For-profit/nonprofit differences in center-based child care quality: Results from the National Institute of Child Health and Human Development Study of Early Child Care and Youth Development

Abstract

In secondary analyses of National Institute of Child Health and Human Development Study of Early Child Care and Youth Development data, multiple indicators of quality (caregiver wages and turnover; child/staff ratio; caregiver education and professionalism; positive caregiving) were compared between child care centers by sector (for-profit/nonprofit) and subsector (for-profit independent/chain, nonprofit church/nonchurch) at multiple points from infancy through prekindergarten. Nonprofit centers evidenced higher caregiver wages and education at most ages and better quality child/staff ratios, turnover, caregiver professionalism, and positive caregiving for toddlers and preschoolers. Subsector differences in preschool classrooms were more complex. In general, quality was higher in nonprofit non-religiously affiliated centers, intermediate in nonprofit religiously affiliated and for-profit independent centers, and lower in for-profit chains, but differences were not found on every indicator or between every group. Further, for-profit chain status predicted lower quality positive caregiving, controlling for family characteristics, staff, and structural quality, at 54 months, but not 36 months. Results support and extend prior research by controlling for family characteristics. Policy implications regarding supply- and demand-side quality-improvement strategies that address market competition and parent choice across subsectors are discussed.

Full text

For-profit/nonprofit differences in center-based child care quality:
Results from the National Institute of Child Health and Human
Development Study of Early Child Care and Youth Development
☆
Laura Stout Sosinsky
a,
⁎
,1
, Heather Lord
a
, Edward Zigler
b,2
a
Department of Psychology, Yale University, USA
b
Yale Center in Child Development and Social Policy, Yale University, USA
Available online 20 July 2007
Abstract
In secondary analyses of National Institute of Child Health and Human Development Study of Early Child Care and Youth
Development data, multiple indicators of quality (caregiver wages and turnover; child/staff ratio; caregiver education and
professionalism; positive caregiving) were compared between child care centers by sector (for-profit/nonprofit) and subsector
(for-profit independent/chain, nonprofit church/nonchurch) at multiple points from infancy through prekindergarten. Nonprofit
centers evidenced higher caregiver wages and education at most ages and better quality child/staff ratios, turnover, caregiver
professionalism, and positive caregiving for toddlers and preschoolers. Subsector differences in preschool classrooms were more
complex. In general, quality was higher in nonprofit non-religiously affiliated centers, intermediate in nonprofit religiously affiliated
and for-profit independent centers, and lower in for-profit chains, but differences were not found on every indicator or between every
group. Further, for-profit chain status predicted lower quality positive caregiving, controlling for family characteristics, staff, and
structural quality, at 54 months, but not 36 months. Results support and extend prior research by controlling for family characteristics.
Policy implications regarding supply- and demand-side quality-improvement strategies that address market competition and parent
choice across subsectors are discussed.
© 2007 Elsevier Inc. All rights reserved.
Keywords: Child care quality; Child care centers; For-profit/nonprofit child care; Sector; Auspice; Structural quality; Process quality
1. Introduction
Child care is second only to parents and the immediate family as a primary early developmental context. More
children are experiencing ever-greater quantities of nonparental child care at ever-younger ages (National Institute of
Child Health and Human Development Early Child Care Research Network [NICHD ECCRN], 2006; Shonkoff &
Journal of Applied Developmental Psychology 28 (2007) 390 –410
☆
Results from this article were presented at the biennial meeting of the Society for Research in Child Development, April 7–10, 2005, Atlanta,
Georgia. Our appreciation goes to Sandra Bishop-Josef and Walter Gilliam at the Edward Zigler Center in Child Development and Social Policy, as
well as Sarah Friedman and the journal reviewers for their comments on earlier drafts of this article.
⁎Corresponding author. Yale Child Study Center, Yale School of Medicine, 230 South Frontage Road, New Haven, Connecticut 06520, USA.
E-mail address: laura.sosinsky@yale.edu (L.S. Sosinsky).
1
Laura Stout Sosinsky is now at the Yale Child Study Center, Yale School of Medicine.
2
The Yale Center in Child Development and Social Policy is now the Edward Zigler Center in Child Development and Social Policy.
0193-3973/$ - see front matter © 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.appdev.2007.06.003

Phillips, 2000). The use of center-based child care has increased, with one-third of the 12.2 million young children of
working mothers in 1999 in child care centers, up from 6% in 1965 (Helburn & Bergmann, 2002; Smolensky &
Gootman, 2003). Center-based care is more common among preschoolers than among infants and toddlers, ranging from
27% of 3-year-olds to more than half of 5-year-olds. Conversely, 17% of infants of working mothers and one-quarter of
2-year-olds were in centers in 1999 (Smolensky & Gootman, 2003). Most significantly, higher quality child care is
associated with better language development, cooperation, and school readiness, whereas poorer quality care is linked to
unfavorable child outcomes (Helburn, 1995; NICHD ECCRN, 1996, 2000, 2002, 2006; Vandell & Wolfe, 2000).
Given the impact on development and scope of use, much attention has focused on child care centers and the quality
of care they provide. Center-based child care in the United States is provided in a mixed-sector market of for-profit and
nonprofit providers. Providers in each sector may be independent or fall under the auspices of one of a variety of
organizations, ranging from religious or educational institutions to government agencies to corporations, which may
confer support and patronage to the center as well as prescribe control and approval of center operations. There is no
inherent reason why sector/subsector and quality should be related, and evidence indicates that high-quality centers can
be found in all sectors and subsectors of the market (Helburn, 1995). However, there is widespread concern that
subsector differences in clientele, funding sources, governing structures, and regulatory accountability may have
implications for program quality through a center's capacity or incentive to meet quality standards (Friesen, 1995;
Kagan, 1991; Morris & Helburn, 2000). Prior empirical examinations of quality by sector show evidence of higher
quality in nonprofit centers, but results are limited or inconsistent (Morris & Helburn, 2000).
The purpose of this article is to examine differences in center-based child care quality by for-profit and nonprofit
sector and subsector. To do so, we conduct a secondary analysis of data from the National Institute of Child Health
and Human Development Study of Early Child Care and Youth Development (NICHD SECCYD), the largest
observational study of early child care conducted to date. Several recent child care market and policy trends support the
value of such an investigation, which has at its core the goal of understanding the contributors to the quality of care
young children experience. First, most proposals for reforming the provision of child care in the United States assume
a continuation of mixed-sector providers. Current tax subsidy policies and political emphasis on parental choice
demonstrate an implicit reliance on market competition to demand quality from the market. There is concern, however,
that parents are ill-equipped to discern high-quality from low-quality care (Blau, 2001) and that consumer demand
alone cannot improve the supply of good-quality child care. In an era of rapid expansion of for-profit providers,
increasing privatization of care (Neugebauer, 2006), and more numerous exemptions for faith-based providers
(Henriques, 2006), continued examination of quality variations in the mixed-sector market is crucial.
1.1. The mixed-sector market and quality of care
1.1.1. For-profit centers
For-profit centers are estimated to constitute more than half of all child care centers across all ages of children served
(Helburn & Bergmann, 2002). Independent providers operate the majority of for-profit centers (about 70%), whereas
national chains operate the remainder. Independent providers, often termed mom and pop providers, are typically small
for-profit businesses owned and operated by an individual family. National chains (e.g., Bright Horizons Family
Solutions, Knowledge Learning Corporation) constitute 13% of all centers and 28% of for-profit centers (Helburn,
1995; Neugebauer, 2000). Combined, the top four largest of these chains had the capacity to care for almost half a
million children in 2005 (Neugebauer, 2006). Rapid economic expansion in the early 1990s created a competitive need
for employers of highly skilled workers to offer child care services as an employee benefit. Much of this demand is met
by outsourcing to for-profit chains, which have grown at a 25% annual rate. Bright Horizons Family Solutions, the
largest of these chains with 620 centers to date, specializes in employer-sponsored centers and is built around a business
model that considers employers, rather than parents, as the customer (Brown, 2001).Other chains market their services
more directly to parents (e.g., LaPetite Academy).For-profit centers have a competitive disadvantage compared with
nonprofits in so far as they cannot accept charitable contributions and have to carry facility costs that can be twice as
high as those of nonprofits (Helburn, 1995).
1.1.2. Nonprofit child care centers
About half of all centers across all child ages are nonprofits (Helburn & Bergmann, 2002). Nonprofit centers include
those operated by religious organizations, community agencies, private schools and colleges, cooperatives, nonprofit
391L.S. Sosinsky et al. / Journal of Applied Developmental Psychology 28 (2007) 390–410

