Early Childhood Research Quarterly 28 (2013) 936–946
Contents lists available at ScienceDirect
Early Childhood Research Quarterly
Examining the factor structure of the Family Child Care Environment Rating
Scale—Revised
Diana Schaack a,∗ , Vi-Nhuan Le b , Claude Messan Setodji b
a
Center for the Study of Child Care Employment (CSCCE), Institute for Research on Labor and Employment (ILRE), University of California at Berkeley, 2521 Channing Way #5555,
Berkeley, CA 94720-5555, United States
b
RAND Corporation, 1776 Main Street, Santa Monica, CA 90401, United States
a r t i c l e
i n f o
Keywords:
FCCERS-R
Factor analysis
Measurement properties
Family child care
Quality rating and improvement system
a b s t r a c t
The purpose of this study was to examine the factor structure of the Family Child Care Environment
Rating Scale—Revised (FCCERS-R) in high-stakes contexts. The results of an exploratory factor analysis
revealed three dimensions of quality on the FCCERS-R: (1) Activities/Materials, (2) Language/Interaction,
and (3) Organization. This study also explored whether abridged versions of the FCCERS-R could serve as
a proxy for the full instrument. In addition to subsets of FCCERS-R items created from the factor structure,
purposively and randomly chosen item subsets were created. The purposively chosen subsets included
6-, 9-, and 12-item scales comprised of the items with the highest factor loading across the three factors,
whereas the randomly chosen subsets consisted of 12 items. Results of a discriminant analysis showed
that the factor subsets were poorer proxies for the total FCCERS-R score than were the other subsets,
which demonstrated comparable internal consistencies and discriminant power as the full FCCERS-R
when classifying homes into general quality categories. Implications for adopting shorter versions of the
FCCERS-R are discussed.
© 2013 Elsevier Inc. All rights reserved.
As social policies have shifted over the past several decades in
the United States, increasing numbers of young children are being
cared for in licensed family child care homes. It is estimated that
nearly a quarter of all children spend time in family child care by
the time they reach kindergarten (Johnson, 2005). In addition, a
number of studies have found associations between the quality of
care that children experience in their family child care home and
their social, cognitive, and language skills (Clarke-Stewart, Vandell,
Burchinal, O’Brian, & McCartney, 2002; Kontos, Howes, Shinn, &
Galinsky, 1995; Loeb, Fuller, Kagan, Carrol, & Carroll, 2004; NICHD
ECCRN & Duncan, 2003). Yet many young children are in family
child care settings that are considered minimal or even poor quality
(Kontos et al., 1995; Layzer & Goodson, 2006; Whitebook & Sakai,
2004), and lower-income children are most frequently enrolled in
the lowest-quality family child care homes (Kryzer, Kovan, Phillips,
Donagall, & Gunnar, 2007; Layzer & Goodson, 2006; Raikes, Raikes,
& Wilcox, 2005). These are also the children who are most likely
to benefit from enriching and nurturing early care and education
experiences (Peisner-Feinberg et al., 2001).
∗ Corresponding author. Tel.: +1 510 643 8293.
E-mail addresses: diana.schaack@berkeley.edu (D. Schaack),
vinhuan@gmail.com (V.-N. Le).
0885-2006/$ – see front matter © 2013 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.ecresq.2013.01.002
For these reasons, states and local communities are attempting
to improve the quality of family child care (Paulsell, Porter, & Kirby,
2010). Many state and local improvement initiatives are hinged on
the assessment of a family child care home’s quality, and administer
the Family Child Care Environment Rating Scale—Revised (FCCERSR; Harms, Cryer, & Clifford, 2007) as part of a battery of assessments
to evaluate children’s experiences in these settings. The results of
the FCCERS-R, along with other indices of family child care quality, are then used to identify family child care homes in need of
improvement. Currently 25 states include the FCCERS-R in their
quality rating and improvement system (QRIS), where the FCCERSR scores are used to inform quality improvement activities. Under
these systems, the FCCERS-R scores may have financial implications
for family child care homes, as these scores can be used to award
different levels of funding to family child care homes, or be made
public to encourage families to choose higher-scoring programs
(Schaack, Tarrant, Boller, & Tout, 2012).
As the use of the FCCERS-R has evolved from the low-stakes
administration in which it was validated (e.g., research purposes
or as a quality improvement tool for providers and consultants)
to one in which high-stakes decisions are increasingly being
placed on it, there is greater need to establish that the FCCERS-R
demonstrates sound psychometric properties (Tout, Zaslow, Halle,
& Forry, 2009). The Standards for Educational and Psychological
D. Schaack et al. / Early Childhood Research Quarterly 28 (2013) 936–946
Testing also recommend that measures should be administered
under the settings and for the purposes for which they have been
validated (American Educational Research Association (AERA),
American Psychological Association (APA), & National Council on
Measurement in Education (NCME), 1999). Thus, research is needed
to understand how the FCCERS-R functions in applied, high-stakes
contexts.
Research on the consequences of high-stakes testing from the
elementary and secondary grades has shown that when high-stakes
decisions are attached to test scores, teachers can respond in ways
that undermine the validity of inferences of the test scores (Smith
& Fey, 2000). For example, when test scores are included as part of
state accountability systems, teachers have been shown to focus
only on those constructs or skills that can be easily improved
(Koretz, McCaffrey, & Hamilton, 2001). It is possible that an analogous response may be seen when administering the FCCERS-R in
high-stakes settings. For example, providers may focus on easierto-improve items (e.g., modifying the arrangement of the home) at
the expense of harder-to-improve items (e.g., improving instructional or caregiving quality). While such a response can increase a
family child care home’s FCCERS-R scores, targeting the easier-toimprove items may also undermine the validity of the inferences
drawn from the FCCERS-R because the increases in scores are
limited to particular aspects of quality.
1. Previous studies of the factor structure of the
Environmental Rating Scales
Although little is known about the measurement properties of
the FCCERS-R, there have been several studies that have examined the measurement properties of its companion measures,
the Infant–Toddler Environment Rating Scale—Revised (ITERS-R;
Harms, Cryer, & Clifford, 2003) and the Early Childhood Environment Rating Scale—Revised (ECERS-R; Harms, Clifford, & Cryer,
1998). Previous factor analysis studies have reported fewer than
the seven dimensions of quality put forth by the developers
of these measures. For example, Hestenes, Cassidy, Hegde, and
Hansen (2007) found four factors on the ITERS-R relating to Materials/Activities, Safety/Organization, Language/Interactions, and
Parents/Staff. In contrast, Bisceglia, Perlman, Schaack, and Jenkins
(2009) found only one factor. Similarly, other studies have failed
to confirm the expected seven-factor structure of the ECERS-R.
Both Cassidy, Hestenes, Hegde, Hestenes, and Mims (2005) and
Sakai, Whitebook, Wishard, and Howes (2003) reported two distinct factors on the ECERS-R; and while there were differences in
specific factor loadings, both studies found roughly the same factor
structure. Cassidy, Hestenes, Hedge, et al. (2005) identified their
factors as Materials/Activities and Language/Interaction, whereas
Sakai et al. (2003) identified their factors as provisions for Learning and Teaching and Interactions. However, Perlman, Zellman,
and Le (2004) concluded that the ECERS-R was essentially a onedimensional scale.
The fact that there may be fewer than seven dimensions of
quality on the Environmental Rating Scales can have implications
for measurement efficiency and resource allocation. If a subset of
items could be identified that had similar psychometric properties as the full scale (e.g., comparable coefficient alpha estimates
and similar relationships to other measures of quality), then from
a resource allocation standpoint, it may make sense to adopt an
abridged version of the measure. If there were ways to administer a psychometrically sound but shorter version of the FCCERS-R
that would save time to administer, then the cost savings associated with conducting a more efficient instrument could be directed
elsewhere, such as toward quality improvement efforts in resourceconstrained child care programs (Bisceglia et al., 2009). In addition,
937
using shorter versions of the FCCERS-R could yield cost savings,
as training raters on a 12-item scale may be more efficient than
training them on a 42-item scale. In addition, selecting easy and
less time-consuming items to administer on the FCCERS-R, such
as those that capture elements of the physical environment, may
allow raters to simultaneously administer other process quality
measures, yielding efficiency and cost savings to data collection.
