Canadian Journal of Behavioural Science Copyright 2007 by the Canadian Psychological Association

2007, Vol. 39, No. 1,1-22 DOI: 10.1037/cjbs2007001

Development and Psychometric Properties of the

Early Development Instrument (EDI):
A Measure of Children's School Readiness*
MAGDALEN A JANUS and DAVID R. OFFORD
Offord Centre for Child Studies, Department of Psychiatry and Behavioural
Neurosdences, McMaster University

Abstract gnitif et connaissances g^n^rales. Les analyses factorielles

The Early Development Instrument (EDI), a teacher-com- ont pu confirmer la presence des trois premiers facteurs,
pleted measure of children's school readiness at entry to mais ont degag^ la n^cessit6 de modifier les deux
Grade 1, was designed to provide communities with an derniers. Ceci a permis de formuler la version finale de
informative, inexpensive and psychometrically sound l'instrument de mesure du developpement a la petite
tool to assess outcomes of early development as reflected enfance qui est compos^e des cinq facteurs suivants :
in children's school readiness. Its psychometric properties sante physique et bien-etre, competence sociale, maturity
at individual level were evaluated in two studies. Five a affective, langage et developpement cognitif, habilet^s a
priori domains - physical health and well-being, social la communication et connaissances g^n^rales. Ces cinq
competence, emotional maturity, language and communi- domaines offrent de bons indices de consistance interne,
cation, and cognitive development and general knowl- comparable a d'autres instruments. Une seconde etude
edge - were tested in a factor analysis of data on over (N = 82) a demontre la presence d'une bonne concor-
16,000 kindergarten children. The factor analyses upheld dance entre revaluation de parents et des enseignants, la
the first three domains, but revealed the need to develop fidelity entre les evaluateurs, la validite concomitante et
two new ones, resulting in the final version of the EDI la validite convergente. Ces resultats confirment que
consisting of: physical health and well-being, social com- l'instrument de mesure du developpement a la petite
petence, emotional maturity, language and cognitive enfance est un outil psychometrique adequat pour eva-
development, communication skills and general knowl- luer les aptitudes de l'enfant lors de l'entree k I'ecole pri-
edge domains. These final domains showed good reliabil-
ity levels, comparable with other instruments. A separate
study (N = 82) demonstrated consistent agreements in
parent-teacher, interrater reliabilities, concurrent validity,
and convergent validity. These results establish the EDI as For many decades, the average percentage of chil-
a psychometrically adequate indicator of child well-being dren with impairing cognitive and behaviour prob-
at school entry. lems in elementary school remained constant at about
25% (Achenbach, 1991; Offord & Lipman, 1996). This
Resume happened despite an increased awareness of the
L'instrument de mesure du developpement a la petite importance of the early years and more widely avail-
enfance a 6te congu afin de fournir un outil approprie, able intervention programs for young children.
peu couteux et valide sur le plan psychometrique, perme- Although children's problems at school entry may
ttant d'evaluer Ies capacites de 1'enfant a entrer en pre- generally occur at a level that would not necessarily
miere annee primaire. Les propri^tes psychometriques de require clinical treatment, there is evidence that as
l'instrument ont ete evalu^es au cours de deux etudes. much as, or perhaps even more than, 25% of children
Une analyse factorielle menee a partir des donnees experience some difficulties that prevent them from
recueillies auprfes de 16 000 enfants de maternelle a per- taking full advantage of the education offered by
mis de tester la presence de cinq facteurs theoriques : schools (e.g., Frank Porter G r a h a m Child
sant6 physique et bien-etre, competence sociale, maturite Development Center, 1999; Rimm-Kaufman, Pianta, &
affective, langage et communication, developpement co- Cox, 2000). Differences in children's first years of
school have long-term sequelae for their school career
* Lome Sulsky was the Action Editor for this manuscript and later life (Alexander & Entwisle, 1988) since even
Redaction de ce manuscrit: Lome Sulsky minor differences in academic achievement at Grade 1

Revue canadienne des sciences du comportement, 2007, 39:1,1-22

 2 Janus and Offord

tend to intensify over the years rather than converge. are rarely used. However, only such assessments can
A population-based model of health suggests that provide background to broadly cast interventions,
low-risk, small deficiencies in large populations con- called "universal" (Offord et al., 1998), which would
tribute to the burden of ill health more than severe have the advantage of helping all children, and thus
problems in a minority of patients (Rose, 1994). Seen raising the population level of school readiness.
in this context, children's school readiness is a health- In the last decade, the issue of children's readiness
relevant, measurable outcome that has long-term con- for school finally reached the forefront of interest not
sequences for population health. just among academics and educators, but also com-
Because early child development is heavily influ- munities and even politicians. In Canada, the 1997
enced by the quality of stimulation, support, and Speech from the Throne contained the commitment
nurturance in the environments where children grow to "measure and report on the readiness to learn of
up, school readiness can be broadly understood as an Canadian children so that we can assess our progress
outcome of the early years. It is a useful construct in providing our children with the best possible
because it acknowledges the importance of the early start." This goal was picked up by a score of commu-
years for children's future development (Schonkoff & nities across the country, making its way into pro-
Phillips, 2000; Shore, 1997). The developmental out- grams and coalitions Qanus & Offord, 2000; McCain
comes, which could be operationalized as school & Mustard, 1999). More recently, one of the political
achievement, behaviour, and cognitive outcomes, or parties in Canada embraced a set of principles sum-
school drop-out rates, depend on the combination of marized as "QUAD" (Liberal Party of Canada, 2005),
the individual and collective factors. While the indi- to be used within the early child care and learning
vidual variation routinely contributes a larger pro- system, thus ensuring the relevance of developmen-
portion of variance than neighbourhood factors tal outcomes for social programs. QUAD stands for
(Boyle & Lipman, 2002), there is nevertheless a grow- quality, universality, accessibility, and developmental
ing body of evidence suggesting that the neighbour- outcomes and represents a promising opportunity to
hood and societal factors also matter, especially, improve the early years' experiences for children.
though not only, within the context of poverty A great deal of debate has been waged over the
(Brooks-Gunn, Guo, & Furstenberg, 1993; Chase- theoretical basis of school readiness and consequent
Lansdale, Gordon, Brooks-Gunn, & Klebanov, 1997). methods of measurement: When should it be mea-
Data from the National Longitudinal Study of sured, who should be the informant, what should be
Children and Youth in Canada (NLSCY) allow us to included (Love, Aber, & Brooks-Gunn, 1994)?
identify factors having an impact on children's out- Readiness for school and its measurement have
comes beyond children's individual characteristics in received their share of attention in the developmental
three broad areas (C. Hertzman, personal com- and educational literature, and several reviews have
muninication, March 27, 2005): family (including been produced to highlight the difference in
income, education, parenting style); neighbourhood approaches over time (Meisels, 1998, 1999; Phillips &
(including safety and cohesion, socio-demographic Love, 1995; Wenner, 1995).
mix); and society (including support for parenting, In the first half of the 20th century, assessment of
e.g., access to high-quality care arrangements). Even school readiness was virtually synonymous with
small changes in any of these three areas can dramat- decision-making for kindergarten entry or delay. The
ically contribute to the social processes behind the tests used focused on reading and writing, and were
well-being of all children, and change the distribu- intended to identify children who should not start
tion of risk at a given level (Offord et al., 1999). To regular kindergarten classes. These trends can be
paraphrase Rose (1994) and Offord, Kraemer, Kazdin, traced to the history of the definition of school readi-
Jensen, and Harrington, (1998), a large number of ness. In the early formulations, it was an ability to
children at a small risk for school failure may gener- perform indicated, usually cognitive, language or
ate a much greater burden of suffering than a small motor tasks on demand (e.g., Gesell test; Ilg, Ames,
number of children with a high risk. Yet, in the cur- Haines, & Gillespie, 1978). Meisels (1998) classifies
rent climate, school or preschool interventions are these types of definitions as "idealist/nativist" or
implemented based on individual diagnostics usual- "empiricist/environmentalist" perspectives. In the
ly only with serious, clinical cases, providing help to idealist/nativist view, readiness can be seen as a
the few whose impairments are severe. Broad assess- within-the-child phenomenon, whereby a child's
ments of children's development in all relevant areas, readiness for school is achieved through a matura-
such that could provide an overall evaluation of the tional process, with little or no impact from the envi-
range of developmental outcomes in a community ronment (including parents, experiences, etc.). The
 Development and Psychometric Properties of the Early Development Instrument (EDi) 3

child's development proceeds through predictable (Lichtenstein, 1990). It has often been used to deter-
stages and cannot be altered by external influences. mine children's readiness for kindergarten, and fol-
Developmental tests were designed to measure this lowed up with placement decisions (Graue &
concept of readiness; however, by adhering too strict- Shepard, 1989). In Graue and Shepard's study, the
ly to specific goals, they tended to misclassify too developmental age measure on GSRT in kindergarten
many children as not ready. The empiricist/environ- correlated with the Grade 1 report card only at 0.23.
mentalist perspective claims that readiness is a set of About 60% of children identified as not ready were
particular behaviours, skills, and personality traits misdiagnosed based on Grade 1 data. Similarly, no
that are basic precursors to school achievements and differences were detected between children classified
are easily measured. Therefore, testing should focus as ready and unready by the GSRT before kinder-
on external evidence of what the child can do. This garten entry in later measures of Grade 1 remedial
conceptualization of readiness provided a theoretical placement, or academic scores in Grade two and
basis for a number of assessments, which tended to three (Buntaine & Costenbader, 1997). Lichtenstein
be curriculum-based or specific-tasks-oriented. (1990) reports an interrater agreement of placement
Unfortunately, similarly to strict developmental recommendation of 78%, based on 46 cases. Few
tasks, such tests often resulted in inappropriate clas- other psychometric properties of the Gesell are avail-
sification of many children. able in literature.
Currently, kindergarten readiness or school readi- Among some of the most frequently used skill-ori-
ness screening measures are often still utilized to pro- ented measures are such readiness tests as the
vide a basis for decision-making on retention, track- Developmental Indicators for the Assessment of
ing, and services (Meisels, 1998), or to be held as per- Learning (DIAL-R) (Mardell-Czudnowski &
formance standards for schools' accountability (La Goldberg, 1998), and the Brigance Diagnostic
Paro & Pianta, 2000). The measures could be skill-ori- Inventory of Early Development (Brigance, 1992;
ented, tapping into the degree of mastery of specific Glascoe, 1995). Both of these measures require a
skills, or developmentally oriented, assessing the trained professional to administer the assessment to
child's developmental age (Costenbader, Rohrer, & children. The assessments include motor, cognitive/
DiFonzo, 2000). Over and above those measures, conceptual, and language areas in 3 (DIAL-R) or up to
school districts use locally constructed tests, or infor- 13 subtests (Brigance, 1992). Each of the two tests
mal observations. New York State school districts, for offers a parent-completed questionnaire to assess
example, use the four types of kindergarten screen- social skills and development. DIAL-R is reported to
ing in almost equal proportions (May & Kundert, have high interrater and test-retest reliabilities (0.90
1992). and 0.86, respectively), and both sensitivity and
In view of the purposes for which they are com- specificity around 85% (Mardell-Czudnowski &
monly used, kindergarten and school readiness mea- Goldenberg, 1998). A positive predictive value of
sures are usually reviewed and validated from the only 0.53, demonstrated in one study Qacob, Snider,
perspective of their accuracy in identifying children & Wilson, 1988), suggests that if used for identifying
at risk for school failure (e.g., Costenbader et al., children at risk for future academic difficulties, it car-
2000), rather than their adequacy of reflecting the ries a high "false-positive" rate. The Brigance is a cri-
concept of school readiness (Meisels, 1998). Seven of terion-referenced inventory of skills, with psychome-
the many well-known and widely used measures tric data similar to those reported for the DIAL-R. One
will be briefly described below. We will review their study of 95 middle-class white 4-5-year-old children
major domains, psychometric properties, and the (Wenner, 1995) found that referrals to special prob-
training needs for assessment. lems and nonpromotion were correctly predicted
One of the earliest measures of school readiness is with the Brigance scores for 67% of children in the
the Gesell School Readiness Test (GSRT), an assess- sample.
ment of skills that are purportedly achieved solely Yet another school readiness assessment, the
through a maturational process (Ilg et al., 1978). It is Lollipop Test (Chew & Lang, 1990), includes four
administered individually to children as an interview subtests covering recognition and identification of
by a trained examiner, who needs to consider the shapes, colours, pictures, letters, and numbers,
content and manner of the child's response. The tasks administered by trained examiners. Chew and Lang
include writing, drawing, visual and motor coordina- (1990) and Chew and Morris (1989) showed that
tion, and the child's verbal expressions. Currently, Lollipop Test's domains mapped closely to those test-
the GSRT is described as an observational, qualitative ed on DIAL-R and the Metropolitan Readiness Test
tool, with results being interpreted clinically (MRT; Swanson, Payne, & Jackson, 1981) yet required
 4 Janus and Offord