independent providers, and public providers, as well as those affiliated with but not operated by religious organizations.
Religious organizations run approximately one-third of nonprofit centers (15% of all centers) (Helburn & Bergmann,
2002). Within the nonprofit sector, management objectives may be the most relevant point of subsector differences.
Morris and Helburn (2000) argue that the most common motivation underlying the management of most nonprofit
child care centers can be described as “altruistic,”but point out that it can also be disguised profit making for the benefit
of the sponsor. Furthermore, nonprofit centers may vary in the emphasis on goals such as provision of high-quality
services or production of low-cost, affordable services (Morris & Helburn, 2000).
1.1.3. Child care quality and sector
Researchers and professionals categorize characteristics of child care quality into (a) basic health and safety
indicators, (b) staffing indicators (e.g., wages and turnover rates), (c) structural features of the classroom and
characteristics of the caregivers (e.g., ratio of children to adults and caregivers' formal education and specialized
education), and (d) indicators of the experiences of the child in the child care setting, often called process quality and
characterized by the nature of the interactions of the child with caregivers and peers, children's participation in activities,
and experiences such as language stimulation (Cryer, 2003; NICHD ECCRN, 1996, 2002; Vandell & Wolfe, 2000).
Each child care quality characteristic has, to some degree, a unique association with other center characteristics, with
regulatory policies, and with child outcomes. Therefore, to understand sector quality variation, multiple indices of
quality must be considered. Positive, supportive, and stimulating interactions of caregivers with children support
positive child development, and the staffing, classroom, and caregiver characteristics of child care centers impact child
development both directly and indirectly by influencing caregiver interactions with children (NICHD ECCRN, 2002;
Vandell & Wolfe, 2000). Systematic variation in staff and structural indices is useful to understand, because these
features are most amenable to improvement and equalization through regulation, oversight, and workforce devel-
opment, and may have direct and indirect effects on child outcomes.
The importance of child care staff to quality is clear. The number and education of staff are quality indicators in and
of themselves and also affect the staff's ability to provide age-appropriate, developmentally stimulating interactions
with a group of young children. As labor is generally the single largest expenditure for a center, constituting 79% of
costs in nonprofit centers and 62% of costs in for-profit centers (Helburn & Howes, 1996), meeting staffing standards is
a major challenge for child care centers. For example, a for-profit center, which bears the full cost of facilities,
materials, and equipment and depends entirely on parent fees, may need to keep costs down or risk pricing themselves
out of the local market (Kagan, 1991), and achieve this goal by hiring staff with lower levels of education and training
who could be paid lower salaries with fewer benefits. Similarly, centers that aim to provide low-cost nonprofit care to
low-income families, as many church-operated organizations do, may hire less qualified (and less costly) caregivers.
Available evidence indicates that, overall, the majority of child care available in this country is of poor to mediocre
quality, with 10 to 20% of arrangements falling below levels deemed adequate (Helburn, 1995). Wages in child care
have been shown to be positively associated with classroom quality, yet wages are low (Phillips, Mekos, Scarr,
McCartney, & Abbott-Shim, 2000). Low wages also contribute to the staff turnover rates of 30–40% seen in child care
(Child Care Services Association, 2002), which can jeopardize the continued operation of a center, hinder a center's
effort to improve quality, and disrupt child–caregiver relationships (Hofferth, 1996; Phillips, Howes, & Whitebook,
1991; Helburn, 1995; Whitebook & Sakai, 2003).
By sector, the evidence that for-profit centers provide lower quality care than nonprofit centers is somewhat
inconsistent (Friesen, 1995; Helburn & Bergmann, 2002; Kagan, 1991; Morris & Helburn, 2000). Several studies have
found that for-profit care is of lower quality on several indicators when compared with nonprofit center care (Friesen,
1995; Phillips, Howes, & Whitebook, 1992; Whitebook, Howes, & Phillips, 1990). However, analysis of data from the
Cost, Quality, and Child Outcomes Study (CQO) revealed no differences in overall process quality by profit status,
except in the one state of the four included with low licensing standards (Helburn, 1995). This was despite finding
higher structural quality in nonprofit centers. Furthermore, no significant differences by profit status were found in
a follow-up CQO study examining quality characteristics that present an easier opportunity to be observed (i.e.,
furnishings, space, and personal grooming) versus quality characteristics that are harder to examine by those other than
trained child care observers (i.e., language stimulation, supervision, and free play) (Morris & Helburn, 2000).
A possible explanation for this equivocal pattern of associations is that the dichotomous categorization of centers
by profit status masks systematic quality differences by subsectors within the for-profit and nonprofit sectors. For
example, subsector differences in quality were found in the National Child Care Staffing Study (Phillips et al., 1992;
392 L.S. Sosinsky et al. / Journal of Applied Developmental Psychology 28 (2007) 390–410

Whitebook, Howes, & Phillips, 1998) on several indicators in classrooms serving a variety of age groups of children.
Turnover rates were lowest in nonprofit non-religiously affiliated centers and highest in for-profit independent centers.
Turnover rates of for-profit chains and nonprofit church-affiliated centers were in the middle, not different from each
other nor from the other group in their own sector. A similar pattern was observed with respect to ratios of children to
staff in infant and toddler classrooms, although in preschool classrooms the differences were completely along sector
lines, with better ratios in nonprofits than in for-profits regardless of subsector. Caregiver training in early childhood
was highest in nonprofit nonreligious centers, lower in nonprofit religious centers, and lower still in both types of for-
profit centers. Comparisons of quality in caregiving, such as sensitivity and harshness of caregivers and activities
provided to children, showed a consistent pattern across child age. The higher caregiving quality observed in nonprofit
non-religiously affiliated centers was significantly higher than that in for-profit independent centers at all comparisons,
but there were mixed results when it came to the other two subsectors and to quality levels in the middle range. In some
comparisons, the for-profit chains and nonprofit religiously affiliated centers did not differ from each other, from the
other subsector in their sector, or from any other group.
Findings from analyses of CQO data were slightly different. The lack of differences between for-profit and nonprofit
centers was attributed to the low quality in one nonprofit subsector, in that quality in religiously affiliated centers was no
different from that in for-profit centers (Helburn, 1995). The highest quality care was provided by child care centers
operated by public agencies, private schools, and cooperatives (Morris & Helburn, 2000). The lowest quality care was
provided by a mix of for-profit and nonprofit centers, namely, for-profit chains, for-profits in states with minimal
regulations, community agencies, and low-quality church-operated centers. Median quality care was provided by church-
affiliated (but not operated), nonprofit independent, high-quality church-operated, and for-profit independent centers.
These findings reflect the complexity of the implications for quality of the different legal and financial policy
structures that apply to each child care subsector. Most centers operate on extremely small margins of income and
earnings (Helburn & Bergmann, 2002), but differ in their legal ability to accept charitable contributions to supplement
parent fees, and differ in their acceptance of subsidies and other forms of payment (Whitebook et al., 1998). It is
possible that financial structures that allow distribution of profits to center owners may motivate for-profit center
management to maximize profits by cutting costs that will reduce quality. It is also possible that regulatory policies that
exempt some nonprofit faith-based centers from meeting licensing standards may open up the possibility of lower
quality as well. However, all centers need to compete for clientele to stay open, and may compete across subsector
lines. Therefore, the impact of market competition on quality may complicate the picture beyond the impact of
regulatory exemptions or profit motivation. The tightest competition may be among subsectors that serve mainly
unsubsidized children of middle-income working families. These subsectors are most reliant on parent fees to stay in
operation, but cannot charge fees too high to be competitive. Further, their clientele are the parents most in demand of
affordable services, as they are most reliant on their own comparably limited incomes to pay for child care. Centers
serving these families cross the for-profit/nonprofit divide, including for-profit chains and nonprofit independents and
church-operated and church-affiliated centers.
1.1.4. Family characteristics
The focus of the current study is on sector and subsector differences in child care center quality, rather than on the
differences in the families served by each sector and subsector. However, child care use is not random, and there
are often systematic differences in quality by the demographic and socioeconomic characteristics of the region and
of participating families. For example, unsubsidized children of middle-income working parents are more likely to
use for-profit chains, nonprofit nonpublic centers, or church-operated centers. Family income bears a curvilinear
association with quality in center-based child care (NICHD ECCRN, 1997), and some demographic risk factors such
as low maternal education and single parent-headed household are associated with use of lower quality child care
(Huston, Chang, & Gennetian, 2002; NICHD ECCRN, 2006). In the current study, we examine the possibility that
quality differences by sector/subsector may be confounded with systematic differences in the populations of families
served, extending prior research.
1.2. Research questions
In sum, there is some basis for the concern that the auspices of a child care center, which are central to the center's
day-to-day operations, staffing and enrollment policies, organizational structure, sources of revenue and expenditures,
393L.S. Sosinsky et al. / Journal of Applied Developmental Psychology 28 (2007) 390–410

clientele, and organizational objectives, may have a significant impact on the quality of care that each center provides to
young children. However, for-profit/nonprofit classifications may be too simple a typology when considering quality
of care. Instead, it is hypothesized that differences by subsectors may be a key explanatory factor. To retain the
differentiations of greatest importance, minimize the large number of subsectors, and use the simplest typology, we rely
on two primary approaches to classifying child care. Child care centers are first grouped dichotomously, as for-profit or
nonprofit, then categorized into one of four subsectors: (1) for-profit independents; (2) for-profit chains; (3) nonprofit
religiously affiliated or sponsored, referred to for simplicity as nonprofit church centers; and (4) nonprofit centers that
may be independent or sponsored by a larger agency or organization that is not religiously affiliated, referred to as
nonprofit nonchurch centers. Results indicating systematic quality differences by profit status and subsector would
both confirm prior findings (Friesen, 1995; Helburn, 1995; Kagan, 1991; Phillips et al., 1992; Whitebook, Sakai, &
Howes, 1997), expand the research by accounting for family selection biases and by examining subsector's unique
impact on child-level quality after controlling for family and other quality variation, and provide direction to future
research intended to investigate the mechanisms underlying such systematic differences.
The research questions addressed are (1) Do staffing quality indicators (wages and turnover), structural quality
characteristics (child/staff ratio, caregiver education and professionalism), and process quality (observed positive
caregiving) differ by a child care center's for-profit or nonprofit status at 6, 15, 24, 36, and 54 months, controlling for
family characteristics? (2) Do staffing, structural, and process quality characteristics differ by a child care center's
subsector at 36 and 54 months, controlling for family income and risk? (3) Does the subsector of a center retain a
unique association with process quality over and above family characteristics and the center's staff and structural
quality at 36 and 54 months?
2. Method
2.1. Overview and procedures of the NICHD SECCYD
Secondary analyses were conducted on data drawn from Phases I and II of the NICHD SECCYD, a large-scale,
comprehensive, prospective longitudinal study of the effects of early child care arrangements on children's devel-
opment. Mothers who had just given birth during selected intervals in hospitals near 10 geographically diverse sites
were contacted, and after a series of sample selection procedures were applied, a random sample of 1364 families were
enrolled. Although the study was not designed to be nationally representative, the sample was regionally, ethnically,
and socioeconomically diverse. Subject retention was high compared with other large-scale, multisite longitudinal
studies, although it varied somewhat across racial and ethnic categories. One thousand two hundred forty-eight families
(91%) were retained through the 15-month assessment, and 83% were retained at the 54-month assessment. For more
details, see NICHD ECCRN (2001a).
Mothers were interviewed by telephone when their children were 5, 14, 23, 34, and 53 months old about their
child's nonmaternal care arrangement(s). Arrangements were categorized by type, ordered from least to most formal:
mother care, father/partner care, relative care (in the child's home or not), in-home nonrelative care (e.g., nannies),
care in a family day care home, and center-based child care. An arrangement was considered eligible for observation
if used for at least 10 h per week. For children in more than one arrangement, only the primary arrangement was
eligible, defined as the setting in which the child spent the most time or, if the child spent equal time in two settings,
the more formal setting. Child care centers that were classified as the child's second arrangement were deemed
observable if the primary arrangement was parental. Parent permission and caregiver consent were obtained to assess
the child's observable child care arrangement. Two half-day observational visits were made to each child's primary
nonmaternal child care arrangement at 6, 15, 24, and 36 months of age; one visit was made at 54 months of age. Data
collection was conducted by research assistants trained at each site and certified and monitored centrally (NICHD
ECCRN, 2001a).
2.2. Participants
The current study includes all families whose child was in an observed center-based child care arrangement for
which sector data were available at each of the five assessment points. The numbers of included families at 6, 15, 24,
36, and 54 months are, respectively, 89, 79, 105, 245, and 495. Families whose child was in exclusive maternal care or
394 L.S. Sosinsky et al. / Journal of Applied Developmental Psychology 28 (2007) 390–410