Indeed, many states are considering incorporating the Classroom
Scoring and Assessment System (Pianta, La Paro, & Hamre, 2008)
into their QRISs (Tout et al., 2010), and administering a shorter
FCCERS-R would allow states to capture multiple elements of quality efficiently.
Earlier efforts that have attempted to create abridged versions
of the ITERS-R and ECERS-R have reported the subsets to be psychometrically sound. Bisceglia et al. (2009) found that internally
consistent subsets of the ITERS-R could be created by randomly
selecting 12 items from the full ITERS-R scale. Furthermore, the
randomly chosen subsets and the full ITERS-R scale classified centers as “Poor,” “Average,” and “Good” quality in a similar ways.
Similar results were found for the ECERS-R (Perlman et al., 2004).
In contrast, Cassidy, Hestenes, Hedge, et al. (2005) found that
the randomly chosen 12-item subsets had substantially lower
coefficient alpha estimates than both the full ECERS-R scale and
the subset created from combining the two factors identified
by their factor analytic results. The authors contend that this
finding was partially attributable to the two-factor subset simulating the full ECERS-R scale better than the random subsets
did.
2. Purpose of this study
In this study, we report results from an exploratory factor analysis of the FCCERS-R, administered under one state’s QRIS. We also
explored shorter versions of the FCCERS-R scale, guided by the
results of our factor analysis and by random selection of the scale
items. We then examined the extent to which the abridged versions
of the FCCERS-R demonstrated similar properties to the full version
of the instrument. We did this in two ways. First, we assessed the
validity of the subsets by comparing the extent to which the subsets
assigned family child care homes to the same quality categories as
the full FCCERS-R scale.
Second, we explored whether the full FCCERS-R scale and the
subsets showed comparable relations to other regulatable quality
indicators, including adult–child ratios and group sizes as well as
to provider formal education, specialized training, and experience
providing paid care to children. We chose these specific indicators
because previous research has consistently found that family child
care providers offer better process quality, including more stimulating interactions and responsive caregiving, when adult–child
ratios and group sizes in homes are lower (Clarke-Stewart et al.,
2002; NICHD ECCRN, 2000). Research has also found that providers
with more formal education, child-related training, and experience
also demonstrate more age-appropriate interactions with children,
better environmental organization, and more responsive caregiving than do providers with less experience or training (Burchinal,
Howes, & Kontos, 2002; Clarke-Stewart et al., 2002; Forry et al.,
2012). Given the body of evidence attesting to the importance
of these specific regulatable quality indicators to the quality of
providers’ interactions with children, they are commonly included
in states’ QRIS (Schaack et al., 2012). Thus, we examined the relationships between the FCCERS-R subsets and these regulatable
quality indicators. The extent to which the subsets of the FCCERS-R
show similar associations to regulatable quality indicators as the
full scale would provide support for the use of shorter versions of
the FCCERS-R.
938
D. Schaack et al. / Early Childhood Research Quarterly 28 (2013) 936–946
3. Method
3.1. Sample
The sample used for this study consisted of 99 licensed family child care homes that were voluntarily participating in a state’s
QRIS. This sample reflected 100% of the family child care homes
that received a QRIS quality rating from September of 2008, when
the revised edition of FCCERS-R was introduced in the state’s QRIS,
to December of 2010. Family child care homes were eligible to participate in the QRIS if they were located near a low-performing
elementary school and accepted child care subsidies as a form of
payment, or if they were located in the state’s largest city and served
at least one 4-year old child. Eighty-five percent of the child care
homes in this sample met the first criteria, and the remaining 15%
met the second criteria.
Prior to their FCCERS-R assessment, all providers working in
family child care homes in this study participated in a workshop
that provided an overview of the FCCERS-R. During this workshop,
each provider was given a checklist that outlined the criteria on the
FCCERS-R that they were able to use as a self-assessment. For 60%
(n = 59) of homes, the FCCERS-R observation used in this study represented a baseline or first observation as a participant in the state’s
QRIS. The remaining 40% (n = 40) had received at least one prior
FCCERS-R assessment and therefore had prior quality improvement support. All providers received the results of their FCCERS-R
assessment approximately one month after it was administered.
In addition, all family child care homes received incentives for
their participation in the state’s QRIS, which included coaching and
grants for materials and equipment after their initial assessment.
Each home’s overall quality rating, including their FCCERS-R score
was publically available and disseminated to families through the
state’s Office of Child Care Resource and Referral Agencies.
3.2. Measures of quality
3.2.1. FCCERS-R
The FCCERS-R is a 38-item measure of global child care quality organized into seven subscales that pertain to: Space and
Furnishings (6 items); Personal Care Routines (6 items); Listening and Talking (3 items); Activities (11 items); Interactions (4
items); Program Structure (4 items); and Parents and Providers (4
items). Following prior research in child care using the Environment Rating Scales (see Farran & Son-Yarbrough, 2001; La Paro,
Sexton, & Snyder, 1998; Whitebook, Howes, & Phillips, 1990), the
agency administering the QRIS eliminated items on the Parents and
Provider subscale. Many of the indictors on this subscale relied
on self-report, and were considered subject to respondent bias
(Achenbach, McConaughy, & Howell, 1987). In addition, many of
the items were considered too distal to children’s school readiness
outcomes, which was the goal of the state QRIS (Zellman, Perlman,
Le, & Setodji, 2008). Each of the remaining 34 items on the FCCERS-R
were scored on a 7-point Likert scale, where a score of 1 represents
“inadequate,” a score of 3 represents “minimal,” a score of 5 represents “good,” and a score of 7 represents “excellent” quality. An
overall FCCERS-R score was calculated by averaging the scores on
all the administered items, and the subscale scores were calculated
by averaging the scores across each of the items within a subscale.
development completed. Formal education level and number of
ECE credits completed were verified by trained data collectors via
transcripts. Provider qualification data was collected from 100%
of the homes in the study. In nearly 40% of the family child care
homes, there was more than one provider caring for children. For
these homes, the education level, years of experience, and number of ECE credits taken were averaged across providers. Analyses
were also conducted using just the lead provider who identified
himself or herself as the owner of the business. However, analyses
yielded similar results to those presented that averaged qualifications across providers; therefore the average qualifications model
is presented.
3.2.3. Ratios and group sizes
Data collectors also documented the number of providers and
children present during eight different time periods over the course
of two days. This method followed those set forth by Le, Perlman,
Zellman, and Hamilton (2006) who found that when ratios and
group sizes were sampled during six specific morning time points
and two specific afternoon time points and were averaged, the
ratios yielded a representative account over a two-week period. In
this study, morning time points were sample on one day during the
FCCERS-R administration, and afternoon time points were sampled
a second day and adult–child ratios and group sizes were calculated
by averaging across the eight different counts. Adult–child ratios
and group sizes were collected from 100% of the sample.
3.3. Procedures
Each family child care home was assigned a one-month data
collection window. During this time, the state’s QRIS administering
agency dispatched trained data collectors to the family care homes.
Data collectors were required to have a minimum of an associate’s
degree in early childhood education or a related field, and all had
experience working in early childhood settings. Prior to collecting
data, each of the eight data collectors participated in a one-week
training on data collection procedures. In addition, each data collector participated in a two-day training session on the FCCERS-R
conducted by an expert scorer, referred to as a state anchor, who
had been trained by the developers of the FCCERS-R, and who had
been deemed a reliable rater. Each data collector was required, during three consecutive administrations of the FCCERS-R, to score 85%
of the items on the FCCERS-R within one scale point of the state
anchor. The reliability of the data collectors was re-checked after
every tenth FCCERS-R administration, and all data collectors passed
re-reliability checks.