a shorter testing time. Neither the Lollipop nor the bly implicit in the fact that the assessments have to be
MRT have specific "readiness levels" used to classify administered by professionals trained in the specific
children as ready or not; their main purpose is to pre- instrument. The need for an external examiner to
dict first grade academic success from a kindergarten administer the tool, rather than reliance on a report
testing. The ability of both MRT and the Lollipop to by an adult familiar with the child, explains why an
predict grades, and standardized achievement test examiner would not be informed well enough to rate
results in Grades 1, 3, and 4 are similar and moderate the child's social behaviour. From the implementa-
to high in magnitude (Chew & Morris, 1989). tion point of view, need for examiners trained in spe-
The Phelps Kindergarten Readiness Scale cific tools increases costs of assessments.
(Augustyniak, Cook-Cottone, & Calabrese, 2004; It is important to note that only three of the tools
Duncan & Rafter, 2005), a newer addition to the spec- mentioned above, the DIAL-R, Brigance, and Phelps
trum of measures, was developed explicitly to mea- were purportedly validated specifically to screen
sure "academic" readiness of children before entry to children who were not ready. Nevertheless,
kindergarten. It contains six major domains: verbal Costenbader et al. (2000) and Duncan and Rafter
processing, perceptual processing, and auditory pro- (2005) suggest that even these three be used in con-
cessing, evaluating children's language competence, junction with other, more detailed psychoeducational
ability to compare and reproduce shapes, and memo- evaluation of readiness. Together with high imple-
ry. Test-retest reliabilities vary from 0.61 to 0.87 for mentation costs and lack of information on children's
individual domains. Concurrent validity, established social and emotional development, this indicates low
in the fall of the kindergarten year with the cost-effectiveness of these measures.
Woodcock-Johnson III Test of Achievement Any of the assessment tools reviewed above could
(Woodcock, McGrew, & Mather, 2001), is 0.59. be used to provide information for groups of chil-
Predictive validity values for the Phelps' total readi- dren, or even population-level data. Population-level
ness score are available for an eight-month period community reporting theoretically can be achieved
with the Woodcock-Johnson, and vary from 0.39 by aggregating any measurement available for all
(reading subtest) to 0.53 (math subtest). In addition, a individuals in the community, or a representative
recent study demonstrated correlations of .47 and .51 sample, similarly to the way census reporting is car-
between the Phelps' kindergarten score and New ried out (Statistics Canada, 2005). However, most
York State fourth-grade assessments in language and available school readiness assessments provide infor-
mathematics tests, respectively (Augustyniak et aL, mation only on the cognitive and language aspects of
2005). child development. Also, since all of them are imple-
The Peabody Picture Vocabulary Test (PPVT; Dunn mented through a direct assessment with an individ-
& Dunn 1981), a test of receptive vocabulary, has also ual child, it would be extremely costly to include all
occasionally been mentioned as a measure of school kindergarten children in such testing. At this point in
readiness (e.g., Kohen, Brooks-Gunn, Leventhal, & time, direct cognitive assessments are rarely done for
Hertzman, 2002; Zill et al., 2001); however, it is rarely whole populations of young children; rather, schools
used as a sole screening method (Costenbader et al., have resources for assessments of children identified
2000). Within its limited skill testing range, the PPVT as at risk by teachers (Love et al., 1994).
has adequate psychometric properties for screening There appears to be a consensus among education-
purposes (Dunn & Dunn, 1981), and is easy and al and developmental experts that school readiness
quick to administer. should be understood as not merely cognitive skills,
Of the seven assessments briefly reviewed above, but rather as a holistic concept involving several
only two offer an optional measure of children's developmental areas such as cognitive, socio-emo-
socio-emotional development, a parent-completed tional, and physical (Jimerson, Egeland, & Teo, 1999;
questionnaire (Brigance and DIAL-R). None of them Love et al., 1994; Meisels, 1999). Competence in all
allows for school-based evidence of children's rela- these areas will ensure that children are ready to ben-
tionships with peers or social competence with adults efit from educational activities offered in the school
other than parents. Most assessments provide some environment (Janus & Offord, 2000). Therefore,
measure of children's motor coordination, confined assessment of children's cognitive status only is no
to fine motor skills (e.g., drawing, writing letters or longer adequate. Furthermore, making costly mea-
numbers, copying shapes). None account for chil- sures available for populations of children (as
dren's gross motor skills (e.g., running, jumping) or opposed to targeted subgroups) would require far
physical independence. Few studies provided infor- greater investments than is currently feasible.
mation on interrater reliability; however, it is proba- Meisels (1999) describes yet another perspective
 Development and Psychometric Properties of the Early Development Instrument (EDI) 5

on the measurement of school readiness, following between the "readiness to learn" and "school readi-
Love et al. (1994): a "social constructivist" approach, ness" (Kagan, 1992; Kagan & Neuman, 1997). The first
where school readiness is defined with reference to refers broadly to the child's neurosystem being ready
how children's behaviour and development are sup- from birth to process information it is being exposed
ported and what the children should be ready for. to and develop accordingly; the second is a narrower
This approach requires a community-level measure- view reflecting the specific domains of development
ment strategy, where assessment of children's abili- relevant to school-based learning as children mature
ties is only one of the components, and has to be put around the age of 4 to 5 years. Kindergarten atten-
in the context of the children's past few years and the dance is still optional in many districts, yet it provides
realities of where they will be educated. Moreover, children with an undisputed advantage in first-grade
this strategy explicitly involves the community's outcomes (Entwisle & Alexander, 1999). Moreover, the
willingness for action based on the results. By pro- structure of teaching in kindergarten classes is very
viding a strategy, and including a context, the social different from grade one. Kindergarten provides the
constructivist view is the most comprehensive transition between the play-based preschool and
approach to the measurement of children's school home environment to the academically based envi-
readiness. ronment of grade school, and ensures that children
This article reports on the development of a new have the opportunity to consolidate skills relevant to
school readiness measurement tool. The holistic grade-school learning. A school readiness measure
framework of children's outcomes at school entry taken at the beginning of the kindergarten year would
was adopted to provide communities with bench- fall back on the mistaken assumption of a common
marks useful for planning of intervention and pre- core of learning happening before school (Meisels,
vention. By emphasizing the population-level of data 1999). However, children who do poorly at school
interpretation, this tool overcomes the barrier of see- readiness measures taken prior to or at the beginning
ing the assessment of school readiness as an individ- of kindergarten, often do well on similar measures of
ual process labeling a child with a deficit. Because it achievement by the end of the year (Meisels, 1987).
comes at the cusp between early development and Even comprehensive screening of children before
school entry, such an assessment has a potential to school entry rarely provides highly reliable results
mobilize communities into providing opportunities (Pianta & McCoy, 1997). Readiness is a process occur-
accessible to all children. ring over time, and cannot simply be completed by
The driving force behind the design of the current the first day of kindergarten. As Meisels puts it, "...
measure was the desire to provide communities with [since] readiness is a process and schools are by neces-
a feasible, acceptable and psychometrically reliable sity a major contributor to this process, then a period
instrument that could be used for whole populations of common schooling needs to occur in which this
of children to monitor community efforts to improve process can take place" (Meisels, 1999, p. 62).
early years' outcomes over time (Janus & Offord, Therefore, an assessment of children's school readi-
2000). The Early Development Instrument (EDI) is a ness for grade one should ideally be carried out well
relatively short, easy-to-administer tool in the format into the kindergarten year, yet with sufficient time
of a teacher-completed checklist, whose results can before the end of the year to allow the use of the col-
be aggregated to various levels (e.g., groups like girls lected data for grade-one programming.
or boys, children living in a neighbourhood, children The EDI combines several areas that have been
attending regular or immersion programs, as well as identified as relevant to children's school readiness
all children) and therefore easily lends itself to link- (Doherty, 1997; Kagan, 1992): physical health and
ages with other population and community data well-being, social competence, approaches to learn-
(Janus, Walsh, Viveiros, & Offord, 2002). Within the ing, emotional maturity, language development, cog-
theoretical framework of approaches to the measure- nitive development, communication skills, and gen-
ment of school readiness, the EDI is positioned in the eral knowledge. This paper describes the develop-
context of the "social constructivist" approach, by ment, factorial structure, and initial psychometric
providing the "child" component necessary to com- properties of the Early Development Instrument
plete the whole picture of community-based school (EDI): A Population-Based Measure for Communities.
readiness.
The focus of the current tool is on children's readi- Instrument Development
ness to enter grade one, rather than on their ability to Conceptual Framework
start attending school at a kindergarten level. This fol- Kagan (1992) and Doherty (1997) outlined the five
lows conceptually the distinction made by Kagan areas of school readiness as pertaining to: physical
 6 Janus and Offord

TABLE 1
Sanaple of Questions From the EDI

Physical Health and Social Competence Emotional Maturity Language and Cognitive Communication Skills and
Weil-Being Development General Knowledge

How often has the child How would you rate this Would you say that this Would you say that this How would you rate this
arrived to school too tired child's ability to get along child will try to help child is generally interest- child's ability to commu-
to do school work? with peers? someone who has been ed in books? nicate own needs in an
hurt? understandable way?