other types of care were excluded. A small percentage of families could not be included at each assessment point due to
missing data on the center's sector. The families whose child's center-based arrangement was unobserved could
not be included. Reasons for not observing eligible arrangements included caregiver refusal, inability to schedule
observations within the appropriate window, child absence from setting, or recent changes in arrangement (NICHD
ECCRN, 2000). At some time points, child care arrangements of African-American children were less likely to be
observed and mothers in the observed sample had higher levels of education than mothers whose children's child care
arrangements were unobserved (NICHD ECCRN, 2000). The quality of the unobserved arrangements is unknown. As
reported elsewhere, the lowest quality child care settings are underrepresented in these data, and the overall quality of
care observed in the NICHD SECCYD is likely higher and less variable than overall quality of care across the United
States. This is due to sample recruitment procedures and because those parents or providers affiliated with the lowest
quality centers were less likely to be observed (NICHD ECCRN, 2001b). These issues will attenuate the range of the
independent and dependent variables, so that differences among subsectors may be underestimated, and estimated
effect sizes may be reduced.
2.3. Measures
2.3.1. Center for-profit/nonprofit sector and subsector
Center directors reported the profit status and subsector of their center in interviews at each assessment point. At 36
and 54 months, centers were categorized from these director reports as: (1) for-profit independent, (2) for-profit chain,
(3) nonprofit church, and (4) nonprofit nonchurch. Centers at 6, 15, and 24 months were not categorized by subsector
because of the small numbers in each cell.
2.3.2. Child care quality
2.3.2.1. Staffing characteristics
2.3.2.1.1. Monthly wages. Caregivers reported their wages at the hourly rate in U.S. dollars during interviews.
Responses were converted to monthly wages by multiplying by 173.81 (the average number of hours worked in a month).
2.3.2.1.2. Annual staff turnover. In interviews, center directors reported the number of paid classroom staff who
left the program in the last 12 months, including only teachers, assistant teachers and aides, teacher/directors, and any
other nonvolunteer staff who work directly with children. Annual staff turnover was computed as the number of staff
who quit in a year divided by the number of full-or part-time staff. Values range from 0 to 2.5, with higher values
denoting a higher turnover ratio.
2.3.2.2. Structural and caregiver quality
2.3.2.2.1. Child/staff ratio. At 6, 15, 24, and 36 months, up to four 44-minute observational cycles distributed
over 2 days were made of each child's center classroom. At 54 months, up to two 44-minute observational cycles were
made during one visit. Observations of the number of children and caregivers at the beginning and end of each cycle
were averaged over and across cycles, then the ratio of the averaged number of children to the averaged number of
adults was computed.
2.3.2.2.2. Caregiver education. Caregivers reported in interviews their level of education as falling into one of six
levels: (1) none, (2) high school training, (3) Child Development Associate degree (CDA), (4) some college, (5)
bachelor's degree, or (6) graduate degree.
2.3.2.2.3. Caregiver professionalism. Caregivers responded to several questions relating to their professionalism
in the field of early care and education. At 6, 15, 24, and 36 months, the composite variable is calculated as the sum of
items reflecting whether they were members of a professional organization, whether they had a preference for other
work (reverse-scored), the expected longevity of their of child care career, and whether they endorsed professional
reasons for caregiving. At 54 months, the composite also includes whether they held certification in early care and
education or child development; due to this minor variation in number of items, the 54-month variable was rescaled for
comparison to the earlier variables.
2.3.2.3. Process quality. The Observational Record of the Caregiving Environment (ORCE; NICHD ECCRN, 1996,
2000), developed by the NICHD SECCYD investigators and systematically adapted to be age appropriate for the
395L.S. Sosinsky et al. / Journal of Applied Developmental Psychology 28 (2007) 390–410

children being observed and for use in a variety of child care settings, was used to assess the quality of the center
caregiving environment at each assessment point. At 6, 15, 24, and 36 months, up to four 44-minute observational
cycles distributed over 2 days were made of each child's center classroom, and at 54 months up to two 44-minute
observational cycles were made during one visit to each classroom. During these ORCE cycles, trained observers
focused on the target child's behavior, activities, and interactions with other people. Qualitative ratings rate the
quality of the caregiver's behaviors in relation to the child's behaviors on a 4-point scale from 1 = not at all
characteristic to 4 = highly characteristic (Cronbach's α= .88). Interobserver reliability was good (.86 with “gold
standard”videotapes and .89 with other observers in the field). Validity was acceptable, as correlations with additional
behavioral frequency and global quality ratings ranged between .40 and .80.
A priori composite variables of qualitative ratings of caregiver's interactions with the target child were created by
the original study investigators at each assessment point. The variables, labeled positive caregiving, use slightly
different observed items at each assessment point. At 6, 15, and 24 months, the positive caregiving composite variable
was the sum of five items: sensitivity and responsiveness to nondistressed communication, characterized by prompt
and appropriate responses to the child's social signals; positive regard, or quality and quantity of expressions to the
child that connote the caregiver's positive feelings toward the child; stimulation of cognitive development, char-
acterized by the quality and frequency of caregiver effort to facilitate the child's cognitive development; detachment
(reverse-scored), characterized by the degree to which the caregiver is emotionally and physically uninvolved with the
child and unaware of the child's needs for appropriate interaction; and flatness of affect (reverse-scored), or the
frequency with which the caregiver lacks animation in facial and vocal expression and tone in interaction with the child
and/or with others. Cronbach's αat 6 months was .89; at 15 months, .88; and at 24 months, .87 (NICHD ECCRN,
1996, 2000). At 36 months, the positive caregiving composite variable was calculated as the sum of seven items,
including the five items employed in the infant and toddler composites plus fostering of child exploration, characterized
by the quality of opportunities for and encouragement of the child's exploration of objects and the environment, and
intrusiveness (reverse-scored), characterized by the degree to which the caregiver imposes her or his agenda on the
child as opposed to interacting in a way that provides the child with a sense of control (Cronbach's α= .83) (NICHD
ECCRN, 2000). At 54 months, the positive caregiving composite was the sum of four items: caregivers' sensitivity,
stimulation of cognitive development, intrusiveness, and detachment, with the latter two items reverse-scored
(Cronbach's α= .72) (NICHD ECCRN, 2003).
2.3.2.4. Family characteristics
2.3.2.4.1. Income-to-needs ratio. In interviews at each assessment, mothers reported maternal income and
nonmaternal income (all household income other than the mother's, including any father's/partner's earnings, unearned
income, and government benefits). Income-to-needs ratio, considered to be a reliable indicator of economic ease or
hardship faced by a family, was calculated as the total family income divided by the appropriate poverty threshold for
the household as determined by the U.S. Department of Labor.
2.3.2.4.2. Maternal education. In an interview 1 month following the child's birth, mothers reported their highest
level of education. Responses were categorized with values up to 12 indicating the number of years of education
completed, 12 indicating a high school diploma or equivalent, 14 indicating an associate's degree or vocational
school beyond high school or some college but no degree, 16 indicating a bachelor's degree, 18 indicating a master's
degree or some graduate work, 19 indicating a law degree, and 21 indicating a doctoral degree or more than a master's
degree.
2.3.2.4.3. Maternal minority status. In interviews 1 month postpartum, mothers reported their race falling into
one of five categories (American Indian/Eskimo/Aleutian, Asian/Pacific Islander, black or African-American, white, or
other) and ethnicity as being Hispanic or not. Variables were combined to create a dichotomous variable indicating
minority status, with the value of 0 indicating maternal self-identification as white non-Hispanic and the value of 1
indicating maternal self-identification as nonwhite and/or Hispanic.
2.3.2.4.4. Single-parent household. At each assessment point, mothers reported their marital and living status as
one of eight categories that were collapsed into a dichotomous variable, with a value of 0 indicating a two-parent
household and a value of 1 indicating a single-parent household. Mothers who reported being married or partnered
and living together were classified as not being in a single-parent household, whereas mothers who reported other
family compositions (i.e., separated and not living together, never married) were classified as being in single-parent
households.
396 L.S. Sosinsky et al. / Journal of Applied Developmental Psychology 28 (2007) 390–410