All FCCERS-R observations were unannounced, although
providers were aware of their one-month data collection window.
All observations were conducted between approximately 8:30 a.m.
and 1:30 p.m. During the FCCERS-R administration, data collectors
documented adult–child ratios and group sizes during six time
points. The data collector returned to the family child care home
for a second visit in the afternoon to collect training and education
data, and to document adult–child ratios and group sizes during
two additional time points.
4. Results
4.1. Descriptive statistics of the quality indicators
3.2.2. Provider qualifications
To obtain information about the family child care provider’s
qualifications, a survey was administered to each provider querying them about their highest level of education, their years of paid
experience caring for and educating children ages birth to five
(including experience in other child care settings), and number of
course credits taken in early childhood education (ECE) or child
Table 1 presents the descriptive statistics for the FCCERS-R total
score and subscales. As shown in Table 1, the mean FCCERS-R
score was 4.38. Forty percent of the family child care homes had
taken part in the QRIS for at least a year, which included funding mechanisms to support programs to improve quality. Analyses
were conducted comparing results for homes that had received
D. Schaack et al. / Early Childhood Research Quarterly 28 (2013) 936–946
939
Table 1
Descriptive statistics, coefficient alphas, and intercorrelations among the FCCERS-R total score and subsets.
FCCERS-R scale
Mean
SD
Coefficient alpha
(1) Space and furnishings
(2) Personal care routines
(3) Listening and talking
(4) Activities
(5) Interaction
(6) Program structure
(7) Total FCCERS-R
4.26
2.89
5.51
4.15
5.78
5.41
4.38
.88
.73
1.19
.84
1.38
1.31
.74
.44
.35
.59
.70
.81
.52
.87
Range
2.33–6.50
1.33–5.17
1.67–7.00
2.36–6.55
1.50–7.00
1.67–7.00
2.55–6.30
Correlations
(1)
(2)
(3)
(4)
(5)
(6)
.41
.50
.62
.55
.47
.80
.33
.38
.35
.20
.58
.61
.69
.41
.77
.46
.36
.75
.36
.66
.66
Note: All correlations are significant at the .01 level.
prior FCCERS-R assessments (and quality improvement support)
and homes that had only received an initial FCCERS-R assessment
(and had not yet received quality improvement support). Analyses yielded highly similar results; therefore the combined sample
model is presented. While other research studies (Helburn, 1995)
have found lower FCCERS-R scores than found in this sample, the
higher scores may reflect the fact that family child care providers in
states with QRISs attend to the items on the FCCERS-R more so than
in states without QRISs and therefore may have higher scores. In
addition, all providers attended a training on the FCCERS-R prior to
being assessed, which may have resulted in higher scores than generally found in the population of family child care homes. Therefore,
inferences from this study should be restricted to states with QRISs
or that use the FCCERS-R as an accountability or improvement tool
across the state.
Mirroring the education levels of family child care providers
nationally (Herzenberg, Price, & Bradley, 2005), 5% of the family
child care providers in this sample had an associate’s degree, 15%
had a bachelor’s degree, and 2% held a masters degree or higher.
Providers had, on average, eight years (SD = 8.02) of paid experience
working with children, and the typical provider had completed
almost 11 ECE credits (SD = 14.79). In addition, low adult–child
ratios were found in this sample, with an average of 3.71 children
for every provider (SD = 1.28). Group size was also small, with an
average of almost 6 children per family child care home (SD = 3.21).
4.2. Factor analysis of the FCCERS-R
Table 1 presents the coefficient alpha estimates and intercorrelations among the subscales and the overall FCCERS-R. Although
the internal consistency estimates for a given scale is partly a function of the number of items, such that higher alpha estimates are
generally found on scales with more items (Streiner & Norman,
1989), in our sample, some subscales with fewer items (e.g., Interaction) showed higher internal consistency estimates than did other
subscales with more items (e.g., Personal Care Routines). This suggested that several distinct aspects of quality were being captured
within the same subscale.
Table 1 also shows that the intercorrelations among the subscales range from 0.20 to 0.69, with a median of 0.50. Although some
subscales were moderately correlated to each other, suggesting
some redundancy among the items, other subscales were modestly correlated, suggesting that there is likely to be more than one
dimension of quality identified in a factor analysis. The item intercorrelations supported this notion, as the median intercorrelation
was relatively low at 0.19.
Table 2 shows the descriptive statistics for the individual
FCCERS-R items. For the most part, there was no evidence of floor
or ceiling effects. The exceptions were safety practices, which had
a low mean of 1.19 points, and greeting/departing and group time,
which had high means of 6.84 and 6.50 points, respectively. We
conducted sensitivity analyses with and without these items, and
with their log transformations, but results were robust to these
alternative specifications. Thus, we retained the items in their original form in the analysis.
Most items were completely observed, with the exception of
two items relating to provisions for children with disabilities and
use of TV, video, and computers. These two items had 85% and 31%
missing responses, respectively. We eliminated the item relating
to provisions for children with disabilities from analysis because of
the large percentage of missing data, which indicated that most of
the family child care homes in our sample did not care for children
with identified disabilities. We retained the item relating to the use
of TV, video, and computers because extended TV watching is more
likely to occur in lower-quality family child care homes (Layzer &
Goodson, 2006). Following the procedures of Truxillo (2005), we
imputed missing data using full-information maximum likelihood
parameter estimation. This resulted in analysis on 33 FCCERS-R
items.
Because we expected the factors to be correlated, we conducted
an exploratory factor analysis with promax rotation. Table 2 provides the results of the factor analysis. By the Kaiser criterion, we
retained three factors with mean communalities among items calculated at 0.52. Following Tinsley and Tinsley (1987), we considered
items with a factor loading of 0.30 or greater when interpreting the
factors. Although items could have factor loadings of at least 0.30
on two different factors, for all items, the difference in factor loadings was at least 0.10, so the item was retained on the factor with
the highest loading.
The first factor, which we labeled Activities/Materials, had an
eigenvalue of 6.85, and explained 40% of the variance. Items on
this factor captured aspects of the family child care home primarily
related to children’s play, including the types of materials available, the areas devoted to play, and the interactions that occurred
while children were engaged in play with the materials. It was
comprised of 12 items and had an internal consistency estimate
of 0.85. The second factor, which we labeled Language/Interaction,
had an eigenvalue of 2.16, and explained 12% of the variance. This
scale assessed the extent to which providers promoted children’s
language use and the emotional tone of providers’ interactions
with children. It was comprised of six items and had an internal
consistency estimate of 0.86. The final factor, which we labeled
Organization, had an eigenvalue of 1.58, and explained 9% of the
variance. The scale included items about scheduling, arrangement
of furniture, and use of space. It was comprised of 10 items and had
an internal consistency estimate of 0.64.
4.3. Creating shortened versions of the FCCERS-R
The factor analysis results raise the possibility that subsets
created from the three factors may be more efficient versions of
the full FCCERS-R. We explored several ways of creating abridged
versions of the full FCCERS-R. First, we created three subsets corresponding to each of our identified factors. The first subset was
940
D. Schaack et al. / Early Childhood Research Quarterly 28 (2013) 936–946
Table 2
Item loadings of the exploratory factor analysis.