How would you rate this Would you say that this Would you say that this Would you say that this How would you rate this
child's ability to manipu- child accepts responsibili- child gets into physical child is showing aware- child's ability to tell a
late objects? ty for actions? fights? ness of rhyming words? story?

How would you rate this Would you say that this Would you say that this Would you say that this How would you rate this
child's level of energy child is able to solve day- child can't sit still, is rest- child is interested in child's ability to take part
throughout the school to-day problems by him- less? games involving num- in imaginative play?
day? self? bers?

well-being and appropriate motor development; researchers. The first draft of the EDI was reviewed
emotional health and a positive approach to new by a group of educators, early years' professionals,
experiences; age-appropriate social knowledge and and academics with expertise in the field. Changes
competence; age-appropriate language skills; and were made to the draft, and subsequently four focus
age-appropriate general knowledge and cognitive groups with kindergarten teachers were conducted.
skills. There is adequate evidence in literature to indi- For several questions, wording was changed; others
cate that each area has an important impact on chil- were dropped and some added, based on teachers'
dren's adjustment to school and short- or long-term recommendations. Table 1 contains examples of ques-
school achievement (Doherty 1997; Jimerson et al., tions in each domain. In addition, some answer/scor-
1999; Love et al., 1992). This view was confirmed in a ing options were modified in response to feedback
discussion held with educators and early childhood from teachers. In particular, items referring to specific
experts, who requested that each domain be repre- skills were provided with only yes/no options,
sented in the new instrument to provide a compre- rather than along the continuum. Teachers indicated
hensive assessment of children's school readiness. to us that these were a better reflection of children's
school readiness. Conversely, answer options to sev-
Item Selection eral questions on children's overall skills were
The items for the EDI were derived from existing expanded to five, as these were perceived to be more
instruments, key informant interviews, and focus variable. An EDI guide, accompanying the instru-
groups, as suggested by Streiner and Norman (1995). ment, was developed to provide brief explanations
A review of some commonly used teacher and par- and anchors for the items.
ent-completed tools was carried out and items for the
instrument were chosen to fit specific areas. An initial Instrument Description
base of 128 questions was created, over 60% of which The first page of the instrument requests informa-
were modified from the items in the Canadian tion on child demographic variables (gender, date of
National Longitudinal Study of Children and Youth birth, language), as well as on selected variables
(NLSCY). The NLSCY is a federally funded study of a related to the child's school-based designations (e.g.,
representative sample of Canadian children. The English as a second language, special needs, type of
items relevant to child behaviour and language and class), and the completion date.
cognitive areas in the NLSCY were based on a number Pages 2 to 7 of the EDI contain questions relevant to
of standardized instruments and consultations with the five domains of school readiness: physical health
experts (NLSC Project Team, 1995). Because it was and well-being; social competence; emotional maturi-
apparent at the time that the NLSCY did not ade- ty; language and cognitive development; and commu-
quately cover all the areas relevant to school readi- nication skills and general knowledge. Most of these
ness (Morongiello, 1997), new questions were con- are "core" questions, which means they directly con-
structed by the authors for the missing areas, based tribute to one of the five domains. There are also
on Doherty (1997), and field-tested with teachers and questions related to children's special skills and spe-
 Development and Psychometric Properties of the Early Development Instrument (EDI) 7

cial problems. Finally, the last page of the instrument pie questions.) All answers are scored on a 3-point
contains questions about children's prekindergarten scale: often or very true (10), sometimes or somewhat
experience (early intervention, child care, preschool, true (5), and never or not true (0).
etc.). Only questions in the five core domains are used
to score children's school readiness. Emotional Maturity
The questionnaire takes between 7 to 20 minutes This domain contains 28 items' and covers proso-
to complete. It is recommended that it be completed cial behaviour, aggression, inattention and hyperac-
in the second half of the kindergarten year, to give tivity, and anxious behaviours. All answers are scored
teachers the opportunity to get to know children in on a 3-point scale: often or very true (10), sometimes
their class. or somewhat true (5), and never or not true (0).
The descriptions below refer to the domains in the
finalized instrument. The total number of core ques- Language and Cognitive Development
tions in the final version of the instrument is 103 (full This domain contains 26 items and refers to the
version of the instrument is available from authors child's ability to use language correctly and covers
upon request, or at the website). cognitive aspects of language and numeracy, in sev-
All core questions are scored from 0 (lowest score) eral areas: basic literacy and numeracy skills, interest
to 10 (highest score). The domain score is calculated and memory, and more complex literacy. All answers
as a mean score of all the valid answers. Thus, scores are scored on a 2- point Scale: "yes" (10) if a child pos-
for each domain have the same minimum and maxi- sesses a skill and "no" (0) if she/he does not.
mum values, even though there are different num-
bers of items. As the feedback from focus groups Communication Skills and General Knowledge
indicated, this way of scoring and presenting the This domain has eight questions and covers the
results proved to be easier to communicate to audi- child's ability to clearly communicate his/her own
ences with little or no research background. needs and thoughts in a way that is understandable
No more than 30% of missing answers are allowed to both adults and other children, the ability to
per domain. If more than one domain is missing, the understand others, to articulate clearly, as well as
questionnaire is not considered complete and is dis- aspects of general knowledge. In contrast to the pre-
carded from analyses. On average, this occurs in no vious domain, this one is about effective communica-
more than 3% of cases. There is no total score on the tion regardless of the grammatical correctness. Seven
EDI. answers are scored on a 5-point scale from very poor
(0), to excellent (10), in 2.5 increments (0, 2.5, 5, 7.5,
Physical Health and Well-Being and 10). One answer is scored on a 3-point scale
This domain contains 13 items and refers to chil- (often, 10, sometimes, 5, and never, 0).
dren's physical preparedness for the school day, fine
and gross motor skills, energy level throughout the Additional Questions
day, and physical independence (examples are in The three additional sections of the EDI cover chil-
Table 1). Ten questions are answered on a 5-point dren's special skills, special problems, and aspects of
scale (from never to always, or excellent to very the prekindergarten history. Seven general areas in
poor), scored from 10 (best) to 0 (worst) in 2.5 point which young children could demonstrate special
intervals: 10, 7.5, 5, 2.5, and 0. Three questions, about skills are listed: numeracy, literacy, arts, music, athlet-
the child's washroom independence, hand prefer- ics/dance, problem-solving, and other. They are sim-
ence, and level of coordination, are answered in a ply scored as "yes" (1) and "no" (0), and summed up,
yes/no format. "Yes" is scored as 10 and "No" as 0. so for each child there is a total score indicating the
number of special skills they demonstrate. Nine spe-
Social Competence cial problem areas are listed: physical, visual, hear-
This domain contains 26 items and covers the fol- ing, speech, learning, emotional, behavioural, home
lowing areas: competence and cooperation in work- environment, and other. These are scored in the same
ing together with others, ability to remember and fol- way as special skills. For the prekindergarten history,
low rules, curiosity and eagerness, approaches to questions about the following areas are asked: child
learning and problem-solving. (See Table 1 for exam- attendance at any early intervention program, pre-
school, language or religion classes. Junior
1 Two additional items were added to this domain in the subse- Kindergarten level, and participation in non parental
quent year, thus bringing the total number of items to 30. This care. The prekindergarten history items are stand-
addition did not change the factor structure of the instrument. alone questions.
 8 Janus and Offord

TABLE 2
Description of Sample in Each of the Studies

Study Valid N Age % Girls % with English as % Mothers with % Living with
(in years) a second language more than high married biological
(ESL) school education parents

2 82 5.6±0.36 50.0 6.9 79.3 83.3

1 Total:16,074 N/a N/a

JK 4,934 4.9±0.28 50.2 31.9

SK 11,140 5.9±0.30 48.4 27.7

TABLE 3
Average Sociodemographic Characteristics for School Areas Participating in Study 1

Site Average 1996 family income {$) Unemployment rate % No high school diploma
75,291 8.5 16.0
70,331 9.4 16.4
66,343 10.9 19.7
37,901 14.1 26.6
45,348 13.2 31.3
24,193 15.4 43.7
Canada 1996 54,583 10.10 22.7

Study 1 areas for which census data were available. The mean
Method SES indicators were computed for each site and are
Participants. The EDI was implemented in six sites presented in Table 3, alongside Canadian national
and completed for 16,583 students. Of those, 16,074 averages from the 1996 census. For three of the sites,
or 97% of questionnaires were complete (had no the SES indicators were better than Canadian aver-
more than one domain and no more than 30 answers ages, and for the remaining three, they were lower.
in total missing). The sites comprised three large Part of the data were collected in Ontario, Canada,
urban (N = 15,319) and three smaller rural areas (N = where in many sites children can start kindergarten
755). Thus, the rural sites contributed 5% of the sam- at a younger age level, called "Junior Kindergarten."
ple, while the urban sites contributed 95%. Statistics These children turn four years old in the calendar
Canada (2005) reports the distribution of the year they enter school. The majority of children in
Canadian population to be 80% urban and 20% rural Canada, however, start school at the 5-year-old level,
All schools within the school boards were involved called "Senior Kindergarten." Since the sample
with an exception of one site where only about 25% included children at both kindergarten levels, the
of schools participated. As indicated in Table 2, there reporting will be split, where appropriate, into the
were approximately equal proportions of boys and Junior (jK) and Senior levels (SK).
girls in the sample, and for about 30% of children,
English was not their first language. No other demo- Analyses. The data were analyzed using several
graphic data were available on the children. techniques to confirm the a priori domain/factor
Although information on the individual socioeco- structure. A confirmatory factor analysis was comput-
nomic status of the families of the children in the ed on the full sample using principal axis factoring
sample was not available, the neighbourhood SES extraction method with promax rotation, allowing
indicators (average income, unemployment rate, and factors extracted to be correlated. Because of the nat-
high school education) were established for the enu- ural clustering of the data by classroom, the within-
meration areas in which participating schools were and between-classroom factor structure was explored.
located, based on the 1996 Canadian census data, A multilevel confirmatory factor analyses, developed
accessed through the DMTI Spatial Inc. Digital Data. by Muthen (1994), which involves a simultaneous
Enumeration areas were the smallest geographical analysis of both the within- and between-group factor
 Development and Psychometric Properties of the Early Development Instrument (EDI) 9

