spanish for ethnically diverse children living in poverty. report. Lewisville, NC: Kaplan Early Learning.
Early Education & Development, 22(3), 520–547. https:// Hill, C. J., Gormley, W. T., & Adelstein, S. (2015). Do the
doi.org/10.1080/10409289.2011.565722 short-term effects of a high-quality preschool program
Culclasure, B., Fleming, D. J., Riga, G., & Sprogis, A. persist? Early Childhood Research Quarterly, 32, 60–79.
(2018). An Evaluation of Montessori Education in South https://doi.org/10.1016/j.ecresq.2014.12.005
Carolina’s Public Schools. The Riley Institute at Furman Hines, C., & Winsler, A. (2016). School quality and mobility
University. between public school prekindergarten and kindergarten:
Cunha, F., Heckman, J., Lochner, L., & Masterov, D. V. Predictors and outcomes [Paper presentation]. Paper
(2006). Interpreting the evidence of life cycle skill forma- Presented at the Annual Conference of the American
tion. In F. Hanushek & F. Welch (Eds.), Handbook of the Educational Research Association, Washington, DC.
economics of education. (pp. 697–812). Elsevier. Hohmann, M., & Weikart, D. P. (2002). Educating young
Debs, M., & Brown, K. E. (2017). Students of color and children: Active learning practices for preschool and child
public Montessori Schools: A review of the literature. care programs. Ypsilanti, MI: High/Scope Press.
Journal of Montessori Research, 3(1), 1–15. https://doi. Human Resources Research Organization, & Harcourt
org/10.17161/jomr.v3i1.5859 Assessment. (2007). Technical report for 2006 FCAT test
Denton, C. A., Hasbrouck, J. E., Weaver, L. R., & Riccio, C. administrations: Reading and mathematics. Harcourt.
A. (2000). What do we know about phonological aware- Jenkins, J. M., Duncan, G. J., Auger, A., Bitler, M., Domina,
ness in Spanish? Reading Psychology, 21, 335–352. T., & Burchinal, M. (2018). Boosting school readiness:
https://doi.org/10.1080/027027100750061958 Should preschool teachers target skills or the whole child?
Dohrmann, K. R., Nishida, T. K., Gartner, A., Lipsky, D. K., Economics of Education Review, 65, 107–125. doi: 0.1016/
& Grimm, K. (2007). High school outcomes for students j.econedurev.2018.05.001 https://doi.org/10.1016/j.econe-
in a public Montessori program. Journal of Research in durev.2018.05.001
Childhood Education, 22(2), 205–217. https://doi.org/10. Jenkins, J. M., & Duncan, G. J. (2017). Do pre-kindergarten
1080/02568540709594622 curricula matter. Brookings Institution and The Center
Duncan, G. J., Dowsett, C. J., Claessens, A., Magnuson, K., for Child and Family Policy.
Huston, A. C., Klebanov, P., Pagani, L. S., Feinstein, L., Kayili, G., & Ari, R. (2011). Examination of the effects of
Engel, M., Brooks-Gunn, J., Sexton, H., Duckworth, K., & the Montessori method on preschool children’s readiness
Japel, C. (2007). School readiness and later achievement. to primary education. Educational Sciences: Theory and
Developmental Psychology, 43(6), 1428–1446. https://doi. Practice, 11, 2104–2109.
org/10.1037/0012-1649.43.6.1428 Kayılı, G. (2018). The effect of Montessori method on cog-
Duncan, G. J., & Magnuson, K. A. (2013). Investing in pre- nitive tempo of kindergarten children. Early Child
school programs. Journal of Economic Perspectives, 27(2), Development and Care, 188(3), 327–335. https://doi.org/
109–132. https://doi.org/10.1257/jep.27.2.109 10.1080/03004430.2016.1217849
Duncan, G. J., and National Institute of Child Health and King, G., & Nielsen, R. (2019). Why Propensity Scores
Human Development Early Child Care Research Should Not Be Used for Matching. Political Analysis,
Network. (2003). Modeling the impacts of child care 27(4), 435–454. https://doi.org/10.1017/pan.2019.11
quality on children’s preschool cognitive development. Kirkham, J. A., & Kidd, E. (2017). The effect of Steiner,
Child Development, 74(5), 1454–1475. https://doi.org/10. Montessori, and national curriculum education upon
1111/1467-8624.00617 children’s pretence and creativity. The Journal of Creative
Elze, M. C., Gregson, J., Baber, U., Williamson, E., Sartori, Behavior, 51(1), 20–34. https://doi.org/10.1002/jocb.83
S., Mehran, R., Nichols, M., Stone, G. W., & Pocock, S. J. Laski, E. V., Vasilyeva, M., & Schiffman, J. (2016).