Factor scale
Factor 1: Activities/Materials
Fine motor
Using books
Free play
Provision for relaxation and comfort
Dramatic play
Music and movement
Math/number
Blocks
Nature/science
Promoting acceptance of diversity
Display for children
Space for privacy
Factor 2: Language/Interaction
Provider–child interaction
Discipline
Interactions among children
Helping children use language
Helping children understand language
Supervision of play and learning
Factor 3: Organization
Arrangement of indoor space for child care
Art
Furniture for routine care, play, and learning
Meals/snacks
Group time
Health practices
Nap/rest
Sand and water play
Use of TV, video, and/or computer
Schedule
Items that did not load on any factor
Active physical play
Diapering/toileting
Greeting/departing
Indoor space used for child care
Safety practices
Excluded itema
Provisions for children with disabilities
FCCERS-R scale
Mean
SD
Range
Activities
Listening and talking
Program structure
Space and furnishings
Activities
Activities
Activities
Activities
Activities
Activities
Space and furnishings
Space and furnishings
4.02
4.37
5.53
4.83
4.73
4.76
4.56
3.49
5.06
4.34
4.78
4.61
2.13
1.98
1.86
1.60
1.44
1.47
1.46
1.19
1.51
1.55
1.60
2.31
1.00–7.00
2.00–7.00
1.00–7.00
2.00–7.00
1.00–7.00
2.00–7.00
2.00–7.00
1.00–7.00
2.00–7.00
2.00–7.00
1.00–7.00
1.00–7.00
Interaction
Interaction
Interaction
Listening and talking
Listening and talking
Interaction
6.06
5.72
6.10
5.90
6.04
4.75
1.58
1.82
1.11
1.34
1.42
2.24
2.00–7.00
1.00–7.00
2.00–7.00
1.00–7.00
2.00–7.00
1.00–7.00
Space and furnishings
Activities
Space and furnishings
Personal care routines
Program structure
Personal care routines
Personal care routines
Activities
Activities
Program structure
2.16
2.74
4.15
1.78
6.50
2.28
3.60
5.43
4.72
4.48
1.79
1.96
1.75
1.53
1.27
1.46
2.65
2.04
2.44
2.16
1.00–7.00
1.00–7.00
1.00–7.00
1.00–7.00
1.00–7.00
1.00–7.00
1.00–7.00
1.00–7.00
1.00–7.00
1.00–7.00
Activities
Personal care routines
Personal care routines
Space and furnishings
Personal care routines
2.02
1.71
6.84
5.03
1.19
.99
1.33
.87
1.17
.52
1.00–6.00
1.00–7.00
2.00–7.00
2.00–6.00
1.00–4.00
Program structure
5.06
2.35
1.00–7.00
Factor 1 loadings
Factor 2 loadings
.81
.69
.68
.64
.64
.64
.58
.53
.51
.44
.36
.49
.36
.34
.38
.39
.41
.37
.37
.34
.34
.85
.82
.78
.74
.62
.60
.37
.44
.36
.35
.48
.32
.58
.52
.50
.50
.38
.36
.36
.34
.34
.30
N/A
N/A
.31
N/A
Factor 3 loadings
Notes: Empty cells represent factor loadings less than 0.30.
a
This item was excluded from the factor analysis because of high rates of missing responses.
comprised of the 12 items comprising the Activities/Materials factor, the second subset was comprised of the 6 items comprising the
Language/Interaction factor, and the third subset was comprised of
the 10 items comprising the Organization factor.
Next, we explored several subsets that had roughly the same
number of items as each of the factors, but instead of limiting the
items to a particular factor, we included items from all three factors. For example, we created a subset that was comprised of the
two items with the highest loadings on each factor. This subset,
which we labeled the 6-item scale, was comprised of the following
items: fine motor, using books, provider–child interaction, discipline, arrangement of indoor space for child care, and art. We then
created two additional subsets in an analogous manner; that is,
we created a subset comprised of the three items with the highest
loadings on each factor, then created another subset comprised of
the four items with the highest loadings on each factor. We labeled
these subsets 9-item and 12-item scales, respectively. The 9-item
scale was comprised of the same items on the 6-item scale, along
with additional items relating to free play, interactions among children, and furniture for routine care, play, and learning. The 12-item
scale was comprised of the same items on the 9-item scale, along
with additional items relating to provision for relaxation and comfort, helping children use language, and meals/snacks.
Finally, we randomly selected three random subsets of items
consisting of 12 randomly chosen items following the method used
by Scarr, Eisenberg, and Deater-Deckard (1994). To create the random subsets of items, we used a random number generator to
assign each item a number, rank ordered the items by their assigned
numbers, then chose the 12 highest-ranked items. We repeated this
procedure three times, resulting in three random subsets of items.
There was little item overlap between the three randomly chosen
subsets and the 12-item scale, with the first two random subset
sharing two items in common with the 12-item scale, and the other
random subset sharing only one item in common with the 12-item
scale. None of the random subsets included safety practices.
Table 3 presents the descriptive statistics, coefficient alphas,
and intercorrelations among the subsets and the total FCCERS-R
score. As expected, all the subsets were highly correlated with the
total FCCERS-R score, with correlations ranging from 0.76 to 0.93.
Notably, despite having fewer items, the 6-item and 9-item scales
had coefficient alpha estimates that were slightly higher than the
12-item random subsets.
4.4. Discriminant analysis
We next explored the extent to which the subsets assigned family child care homes to the same quality categories as the total
FCCERS-R score. We assigned family child care homes to quality
designations using two different methods. First, following the same
methods used by previous studies that examined the factor structures of the ITERS-R and ECERS-R (see e.g., Bisceglia et al., 2009;
Perlman et al., 2004), we assigned homes to three quality levels
based on their overall FCCERS-R scores. Family child care homes
with a total FCCERS-R score between 1 and 3 were classified as
D. Schaack et al. / Early Childhood Research Quarterly 28 (2013) 936–946
941
Table 3
Descriptive statistics, coefficient alphas, and intercorrelations among the FCCERS-R total score and the shortened versions of the FCCERS-R.
Scale
Mean
SD
Coefficient alpha
(1) Activities/Materials
(2) Language/Interaction
(3) Organization
(4) 6-Item scale
(5) 9-Item scale
(6) 12-Item scale
(7) Random subtest 1
(8) Random subtest 2
(9) Random subtest 3
(10) Total FCCERS-R
4.59
5.88
3.74
4.21
4.59
4.84
4.15
4.43
4.33
4.38
1.03
1.24
.92
1.18
1.04
.98
.73
.78
.55
.74
.85
.86
.64
.70
.76
.81
.60
.65
.64
.87
Range
2.67–7.00
1.50–7.00
1.50–6.00
1.50–7.00
1.89–7.00
1.83–6.50
2.58–6.42
2.75–6.42
2.67–6.08
2.55–6.30
Correlations
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
.50
.42
.73
.78
.81
.84
.76
.80
.84
.47
.78
.80
.81
.56
.64
.66
.77
.61
.65
.65
.66
.72
.67
.76
.96
.95
.80
.87
.74
.86
.98
.83
.85
.80
.91
.83
.85
.81
.93
.85
.86
.89
.78
.89
.90
Note: All correlations are significant at .01 level.
“Poor,” family child care homes with a score greater than 3 but less
than 5 were classified as “Average,” and family child care homes
with a score equal to or greater than 5 were classified as “Good.”
Overall, there were 5 family child care homes classified as Poor, 77
classified as Average, and 17 classified as Good.
Next, we classified full FCCERS-R scores based on the quality
designations outlined in the state’s QRIS. The state’s QRIS uses a
point system whereby a home is assigned 0 points if the FCCERS-R
score falls below 3.50, 2 points if the FCCERS-R score falls between
3.50 and 3.99, 4 points if the FCCERS-R score falls between 4.00 and
4.69, 6 points if the FCCERS-R score falls between 4.70 and 5.49, 8
points if the FCCERS-R score falls between 5.50 and 5.99, and 10
points if the FCCERS-R score is at or above 6.00 (Tout et al., 2010).
Under the state’s quality classification system, 11 homes received
0 points, 17 homes received 2 points, 35 homes received 4 points,
30 homes received 6 points, 5 homes received 8 points, and 1 home
received 10 points.