TABLE 4
Summary of the Factor Analysis (Principal Axis Factoring)

Factor Physical Health and Social Competence Emotional Maturity Language and Communication
Well-Being N of items. Mean N of items. Mean Cognitive Skills and General
N of items. Mean loading (range) loading (range) Development Knowledge
loading (range) N of items. Mean N of items. Mean
loading loading
(range) (range)
1 17, 0,641
(0.411-0,938)
2 16, 0,557
(0,204-0,961)
3 8, 0,861
(0,426-1,061)
4 8, 0,836
(0,769-0,962)
5 1, 0,119 8, 0,627
(0,448-0,876)
6 7,0,691
(0,500-0,833)
7 5, 0,787
(0,704-0,858)
8 6, 0,633
(0,310-0,924)
9 4, 0,835 1, 0,211
(0,515-1,050)
10 4, 0,611
(0,504-0,702)
11 5, 0,511
(0,302-0,697)
12 5, 0,570 1, 0,265
(0,333-0,819)
13 2, 0,894
(0,876-0,913)
14 3, 0,310
(0,271-0,387)
Total 13, 0,572 26, 0,647 28, 0,682 26, 0,576 8, 0,861

Cumulative % 4,8 37,7 48,2 58,9 63,1

variance explained

structure using the Mplus software (Muthen & and gender relationship with the EDI scores were car-
Muthen, 2004), was employed to assess the factor ried out.
structure for each domain. In order to assess the need
for further multilevel analyses, the proportion of vari- Results
ance between teachers or the intraclass correlation Factor structure. The principal axis factoring analy-
coefficients (ICC) obtained in the above procedure sis revealed 14 factors, with eigenvalues greater than
were examined. Finally, the average teacher reliability one. This was expected, since some of the broad
(indicating consistency levels) for each domain was domains covered more than one distinct factor, and
assessed using the unconditional multilevel models forcing the distribution into only five would have
with the hierarchical linear modelling (HLM) method- been counterproductive (Gorsuch, 1983), The 14 fac-
ology. Software used included SPSS and Mplus, tors were aggregated into the five domains based on
In addition, the internal consistency indicators the conceptual framework (Table 4), For all but three
(Cronbach's alpha) for the EDI domains were com- items, the highest loading belonged to a factor within
puted, and the convergent validity analyses on age the predicted domain. However, even for these three.
 10 Janus and Offord

TABLE 5
Indices of Fit for Multilevel Models

Domain Between-teacher Wi thin-teacher

CFI RMSEA SRMR CFI RMSEA SRMR Difference
in CFI
Physical Health and Well-Being 0.699 0.248 0,190 0.739 0.140 0,085 0.105
Social Competence 0,414 0.224 0,202 0.502 0.151 0,141 0.065
Emotional Maturity 0,657 0,200 0.092 0,724 0,130 0,091 0.001
Language and Cognitive Development 0,626 0.157 0.091 0,739 0,098 0,073 0.018
Communication Skills and General Knowledge 0,955 0.170 0.015 0,972 0,101 0.019 0.004

Note. CFI = comparative fit index; RMSEA = root mean square error approximation; SRMR = standardized root mean residual.

TABLE 6
Results of the Multilevel Confirmatory Analyses
domain. The values for "between" and "within" com-
parisons are very close, regardless of the models
Intraclass Teacher employed (between or within). This indicates that the
coefficients reliability factor structure within classrooms is similar to the
(ICC) estimates
structure between classrooms.
Physical Health and Well-Being 0.305 0.843
Social Competence 0.196 0.759 Teacher reliability. Intraclass correlations for the five
Emotional Maturity 0.206 0,770 domains are in Table 6. The ICCs for all the items var-
Language and Cognitive ied from the minimum of 0.017 to the maximum of
Development 0,245 0,804
Commurucation Skills and
0,400, with 57% of items at 0,200 and less, indicating
Ceneral Knowledge 0.258 0,813
low levels of variability between classrooms or teach-
ers. In the case of all items and domain scores, the
majority of variance came from children (0,600 to
the second highest loading belonged to the predicted
0,983).
domain. Seven items were retained despite loading
less than 0.3 on a factor, due to perceived importance Average teacher consistency in each domain, esti-
by teachers participating in the focus groups (three of mated with the HLM reliabilities, varied from 0,76 to
those were the ones that did not separate as expect- 0.84 (Table 6).
ed). These items were: independent in washroom,
well-coordinated, sucks a finger, knows how to han- Internal consistency. The internal consistency of the
dle a book, interested in books, interested in reading, specified domains was explored using Cronbach's
remembers things easily. All remaining items loaded alpha. All five domains showed satisfactory internal
0.3 or higher on the factors.^ consistency levels: Physical Health and Well-Being
The 14-factor solution accounted for 63.1% of the 0.84; Social Competence 0.96; Emotional Maturity
variance. The factors contributed to the five domains 0,92; Language and Cognitive Development 0.93; and
in the following way: Physical Health and Well- Communications Skills and General Knowledge 0.95.
Being, Factors 7, 10, and 14 (and one item from 5),
4,8% of variance; Social Competence, Factors 1, 9, Relationship to age, gender, and English as a second
and 12, 32,9% of variance; Emotional Maturity, language status. The EDI was intended to be an instru-
Factors 4, 5, 6, and 11, 10.5% of variance; Language ment based on the child's developmental status and
and Cognitive Development, Factors 2, 8, and 13 (and not achievement in relation to specific curriculum
one items each from 9 and 12), 10.7% of variance; objectives. Therefore, it was imperative that it should
Communications Skills and General Knowledge, be sensitive to the child's age and gender. One-way
Factor 3, 4.2% of variance (Table 4). analysis of variance (ANOVA) was used to compare
The Muthen procedure for exploring between- the five domain scores for girls and boys. Table 7
and within-group factor variance confirmed the fac- shows the means and standard deviations, separately
tor structure. Table 5 shows the fit indices for each for the cohort of 4-year-olds and 5-year-olds. Girls
were rated on average significantly higher than boys
in all five domains.
2 The actual loadings are available from the first author. Correlations of the EDI domains with age were
 Developnnent and Psychometric Properties of the Early Development Instrument (EDI) 11
TABLE 7
Gender and Age in the EDI Scores

SK (5-year-olds) JK (4-year-olds)
N Mean Standard f ratio* N Mean Standard F ratio" Correlation
Deviation Deviation with age'
Physical health Boys 5,628 8.39 1.15 172.00 2,448 7,87 1,25 107,12
and well-being ,231
Girls 5,267 8.66 1,03 2,470 8,23 1,16
Social Boys 5,278 7.80 1.94 438,17 2,210 7,23 2,09 207,13
competence ,170
Girls 4,965 8.54 1,62 2,220 8.07 1.78
Emotional Boys 5,590 7,41 1.65 717,49 2,434 6,93 1.69 278,65
,162
maturity Girls 5,253 8,20 1.38 2,457 7.68 1.44
Language and Boys 5,634 7,69 2,10 187,42 2,449 5.93 2,37 102.22
,377
cognitive development Girls 5,275 8,21 1,90 2,471 6,60 2,25
Communication Boys 5,609 6,85 2,20 125,63 2,449 6,20 2,35 52,29
,192
and general Girls 5,247 7.31 2,12 2,470 6,69 2,32
knowledge
•All statistically significant at p < .0001.

TABLE 8
EDI Domains by ESL and JK/SK

JK SK
ESL N Mean Standard f-stat p-value N Mean Standard F-stat p-value
EDI Domain Status Deviation Deviation
Physical No ESL 3,096 8.18 1,19 111,92 < 0.001 7,625 8,56 1.11 48,45 < 0,001
ESL 1,448 7.78 1,22 2,928 8,39 1.08
Social No ESL 2,814 7.86 1,93 107,26 < 0.001 7,228 8.21 1.83 22,59 < 0,001
ESL 1,275 7.18 1,98 2,745 8.01 1.79
Emotional No ESL 3,077 7.45 1,64 107.17 < 0.001 7,593 7.82 1.61 20,59 < 0,001
ESL 1,440 6.93 1,46 2,914 7.67 1.45
Language/ No ESL 3,098 6.68 2,21 359.03 < 0.001 7,637 8.10 1,94 165,09 < 0,001
Cognitive ESL 1,448 5,33 2.30 2,930 7.54 2,20
Communication / No ESL 3,097 7,47 1.80 3,279.51 < 0,001 7,587 7.76 1,83 4,189,20 < 0,001
General Knowledge ESL 1,448 4,19 1.79 2,927 5.17 1,87

also all statistically significant (Table 7; both cohorts Love, 1995) and accounted for 63% of variance.
analyzed together), and fairly low, with exception of However, two domains, covering the language, com-
the Language and Cognitive domain, where the cor- munication, and cognitive abilities, did not emerge as
relation reached a moderate range, as was expected. the a priori hypothesized categories (language with
These results demonstrate the EDI's expected sensi- communication, and cognitive development sepa-
tivity to age and gender. rately). Considering the range of abilities that are
Differences between children with and without supposed to contribute to each domain in the theoret-
ESL status were also explored with an ANOVA (the ical models, it was to be expected that the factor
gender distributions did not differ significantly analyses would reveal more than five factors. This
between the groups either for JK or SK level). As multifactorial structure Of the domains needs to be
expected, the EDI scores were lower for children for explored further. It was crucial, however, that most, if
whom English was a second language, both at the 4- not all, items showed clear contributions to the set of
year and 5-year-old level. The largest discrepancies factors belonging to a particular domain. In fact, all
between the two groups were in the Communication but three of the items loaded on the expected factors.
Skills and General Knowledge domain (Table 8), Factor analysis experts suggest removing lowest-
loading items (Gorsuch, 1983); however, we made the
Discussion decision to keep the seven items that loaded the low-
The factor solution replicated the domains of est on the finalized version of the EDI. This decision
school readiness found in literature (e,g,, Phillips & was dictated by the need to preserve the relevance of
 12 Janus and Offord