2.4. Data analysis plan
First, descriptive statistics were calculated and correlations computed to describe child care type usage overall, the
samples of observed child care centers by sector and subsector, and associations among child care variables and family
characteristics. Second, analyses of covariance (ANCOVAs) were conducted to compare quality on each of six
characteristics between for-profit to nonprofit center classrooms for five age groups (6, 15, 24, 36, and 54 months), with
family income, maternal education, maternal minority status, and household structure entered as covariates to control
for family selection biases and data collection site dummy variables entered as covariates to control for regional
differences. Cohen's dstatistics were computed for all significant group differences; when covariates were signif-
icantly correlated with the dependent variable in bivariate associations, dstatistics were computed using a formula that
takes into account the effect of covariates (Cohen, 1988; Cortina & Nouri, 2000). Third, a series of 12 ANCOVAs were
conducted to compare each quality characteristic among for-profit independent, for-profit chain, nonprofit church,
and nonprofit nonchurch centers at the two assessment points for which sufficient subsector data were available (36 and
54 months). Post hoc Bonferroni ttests were used to adjust for multiple comparisons. Again, family characteristics
and data collection site dummy variables were entered as covariates. Effect size estimates were not computed for these
four-group comparisons, as this type of omnibus analysis with two or more degrees of freedom in the numerator does
not reveal “where the action is”(McCartney & Rosenthal, 2000, p. 174). Finally, to examine whether a center's
subsector is uniquely associated with process quality, hierarchical multiple regressions were employed separately at 36
and 54 months, with the composite positive caregiving quality variable included as the dependent variable, and data
collection site dummy variables, family characteristics, staffing characteristics, structural characteristics, and dummy
codes for subsectors entered in four separate steps.
3. Results
3.1. Preliminary analyses
Statistics describing child care arrangements overall and the sector and subsector of center-based classrooms by
child age are summarized in Table 1. Center-based child care was used by 9% of the sample at 6 months and by more
than half of the sample at 54 months. At 6, 15, 24, and 36 months, children in center-based arrangements were about
Table 1
Frequencies and percentages of child care arrangements by child age
6 months 15 months 24 months 36 months 54 months
Type of primary arrangement
a
Parental 636 (48) 576 (45) 510 (41) 419 (34) 264 (23)
Center 117 (9) 143 (11) 218 (18) 369 (30) 616 (54)
Home-based 552 (42) 529 (42) 498 (40) 431 (35) 254 (22)
Other 11 (1) 21 (2) 13 (1) 10 (1) 2 (0)
Total n1316 1269 1239 1229 1136
Centers by sector
b
For-profit (FP) 41 (46) 40 (51) 59 (56) 118 (48) 156 (32)
Nonprofit (NP) 48 (54) 39 (49) 46 (44) 127 (52) 339 (68)
Centers by subsector
b
FP independent 90 (37) 123 (25)
FP chain 28 (11) 33 (7)
NP church 80 (33) 180 (36)
NP nonchurch 47 (19) 159 (32)
Total included 89 79 105 245 495
Note. Values are expressed as n(%).
a
Type of arrangement, parent report.
b
Sector/subsector of those center-based arrangements included in current study; centers without observational quality data and/or director-reported
sector/subsector data are not included.
397L.S. Sosinsky et al. / Journal of Applied Developmental Psychology 28 (2007) 390–410

evenly divided between for-profit and nonprofit child care centers. At 54 months, use of nonprofit centers accounted for
almost 70% of all center-based arrangements. Nonprofit nonchurch centers were also further subcategorized at 54
months: 30% were in the public sector (i.e., state or local government, public school, board of education); about 30%
were sponsored by a university, private school, corporation, or community organization; and the remaining centers
could not be further categorized as a group. As other research has found quality differences between public and other
nonprofit nonsectarian centers (Helburn, 1995), quality on all six variables was compared between these two groups.
Only two differences were found (caregivers in public centers earned higher wages and were more professional
compared to caregivers in nonpublic centers). To retain the entire subgroup sample and for clarity, therefore, these were
kept grouped together as nonprofit nonchurch centers.
Statistics describing overall means, standard deviations, and ranges of family characteristics and caregiver staff and
structural quality indices are summarized in Table 2. Correlations between family and child care variables within child
age group are listed in Table 3. Analyses of family characteristics indicated that the sample on average was not
particularly disadvantaged. The mean income-to-needs ratio was between 3.6 and 3.7 across the five assessment points,
with large standard deviations (between 3.0 and 3.2). As the income-to-needs ratio of a family at the poverty line would be
1.0 and that of a family at 200% of the federal poverty level would be 2.0, this is overall not a predominately low-income
sample. The mean educational level of mothers across those included at any assessment point was 14.7 (SD = 2.5),
indicating some college education for mothers on average; the range was wide, from 7 years of education through earned
doctoral or other advanced degrees. About 18% of mothers included at any assessment point were minority. The
percentage of households that were headed by a single parent ranged from 8 to 21% across the five assessment points.
Table 2
Descriptive statistics
1 month 6 months 15 months 24 months 36 months 54 months
Family income
M(SD) 5.3 (3.7) 4.5 (3.8) 4.5 (3.4) 4.5 (3.8) 4.1 (3.8)
Range .4–19.2 .3–20.4 .5–21.5 .1–28.5 .1–56.7
Maternal education
M(SD) 14.7 (2.5)
Range 7–21
Maternal minority status
n(%) 118 (18.2)
Single-parent household
n(%) 7 (8%) 12 (15%) 22 (21%) 43 (18%) 65 (13%)
Caregiver monthly wages (in dollars)
M(SD) 1064 (335) 1002 (238) 1050 (268) 1260 (543) 1777 (846)
Range 375–2129 739–1738 739–1856 621–4735 739–6605
Center turnover
M(SD) .4 (.3) .3 (.3) .3 (.3) .3 (.2) .2 (.3)
Range .0–1.2 .0–1.7 .0–1.5 .0–1.7 .0–2.5
Child/staff ratio
M(SD) 3.3 (1.8) 3.8 (1.5) 5.5 (1.9) 7.2 (2.4) 9.1 (4.0)
Range .9–12.7 1.6–8.6 1.6–12.4 2.0–13.4 1.4–30.0
Caregiver education
M(SD) 3.1 (1.1) 2.9 (.8) 3.1 (.9) 3.2 (.9) 3.6 (1.4)
Range 1.0–6.0 1.0–5.0 1.0–5.0 2.0–6.0 1.0–6.0
High school or less, n(%) 24 (25.6) 18 (23.2) 21 (22.0) 44 (20.7) 141 (29.0)
CDA, n(%) 37 (45.1) 39 (56.5) 42 (48.8) 102 (47.9) 42 (9.4)
Some college, n(%) 10 (12.2) 10 (14.5) 17 (19.8) 43 (20.2) 127 (28.3)
BA, n(%) 10 (12.2) 2 (2.9) 6 (7.0) 21 (9.9) 102 (22.8)
MA or doctorate, n(%) 1 (1.2) 0 (.0) 0 (.0) 3 (1.4) 36 (8.0)
398 L.S. Sosinsky et al. / Journal of Applied Developmental Psychology 28 (2007) 390–410

Table 3
Correlations between family and child care variables
12345678910
6 months
1 Family income
2 Maternal education .49⁎⁎
3 Maternal minority (0 = no, 1 = yes) −.14 −.26⁎
4 Single parent (0 = no, 1 = yes) −.32⁎⁎ −.38⁎⁎ .22⁎
5 CG wages .25⁎.25⁎−.09 −.17
6 Turnover .11 .08 .08 −.18 −.06
7 Child/staff ratio −.20 −.22⁎−.14 .10 −.31⁎⁎ −.02
8 CG education .03 .31⁎⁎ −.02 −.11 .61⁎⁎ −.07 −.22
9 CG professionalism .07 −.13 .10 .09 .12 .04 −.22⁎.07
10 Positive caregiving .02 .12 .11 .08 −.12 −.09 −.19 −.04 .20
11 Sector (0 = FP, 1 = NP) .06 .18 −.03 −.07 .29⁎−.03 −.07 .21 .12 −.04
15 months
1 Family income
2 Maternal education .51⁎⁎
3 Maternal minority (0 = no, 1 = yes) −.13 −.19
4 Single parent (0 = no, 1 = yes) −.35⁎⁎ −.39⁎⁎ .09
5 CG wages .22 .34⁎⁎ .02 −.21
6 Turnover .20 −.05 −.11 .08 −.06
7 Child/staff ratio −.11 −.22 −.11 .06 −.41⁎⁎ .07
8 CG education −.16 .11 −.05 .04 .36⁎⁎ −.08 −.03
9 CG professionalism .03 −.07 .04 −.04 .08 .01 −.25⁎.29⁎
10 Positive caregiving −.07 .10 −.04 .14 .01 −.04 −.26⁎.30⁎.15
11 Sector (0 = FP, 1 = NP) −.19 .07 .12 .19 .06 −.27 −.24 .55⁎⁎ −.03 .19
24 months
1 Family income
2 Maternal education .59⁎⁎
3 Maternal minority (0 = no, 1 = yes) −.12 −.18
4 Single parent (0 = no, 1 = yes) −.40⁎⁎ −.38⁎⁎ .08
5 CG wages .37⁎⁎ .37⁎⁎ .06 .19
6 Turnover −.00 .03 .00 −.07 −.04
7 Child/staff ratio −.13 −.26⁎−.06 .19 −.17 −.08
8 CG education .43⁎⁎ .36⁎⁎ −.11 −.16 .41⁎⁎ .12 −.27⁎
9 CG professionalism −.03 −.02 −.08 .05 .18 .11 −.37⁎⁎ .23⁎
10 Positive caregiving .14 .23⁎−.04 −.11 .15 .16 −.20 .03 .24⁎
11 Sector (0 = FP, 1 = NP) .09 .11 −.05 −.03 .12 −.24 −.16 −.08 .15 .38⁎⁎
36 months
1 Family income
2 Maternal education .50⁎⁎
3 Maternal minority (0 = no, 1 = yes) −.19⁎⁎ −.19⁎⁎
4 Single parent (0 = no, 1 = yes) −.32⁎⁎ −.23⁎⁎ .15⁎
5 CG wages .09 .10 −.06 −.03
6 Turnover .05 −.01 .07 −.09 −.24⁎⁎
7 Child/staff ratio .03 −.08 −.03 −.10 −.20⁎⁎ .04
8 CG education .14⁎.14⁎.00 −.07 .41⁎⁎ −.07 −.28⁎⁎
9 CG professionalism .07 −.02 −.04 .15⁎.20⁎⁎ .06 −.05 .26⁎⁎
10 Positive caregiving .01 .09 .02 −.09 .26⁎⁎ −.17⁎−.10 .01 .04
11 Sector (0 = FP, 1 = NP) −.05 −.07 .07 .08 .14⁎−.14⁎−.29⁎⁎ .13 .06 .10
54 months
1 Family income
2 Maternal education .46⁎⁎
3 Maternal minority (0 = no, 1 = yes) −.17⁎⁎ −.22⁎⁎
4 Single parent (0 = no, 1 = yes) −.22⁎⁎ −.23⁎⁎ .21⁎⁎
(continued on next page)
399L.S. Sosinsky et al. / Journal of Applied Developmental Psychology 28 (2007) 390–410