We conducted discriminant analyses to explore the correspondence in quality designations between the subsets and the full
FCCERS-R instrument. To assign the prior probabilities, we assumed
that the distribution of family child care homes was proportionally
weighted across the quality categories in the same manner as the
sample distribution. This method is similar to that employed by
Perlman et al. (2004). Table 4 provides the correspondence between
the three-category classifications (Poor, Average, and Good) based
on the full FCCERS-R instrument and the classifications based on
the subsets. The accuracy of classifications was consistent across
the three-factor subsets, with the factor subsets matching the
full FCCERS-R designations approximately three quarters of the
time. However, across all three-factor subsets, there was a trend
toward systematic under-prediction. For example, whereas the full
FCCERS-R scale classified 17 family child care homes as Good, the
Language/Interaction subscale classified all of these homes as Average. Similarly, approximately half of the family child care homes
that were designated by the full FCCERS-R scale as Good were
classified as Average by the Activities/Materials and Organization
subsets.
Classification accuracy with the three-category quality levels
was better among the random subsets and the 6-, 9-, and 12-item
scales. For these subsets, the accuracy rates ranged from 83% to
91%, with the highest accuracy rate found on the 12-item subset.
Misclassifications were most likely to occur on the Good classification, with a trend toward under-prediction. Namely, family child
care homes that were designated by the full FCCERS-R instrument
as Good were classified by the subsets as Average.
Table 6 provides the correspondence between the state’s quality point classifications and the classifications based on the subsets.
The accuracy of classifications varied greatly by subset. Misclassification rates among the subsets derived from the factors on
the state’s quality point system were notably higher than the
misclassification rates using the three-category quality classifications. The factor subsets matched the full FCCERS-R designations
for only 44–55% of the time. All three-factor subsets, Activities/Materials, Language/Interaction, and Organization tended to
over-predict on the lower end of the point scale and under-predict
at the higher end of the point scale. While the full FCCERS-R identified homes as being at the lower end of the point scale, the factor
subsets classified many of these same programs as being in the
mid-range of the point system. Conversely, whereas the FCCERSR scale classified programs at the high end of the continuum, the
factor subsets classified many of these same programs within the
mid-range of the quality point continuum.
Classification accuracy with the state’s quality point system was
also higher for the randomly selected subsets and with the 6-, 9-,
and 12-item subsets than with the factor subsets, as classification
accuracy ranged from 60% to 69%. Similar trends were noted in
misclassifications as with the factor subsets. Namely, the 6-, 9and 12-item subsets and the randomly selected subsets tended
to over-predict on the lower end of the quality-point scale, and
under-predict at the higher end.
4.5. Relationships with other indicators of quality
Given the potential of the subsets to serve as proxies for the full
FCCERS-R scale, we next examined whether the subsets showed
the same relationships to other indicators of quality as the total
FCCERS-R scale. Table 6 provides the correlations between the
factor subsets, random subsets, the 6-, 9-, and 12-item scales,
and the total FCCERS-R scale with adult–child ratios, group size,
and providers’ years of experience, ECE credits, and degree status
(defined as whether or not the provider obtained at least an associate’s degree). As shown in Table 5, the subsets generally showed
the same magnitude of correlations with the other quality indicators as the full FCCERS-R scale. However, for both the subsets
and the total FCCERS-R score, no significant relationships were
observed with the other regulatable quality indicators.
5. Discussion
This exploratory study examined the measurement properties
of the FCCERS-R, a widely used scale for the measurement of child
care quality in family child care homes. Although the results of this
study are preliminary, and replication in larger samples and in more
diverse regulatory contexts are needed, our results suggest that the
FCCERS-R captured several distinct aspects of quality. However, it
did not appear to capture the seven dimensions of quality hypothesized by the developers. Instead, the exploratory factor analysis
identified three dimensions of quality: Activities/Materials, Language/Interaction, and Organization. All three factors were highly
correlated with the full FCCERS-R scale, and the two former factors
942
D. Schaack et al. / Early Childhood Research Quarterly 28 (2013) 936–946
Table 4
Accuracy of classification predictions of the three-category quality levels using the FCCERS-R subsets.
Subsets
Subtest classification
Group membership based on total FCCERS-R score
Poor (n = 5)
Average (n = 77)
Percent correct
Good (n = 17)
Activities/Materials
Poor
Average
Good
0
0
0
5
68
10
0
9
7
75.75
Language/Interaction
Poor
Average
Good
3
0
0
2
73
17
0
4
0
76.76
Organization
Poor
Average
Good
1
0
0
4
68
11
0
9
6
75.75
6-Item scale
Poor
Average
Good
4
1
0
1
73
8
0
3
9
86.87
9-Item scale
Poor
Average
Good
4
2
0
1
73
6
0
2
11
88.89
12-Item scale
Poor
Average
Good
4
0
0
1
74
5
0
3
12
90.91
Random subset 1
Poor
Average
Good
2
1
0
3
73
6
0
3
11
86.87
Random subset 2
Poor
Average
Good
3
1
0
2
73
8
0
3
9
85.86
Random subset 3
Poor
Average
Good
2
2
0
3
70
7
0
5
10
82.83
had internal consistency estimates that were comparable to the full
FCCERS-R scale.
We also explored the possibility that we could create abridged
versions of the FCCERS-R scale that would show similar measurement properties as the full FCCERS-R instrument. In addition to the
subsets created from the factor structure, we also explored randomly chosen subsets and purposively chosen subsets (i.e., 6-, 9-,
and 12-item scales comprised of the items with the highest factor loading across the three factors). We examined the extent to
which these subsets accurately classified family child homes based
on three-quality categories (Poor, Average, and Good) and based
on the state’s QRIS point system, which utilizes six quality categories. We found that classification accuracy for the factor subsets
was particularly poor, with systematic under-prediction for the
three-quality categories, and high misclassification rates for the
state’s QRIS point system. In contrast, the classification accuracy for
the purposively chosen subsets and the randomly chosen subsets
were higher, particularly on the three-quality category designations, where classification accuracy was over 90% for the 12-item
subset. The classification rates also became less accurate when
using the more granulated quality categories found in the state’s
QRIS point system. This suggests that within this state’s specific
QRIS, the randomly chosen and purposively chosen subsets may
not be accurate at assigning QRIS points. However, it is possible
that these subsets may be more accurate at assigning QRIS points
in states that utilize fewer FCCERS-R quality classifications, as is the
case with the majority of QRIS operating throughout the country
(Tout et al., 2010).
Despite the high internal consistency estimates and the strong
correlations with the total FCCERS-R score, the factor subsets
turned out to be poorer proxies for the full FCCERS-R scale than
the other subsets examined. That is, the randomly chosen subsets
and the 6-, 9-, and 12-item scales were relatively good proxies for
the full FCCERS-R instrument, assigning family child care homes to
the same quality categories as the full FCCERS-R scale in upwards
of 83% of the cases when using the three-quality classifications.
In addition, these subsets showed the same relationships to other
quality indicators as the total FCCERS-R score.
Although the results should be interpreted cautiously because
of the small sample size, the findings are consistent with our
expectations, as child care quality is generally perceived to be
multidimensional (Marshall, 2004). Each individual factor subset
assessed only one specific aspect of quality, either activities and
materials, or language and interactions, or the organization of the
family child care home. Because each factor subset was narrowly
defined, using a single factor subset alone as a proxy for the full
FCCERS-R instrument underrepresented the full range of quality
constructs measured by the FCCERS-R. This result is consistent
with the findings of Bisceglia et al. (2009) as well as with Cassidy,
Hestenes, Hedge, et al. (2005), who reported that a combination
of items that assess both process and structural features of quality is best for simulating the full instrument. This may explain why
the random subsets and the 6-, 9-, and 12-item scales were superior proxies for the total FCCERS-R scale. The random subsets and
6-, 9-, and 12-item scales assessed a wider range of quality constructs than the factor subsets because they included items across
all three factors. Thus, the random subsets and 6-, 9-, and 12-item
scales were more similar in content to the FCCERS-R scale than the
factor subsets.