TABLE 9
Parent Interview Questions and Their Relevance to the EDI Domains

Question Responses Association with EDI domains

How many times has your family consulted or Number Physical Health and Well-Being
visited a health professional within the past year?
In general, would you say that your child's health is: 1-poor, 2-fair, 3-good, 4-very good, Physical Health and Well-Being
5-excellent
In your opinion, how physically active is your child 1-not active at all, 2-a bit active, Physical Health and Well-Being
compared to other children the same age and sex? 3-average, 4-active, 5-very active
Would you describe your child as being usually: 1-happy and interested in life, 2-some- Physical Health and Well-Being
what happy, 3-somewhat unhappy, Social Competence
4-unhappy with little interest in life, Emotional Maturity
5-so unhappy that life is not worthwhile
How would you describe his/her usual ability to 1-able to think clearly and solve prob- Social Competence
think and solve day-to-day problems? Is she/he: lems, 2-having a little difficulty, 3-hav- Emotional Maturity
ing some difficulty, 4-having a great
deal of difficulty, 5-unable to think or
solve problems
Apart from school, about how many days a week 1-never, 2-1 day/wk, 3-2 to 3 days/wk, Social Competence
does she/he do things with other children? 4-4 to 5 days/wk, 5-6 to 7 days/wk, Emotional Maturity
During the past six months, how well has your child 1-very well, no problems, 2-quite well,
been getting along with other children such as hardly any problems, 3-pretty well,
friends or classmates (excluding siblings)? occasional problems, 4-not well at all,
constant problems
Since starting school in the fall, how well has she/he As above Social Competence
been getting along with her/his teachers? Emotional Maturity
During the past six months, how well has she/he As above Social Competence
been getting along with her/her parents? Emotional Maturity
With regard to how she/he feels about school, how 1-almost never, 2-rarely, 3-sometimes, Social Competence
often does she/he look forward to going to school? 4-often, 5-almost always Emotional Maturity

How often does your child look at books, comics, 1-rarely, 2-less than once a month, Social Competence
magazines, etc. on his/her own? 3-once a month, 4-a few times a months, Emotional Maturity
5-once a week, 6-a few times a week, Language and Cognitive
7-daily, 8-many times each day Development
Communication Skills
and General Knowledge
How often does she/he play with markers or pencils As above Language and Cognitive
doing real or pretend writing? Development
Communication Skills and General
Knowledge
Currently, how often do you or another adult read As above Language and Cognitive
aloud to him/her or listen to him/her read or Development
attempt to read aloud? Communication Skills and
General Knowledge
How old was she/he when you started (to the near- Age in months Language and Cognitive
est month of age)? Development
Communication Skills and General
Knowledge
 Development and Psychometric Properties of the Early Development Instrument (EDI) 13

the questionnaire and its coverage to the community 1999). This gap has been shown to persist into later
of teachers and educators. years of school (Herbert & Stipek, 2005; Sheehan,
If the data are correlated and clustered, as they Cryan, Wiechel, & Bandy, 1991). Clearly, it is an
are in the present study, the factor analyses of important educational issue, and as such it is receiv-
between-classroom data and within-classroom data ing attention of practitioners (e.g., Spence, 2005). Age
could show different results. While the factor analy- also has an impact on the EDI scores: In four
ses methodology used accounted for the correlated domains, for a year increase in age, the scores
factors, it did not account for clustering. The four- increased on average by 0.5 points. In Language and
step Muthen procedure (Muthen, 1994) enables us Cognitive Development, the scores increased by
to detect differences between the two factor struc- almost two points between 4- and 5-year-olds.
tures, if there are any. The finding that there was Children with an ESL status had lower scores than
very little difference between the fit coefficients in children for whom English was the first language. As
the two models allows us to say that for each the school's instruction language is English, it is not
domain, the EDI factor structure between classrooms suprising that children with worse command of
is similar to the factor structure within classrooms. English have difficulties (Schwartz & Stiefel, 2006).
Because the clustering within classroom is an ESL learners routinely struggle with acquiring the
unavoidable natural phenomenon that is replicated competence in the language of instruction (Roessingh
when the EDI is used in the communities, it is & Kover, 2003). Combined with gender and age dif-
important to assess the differences between levels of ferences, these suggest that the composition of
analyses. Muthen's multilevel confirmatory factor kindergarten classes is an important factor to be con-
analyses methodology is suggested as the most ade- sidered in planning educational activities.
quate to test whether the structure of a construct
differs across levels of analyses (Dyer, Hanges, & Study 2
Hall, 2005). We chose this method of testing the Method
impact of clustering, rather than a random selection Participants. Teachers in 10 schools in two large
of a student per classroom, or averaging results per urban settings sent a letter describing the study to all
classroom, because 1) neither of the other two parents of Senior Kindergarten children (that is, chil-
would account for the variability occurring within dren who have their fifth birthday in the year of
students, 2) there are arguments in literature sug- entry to school). Of the 117 letters sent out, 100 were
gesting that factor-analysis of means can produce returned (85%) with parental agreement to partici-
misleading results (Dyer et al., 2005). pate. Unfortunately, due to circumstances beyond the
The consistency of teachers' ratings was explored control of the research team, only 85 of the 100 could
with the ICCs. The low ICCs indicated that the majori- be contacted. For 82 families, complete data were col-
ty of variance among the item and domain scores lected from both parent and teacher. Fifty-three chil-
was due to the variability of children within class- dren in seven schools attended kindergarten at
rooms rather than between classrooms. The high school (half-time), and a kindergarten-age program
average teacher reliability for each domain indicated at a child-care centre (half-time).
that despite the fact that one teacher contributed the
scores for children in the class, their ratings for indi- Measures. The EDI was completed by school teach-
vidual children were sufficiently different to warrant ers and parents for all 82 children, and by child-care
the claim that the data were reliable at the individual teachers for 53 children. Children's receptive vocabu-
level. lary was directly assessed with the Peabody Picture
The internal consistency of finalized scales was Vocabulary Test (Dunn & Dunn, 1981). The PPVT was
acceptable. Convergent validity, as shown by associa- administered to children within less than two weeks
tions of EDI scores with age and gender, was accept- from the teacher completing the EDI.
able, though it requires further investigation with Parents were interviewed to provide family back-
another sample to allow for inclusion of socioeco- ground information, including parent education and
nomic variables. marital status. They also answered additional ques-
The magnitude of differences between boys and tions about the child's health and behaviour. These
girls was especially large in the social and emotional were used to establish the external validity of the
domains, where 5-year-old boys (SK group) scored on teacher-completed EDI. The questions, answer
average lower than 4-year-old girls (JK group). This options, and coding, and their relevance to EDI
appears to be a consistent difference between boys domains are listed in Table 9.
and girls, also found in other populations (e.g., Zill,
 14 Janus and Offord
TABLE 10
Reliability and Validity Data on the EDI (Pearson Correlations)

Physical Health Social Emotional Language and Communication

& Well-Being Competence Maturity Cognitive and Ceneral
Development Knowledge
School-child care teacher .69 .80 .77 .72 .53
correlations {N = 53)
P .00 .00 .00 .00 .00
Parent-teacher correlations .36 .50 .36 .64 .41
(N = 82)
P .00 .00 .00 .00 .00
Correlations with PPVT language .05 .01 .13 .31 .47
assessment (N = 82)
P NS NS NS .00 .00

TABLE 11
Associations Between Teacher EDI Ratings on Social Competence and Emotional Maturity and Parent Ratings Related to Child
Socio-Emotional Competence (N = 82)

Thinking/ Frequency of Getting along Getting along Getting along Looking

Problem playing with with children with teacher with forward
Happiness solving children at school at school parents to school
Social Pearson r .36 .31 .39 .11 .48 .33 .31
competence P .00 .01 .00 .35 .00 .00 .01

Emotional Pearson r .25 .30 .27 .14 .42 .21 .24

maturity P .03 .01 .01 .20 .00 .06 .03

TABLE 12
Associations Between Teacher EDI Ratings on Language and Cognitive Development and Communication Skills and General
Knowledge and Parent Ratings Related to Child Cognitive Competence (N = 82)

Interest in Interest in Frequency of reading Age started being

books writing with child read to
Language and Pearson r .24 .22 .25 .20
cognitive development P .03 .05 .02 .08

Communication Pearson r .16 .21 .16 .26

and general P .15 .07 .14 .02
knowledge

Results all were statistically significant (Table 10). The lowest

Interrater reliability. In order to investigate the level agreement between parents and teachers occurred in
of agreement between two independent observers Physical Health and Well-Being and Emotional
completing the EDI, the EDI ratings were compared Maturity; the highest (0.64) in the domain of
between school kindergarten teachers and early Language and Cognitive Development.
childhood educators (ECE), and between school
teachers and parents. Concurrent test-criterion relationship. Correlations of
The correlations between teachers and ECE ranged the EDI language-related domains. Language and
from 0.53 to 0.8 (Table 10), and all were statistically Cognitive Development scale and the
significant. Correlations between parent and teacher Communication Skills, with PPVT scores were statisti-
ratings ranged from 0.36 to 0.64 (average of 0.45) and cally significant, though low to moderate (0.31 and
 Development and Psychometric Properties of the Early Development Instrument (EDI) 15

0.47, respectively, Table 10). These associations pro- agreement rates, although there is a fairly high rate of
vide the evidence for test-criterion validity for the agreement between parents (Grietens et al., 2004), It
two domains (Joint Committee on Standards for has been argued that respondents hold differing
Educational and Psychological Testing, 1999), that is, thresholds and standards (Grietens et al,, 2004),
that these two different measures, purportedly mea- resulting in low agreement. Low concordance could
suring the same concept, indeed do so. PPVT scores also be attributed to unique variance contributing to
were not correlated with the remaining three EDI the ratings (Dishion, French, & Patterson, 1995):
domains. Schools may elicit different behaviour patterns in
children than do home settings. Moreover, some
Association with parent interviews. Parent-rated behaviours, especially problem behaviours, have low
aspects of child health and behaviour (listed in Table frequency or low visibility (Campbell, 2002; Deng,
9) were correlated with teacher ratings on relevant Liu, & Roosa, 2004), which makes them hard to
EDI domains. notice reliably. All of these possibilities are likely
Of the four parent-based variables relevant to reflected in the interrater agreements on particular
Physical Health and Well-Being (Items 1-4 in Table 9), domains of the EDI, First, agreements between the
only the correlation of the parent rating of the child's two teachers are higher than between the teacher and
overall health was statistically significant (r = 0.34, p parent. This suggests that 1) children behave similar-
< 0.05). ly in educational settings, but differently at home, in
Six of the seven parent-based variables relevant to particular in terms of their emotional expressions,
Social Competence and Emotional Maturity domains and 2) school teachers and teachers in early child-
were statistically significantly correlated with teacher hood educational settings a have similar perspective
ratings on the EDI (Table 11), in assessing children's behaviour. This second find-
Teacher ratings of the child in Language and ing is especially important since the results of the EDI
Communication domains of the EDI were significant- are frequently aggregated across different teachers
ly correlated with three out of four parent-based and this basic trust in teachers' reliability is crucial.
items - interest in books, writing, and frequency of Also, these similarities indicate that the concepts cap-
reading with adult - while the Communications tured by the EDI are clear and easily assessed by
Skills score was significantly correlated with one out trained educators. Second, low parent-teacher agree-
of four - the age at which the child was first being ment (r = 0,36) in the Emotional Maturity domain
read to (Table 12). All correlations, however, were in may well reflect the low - and variable - frequency of
the expected direction. problem behaviours, especially internalizing ones
It is also important to note that there were only like anxiety (Cambell, 2002), which are part of that
three statistically significant correlations between a domain. Similar results were found in research on the
parent-based variable and EDI domain not directly reliability of the Strength and Difficulties
relevant to this variable. These were: Language and Questionnaire (SDQ; Goodman, 2001), Two scales of
Cognitive Development with frequency of seeing the SDQ, Emotional Symptoms and Prosocial
other children, r = 0.26, p = 0.002, and liking school, r Behaviour, which are conceptually close to the
= 0.30, p = 0.009, and Communication Skills and Emotional Maturity domain of the EDI, had the low-
General Knowledge with the ability to think and est parent-teacher agreement rates (0,27 and 0,25,
solve problems, r = 0.33, p = 0,003, None of the par- respectively) in a community sample of over 7,000
ent-based items not relevant to either Physical Health children.
and Well-Being, or Social Competence, or Emotional A fairly high parent-teacher agreement was
Maturity were correlated with teacher ratings in achieved for the Language and Cognitive
these domains. Development domain, which includes letter knowl-
edge, number knowledge, memory, and basic read-
Discussion ing and writing skills. This agreement is higher than
Interrater agreements on the EDI domains were expected based on the evidence from largely behav-
moderate to high for the two teacher ratings, and low iour-based scales (see above). However, parent rat-
to moderate for parent-teacher ratings. Agreement ings have not been commonly used to evaluate chil-
between multiple respondents on children's behav- dren's cognitive ability, in particular for school-age
iour is notoriously low (e.g., Boyle et al., 1996; children. There is some evidence that parents tend to
Verhulst & Akkerhuis, 1989; Culp, Howell, Culp, & overestimate their children's development (Deimann,
Blankemeyer, 2001; Winsler & Wallace, 2002). In par- 2005; Glascoe & Sandier, 1995). Maternal predictions
ticular, teachers and parents appear to have low of their 4-year-olds' performance on 96 test items
 16 Janus and Offord