Caregiver monthly wages ranged widely, but overall were low. Mean annual staff turnover was similar to that found
in other studies. Mean levels of child/staff ratios typically met professional recommendations, but encompassed a
wide range including some ratios that were far higher than recommended by professional early care and education
associations. Caregiver education also ranged widely, with the typical level at each age being a CDA.
Income-to-needs ratio and other family characteristics generally did not differ significantly between families
using for-profit and nonprofit centers. The only exceptions were in regard to minority status at 36 months and
income, maternal education, and minority status at 54 months. At the 36-month assessment point, mothers who
were of minority race or ethnicity were more likely to have their children in nonprofit nonchurch centers than in
other subsectors, χ
2
(3) = 8.53, pb.05. At 54 months, families who used nonprofit centers had significantly lower
incomes than families using other subsectors, F(1, 482) = 11.14, pb.01, d= .33, and mothers whose children were
in for-profit centers had higher levels of education, F(1, 484) = 13.06, pb.001, d= .38. Across the four subsectors
at 54 months, families using nonprofit nonchurch programs had significantly lower incomes compared with all
other groups (F(3, 444) = 7.90, pb.001). Mothers of children in for-profit independent centers had the highest
levels of education compared with all other subsectors, followed by mothers with children in nonprofit church-
affiliated centers, whose education in turn was significantly higher than that of mothers with children in either for-
profit chains or nonprofit nonchurch centers; the latter two subsectors did not differ with respect to maternal
education. Similar to 36 months, mothers with children in nonprofit nonchurch centers were more likely to be
minority than in other subsectors, χ
2
(3) = 42.30, pb.001. All income and risk comparisons controlled for data
collection site.
3.2. Main analyses
3.2.1. Comparison of quality by sector
To address the first research question regarding differences in quality characteristics between for-profit and
nonprofit centers at 6, 15, 24, 36, and 54 months, a series of ANCOVAs were conducted controlling for family
characteristics and data collection site. Results are summarized in Table 4.
3.2.1.1. Staffing quality characteristics. Caregivers' reported monthly wages were significantly higher in nonprofit
child care centers than in for-profit centers at 6 months (mean difference: $180; F(1, 62) = 6.84, pb.05, d=.57), 24
months (mean difference: $110; F(1, 68) = 5.48, pb.05, d= .45), 36 months (mean difference: $190; F(1, 187) = 6.37,
pb.05, d= .26), and 54 months (mean difference: $357, F(1, 386) = 16.41, pb.001, d=.38). Annual turnover rates
were statistically significantly higher in for-profit compared with nonprofit centers at 24 and 36 months, with trends in
the same direction at 15 and 54 months. Specifically, the mean turnover rate in for-profits was .13 higher at 24 months,
F(1, 80) = 5.12, pb.05, d=.55, and .07 higher at 36 months, F(1, 210) = 4.23, pb.05, d= .28.
3.2.1.2. Structural quality characteristics. Structural characteristics of nonprofit center classrooms were either of
higher quality than or no different from those of for-profit centers. Significantly lower child/staff ratios were observed
in toddler and preschool nonprofit center classrooms as compared with for-profit centers. Specifically, at 24 months,
nonprofit classrooms had, on average, one fewer child per staff member than for-profit classrooms (4.9:1 vs. 5.8:1;
F(1, 69) = 6.81, pb.05, d= .50); at 36 months, nonprofit classrooms had almost one and a half fewer children per
staff member when compared with their for-profit counterparts (6.6:1 vs. 7.9:1; F(1, 199) = 18.55, pb.001, d=.58);
Table 3 (continued)
12345678910
5 CG wages −.04 .01 −.00 −.08
6 Turnover −.01 −.04 .00 .12⁎⁎ −.25⁎⁎
7 Child/staff ratio .02 .02 −.06 −.05 −.04 −.04
8 CG education −.12⁎.09 .00 .10⁎.38⁎⁎ −.09 −.03
9 CG professionalism .10⁎.06 .03 −.00 .27⁎⁎ −.04 −.04 .36⁎⁎
10 Positive caregiving .14⁎⁎ .12⁎−.05 −.11⁎.10⁎.11⁎−.21⁎⁎ .12⁎.13⁎⁎
11 Sector (0 = FP, 1 = NP) −.15⁎⁎ −.18⁎⁎ .06 .03 .17⁎⁎ −.10⁎−.23⁎⁎ .13⁎⁎ .02 .13⁎⁎
Note. CG = caregiver; FP = for-profit; NP = nonprofit.
⁎pb.05. ⁎⁎pb01.
400 L.S. Sosinsky et al. / Journal of Applied Developmental Psychology 28 (2007) 390–410

and at 54 months, nonprofit classrooms had, on average, two fewer children per staff member than for-profit
classrooms (8.5:1 vs. 10.5:1; F(1, 470) = 26.56, pb.001, d= .48). Ratios in classrooms for 6- and 15-month-olds
were not significantly different by profit status. Caregivers in nonprofit classrooms had significantly higher levels of
education compared with caregivers in for-profit classrooms at all points except at the 15-month assessment: at 6
months, F(1, 66) = 3.60, pb.05, d= .50; at 24 month, F(1, 66) = 9.73, pb.01, d= .30; at 36 months, F(1, 188) = 5.08,
pb.05, d= .33; and at 54 months F(1, 424) = 8.80, pb.01, d= .30. Caregivers' levels of professionalism were higher
in nonprofit classrooms than in for-profit classrooms at 24 months, F(1, 66) = 6.75, pb.05, d= .42; differences in
caregivers' professionalism by profit status were not observed at 6, 15, 36, and 54 months.
Table 4
Estimated marginal means of child care quality of for-profit and nonprofit child care centers by child age
Wages (in dollars) Turnover Ratio
Est. M(SE)dEst. M(SE)dEst. M(SE)d
6 months
FP (n= 41) 973 (47)
a
.57 .39 (.05)
a
3.4 (.3)
a
NP (n= 48) 1153 (45)
b
.33 (.05)
a
3.2 (.2)
a
15 months
FP (n= 40) 984 (38)
a
.40 (.05)
a
4.0 (.2)
a
NP (n=39) 1029 (38)
a
.25 (.05)
a
3.6 (.2)
a
24 months
FP (n= 59) 1007 (30)
a
.45 .33 (.04)
a
.55 5.8 (.2)
a
.50
NP (n= 46) 1117 (35)
b
.20 (.04)
b
4.9 (.3)
b
36 months
FP (n= 118) 1165 (53)
a
.26 .28 (.02)
a
.28 7.9 (.2)
a
.58
NP (n= 127) 1355 (51)
b
.21 (.02)
b
6.6 (.2)
b
54 months
FP (n= 156) 1532 (73)
a
.38 .22 (.02)
a
10.5 (.3)
a
.48
NP (n= 339) 1889 (51)
b
.17 (.02)
a
8.5 (.2)
b
Education Professionalism Positive caregiving
Est. M(SE)dEst. M(SE)dEst. M(SE)d
6 months
FP (n= 41) 2.8 (.2)
a
.50 8.8 (.4)
a
13.4 (.5)
a
NP (n= 48) 3.3 (.2)
b
9.3 (.4)
a
13.5 (.4)
a
15 months
NP (n= 39) 2.9 (.1)
a
9.3 (.4)
a
12.7 (.4)
a
FP (n= 40) 2.9 (.1)
a
8.7 (.3)
a
13.9 (.5)
a
24 months
FP (n= 59) 2.9 (.1)
a
.30 8.6 (.3)
a
.42 12.3 (.3)
a
.87
NP (n=46) 3.4 (.1)
b
9.8 (.3)
b
14.1 (.4)
b
36 months
FP (n= 118) 3.1 (.1)
a
.33 9.1 (.2)
a
19.0 (.3)
a
NP (n= 127) 3.4 (.1)
b
9.3 (.2)
a
19.7 (.3)
a
54 months
FP (n= 156) 3.3 (.1)
a
.30 9.7 (.1)
a
11.5 (.2)
a
.38
NP (n= 339) 3.8 (.1)
b
9.8 (.1)
a
12.4 (.2)
b
Note. FP = for-profit; NP = nonprofit. Estimated marginal means take into account the contribution of data collection site, family income, maternal
education, maternal minority status, and single parent-headed household as covariates. The positive caregiving composite variable was computed
slightly differently at each assessment point. The possible range of values at 6, 15, and 24 months is 5–20; at 36 months, 7–28; and at 54 months, 4–16.
a,b
Means with different superscripts differ significantly, pb.05. Means with the same superscript are not significantly different, pb.05.
401L.S. Sosinsky et al. / Journal of Applied Developmental Psychology 28 (2007) 390–410