If confirmed with larger and more diverse samples, the results of
this study have potential implications for cost savings in the administration of the FCCERS-R. Our study suggests that as few as six items
can be administered, and the resulting subset can serve as a reasonable proxy for the full scale. The 6-item scale, for example, was
comparable if not better than the 12-item random subsets in terms
of its classification accuracy with the three-quality categories. The
D. Schaack et al. / Early Childhood Research Quarterly 28 (2013) 936–946
943
Table 5
Accuracy of classification predictions of the state’s QRIS point system using the FCCERS-R subsets.
Subtest
Subset classification
Point rating based on total FCCERS-R score
Percent
0 (n = 11)
2 (n = 17)
4 (n = 35)
6 (n = 30)
8 (n = 5)
10 (n = 1)
Correct
Activities/Materials
0
2
4
6
8
10
5
5
1
0
0
0
4
5
6
0
0
0
2
7
14
12
0
0
0
0
14
18
4
1
0
0
0
0
1
0
0
0
0
0
0
1
44.44
Language/Interaction
0
2
4
6
8
10
7
3
0
0
0
0
3
6
3
0
0
0
1
7
22
11
1
0
0
1
10
19
4
1
0
0
0
0
0
0
0
0
0
0
0
0
54.55
Organization
0
2
4
6
8
10
5
3
0
0
0
0
0
0
2
1
1
0
6
13
26
9
9
0
0
1
7
19
19
3
0
0
0
1
1
2
0
0
0
0
0
0
51.52
6-Item scale
0
2
4
6
8
10
9
3
0
0
0
0
1
6
3
0
0
0
1
7
25
8
0
0
0
1
7
22
2
0
0
0
0
0
3
1
0
0
0
0
0
0
65.66
9-Item scale
0
2
4
6
8
10
9
2
0
0
0
0
2
7
2
0
0
0
0
7
29
9
0
0
0
1
4
20
3
0
0
0
0
1
2
1
0
0
0
0
0
0
67.68
12-Item scale
0
2
4
6
8
10
9
1
0
0
0
0
2
9
5
0
0
0
0
7
24
7
0
0
0
0
6
22
1
0
0
0
0
1
4
1
0
0
0
0
0
0
68.69
Random subset 1
0
2
4
6
8
10
8
2
0
0
0
0
3
6
5
1
0
0
0
7
23
8
0
0
0
2
7
19
2
1
0
0
0
2
3
0
0
0
0
0
0
1
59.60
Random subset 2
0
2
4
6
8
10
6
1
0
0
0
0
5
5
6
0
0
0
0
10
26
7
0
0
0
1
3
22
2
0
0
0
0
1
3
1
0
0
0
0
0
0
62.22
Random subset 3
0
2
4
6
8
10
10
3
0
0
0
0
1
9
4
1
0
0
0
5
23
7
0
0
0
0
8
21
0
0
0
0
0
1
5
1
0
0
0
0
0
0
68.69
Table 6
Correlations between selected FCCERS-R subsets and total FCCERS-R score with regulatable quality indicators.
Total
FCCERS-R
Ratios
Group size
Years of experience
ECE credits
A.A. degree or higher
−.01
−.18
.12
.10
−.04
Activities/
Materials
.01
−.11
.14
.12
−.08
Language/
Interaction
.01
−.18
.10
.01
−.06
Organization
−.03
−.11
−.07
.08
.02
6-Item
scale
−.05
−.20
.11
.12
−.07
9-Item
scale
−.06
−.20
.12
.10
−.07
12-Item
scale
−.04
−.20
.14
.08
−.07
12-Item random
subsets
1
2
3
−.05
−.02
.14
.11
−.06
−.05
−.15
.12
.15
−.04
−.03
−.08
.18
.07
−.08
944
D. Schaack et al. / Early Childhood Research Quarterly 28 (2013) 936–946
6-item scale also showed a high internal reliability estimate and
similar relationships to other quality indicators as the full FCCERSR scale. However, it is important to emphasize that the 6-item scale
was purposively constructed, so that items with the highest factor
loadings were selected for inclusion. Other subsets created from
a different set of six items may not be good proxies for the total
FCCERS-R score. For example, the Language/Interaction factor was
also comprised of six items, and had an even higher internal consistency estimate than the 6-item scale, yet proved to be a poorer
proxy for the total FCCERS-R score.
The contrast between the Language/Interaction subset and the
6-item scale in terms of their psychometric properties highlights
the tension associated with creating measures that attempt to capture both breadth and depth. All else being equal, measures that
have higher internal consistency estimates will be more sensitive
to detecting the effects of quality than less reliable measures (Light,
Singer, & Willett, 1990). Yet, measures that have high internal consistency estimates may also be focusing on very specific or narrow
aspects of quality. Thus, there may be trade-offs between creating
measures that represent the full range of quality, while also maintaining high internal consistency estimates for those measures.
To help evaluate these trade-offs, it will be important to understand the functioning of subsets of the FCCERS-R in relationship
to outcomes of interest to states. In this study, we were only able
to examine the full FCCERS-R and FCCERS-R subsets in relation to
regulatable quality indices measured in most states’ QRIS. While
the subsets showed similar associations as the full FCCERS-R, it
is unclear whether analogous relationships would be observed if
child outcomes had been examined instead. It may be the case
that subsets of items that specifically measure care and instructional processes demonstrate stronger relationships to children’s
learning and development than subsets that are primarily composed of items related to physical space and health practices, which
are more distal to children’s learning. In recent research examining
the functioning of the ITERS-R, items included in the Interaction
scale demonstrated stronger relationships with children’s cognitive development than did items on other subscales (Setodji, Le, &
Schaack, in press). If similar results are observed with the use of
the FCCERS-R in family child care homes, then it may make sense
for states to select the more narrow Language/Interaction subset
and consider the trade-offs with respect to classification accuracy,
relative to the full FCCERS-R instrument.
Within the context of this study, we did not have access to
other measures of process quality, as they were not used in the
state’s QRIS and were only able to relate FCCERS-R subsets to structural quality variables. In future studies, FCCERS-R factors should
be examined in relation to measures of process quality, such as
instructional support and emotional climate, which have been
shown to be associated with children’s developmental outcomes
(Pianta et al., 2008). If specific FCCERS-R factors or items are redundant with other process quality measures, states may opt to select
shorter versions of the FCCERS-R that are not overlapping in constructs.
It is also important to consider potential implications of administering a subset of FCCERS-R items within accountability and
quality improvement contexts. It is certainly possible that under
these high-stakes conditions, providers may focus on changing
practices in ways that comport only with the items selected. Consequently, it may be important for providers to be blind to the
items assessed to help ensure that they focus on improving multiple aspects of quality. Sampling fewer items also limits the breadth
of information that can be used by providers and coaches to inform
quality improvement efforts. One strategy to overcome this limitation is to train coaches on the FCCERS-R, and to have the coaches and
providers conduct a self-assessment using the full range of FCCERSR items. Generating quality improvement goals from facilitated
self-assessments may be more aligned with sound adult learning principles, and may therefore be more effective at changing
practices than generating them from top-down high-stakes assessments (Palsha & Wesley, 1998). Such a strategy may also serve to
reduce the costs of assessment and monitoring efforts by reducing rater training time, and allow for more resources to be directed
toward quality improvement efforts or administering other measures of quality more strongly associated with children’s well-being
(Pianta et al., 2008). It is also possible that administering fewer
items on the FCCERS-R may reduce the time needed to conduct
observations and therefore reduce provider stress associated with
the administration of the FCCERS-R (Pope, Denny, Homer, & Ricci,
2006). This, in turn, may prompt more family child care homes to
participate in quality enhancement projects.