were highly correlated with the children's actual per- ble that the lack of power was caused by associating
formance, yet the "errors" in judgment were mostly a single aspect of a spectrum with an EDI domain
overestimates (Hunt & Paraskevopoulos, 1980), A combining many components. Moreover, as argued
study of kindergarten-age children with develop- by Dishion et al, (1995), school context influences
mental disabilities demonstrated that parent and child behaviours, which may differ from those
teacher ratings of children's language development observed at home. Parents rarely have a chance to
were positively and significantly correlated (Sigafoos observe their child in an environment with 20 peers,
& Pennell, 1995), in particular in the area of expres- rather than just one or two, and therefore their per-
sive language. Moreover, maternal education con- ception may not be the same as that of the teacher.
tributes to the accuracy in assessment of their chil-
dren's abilities (Hunt & Paraskevopoulos, 1980), General Discussion
Almost 80% of mothers in our study were well edu- The Early Development Instrument was designed
cated, which most likely contributed to their knowl- to fill the gap in the population-level measurement of
edge about their children's cognitive abilities. children's school readiness with a tool that is feasible
The language-related EDI domains were signifi- and quick to complete, informative, and psychomet-
cantly associated with directly tested children's rically adequate, while at the same time lending itself
receptive vocabulary. Since receptive vocabulary is a well to aggregation for social reporting. The analyses
part of the larger assessment of children's IQ and cor- in this paper suggest that the EDI's psychometric
relates well with composite IQ measures (Dunn & properties are acceptable and comparable with other
Dunn, 1981), PPVT scores are often taken as a proxy instruments measuring children's behaviour (e,g.,
of a child's intelligence. Significant correlations with CBCL; Achenbach, 1991) and academic skills (e,g,,
the appropriate EDI domains indicate good criterion PKRT; Duncan & Rafter, 2005),
validity on these domains. Nevertheless, further evi- Internal consistency of the EDI scales ranged from
dence is needed to ascertain that other areas of child 0,84 to 0,96; the 14-factor solution replicated the
cognitive development (number concepts, problem- domains of school readiness suggested in literature
solving, expression, memory) are accurately reflected (Phillips & Love, 1995), and accounted for 63% of
in EDI scores. At the same time, the lack of correla- variance. The interrater reliability correlations were
tions between PPVT and the three remaining EDI moderate (0,53) to high (0.80), While not reported
domains clearly provides evidence of the discrimi- here (Duku & Janus, 2004), the test-retest correlations
nant validity of the EDI domains. were also high (0.82-0,94). Validity investigations
The patterns of correlations between parent-based encompassed several analyses. Parent-teacher agree-
variables and relevant teacher-reported EDI domains ments on the EDI were moderate (0,36-0,64),
further indicate that the domains discriminate among Concurrent test-criterion validity of the EDI, as
the aspects of school readiness. Although the magni- explored in comparisons with direct language test
tude of correlations was, at best, moderate (0,24-0,48), and parent interview about children's behaviour
they were all in the expected direction. Unlike the demonstrated low to moderate, yet consistent, rela-
correlations between parent ratings on the EDI and tionships.
teacher ratings on the EDI, where the same questions The age and gender difference patterns demon-
were asked of different observers, these parent vari- strated in other large samples of kindergarten chil-
ables are based on interview questions in general dren were also replicated by the EDI results, Zill
areas relevant to the specific EDI domains. The corre- (1999) found that boys and children with birthdays
lations suggest that there is a certain small (0,06-0,23) late in the year were more likely to have problems in
amount of shared variance among the variables. kindergarten; male gender, and younger age at
Because most parent interview variables were based school entry significantly contributed to "school
on much narrower concepts than the EDI domains, unreadiness" in Farkas and Hibel's (2005) analysis of
the low-level associations are not surprising, the ECLS-K data in the U,S, Interestingly, among
A case in the above point is provided by parents' kindergarten children in the sample analyzed by
judgment of their children getting along with school Farkas and Hibel, boys were significantly older at
friends, which was the only explored aspect of the entry than girls, a finding interpreted by the authors
socio-emotional skills not significantly correlated as a possible "strategizing" effort by parents. In juris-
with teacher EDI (rs = 0,11 and 0,14), Social dictions where rules about the age of school entry are
Competence and Emotional Maturity domains each less uniformly observed, the EDI scores need to be
cover a spectrum of related concepts, not only the grouped by actual age intervals rather than by "grade
"getting along" or "prosocial" behaviours. It is possi- level." This procedure is currently being used in
 Development and Psychometric Properties of the Early Development Instrument (EDI) 17

Australia (Goldfeld et al., 2006), The EDI scores were Fantuzzo, 2004; Goodman, 2001; Lutz, Fantuzzo, &
consistently lower for children with the ESL status. McDermott, 2002). Together with a moderate predic-
Lack of proficiency in the language of instruction fre- tive validity of the EDI from kindergarten to third
quently contributes to children's lower achievement grade (Gaskin, Duku, & Janus, 2005), also compara-
in school (Fontaine, Torre, & Grafwallner, 2006), ble with other measures (LaParo & Pianta, 2000),
There is evidence, however, that foreign-born ESL these properties suggest that the EDI could be a use-
children have better academic achievement than ful addition to the spectrum of measures available to
native-born children (Schwartz & Stiefel, 2006), students of children's behaviour and school adjust-
Moreover, as Bialystok points out, the quality of ment in the preschool and early school years.
home environment and its promotion of reading and The major advantage of the EDI is its combination
learning will have an impact on the school achieve- of several domains of child development into one
ment of children with the ESL status (Bialystok, 2001), comprehensive instrument, which sets it apart from
Unfortunately, in our study we were not able to con- the other available measures of school readiness.
trol for either of these factors, and therefore this issue Questions are based on behaviours and skills easily
has to be explored further. observable in a school setting, and responses are
Associations with various other measures were rated based on observed frequency of behaviours or
usually only statistically significant where there was presence of skills, rather than on the child's perfor-
a strong theoretical basis for them to be so. For exam- mance in relation to a specific group (e.g., "top half of
ple, direct language tests were not significantly corre- the class"). These properties make teachers experts in
lated with the noncognitive EDI domains; parent rat- providing the information on children without the
ings of child getting along with friends at school was necessity of additional training. On the other hand,
not significantly correlated to social and emotional teacher ratings could be subject to individual bias,
competence rated by the teacher on the EDI, and the due to characteristics of teacher, child, school, or
cognitively oriented EDI domains were not, as a rule, interactions of all three (Pianta, Steinberg, & Rollins,
correlated significantly with parent ratings of chil- 1995). Although it is impossible to fully address the
dren's social competence. These findings emphasize question of teacher bias with the data from studies
the discriminatory character of the instrument, and reported in this paper, two findings raise the confi-
underline our view of reporting on each domain sep- dence in teachers' fairly uniform standards of
arately, rather than producing a composite total answers: interrater reliabilities, with both other
score, which could obscure real differences. teachers and parents, and the high teacher consisten-
Unlike the many existing assessments of school cies. Elsewhere, teacher ratings of overall summary
readiness, the EDI has not been validated for screen- skills were reported to have only moderate associa-
ing at an individual or diagnostic level. In contrast to tion with later outcomes (Mashburn & Henry, 2004;
an instrument like the Child Behaviour Checklist Meisels, Bickel, Nicholson, Xue, & Atkins-Burnett,
(CBCL; Achenbach, 1991), for example, which has set 2001). Nevertheless, a recent study suggests that the
thresholds indicating clinical diagnoses, an EDI score population context should be taken into account in
in a certain range cannot be taken as indicative of a assessing the appropriateness of the methodology
clinical problem. However, even the CBCL author used: Teacher ratings, while not specific enough to
warns of equating the CBCL scores with particular warrant early identification, are valid enough to sug-
disorders, and instead recommends integrating the gest intervention models (Crooks & Peters, 2005), The
CBCL "descriptions of the child" with other types of teacher measures used in the cited studies all con-
data on the child and family in order to arrive at a tained less than 20 items of varying generality. It
diagnosis (Achenbach, 1991). Measurement experts appears that although the EDI is longer, it may be a
suggest that a test used for decision-making at an compromise between multiple, costly, standardized
individual level needs to be more reliable than one assessments and brief rating scales, as it provides
used for group-level analysis and research (Streiner anchored teacher ratings of detailed competencies.
& Norman, 1995). Establishing a diagnostic use for Several limitations have to be noted here. One of
the EDI would considerably increase its costs, and the studies had a small and moderately variable sam-
thus the availability for population-level use. With ple. In particular, very few of the children were non-
the exception of clinical identification, the EDI psy- English speakers. Although parent country of birth
chometric properties described here are at similar does not have a significant impact on children EDI
levels as those of other teacher questionnaires used scores (Janus & Duku, 2006; Janus, Offord, & Walsh,
for assessment of behaviour of preschool and early 2001), children for whom language of instruction is a
school-age children (e.g., Bulotsky-Shearer & second language face obvious disadvantages enter-
 18 Janus and Offord