3.2.1.3. Positive caregiving. Levels of positive caregiving were observed to be higher in nonprofit center class-
rooms compared with for-profit center classrooms at 24 months, F(1, 69) = 13.48, pb.001, d=.87,and54months,
F(1, 470) = 15.51, pb.001, d= .38, with a trend in the same direction at 15 months. No differences in positive
caregiving were observed at 6 and 36 months.
3.2.2. Comparison of quality by subsector
To address the second research question regarding differences in quality characteristics by subsector (e.g., for-profit
independents/chains, nonprofit church/nonchurch centers) at 36 and 54 months, a series of ANCOVAs were conducted
controlling for family characteristics and data collection site. Results are summarized in Table 5 and Fig. 1.
3.2.2.1. Staffing quality characteristics. Caregivers who worked in nonprofit nonchurch child care centers reported
significantly higher monthly wages compared with caregivers in all other subsectors at 36 months, F(3, 185) = 7.34,
pb.001, and 54 months, F(3, 384) = 15.25, pb.001. The other three subsectors did not differ from one another at 36
months, but did at 54 months. Specifically, at 54 months wages reported by caregivers in nonprofit church centers were
higher than those reported by caregivers in for-profit chains, though not significantly higher than those of caregivers
in for-profit independent centers, and caregivers' wages did not differ between for-profit independent and non-
profit church-affiliated centers. Annual turnover rates were statistically significantly higher in for-profit chain centers
compared with all other subsectors at both 36 months, F(3, 208) = 4.79, pb.01, and 54 months, F(3, 427) = 4.88,
pb.01. Levene's tests of equality of error variances were significant at pb.05 for caregiver wages and turnover at
both 36 and 54 months, demonstrating that variability about the estimated means was not equal across groups.
3.2.2.2. Structural quality characteristics. Differences in structural quality characteristics across subsectors were
complex. At 36 months, child/staff ratios differed across subsector, F(3, 197) = 8.46, pb.001. Post hoc analyses using
Bonferroni ttests indicated that child/staff ratios in nonprofit nonchurch classrooms (5.9:1) were significantly lower
than those in both for-profit subsectors (7.8:1 in for-profit independents and 8.1:1 in for-profit chains). Nonprofit
church center classroom child/staff ratios were in the middle (7.0:1) and not significantly different from ratios in any
other subsector. Child/staff ratios in the two for-profit subsectors were not statistically different from one another.
Child/staff ratios differed across subsector at 54 months, F(3, 432) = 9.20, pb.001. Ratios were lowest in nonprofit
nonchurch classrooms (8.3:1), which were not statistically different from ratios in nonprofit church-affiliated
classrooms (8.8:1) but were statistically lower than ratios in the two for-profit subsectors. In turn, nonprofit church
classroom ratios were lower than those of for-profit independent classrooms (10.5:1), but not statistically different from
ratios of for-profit chains (10.6:1). As at 36 months, child/staff ratios in the two for-profit subsectors were similar to
one another.
Table 5
Estimated marginal means of child care quality characteristics among preschool classrooms by subsector
Wages (in dollars) Turnover Ratio Education Professionalism Positive caregiving
36 months
FP independent (n=90) 1212 (60)
a
.24 (.03)
a
7.8 (.3)
a
3.1 (.1)
a
9.3 (.2)
ab
19.1 (.4)
a
FP chain (n= 28) 1037 (102)
a
.40 (.05)
b
8.1 (.4)
a
3.0 (.2)
ab
8.5 (.4)
a
19.0 (.6)
a
NP church (n=80) 1231 (60)
a
.22 (.03)
a
7.0 (.3)
ab
3.3 (.1)
ab
9.0 (.2)
a
19.5 (.4)
a
NP nonchurch (n= 47) 1609 (86)
b
.19 (.04)
a
5.9 (.4)
b
3.6 (.2)
b
10.0 (.3)
b
20.0 (.5)
a
54 months
FP independent (n=123) 1628 (79)
a,b
.19 (.02)
a
10.5 (.4)
a
3.4 (.1)
a
9.8 (.2)
a,b
11.8 (.2)
a,b
FP chain (n= 33) 1184 (150)
a
.34 (.05)
b
10.6 (.7)
a,b
3.0 (.3)
a
9.1 (.3)
a
10.7 (.4)
a
NP church (n=180) 1684 (66)
b
.15 (.02)
a
8.8 (.3)
b,c
3.4 (.1)
a
9.6 (.1)
a
12.5 (.2)
b
NP nonchurch (n= 159) 2152 (76)
c
.20 (.02)
a
8.3 (.3)
c
4.2 (.1)
b
10.2 (.2)
b
12.5 (.2)
b
Note. FP = for-profit; NP = nonprofit. Estimated marginal means take into account the contribution of data collection site, family income, maternal
education, maternal minority status, and single parent-headed household as covariates. The positive caregiving composite variable was computed
slightly differently at each assessment point. The possible range of values at 36 months is 7–28, and at 54 months, 4–16.
a–c
Means with different superscripts differ significantly, pb.05, in the Bonferroni ttest. Means with the same superscript are not significantly
different, pb.05 in the Bonferroni ttest.
402 L.S. Sosinsky et al. / Journal of Applied Developmental Psychology 28 (2007) 390–410

Fig. 1. Estimated marginal means
1
of child care quality characteristics among preschool classrooms by subsector. (A) Caregiver monthly wages.
(B) Center annual turnover rate. (C) Child/staff ratio. (D) Caregiver education. (E) Caregiver professionalism. (F) Positive caregiving.
1
Estimated
marginal means take into account the contribution of data collection site, family income, maternal education, maternal minority status, and single
parent-headed household as covariates.
2
The positive caregiving composite variable was computed slightly differently at each assessment point.
The possible range of values at 36 months is 7–28, and at 54 months, 4–16.
a–c
Means with different letters differ significantly, pb.05. Means with
the same letter are not significantly different at pb.05.
403L.S. Sosinsky et al. / Journal of Applied Developmental Psychology 28 (2007) 390–410

Caregivers in nonprofit nonchurch center classrooms at 36 months had significantly higher levels of education
compared with caregivers in for-profit independent centers. Education levels of caregivers in nonprofit church-
affiliated and for-profit chains were the same compared with each other and all other groups, F(3, 186) = 3.04, pb.05.
At 54 months, education levels were highest among caregivers in nonprofit nonchurch centers compared with all other
subsectors, which were the same as one another, F(3, 422) = 10.86, pb.001.
With respect to caregiver professionalism, at both preschool assessment points, caregivers in nonprofit nonchurch
centers had significantly higher levels than caregivers in all other subsectors except for-profit independent centers.
Caregivers in the other three subsectors (for-profit independents and chains and nonprofit church-affiliated centers) did
not differ from one another with respect to professionalism at both 36 months, F(3, 186) = 3.88, pb.05, and 54
months, F(3, 431) = 4.78, pb.01.
3.2.2.3. Positive caregiving. Observed positive caregiving differed across subsector at 54 months but not at 36
months, F(3, 432) = 8.25, pb.001. At 54 months, levels of positive caregiving were lower in for-profit chains than in
the two nonprofit subsectors, though not significantly lower than in for-profit independent centers. This latter group
was in the middle, and not statistically different in positive caregiving from any other subsector.
3.2.3. Unique contribution of subsector to process quality
A series of hierarchical multiple regressions were used to address the final research question of whether center
subsector contributed to the prediction of process quality beyond the contributions of family characteristics, staffing,
Table 6
Hierarchical multiple regression of center subsector on positive caregiving in 54-month classrooms controlling for staffing, structural, and caregiver
characteristics stepwise changes (dependent variable: 54-month positive caregiving composite)
Model step R
2
Adj. R
2
R
2
-ch F-ch
(df)
p
Step 1. Data collection site
a
.075 .053 .075 3.44
(9, 381)
b.001
Step 2. Family characteristics .091 .060 .016 1.69
(4, 377)
.151
Step 3. Staffing characteristics .115 .079 .023 4.91
(2, 375)
.008
Step 4. Structural quality characteristics .167 .127 .052 7.77
(3, 372)
b.001
Step 5. Center subsector .193 .147 .026 4.02
(3, 369)
.008
Standardized model coefficients on entry and in final step
Model and reference group Variable
a
β
entry
β
final
Income-to-needs ratio .08 .08
Maternal education .05 .05
Maternal minority status −.04 −.04
Single-parent household −.02 −.03
Caregiver wages .14⁎⁎ .05
Center staff turnover −.05 −.03
Child/staff ratio −.16⁎⁎ −.13⁎
Caregiver education .05 .05
Caregiver professionalism .16⁎⁎ .16⁎⁎
A. For-profit independent For-profit chain −.12⁎
Nonprofit church .12
Nonprofit nonchurch .03
B. For-profit chain For-profit independent .20⁎
Nonprofit church .34⁎⁎
Nonprofit nonchurch .24⁎
C. Nonprofit church For-profit independent −.11
For-profit chain −.18⁎⁎
Nonprofit nonchurch −.08
D. Nonprofit nonchurch For-profit independent −.03
For-profit chain −.14⁎
Nonprofit church .08
Note. ⁎pb.05. ⁎⁎pb.01.
a
Standardized coefficients are not presented for individual site dummy variables in Step 1.
404 L.S. Sosinsky et al. / Journal of Applied Developmental Psychology 28 (2007) 390–410