(2017). Comparison of Propensity Score Methods and Longitudinal comparison of place-value and arithmetic
14 A. ANSARI AND A. WINSLER

knowledge in Montessori and non-Montessori students. knowledge on pre-Kindergarten effects. A consensus state-
Journal of Montessori Research, 2(1), 1–15. https://doi. ment. Brookings Institution and The Center for Child
org/10.17161/jomr.v2i1.5677 and Family Policy.
LeBuffe, P. A., & Naglieri, J. A. (1999). Devereux early child- Phillips, D., Gormley, W., & Anderson, S. (2016). The
hood assessment. Kaplan Press. effects of Tulsa’s CAP Head Start program on middle-
Lillard, A. S. (2016). Montessori: The science behind the school academic outcomes and progress. Developmental
genius. Oxford University Press. Psychology, 52(8), 1247–1261. https://doi.org/10.1037/
Lillard, A. S. (2012). Preschool children’s development in dev0000151
classic Montessori, supplemented Montessori, and con- Pianta, R., Howes, C., Burchinal, M., Bryant, D., Clifford,
ventional programs. Journal of School Psychology, 50(3), R., Early, D., & Barbarin, O. (2005). Features of Pre-
379–401. https://doi.org/10.1016/j.jsp.2012.01.001 Kindergarten Programs, Classrooms, and Teachers: Do
Lillard, A. S., & Else-Quest, N. (2006). The early years: They Predict Observed Classroom Quality and Child-
Evaluating Montessori education. Science ( Science.), Teacher Interactions?. Applied Developmental Science, 9(3),
313(5795), 1893–1894. https://doi.org/10.1126/science.1132362 144–159. https://doi.org/10.1207/s1532480xads0903_2
Lillard, A. S., Heise, M. J., Richey, E. M., Tong, X., Hart, A., Pianta, R. C., Whittaker, J. E., Vitiello, V., Ansari, A., &
& Bray, P. M. (2017). Montessori preschool elevates and Ruzek, E. (2018). Classroom process and practices in
equalizes child outcomes: A longitudinal study. Frontiers public pre-K programs: Describing and predicting educa-
in Psychology, 8, 1783. https://doi.org/10.3389/fpsyg.2017. tional opportunities in the early learning sector. Early
01783 Education and Development, 29(6), 797–813. https://doi.
Lipsey, M. W., Farran, D. C., Durkin, K., & Hofer, K. org/10.1080/10409289.2018.1483158
(2018). Effects of the Tennessee prekindergarten program Preschool Curriculum Evaluation Research Consortium.
on children’s achievement and behavior through third (2008). Effects of preschool curriculum programs on
grade. Early Childhood Research Quarterly, 45, 155–176. school readiness (NCER 2008–2009). Washington, DC:
https://doi.org/10.1016/j.ecresq.2018.03.005 National Center for Education Research, Insti- tute of
Lopata, C., Wallace, N. V., & Finn, K. V. (2005). Education Sciences.
Comparison of academic achievement between Puma, M., Bell, S., Cook, R., Heid, C., Broene, P., Jenkins,
Montessori and traditional education programs. Journal F., & Downer, J. (2012). Third grade follow-up to the
of Research in Childhood Education, 20(1), 5–13. https:// head start impact study: Final report. U.S. Department of
doi.org/10.1080/02568540509594546 Health and Human Services.
Mallett, J. D., & Schroeder, J. L. (2015). Academic achieve- Rathunde, K., & Csikszentmihalyi, M. (2005). Middle school
ment outcomes: A comparison of Montessori and non- students’ motivation and quality of experience: A com-
Montessori public elementary school students. Journal of parison of Montessori and traditional school environ-
Elementary Education, 25, 39–53. ments. American Journal of Education, 111(3), 341–371.