ing the school system. Second, only limited data on identification of inequalities related to characteristics
families' socioeconomic status were available. that may be remediable by appropriate policies
Unfortunately, in this respect the validation of the (Starfield, 2002). The EDI is the first tool for which
EDI in this study is very similar to validation of other community aggregation is feasible due to its low cost,
instruments, which often have been criticized for relevance in covering the five developmental
small samples. These limitations are addressed in the domains, and proven psychometric properties.
next study (Janus & Duku, 2006). Moreover, as the
EDI is currently used in many communities, local The authors would like to acknowledge The Founder's
researchers are encouraged to include some valida- Network, Invest in Kids Foundation, and Human
tion components in their projects. Resources Development Canada for their financial sup-
Throughout the process of the EDI development port. We would like to thank teachers and parents in
the engagement of representatives of the communi- Toronto and Calgary for volunteering their time, and Eric
ties of stakeholders - teachers, early childhood edu- Duku for his invaluable statistical expertise.
cators, child health professionals, and parents - was Correspondence concerning this article should be
highly valued. Their engagement helped to achieve addressed to Dr. Magdalena Janus, Offord Centre for
an instrument relevant to the community. On the Child Studies, McMaster University, 1200 Main St. W.,
other hand, the emphasis that we put on ensuring Patterson Bldg. Rm. 215, Hamilton, Ontario, Canada
this participation may have somewhat impeded the L8N 3Z5.
achievement of psychometric elegance, for example,
in keeping some items due to their highly perceived References
importance by teachers despite their low factor load- Achenbach, T. M. (1991). Manual for the Child Behavior
ing, or having an inconsistent number of answer Checklist/4-18 and 1991 Profile. Burlington, VT:
options in domains. Although it may be questioned, University of Vermont.
we believed that this approach was crucial for com- Alexander, K. L., & Entwisle, D. R. (1988). Achievement
munities' perceived ownership of the measure. in the first 2 years of school: Patterns and processes.
Developing the EDI with the stakeholders' input Monographs ofthe Society for Research in Child
appears to have been a sound decision, as interest in Development, 53(2), 1-157.
measurement and improvement of children's school Augustyniak, K. M., Cook-Cottone, C. P., & Calabrese, N.
readiness in many communities in Canada has con- (2004). The predictive validity of the Phelps
tinued to be widespread. Kindergarten Readiness Scale. Psychology in the
In conclusion, while clearly some of the analyses Schools, 41, 509-516.
have to be repeated with larger or different samples, Bialystok, E. (2001). Bilingualism in development: Language,
the EDI's psychometric properties have proven to be literacy, and cognition. New York: Cambridge
acceptable. The EDI's simplicity, ease of use, and low University Press.
cost all lend themselves easily to community-wide Boyle, M. H., & Lipman, E. L. (2002). Do places matter?
implementation. Data collected for whole populations Socioeconomic disadvantage and behavioral problems
of children have the advantage of giving the commu- of children in Canada. Journal of Consulting and Clinical
nity the true picture and, especially in conjunction Psychology, 70(2), 378-389.
with other locally relevant data, allow for making Boyle, M. H., Offord, D. R., Racine,Y .A., Szatmari, P.,
useful recommendations (Janus et al., 2002) and pro- Sanford, M., & Fleming, J. E. (1996). Interviews versus
vide a baseline for future assessments of progress. It checklists: Adequacy for classifying childhood psychi-
is, however, important to point out that while the EDI atric disorder based on adolescent reports.
already has a demonstrated ability to distinguish International Journal of Methods in Psychiatric Research,
among communities Qanus et al., 2002), and is related 6, 309-319.
to neighbourhood collective efficacy (Seward & Brigance, A. (1992). Brigance Diagnostic Inventory of Early
Loomis, 2005), its sensitivity to changes in communi- Development-Revised. North Billerica, MA: Curriculum
ties due to reform still remains to be investigated. Associates.
The only types of individual measurements that Brooks-Gunn, G. J., Cuo, G., & Furstenberg, F. F. (1993).
have been commonly aggregated to represent popu- Who drops out and who continues beyond high
lation-level results are school-based achievement school? A twenty-year follow-up of black urban youth.
tests (e.g., Willms & Jacobsen, 1990). This is usually Journal of Adolescence, 3, 271-294.
done for later grades, and rarely, if ever, to provide Bulotsky-Shearer, R., & Fantuzzo, J. (2004). Adjustment
communities with a useful overview for implement- scales for preschool intervention: Extending validity
ing action. Yet, the population-level analyses allow and relevance across multiple perspectives. Psychology
 Development and Psychometric Properties of the Early Development Instrument (EDI) 19

in the Schools, 4(7), 725-735. Duncan, J., & Rafter, E. M. (2005). Concurrent and predic-
Butaine, R., & Costenbader,, V. (1997). The effectiveness of tive validity of the Phelps Kindergarten Readiness
a transitional prekindergarten program on later acade- Scale-II. Psychology in the Schools, 42, 355-359.
mic achievement. Psychology in the Schools, 34(1), 41-50. Dunn, L. M., & Dunn, L. M. (1981). Peabody Picture
Campbell, S. B. (2002). Behavior problems in preschool chil- Vocabulary Test-Revised. Manual for Forms L and M.
dren: Clinical and developmental issues. New York: Circle Pines, MN: American Guidance Service.
Guilford Press. Dyer, N. G., Hanges, P .J., & Hall, R. J. (2005). Applying
Chase-Lansdale, P. L., Gordon, R. A., Brooks-Gunn, J., & multilevel confirmatory factor analysis techniques to
Klebanov, P K. (1997). Neighborhood and family influ- the study of leadership. The Leadership Quarterly, 16,
ences on the intellectual and behavioral competence of 149-167.
presechool and early school-age children. In G. J. Entwisle, D. R., & Alexander, K. L. (1999). Early schooling
Brooks-Gunn & J. L. Duncan (Eds.), Neighborhood and social stratification. In R. C. Pianta & M. J. Cox
poverty (pp. 79-118). New York: Russell Sage (Eds.), The transition to kindergarten (pp. 13-39).
Foundation. Baltimore, MD: Paul H. Brookes.
Chew, A. L., & Lang, W. S. (1990). Predicting academic Farkas, G., & Hibel, J. (2005, October). Being unready for
achievement in kindergarten and first grade from school: Factors affecting risk and resilience. Paper present-
prekindergarten scores on the Lollipop Test and Dial. ed at the National Symposium on Family Issues,
Educational and Psychological Measurement, 50, 431-437. Pennsylvania State University.
Chew, A. L., & Morris, J. D. (1989). Predicting later acade- Fontaine, N. S., Torre, D. L., & Grafwallner, R. (2006).
mic achievement from kindergarten scores on the Effects of quality early care on school readiness skills
Metropolitan Readiness Tests and the Lollipop Test. of children at risk. Early Child Development and Care,
Educational and Psychological Measurement, 49, 461-465. 176(1), 99-109.
Costenbader, V., Rohrer, A. M., & DiFonzo, N. (2000). Gaskin, A., Janus, M., & Duku, E. (2005, April).
Kindergarten screening: A survey of current practice. Concurrent and predictive validity of the Early
Psychology in the Schools, 37(4), 323-332. Development Instrument (EDI). Poster presentation,
Crooks, C. V., & Peters, R. D. (2005). Predicting academic 17th Annual Research Day, Department of Psychiatry
difficulties: Does a complex, multidimensional model & Behavioural Neurosciences, McMaster University,
outperform a unidimensional teacher rating scale? Hamilton, ON.
Canadian Journal of Behavioural Science, 37(3), 170-180. Glascoe, F. P., & Sandier, H. (1995). Value of parents' esti-
Culp, R. E., Howell, C. S., Culp, A. M., & Blankemeyer, mates of children's developmental ages. The Journal of
M. (2001). Maltreated children's emotional and behav- Pediatrics, 127, 831-835.
ioral problems: Do teachers and parents see the same Goldfeld, S., Brinkman, S., Sayers, M., Liddelow, T.,
things? Journal of Child and Family Studies, 10(1), 39-50. Daws, A., & Coutts, M., et al. (2006, April). A national
Deimann, P. (2005). Mothers' evaluation of their kinder- approach to measuring child development in Australia:
garten children's developmental status. Zeitschrift fur Process challenges and policy implications. Paper present-
Entwicklungspsychologie und Padagogische Psychologie, ed at the Measuring Early Child Development
37,122-134. Conference, Vaudreuil, QC.
Deng, S., Liu, X., & Roosa, M. W. (2004). Agreement Goodman, R. (2001). Psychometric properties of the
between parent and teacher reports on behavioral strengths and difficulties questionnaire. Journal of
problems among Chinese children. Developmental and American Academy of Child and Adolescent Psychiatry,
Behavioral Pediatrics, 25(6), 407-413. 40(11), 1337-1345.
Dishion, T. J., French, D. C, & Patterson, G. R. (1995). The Gorsuch, R. L. (1983). Factor analysis. Hillsdale, NJ:
development and ecology of antisocial behavior. In D. Erlbaum.
Cicchetti & D. J. Cohen (Eds.), Developmental psy- Graue, M. E., & Shepard, L. A. (1989). Predictive validity
chopathology (pp. 421-471). New York: Wiley. of the Gesell School Readiness Tests. Early Childhood
Doherty, G. (1997). Zero to six: The basis for school readiness. Research Quarterly, 4, 303-315.
Ottawa, ON: Human Resources Development Canada. Grietens, H., Onghena, P., Prinzie, P., Gadeyne, E.,
R-97-8E. VanAssche, V., Ghesquiere, P., et al. (2004).
Duku, E., & Janus, M. (2004, April). Stability and reliability Comparison of mothers', fathers', and teachers'
of the early development instrument: A population-based reports on problem behavior in 5- to 6-year-old chil-
measure for communities (EDI). Poster presentation, 16th dren. Journal of Psychopathology and Behavioral
Annual Research Day, Department of Psychiatry and Assessment, 26(2), 137-145.
Behavioural Neurosciences, McMaster University, Herbert, J., & Stipek, D. (2005). The emergence of gender
Hamilton, ON. differences in children's perceptions of their academic
 20 Janus and Offord