and structural quality. The dependent variables were the positive caregiving composite variables observed at 36 and 54
months. Data collection site dummy variables were entered in Step 1, family characteristics were entered together
in Step 2 (income-to-needs ratio, maternal education, maternal minority status, and single-parent household status),
two staffing quality characteristics were entered together in Step 3 (wages and turnover), three structural quality
characteristics were entered as a group in Step 4 (child/staff ratio, caregiver education, and caregiver professionalism),
and dummy variables coding the four subsector groups were entered in Step 5. Four models were analyzed in the final
steps to examine all possible comparisons of the four subsector groups. The reference group in Model A was for-profit
independent centers, the reference group in Model B was for-profit chains, the reference group in Model C was
nonprofit church centers, and the reference group in Model D was nonprofit nonchurch centers.
Findings at 36 months are presented in text only. At 36 months, only two model steps significantly contributed to the
prediction of variance in positive caregiving. After data collection site (R
2
= .12, F(9, 184) = 2.78, pb.01), the third
model step in which staffing characteristics were entered predicted an additional 6% of the variance in positive
caregiving (R
2
= .19, R
2
= .06, F(1, 178) = 6.70, pb.01). Specifically, caregiver wages positively contributed to
positive caregiving, β= .20, pb.05.
Table 6 displays stepwise changes in prediction of overall process quality at 54 months and the standardized
regression coefficients on entry and in the final step. At 54 months, all model steps except family characteristics
significantly contributed to variance in positive caregiving. Data collection site predicted 8% of the variance in positive
caregiving. In the third model step, staff characteristics predicted an additional 2% of variance in positive caregiving.
Specifically, higher caregiver wages contributed positively to positive caregiving, β= .14, pb.01. Structural quality
characteristics significantly improved prediction of process quality by 5%. Specifically, on entry, child/staff ratio was
a significant negative predictor of process quality, β=−.16, pb.01, and caregiver professionalism was a significant
positive predictor of process quality, β= .16, pb.01.
The addition of subsector dummy variables in the final step of the 54-month model significantly improved
prediction of process quality an additional 3%. All five steps together accounted for a total of 15% of the variance in
process quality. Coefficients indicate that status as a for-profit chain accounted for the contribution of a center's
subsector to prediction of process quality after accounting for the contribution of site, family, staff, and structural
characteristics. In Model A, compared with for-profit independent center status, for-profit chain status predicted lower
quality in positive caregiving, β=−.12, pb.05. Also, there was a trend toward status as a nonprofit church center
being associated with higher positive caregiving quality compared with status as a for-profit independent. In Model B,
coefficients indicate that, compared with being a for-profit chain center, being in any of the other three subsectors
significantly predicted higher process quality, β
FP independent
= .20, pb.05, β
NP church
= .34, pb.01, β
NP nonchurch
= .24,
pb.05. In Model C, compared with nonprofit church status, for-profit chain status predicted lower positive
caregiving levels, −.18, pb.01, and there was a trend toward lower quality in for-profit independent centers. Finally, in
Model D, compared with nonprofit nonchurch centers, being a for-profit chain predicted lower levels of positive
caregiving, β=−.14, pb.05.
4. Discussion
4.1. Summary of findings
Differences in quality found between nonprofit and for-profit child care centers in the current secondary analyses of
data from the NICHD SECCYD by and large supported prior research findings that nonprofit center quality is generally
higher (e.g., Friesen, 1995; Phillips et al., 1992; Whitebook et al., 1990, 1997). Differences were not found on every
indicator of quality at each of the five age points. However, when found, significant group differences were consistently
in the direction of higher quality care provided by nonprofit centers compared with for-profit centers. Further
breakdown of centers by subsector (for-profit chain and independent centers, nonprofit religiously and non-religiously
affiliated centers) revealed a complementary but more complex pattern of differences.
By sector, caregiver wages and caregiver education were higher in nonprofit classrooms than in for-profit
classrooms serving four of five age groups studied. In classrooms serving children 2 years old or older, caregivers in
nonprofit centers had fewer children per adult to care for than their counterparts in for-profit classrooms. Turnover rates
were significantly lower in nonprofit centers at 24 and 36 months, and caregiver professionalism was significantly
higher in nonprofit toddler classrooms. Caregiver interactions with children were observed to be more positive in
405L.S. Sosinsky et al. / Journal of Applied Developmental Psychology 28 (2007) 390–410

nonprofit classrooms than for-profit classrooms for toddlers and older preschoolers. The magnitude of these differ-
ences, when found, ranged from medium to large (Cohen, 1988).
By subsector, wages were highest for caregivers in nonprofit nonreligious centers at both preschool assessment
points. Additional subsector differences in wages were found at 54 months, with for-profit independent and nonprofit
church centers occupying a middle ground. Turnover rates were highest (poorest) in for-profit chain centers at both
points as well. Child/staff ratios and caregiver education levels showed a variable ordering across subsectors, with
nonprofit non-religiously affiliated centers generally high, but typically not significantly higher than nonprofit church-
affiliated centers and, in one comparison, not higher than one for-profit subsector. Church-affiliated nonprofits
provided intermediate-quality ratios and caregiver education levels, which mostly were not different from those of
other nonprofits but also not different from those of one or both for-profit groups. Regarding the professionalism of
caregivers, for-profit independent centers were in the middle but similar, similar to both the higher levels found in
nonprofit nonreligious centers and to the lower levels found in for-profit chains and nonprofit church centers. Quality
of caregiver interactions with children at 54 months (though not at 36 months) was higher in both nonprofit subsectors
compared with for-profit chains, but again, for-profit independent centers stood midway, similar to all other subsectors.
These findings are similar in some ways to findings from the Cost, Quality, and Child Outcomes study and the National
Child Care Staffing Study (Phillips et al., 1992), in that nonprofit nonsectarian centers were generally of the highest
quality but results were otherwise mixed.
The current findings extend prior research on mixed-sector child care center quality by addressing whether
subsector classification is uniquely associated with the positive caregiving children directly experience, after con-
trolling for subsector differences in family characteristics, staff wages and turnover, and structural quality. Indeed,
evidence was found for a unique effect of subsector, namely that categorization as a for-profit chain was associated with
lower positive caregiving compared with categorization as any other subsector. This was found in the older preschool
classrooms but not the younger, which parallels the lack of subsector differences in positive caregiving at 36 months
before controlling for staff and structural quality.
The fact that sector/subsector differences were not found in every comparison has two general implications for the
relationship of profit status to quality of care provided for children of different ages. The lack of significant differences
on most quality indicators in infant classrooms may be explained by limited statistical power to detect differences
because of smaller sample sizes, lower overall quality in infant rooms, as suggested by other studies (Helburn, 1995),
or a difference in the types of centers that do and do not serve infants. Nonprofit centers are less likely to offer
infant care in general. Infant care is the most expensive to deliver because of the very low child/staff ratios required.
Examination of policies to ensure the delivery of high-quality care for infants across sectors is an important issue for
further research.
It is possible the lack of process quality differences in 3-versus 4-year-old preschool classrooms is due to the slight
difference in the construction of the positive caregiving variable or other methodological issues, or that process quality
differences may not be as robust in 3-year-old classrooms. It may also be that quality differences across subsector
observed in 4-year-old classrooms are not as detectable in 3-year-old classrooms because of a shift in the proportions of
classrooms in each sector by child age. There is a large increase in nonprofit service provision and use for 4-year-olds
compared with services for younger children. As staff and structural quality is higher in these types of nonprofit centers
in this study and others (Helburn, 1995; Phillips et al., 1992;) and support higher process quality (NICHD ECCRN,
2002), the influx of classrooms in this subsector at this age may increase the ability to detect differences in the more
subtle aspects of quality of care provided in caregivers' interactions with children.
The findings suggest that, of the four subsectors, for-profit chains were often lower in quality and never highest in
quality (though occasionally were the same). As the for-profit chain group was much smaller than the other three
subsectors, future research on this topic would benefit from a broader and larger sampling of for-profit chain child care
centers. However, the fact that the current findings are consistent, though not definitive, suggests that there might be
something unique in the service delivery in for-profit chains that affects quality. For example, there may be unmeasured
differences between for-profit chains and other subsectors in management and hiring practices, availability of
resources, unexamined variations in populations served, or other center features. It may also be that quality is
influenced by a confluence of factors, with the sum of the parts being greater than the whole. It may be that the
comparatively lower wages, higher turnover, lower levels of educational preparation and professional commitment,
and larger numbers of children to care per adult all found in for-profit chains may combine to have a dampening effect
on caregivers' positive interactions with children beyond the direct effects of each.
406 L.S. Sosinsky et al. / Journal of Applied Developmental Psychology 28 (2007) 390–410

Furthermore, child care quality in centers is very heterogeneous, and may be especially so in for-profit chains. This
was clear with respect to caregiver wages and turnover rates and may play a role in other quality indicators as well.
Certainly there are for-profit chains that provide high-quality child care services (e.g., Bright Horizons Family
Solutions). In the current secondary analyses of data from the NICHD SECCYD, specific national for-profit child care
companies could not be compared individually because of potential violations of participant confidentiality agree-
ments, not to mention that further partitioning of the already small subsamples would severely limit power to detect
differences. However, we are in a current period of increasing privatization of child care services and increasing
outsourcing of child care benefits to national chains by corporations and government agencies. Future comparison
across partitions such as private versus public clients of for-profit chains, worksite versus freestanding centers, and
centers that do and do not receive substantial public subsidy funds would update and extend understanding of these
divisions (some such comparisons were conducted in the Cost, Quality, and Outcomes Study) (Helburn, 1995).
Moreover, understanding of the range of variation in quality within subsectors and the correlates and contributors to
quality at the extremes is essential.
4.2. Limitations and future research
Some methodological limitations of this study are related to the overall NICHD SECCYD design. The sample is not
nationally representative and cannot provide reliable estimates of the prevalence of arrangements and their quality by
sector/subsector. Nor is the design experimental, so causal arguments cannot be made. In addition, the sample size of
the full study was large, but sampled families, not centers, so the level of dissagregation across sector/subsector
provided limited power for some analyses. Finally, the birth cohort design does not preclude the selection bias of
families into types of care and quality within type that was not accounted for by the included family socioeconomic
characteristics. Omitted variables regarding family selection or other issues may play a role in quality differences by
sector/subsector.
It is also important to note that the data in the NICHD SECCYD were collected in the first half of the 1990s
and reflect available child care choices. Policy implications should be interpreted in light of the changing early care
and education landscape in the intervening decade. The most notable change is the vast increase in state-funded
prekindergarten services, which now serve 20% of the nation's 4-year-olds, a 43% increase in just 5 years (Barnett,
Hustedt, Hawkinson, & Robin, 2006). State-funded prekindergarten classrooms are often situated in public schools,
with differences from non-school-based classrooms in structural and caregiver factors (Gilliam & Marchesseault, 2005)
and possibly weaker associations between teacher/caregiver education and process quality (Pianta et al., 2005). Such
differences point to the importance of continuing research and examination of these new divisions of early care and
education.
Although not every individual sector/subsector comparison yielded significant differences at each assessment point,
the pattern of findings was consistent. In none of the two-way comparisons was for-profit center quality higher than
nonprofit center quality. Whether the nonsignificant comparisons are due to actual similarity in quality or related to the
limited power and restricted range of variables within this sample is an empirical question that can be addressed in
future research. Studies designed to target these questions could sample centers and measure additional characteristics,
such as assessment of caregiver education separately from field of training, director characteristics (e.g., education,
professionalism, longevity in the field, salary), and financial and operational characteristics (e.g., per child fees and
expenditures, funding and revenue sources, objectives and operational challenges of center owners/sponsors). Per child
expenditures are particularly important points of comparison regarding sector/subsector quality. Differences in costs by
sector/subsector could not be examined in the current analyses, as data collected on costs in the NICHD SECCYD
could not be disaggregated to a level that would allow true comparisons of quality by amount spent per child. Research
in this area is needed to identify explanatory factors for the link between profit status and subsector quality.
4.3. Policy implications and conclusions
The effect sizes reported range from small to large. However, even the findings with small effect sizes have practical
relevance for parents, child care providers, and policymakers for several reasons. First, the implications of the quality
differences found are meaningful with respect to the impact of child care on child outcomes. For example, higher levels
of teacher education have been associated with higher process quality in child care (NICHD ECCRN, 2000), yet
407L.S. Sosinsky et al. / Journal of Applied Developmental Psychology 28 (2007) 390–410