It is also important to consider the broader implications of
using a measure such as the FCCERS-R in high-stakes contexts. The
FCCERS-R by design, is global in nature, and may not adequately
assess the types of care and instructional practices most associated with children’s social development and academic learning
(Snow & Van Hemel, 2008). Indeed, the majority of items relate
to the physical home environment, including the furniture, quantity of learning materials, and the organization of the physical
space, and far fewer items tap into care and instructional processes (Cassidy, Hestenes, Hansen, Hedge, & Shim, 2005). By using
the FCCERS-R as an accountability and monitoring tool, providers
and those tasked with supporting family child care improvement
also appear to be narrowing their focus to the “mechanics” of
early childhood education measured on the FCCERS-R, to the exclusion of strengthening pedagogy and caregiver interactions in ways
that promote child learning and development (Smith, Schneider,
& Kreader, 2010). Consequently careful consideration of the unintended consequences of the use of global measures should be taken
before being administered in accountability contexts.
It is also interesting to note that although we found similar
relationships between the full FCCERS-R and FCCERS-R subsets
with regulatable quality indices, neither the full FCCERS-R scale
nor the subsets were related to adult–child ratios, group-sizes,
and provider training and education variables. The low adult–child
ratios and groups sizes observed in this study may have contributed
to these findings. It is unclear, however, why different dimensions
of provider training and education were unrelated to the FCCERSR or to the subsets. Providers in this study, like most nationally
(Herzenberg et al., 2005), had low levels of formal education and
had completed very few classes related to young children. It is possible that there is a minimum threshold of education that is needed
before relationships between global quality and provider education
can be observed. In this study, we also did not have information
related to the content and quality of the coursework providers had
completed, which may also be important factors in explaining the
lack of relationships found (Whitebook & Ryan, 2011). However,
it is also possible that within high-stakes initiatives such as QRIS,
providers, regardless of their education, are attuned to the constructs measured in the FCCERS-R because of the consequences
of their scores to the levels of funding received by their program.
Therefore, they may comport their programs to meet the FCCERSR criteria, thus attenuating the relationships between regulatable
quality variables and observed quality.
It is also important to note that the Parents and Staff subscale
on the FCCERS-R was not administered in this study because the
state’s QRIS was concerned with respondent bias associated with
self-report measures, especially within the context of high-stakes
assessments. Nonetheless, provider and family needs are an important element of quality and are often included in state QRISs, but the
construct itself has been difficult to measure (Zellman & Perlman,
2006). However, there are efforts underway to develop better
measures of family sensitive care (see Bromer et al., 2011). When
D. Schaack et al. / Early Childhood Research Quarterly 28 (2013) 936–946
such measures are available, it will be important to understand the
relationships between FCCERS-R factors and theses measures.
There are important limitations to this study that need to
be addressed in future research before states consider adopting
abridged versions of the FCCERS-R. Because of the small sample
size, we were unable to conduct a confirmatory factor analysis to
test the robustness of the three-factor solution. Nonetheless, our
results indicted a high level of over-determination of factors (i.e.,
many items with strong loadings within each factor) and moderate to strong communalities among items, which suggests stability
in our solution (Costello & Osborne, 2005; MacCallum, Widaman,
Preacher, & Hong, 2001). It is also important to note that our sample only included family child care homes from one regulatory
climate. Therefore, this study should be considered preliminary,
and additional confirmatory studies should be conducted to determine whether our structure is replicated in other settings, including
within family child care homes in states with differing regulatory
environments governing family child care.
As policymakers increase the consequences attached to the
FCCERS-R, more attention needs to be paid to its psychometric
properties. This study found evidence that the FCCERS-R is a multidimensional measure of quality, assessing several distinct aspects
of family child care quality. Our results also suggest that less than
one fifth of the FCCERS-R items could be administered, and the
resulting subsets would remain psychometrically sound. In addition to conducting confirmatory factor analysis on the factors found
in this study, future studies should also explore the relationships
of these factors as well as abridged versions of the FCCERS-R to
children’s outcomes. Cost–benefit analysis are also needed to determine whether the potential savings associated with a reduced
administration of the FCCERS-R warrants the potential increases
in inaccuracies or misclassifications that may arise from using an
abridged version.
References
Achenbach, T. M., McConaughy, S. H., & Howell, C. T. (1987). Child/adolescent behavioral and emotional problems: Implications of cross-informant correlations for
situational specificity. Psychological Bulletin, 101, 213–232.
American Educational Research Association, American Psychological Association,
& National Council on Measurement in Education. (1999). Standards for educational and psychological testing. Washington, DC: American Educational Research
Association.
Bisceglia, R., Perlman, M., Schaack, D., & Jenkins, J. (2009). Examining the psychometric properties of the Infant Toddler Environment Rating Scale in high stakes
contexts. Early Childhood Research Quarterly, 24(3), 121–132.
Bromer, J., Paulsell, D., Porter, T., Weber, R., Henly, J., Ramsburg, D., et al. (2011).
Family-sensitive caregiving: A key component of quality in early care and education. In M. Zaslow, K. Tout, T. Halle, & I. Martinez-Beck (Eds.), Next steps in the
measurement of quality in early childhood settings. Baltimore, MD: Brookes.
Burchinal, M. R., Howes, C., & Kontos, S. (2002). Structural predictors of child care
quality in child care homes. Early Childhood Research Quarterly, 17(1), 87–105.
Cassidy, D., Hestenes, L., Hansen, J., Hedge, A., & Shim, J. (2005). Revisiting the two
faces of child care quality: Structure and process. Early Education and Development, 16(4), 505–520.
Cassidy, D. J., Hestenes, L. L., Hegde, A., Hestenes, S., & Mims, S. (2005). Measurement
of quality in preschool childcare classroom: An exploratory and confirmatory
factor analysis of the Early Childhood Environment Rating Scale—Revised. Early
Childhood Research Quarterly, 20(3), 345–360.
Clarke-Stewart, A. K., Vandell, D. L., Burchinal, M., O’Brian, M., & McCartney, K. (2002).
Do regulatable features of child-care homes affect children’s development? Early
Childhood Research Quarterly, 17(1), 52–86.
Costello, A. B., & Osborne, J. W. (2005). Best practices in exploratory factor analysis: Four recommendations for getting the most from your analysis. Practical
Assessment Research & Evaluation, 10(7), 1–9.
Farran, D. C., & Son-Yarbrough, W. (2001). Title I funded preschools as a developmental context for children’s play and verbal behaviors. Early Childhood Research
Quarterly, 16(2), 245–262.
Forry, N. D., Iruka, I., Kainz, K., Tout, K., Torquati, J., Susman-Stillman, A., et al. (2012).
Identifying profiles of quality in home-based child care [Issue Brief OPRE 2012-20].
Washington, DC: Office of Planning, Research and Evaluation, Administration
for Children and Families, U.S. Department of Health and Human Services.
Harms, T., Clifford, R. M., & Cryer, D. (1998). Early Childhood Environment Rating Scale
Revised Edition. New York, NY: Teacher’s College Press.
945
Harms, T., Cryer, D., & Clifford, R. M. (2003). The Infant/Toddler Environment Rating
Scale—Revised Edition. New York, NY: Teacher’s College Press.
Harms, T., Cryer, D., & Clifford, R. M. (2007). Family Child Care Environment Rating
Scale Revised Edition. New York, NY: Teacher’s College Press.
Helburn, S. W. (1995). Cost, quality, and child outcomes in child care centers. Technical
report. Denver: Department of Economics, Center for Research in Economics and
Social Policy, University of Colorado at Denver.
Herzenberg, S., Price, M., & Bradley, D. (2005). Losing ground in early childhood education: Declining workforce qualifications in an expanding industry, 1979–2004.
Harrisburg, PA: Keystone Research Center.