Mcclelland, M. M., Cameron, C. E., Connor, C. M., Farris, https://doi.org/10.1086/428885
C. L., Jewkes, A. M., & Morrison, F. J. (2007). Links Raver, C. C., Jones, S. M., Li-Grining, C., Zhai, F., Bub, K.,
between behavioral regulation and preschoolers’ literacy, & Pressler, E. (2011). CSRP’s Impact on Low-Income
vocabulary, and math skills. Developmental Psychology, Preschoolers’ Preacademic Skills: Self-Regulation as a
43(4), 947–959. https://doi.org/10.1037/0012-1649.43.4. Mediating Mechanism. Child Development, 82(1),
947 362–378. https://doi.org/10.1111/j.1467-8624.2010.01561.
Miller, L. B., & Bizzell, R. P. (1984). Long-term effects of x
four preschool programs: Ninth-and tenth-grade results. Rohrer, J. M. (2018). Thinking Clearly About Correlations
Child Development, 55(4), 1570–1587. https://doi.org/10. and Causation: Graphical Causal Models for
2307/1130027 Observational Data. Advances in Methods and Practices in
Montessori, M. (1994). From childhood to adolescence. Clio Psychological Science, 1(1), 27–42. https://doi.org/10.
Press. 1177/2515245917745629
Muller, C., Katz, S. R., & Dance, L. J. (1999). Investing in Schweinhart, L. J., Montie, J., Xiang, Z., Barnett, W. S.,
Teaching and Learning. Urban Education, 34(3), 292–337. Belfield, C. R., & Nores, M. (2005). Lifetime effects: The
https://doi.org/10.1177/0042085999343003 High/Scope Perry Preschool Study through age 40.
Muthen, L. K., & Muthen, B. O. (1998-2013). Mplus User’s Retrieved from http://www.highscope.org/file/Research/
Guide. (7th Ed.). Muthen & Muthen. PerryProject/specialsummary_rev2011_02_2.pdf
National Center for Montessori in the Public Sector. (2014). Weiland, C., & Yoshikawa, H. (2014). Does higher peer
Growth of public Montessori in the United States: 1975- socio-economic status predict children’s language and
2014. http://www.public-montessori.org/white-papers/ executive function skills gains in prekindergarten?
growth-ofpublic-montessori-in-the-united-states-1975- Journal of Applied Developmental Psychology, 35(5),
2014/ 422–432. https://doi.org/10.1016/j.appdev.2014.07.001
Nehring, A. D., Nehring, E. F., Bruni, J. R., & Randolph, White, J. M., Yussen, S. R., & Docherty, E. M. (1976).
P. L. (1992). Learning Accomplishment Profile—Diagnostic Performance of Montessori and traditionally schooled
standardized assessment. Kaplan Press. nursery children on tasks of seriation, classification, and
Phillips, D., Lipsey, M. W., Dodge, K. A., Haskins, R., conservation. Contemporary Educational Psychology, 1(4),
Bassok, D., Burchinal, M. R., Duncan, G. J., Dynarski, 356–368. https://doi.org/10.1016/0361-476X(76)90055-2
M., Katherine, A., Magnuson, K. A., & Weiland, C. Winsler, A., Tran, H., Hartman, S., Madigan, A. L., Manfra,
(2017). Puzzling it out: The current state of scientific L., & Bleiker, C. (2008). School readiness gains made
 APPLIED DEVELOPMENTAL SCIENCE 15

by ethnically-diverse children in poverty attending Prevalence and predictors of placement in elementary

center-based childcare and public school pre-kinder- school gifted education programs. The Urban Review,
garten programs. Early Childhood Research Quarterly, 45(4), 416–447. https://doi.org/10.1007/s11256-013-0259-0
23(3), 314–329. https://doi.org/10.1016/j.ecresq.2008.02. Yoshikawa, H., Weiland, C., Brooks-Gunn, J., Burchinal, M.,
003 Espinosa, L., Gormley, W., Ludwig, J., Magnuson, K., Phillips, D.,
Winsler, A., Karkhanis, D. G., Kim, Y. K., & Levitt, J. & Zaslow, M. (2013). Investing in our future: The evidence base
(2013). Being Black, male, and gifted in Miami: on preschool education. Foundation for Child Development.