competence. Journal of Applied Developmental Convention Paper, Ottawa, ON: Liberal Party of
Psychology, 26(3), 276-295. Canada.
Hunt, J. M., & Paraskevopoulos, J. (1980). Children's psy- Lichtenstein, R. (1990). Psychometric characteristics and
chological development as a function of the inaccura- appropriate use of the Gesell School Readiness
cy of their mothers' knowledge of their abilities. Screening Test. Early Childhood Research Quarterly, 5,
Journal of Genetic Psychology, 136(2), 285-298. 359-378.
Ilg, R, Ames, L., Haines, J., & Gillespie, C. (1978). Gesell Love, J. M., Aber, J. L., & Brooks-Gunn, J. (1994). Strategies
School Readiness Test. New York: Harper & Row. for assessing community progress toward achieving the first
Jacob, S., Snider, K., & Wilson, J. (1988). Validity of the national educational goal. Princeton, NJ: Mathematica
DIAL-R for identifying children with special educa- Policy Research, Inc.
tion needs and predicting early reading achievement. Lutz, M. N., Fantuzzo, J., & McDermott, P (2002).
Journal of Psychoeducational Assessment, 6, 289-297. Multidimensional assessment of emotional and behav-
Janus, M., & Duku, E. (2006). The school entry gap: ioral edjustment problems of low-income preschool
Socioeconomic, family, and health factors associated with children: Development and initial validation. Early
children's school readiness to learn. Unpublished manu- Childhood Research Quarterly, 17(3), 338-355.
script. Mardell-Czudnowski, C, & Goldberg, D. S. (1998).
Janus, M., & Offord, D. (2000). Readiness to learn at Developmental indicators for assessment of learning
school. ISUMA, 1(2), 71-75. (DIAL-3) (3rd ed.). Circle Pines, MN: American
Janus, M., Offord, D., & Walsh, C. L. (2001, April) Guidance Service.
Population-level assessment of readiness to learn at school Mashburn, A. J., & Henry, G. T. (2004). Assessing school
in 5-year-olds: Relationship to child and parent measures. readiness: Validity and bias in preschool and kinder-
Proceedings of the 2001 Biennial Meeting, Society for garten teachers' ratings. Educational Measurement:
Research in Child Development, Minneapolis, MN. Issues and Practice, 23(4), 16-30.
Janus, M., Walsh, C, Viveiros, H., & Offord, D. R. (2002, May, D. C, & Kundert, D. K. (1992). Kindergarten screen-
June). Community, neighbourhood and 5-year-olds' readi- ing in New York State: Tests, purposes and recommen-
ness to learn at school. Poster session presented at the dations. Psychology in the Schools, 29, 35-41.
Head Start 6th National Conference, Washington, DC. McCain, M. N., & Mustard, J. F. (1999). Report on the early
Jimerson, S., Egeland, B., & Teo, A. (1999). A longitudinal years study. Toronto, ON: Province of Ontario, Ontario
study of achievement trajectories: Factors associated Children's Secretariat. ISBN 0-7778-8583-2.
with change. Journal of Educational Psychology, 91(1), Meisels, S. J. (1987). Uses and abuses of developmental
116-126. screening and school readiness testing. Young Children,
Joint Committee on Standards for Educational and 42(4), 6-68.
Psychological Testing. (1999). Standards for educational Meisels, S. J. (1988). Assessing readiness. Ann Arbor, Ml:
and psychological testing. Washington, DC: American Center for the Improvement of Early Reading
Educational Research Association. Achievement. Ciera Report #3-002.
Kagan, S. L. (1992). Readiness past, present, and future: Meisels, S. J. (1999). Assessing readiness. In R. C. Pianta &
Shaping the agenda. Young Children, 48(1), 48-53. M. J. Cox (Eds.), The transition to kindergarten (pp. 39-
Kagan, S. L., & Neuman, M. J. (1997). Defining and imple- 66). Baltimore, MD: Paul H. Brookes Publishing Co.
menting school readiness: Challenges for families, Meisels, S. J., Bickel, D. D., Nicholson, J., Xue, Y, &
early care and education, and schools. In R. P. Atkins-Burnett, S. (2001). Trusting teachers' judg-
Weissberg, T. P. Gullotta, R. L. Hampton, B. A. Ryan, & ments: A validity study of a curriculum-embedded
G. R. Adams (Eds.), Establishing preventive services performance assessment in kindergarten to grade 3.
(pp. 61-96). Thousand Oaks, CA: Sage Publications. American Educational Research Journal, 38(1), 73-95.
Kohen, D. E., Brooks-Gunn, J., Leventhal, T, & Hertzman, Morongiello, B. (1997). Tapping school readiness in the
C. (2002). Neighborhood income and physical and NLSCY: Measurement issues and solutions. Ottawa, ON:
social disorder in Canada: Associations with young Applied Research Branch, Strategic Policy, Human
children's competencies. Child Development, 73(6), Resources Development Canada. T-98-1E.
1844-1860. Muthen, B. O. (1994). Multilevel covariance structure
La Paro, K. M., & Pianta, R. C. (2000). Predicting chil- analysis. Sociological Methods and Research, 22, 376-398.
dren's competence in the early school years: A meta- Muthen, B. O., & Muthen, L. K. (2004). Mplus version 3.0.
analytic review. Review of Educational Research, 70, 443- Los Angeles, CA: Muthen & Muthen.
484. NLSC Project Team. (1995). National longitudinal survey of
Liberal Party of Canada. (2005, March) Supporting children: Overview of survey instruments for 1994-95 data
Canadian Families, Liberal Party of Canada Biennial collection cycle 1 (pp. 95-102). Ottawa, ON: Statistics
 Development and Psychometric Properties of the Early Development Instrument (EDI) 21

Canada Human Resources Development Canada. Research findings for early childhood educators.
Offord, D. R., Kraemer, H. C, Kazdin, A. E., Jensen, R S., Journal of Research in Childhood Education, 6(1), 66-74.
& Harrington, R. (1998). Lowering the burden of suf- Shore, R. (1997). Rethinking the brain: New insights into
fering from child psychiatric disorder: Trade-offs early development. New York: Families and Work
among clinical, targeted, and universal interventions. Institute.
Journal of American Academy of Child and Adolescent Sigafoos, J., & Pennell, D. (1995). Parent and teacher
Psychiatry, 37, 686-94. assessment of receptive and expressive language in
Offord, D. R., Kraemer, H. C, Kazdin, A. E., Jensen, R S., preschool children with developmental disabilities.
Harrington, R., & Gardner, J. S. (1999). Lowering the Education and Training in Mental Retardation and
burden of suffering: Monitoring the benefits of clini- Developmental Disabilities, 30(4), 329-335.
cal, targeted, and universal approaches. In D. Keating Spence, C. M. (2005). Creating a literacy environment for
& C. Hertzman (Eds.), Developmental health and the boys: Ideas for teachers, administrators, and parents.
wealth of nations, (pp. 293-310). New York: The Toronto, ON: Thomson Nelson.
Guilford Press. Starfield, B. (2002). Forward to: Improving children's
Offord, D. R., & Lipman, E. L. (1996). Emotional and health: How population-based research can inform
behavioural problems. In Anonymous, Growing up in policy - The Manitoba Experience. Canadian Journal of
Canada: National longitudinal survey of children and youth Public Health, 93, S7.
(pp. 119-126). Ottawa, ON: Human Resources Statistics Canada (2005). Canada at a glance 2005.
Development Canada. Catalogue number [12-581-XPE]. Ottawa, ON:
Phillips, D., & Love, J. (1995, May). Indicators for school Statistics Canada, Communications Division.
readiness, schooling, and child care in early to middle child-
Streiner, D. L., & Norman, G. R. (1995). Health measure-
hood. Washington, DC: Institute for Research on ment scales: A practical guide to their development and use
Poverty. Special Report No. 60b. (2nd ed.). New York: Oxford LJniversity Press.
Pianta, R. C, & McCoy, S. J. (1997). The first day of Swanson, B., Payne, D., & Jackson, B. (1981). A predictive
school: The predictive validity of early school screen- validity study of the Metropolitan Readiness Tests and
ing. Journal of Applied Developmental Psychology, 18,1- Meeting Street School Test against first grade
22. Metropolitan Achievement Test scores. Educational and
Pianta, R. C, Steinberg, M. S., & Rollins, K. B. (1995). The Psychological Measurement, 41, 575-578.
first two years of school: Teacher-child relationships Verhulst, F C, & Akkerhuis, G. W. (1989). Agreement
and deflections in children's classroom adjustment. between parents' and teachers' ratings of
Development and Psychopathology, 7, 295-312. behavioral/emotional problems of children aged 4-12.
Rimm-Kaufman, S. E., Pianta, R. C, & Cox, M. J. (2000). Journal of Child Psychology & Psychiatry & Allied
Teachers' judgments of problems in the transition to Disciplines, 30(1), 123-136.
kindergarten. Early Childhood Research Quarterly, 15, Wenner, G. (1995). Kindergarten screens as tools for the
147-165. early identification of children at risk for remediation
Roessingh, H., Kover, P. (2003). Variability of learners' or grade retention. Psychology in the Schools, 32, 249-
acquisition of cognitive academic language proficien- 254.
cy: What can we learn from achievement measures? Willms, J. D., & Jacobsen, S. (1990). Growth in mathemat-
TESL Canada Journal, 21(1), 1-21. ics skills during the intermediate years: Sex differences
Rose, G. (1994). The strategy of preventive medicine. Oxford: and school effects. InternationalJournal of Educational
Oxford University Press. Research, 14,157-174.
Schonkoff, J. P., & Phillips, D. A. (2000). From neurons to Winsler, A., & Wallace, G. L. (2002). Behavior problems
neighborhoods. Washington, DC: National Academy and social skills in preschool children: Parent-teacher
Press. agreement and relations with classroom observations.
Schwartz, A. E., & Stiefel, L. (2006). Is there a nativity Early Education and Development, 13(1), 41-58.
gap? New evidence on the academic performance of Woodcock, R. W., McGrew, K. S., & Mather, N. (2001).
immigrant students. Education Finance and Policy, 1(1), Woodcock-Johnsonlll Tests of Achievement. Itasca, IL:
17-49. Riverside.
Seward, C. L., & Loomis, C. (2005, April) Rural neighbour- Zill, N. (1999). Promoting educational equity and excel-
hood collective efficacy influences children's readiness to lence in kindergarten. In R. C. Pianta and M. J. Gox
learn. Poster session. Biennial Meeting, Society for (Eds.), The transition to kindergarten. Baltimore, MD:
Research in Child Development Meeting, Atlanta, GA. Paul H. Brookes.
Sheehan, R., Cryan, J. R., Wiechel, J., & Bandy I. G. (1991). Zill, N., Resnick, G., Kim, K., Hubbel McKey, R., Clark,
Factors contributing to success in elementary schools: C, Pai-Samant, S., et al. (2001). Head Start FACES:
 22 Janus and Offord

Longitudinal finding on program performance, third Received June 2, 2005

progress report. Washington, DC: Research, Revised February 6, 2006
Demonstration, and Evaluation Branch & Head Start Accepted July 31, 2006
Bureau, Administration on Children, Youth and
Families, U.S. Department of Health and Human
Service.