Hestenes, L., Cassidy, D., Hegde, A., & Hansen, J. (2007). Quality in inclusive and
non-inclusive infant and toddler classrooms. Journal of Research in Childhood
Education, 22(1), 69–84.
Johnson, J. O. (2005). Who’s minding the kids? Child care arrangements: Winter 2002.
Washington, DC: U.S. Census Bureau.
Kontos, S., Howes, C., Shinn, M., & Galinsky, E. (1995). Quality in family child care and
relative care. New York, NY: Teacher’s College Press.
Koretz, D. M., McCaffrey, D. F., & Hamilton, L. S. (2001). Toward a framework for validating gains under high-stakes conditions [CSE technical report 551]. Los Angeles,
CA: Center for Research on Evaluation, Standards, and Student Testing.
Kryzer, E., Kovan, N., Phillips, D., Donagall, L., & Gunnar, M. (2007). Toddlers’ and preschoolers’ experiences in family day care: Age differences and behavioral correlates. Early Childhood Research Quarterly, 22(4),
451–466.
La Paro, K. M., Sexton, D., & Snyder, P. (1998). Program quality characteristics in
segregated and inclusive early childhood settings. Early Childhood Research Quarterly, 13(1), 151–168.
Layzer, J. I., & Goodson, B. D. (2006). Care in the home: A description of family child
care and the experiences of families and children that use it: National study for
low-income families, wave 1 report. Cambridge, MA: Abt Associates.
Le, V., Perlman, M., Zellman, G. L., & Hamilton, L. S. (2006). Measuring child–staff
ratios in child care centers: Balancing effort and representativeness. Early Childhood Research Quarterly, 21(3), 267–279.
Light, R. J., Singer, J. D., & Willett, J. B. (1990). By design: Planning research on higher
education. Cambridge, MA: Harvard University Press.
Loeb, S., Fuller, B., Kagan, S. L., Carrol, B., & Carroll, J. (2004). Child care in poor communities: Early learning effects of type, quality, and stability. Child Development,
75(1), 47–65.
MacCallum, R. C., Widaman, K. F., Preacher, K. J., & Hong, S. (2001). Sample size
in factor analysis: The role of model error. Multivariate Behavioral Research, 36,
611–637.
Marshall, N. (2004). The quality of early child care and children’s development.
Current Directions in Psychological Sciences, 13(4), 165–168.
NICHD ECCRN. (2000). Characteristics and quality of child care for
and
preschoolers.
Applied
Developmental
Science,
4(3),
toddlers
116–141.
NICHD ECCRN, & Duncan, G. J. (2003). Does quality of child care affect child outcomes
at age 4 21 ? Developmental Psychology, 39(3), 451–469.
Palsha, S. A., & Wesley, P. W. (1998). Improving quality in early childhood environments through on-site consultation. Topics in Early Childhood Special Education,
18(4), 243–253.
Paulsell, D., Porter, T., & Kirby, G. (2010). Supporting quality in home-based child care.
Final brief. Princeton, NJ: Mathematica Policy Research.
Peisner-Feinberg, E. S., Burchinal, M. R., Clifford, R. M., Culkin, M. L., Howes, C.,
Kagan, S. L., et al. (2001). The relation of preschool child-care quality to children’s cognitive and social developmental trajectories through second grade.
Child Development, 72, 1534–1553.
Perlman, M., Zellman, G., & Le, V. (2004). Examining the psychometric properties of the Early Childhood Environment Rating Scale—Revised. Early Childhood
Research Quarterly, 19(3), 398–412.
Pianta, R. C., La Paro, K., & Hamre, B. K. (2008). Classroom Assessment Scoring System.
Baltimore: Paul H. Brookes.
Pope, B. G., Denny, J. H., Homer, K., & Ricci, K. (2006). What is working? What is not
working? Report on the qualitative study of the Tennessee Report Card and StarQuality Program and Support System. Knoxville, TN: The University of Tennessee
College of Social Work, Office of Research and Public Service.
Raikes, H. A., Raikes, H., & Wilcox, B. (2005). Regulation, subsidy receipt and provider
characteristics: What predicts quality in child care homes? Early Childhood
Research Quarterly, 20(2), 164–184.
Sakai, L. M., Whitebook, M., Wishard, A., & Howes, C. (2003). Evaluating the
Early Childhood Environment Rating Scale (ECERS): Assessing differences
between the first and revised edition. Early Childhood Research Quarterly, 18(4),
427–445.
Scarr, S., Eisenberg, M., & Deater-Deckard, K. (1994). Measurement of quality in child
care centers. Early Childhood Research Quarterly, 9(2), 131–151.
Schaack, D., Tarrant, K., Boller, K., & Tout, K. (2012). Quality rating and improvement systems: Frameworks for early care and education systems change. In S.
L. Kagan, & K. Kaurez (Eds.), Early childhood systems: Transforming early learning
(pp. 71–86). New York, NY: Teacher’s College Press.
Setodji, C.M., Le, V., & Schaack, D. Using Generalized Additive Modeling to Empirically Identify Thresholds Within the ITERS in Relation to Toddlers’ Cognitive
Development. Developmental Psychology, in press.
Smith, M. L., & Fey, P. (2000). Validity and accountability in high-stakes testing.
Journal of Teacher Education, 51(5), 334–344.
Smith, S., Schneider, W., & Kreader, J. L. (2010). Features of professional development
and on-site assistance in child care quality rating improvement systems: A survey
946
D. Schaack et al. / Early Childhood Research Quarterly 28 (2013) 936–946
of state-wide systems. New York, NY: National Center for Children in Poverty,
Columbia University Mailman School of Public Health.
Snow, C. E., & Van Hemel, S. B. (2008). Early childhood assessment: Why, what, and
how. Washington, DC: National Academies Press.
Streiner, D. L., & Norman, G. R. (1989). Health Measurement Scales: A practical guide
to their development and use. New York, NY: Oxford University Press.
Tinsley, H. E. A., & Tinsley, D. J. (1987). Uses of factor analysis in counseling psychology research. Journal of Counseling Psychology, 34(4),
414–424.
Tout, K., Starr, R., Soli, M., Moodie, S., Kirby, G., & Boller, K. (2010, April). The child
care quality rating system (QRS) assessment: Compendium of quality rating systems
and evaluations, OPRE report. Washington, DC: Office of Planning, Research and
Evaluation, Administration for Children and Families, U.S. Department of Health
and Human Services.
Tout, K., Zaslow, M., Halle, T., & Forry, N. (2009). Issues for the next decade of quality rating and improvement systems. Washington, DC: Child Trends and the U.S.
Department of Health and Human Services, Administration for Children and
Families, Office of Planning, Research and Evaluation.
Truxillo, C. (2005, April). Maximum likelihood parameter estimation with incomplete data. In Proceedings of the thirtieth annual SAS(r) Users Group International
Conference Philadelphia, PA.
Whitebook, M., Howes, C., & Phillips, D. (1990). Who cares? Child care teachers and the
quality of care in America. Final report: National Child Care Staffing Study. Oakland,
CA: Child Care Employee Project.
Whitebook, M., & Ryan, S. (2011). Degrees in context: Asking the right questions about
preparing skilled and effective teachers of young children [Preschool Policy Brief,
22]. New Brunswick, NJ: National Institute for Early Education Research.
Whitebook, M., & Sakai, L. (2004). By a thread: How child care centers hold on to
teachers, how teacher build lasting careers. Kalamazoo, MI: The Upjohn Institute
for Employment Research.
Zellman, G. L., & Perlman, M. (2006). Parent involvement in child care settings:
Conceptual and measurement issues. Early Child Development and Care, 176(5),
521–538.
Zellman, G., Perlman, M., Le, V., & Setodji, C. M. (2008). Assessing the validity of
the Qualistar Early Learning Quality Rating and Improvement System as a tool for
improving child care quality. Santa Monica, CA: RAND Corporation.