systematic differences in teacher education were found by profit status at almost all ages, with nonprofit caregivers
having higher educational levels. Likewise, the additional contribution of subsector classification to positive caregiving
was small, but even after accounting for the other factors, subsector explained about as much variance in positive
caregiving as did staff quality. Second, even small effects pertaining to large numbers of children, as is the case with
child care, may be more consequential than large effects applicable to only a few (McCartney & Rosenthal, 2000). As
child care quality in centers is complicated and multiply determined, the fact that subsector contributed to process
quality, at least at the assessment point with the largest number of classrooms, even after controlling for family
selection and for the known staff and structural contributors to process quality (NICHD ECCRN, 2002), denotes
systematic quality variation in the mixed-sector child care system.
The current findings support a multipronged approach toquality improvement to raise and equalize quality in child care
centers. The goal of a high-quality child care supply should be targeted with a combination of regulations and incentives,
while the demand for high quality can be improved through constructive consumer education. Consideration of market
competition factors must be woven throughout both supply- and demand-side strategies, especially as competition applies
within and across sector and subsector but competition's effect on quality manifests differently in each.
Stricter regulation and enforcement have the potential to raise quality, as suggested by comparisons among states
with regulations of varying strengths (Helburn, 1995; Phillips et al., 1992). In the absence of federal quality standards,
state regulations vary not only in the levels at which standards are set, but also in the types of centers to which they
apply. The most common exemption from regulation is for faith-based providers. In the current results, faith-based
center quality was intermediate, being sometimes lower than that of other nonprofit centers, and sometimes higher than
but other times the same as quality in for-profit centers. These quality differences may reflect an attempt to keep
costs down, possibly to keep parent fees low or perhaps to pass along some revenue to the sponsoring organization
(Helburn & Bergmann, 2002). It is notable that faith-based nonprofits did not differ from nonsectarian nonprofits
on positive caregiving despite these staff and structural quality differences. This is in contrast to the findings from
the Cost, Quality, and Child Outcomes Study, and warrants further investigation. Nonetheless, staff and structural
quality indicators are important in and of themselves, as they can have direct impact on child outcomes and also
have implications for the quality and sustainability of the early care and education workforce. The exemption from
regulation enjoyed by many faith-based centers may allow or even encourage lower quality or at least greater
variability in staffing and structure (Henriques, 2006). When considered in combination with the fact that some policies
may in effect encourage use of faith-based providers (Bogle, 2001), revisiting such exemptions is warranted.
Stricter regulation might raise quality, but it likely would also raise costs for providers. Attention to quality
improvement and monitoring in child care centers is vital, but must consider related effects on centers' ability to stay
in operation and deliver good-quality services for all age groups without substantial increases in costs to parents.
Furthermore, stronger regulation alone is not likely to sufficiently raise and even out the quality of care provided in the
mixed-sector center-based child care market. Incentives to providers to raise quality have been proposed. For example,
providers' fees could be tied to accreditation status (Gormley & Lucas, 2000). Some states have or are proposing
consumer guide rating systems in highly interpretable formats (e.g., report cards, star systems) to help parents make
child care decisions and to encourage providers to improve quality (Jacobson, 2005; National Child Care Information
Center, 2002). These ratings have the potential to improve quality by providing consumer information, embedding
ratings in the licensure system, or reimbursing programs on a tiered system. Furthermore, just as regulations should
apply evenly across subsectors, quality improvement incentives must be made available to providers in all subsectors.
Adequate incentives to improve quality may be quite powerful in an industry with such low profit margins and high
overhead costs.
A pervasive assumption underlying much child care policy is that competition will put continual pressure on
providers to improve or be forced out of the market. At its basic level, however, this assumption ignores the
complicated association of market competition with child care quality and service delivery. For example, the subsectors
that compete for children from unsubsidized middle-income children are most dependent on parent fees from families
who are most dependent on comparatively limited incomes with which to pay those fees. Raising fees to raise quality
may result in a center's closure, whereas lowering fees to remain competitive may mean lowering quality. In another
example stemming from the recent explosion in state-funded prekindergarten services and proposals for universal
preschool education, there is concern, especially among for-profit providers, that competition for preschool-aged
children may have the unintended consequence of affecting the availability or quality of care for younger children if
centers lose too many more profitable preschool-aged clients (Zigler, Gilliam, & Jones, 2006; Neugebauer, 2006).
408 L.S. Sosinsky et al. / Journal of Applied Developmental Psychology 28 (2007) 390–410

Georgia and New York addressed this concern in the creation of their universally available, voluntary prekindergarten
programs in which child care is provided in multiple sectors, including nonprofit centers and private for-profit
providers (Barnett, Hustedt, Robin, & Schulman, 2004).
The other problem with the assumption that market competition will raise quality is that it is based on the deeper
assumption that consumers have and rely on perfect information when making their choices. Indeed, asymmetric
information precludes the ability of parents to make an informed decision regarding the selection of the most
appropriate child care center for their child. The child care market meets the conditions for what economists term moral
hazard. In the case of child care, it is difficult for parents to differentiate high-quality from low-quality centers. For
example, Mocan (2001) found that nonprofit centers with very clean reception areas, which are readily observed and
may signal good quality to parents, tended to produce lower levels of quality for harder-to-observe aspects. Likewise,
minority parents' assessment of quality is inflated if teachers in the classroom are of the same race, suggesting
“misplaced trust”(Mocan, 2001). Many parents consider nonprofit status a neutral signal of quality (Mocan, 2001),
whereas this and other research shows nonprofit status to be a positive signal of center quality. However, within the
nonprofit sector, parents may assume that faith-based providers are well intentioned and, thereby, high quality, whereas
research demonstrates lower quality on staffing and structural characteristics compared with other nonprofits.
Ultimately, if left unchecked, not only will parents' inability to reliably distinguish high- and low-quality child care
lead to bad choices for their children, it may affect the quality available in the marketplace. When consumers do not
choose on the basis of actual quality, the excellence of high-quality centers is left unrewarded, prompting high-quality
centers to exit the market and leave primarily centers that provide mediocre quality at best (Mocan, 2001). However,
strategies to educate parents about the quality of child care options may support the potential success of market
competition to raise child care quality by creating a better-informed consumer population and providing further
incentive to providers to meet high-quality standards (Gormley & Lucas, 2000).
References
Barnett, W. S., Hustedt, J. T., Robin, K. B., & Schulman, K. L. (2004). The state of preschool: 2004 state preschool yearbook. New Brunswick, NJ:
National Institute for Early Education Research.
Barnett, W. S., Hustedt, J. T., Hawkinson, L. E., & Robin, K. B. (2006). The state of preschool: 2006 state preschool yearbook. New Brunswick, NJ:
National Institute for Early Education Research.
Blau, D. M. (2001). The child care problem: An economic analysis. New York: Russell Sage Foundation.
Bogle, M., (2001), March). The role of faith-based organizations in child care. Paper presented at the Brookings Institution Sacred Places, Civic
Purposes morning discussion, Washington, DC.
Brown, R. (2001). How we built a strong company in a weak industry. Harvard Business Review,79(1), 51−58.
Child Care Services Association (2002). T.E.A.C.H. Early childhood project 2001–2002 annual report. Chapel Hill, NC: Author.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd Ed.) Hillsdale, NJ: Erlbaum.
Cortina, J. M., & Nouri, H. (2000). Effect size for ANOVA designs. (Sage University papers series on quantitative applications in the social sciences,
07-129). Thousand Oaks, CA: Sage Publications.
Cryer, D. (2003). Defining program quality. In D. Cryer, & R. M. Clifford (Eds.), Early childhood education and care in the USA Baltimore, MD:
Paul H. Brookes.
Friesen, B.K., (1995). A sociological examination of the child care auspice debate. Toronto: Childcare Resource and Research Unit, Centre for Urban
and Community Studies, University of Toronto.
Gilliam, W. S., & Marchesseault, C. M. (2005). From capitols to classrooms, policies to practice: State-funded prekindergarten at the classroom
level. New Haven, CT: Yale University Child Study Center.
Gormley, W. T., Jr., & Lucas, J. K. (2000). Money, accreditation, and child care center quality. New York: Foundation for Child Development.
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.
Helburn, S. W., & Bergmann, B. R. (2002). America's child care problem: The way out. New York: Palgrave MacMillan.
Helburn, S. W., & Howes, C. (1996). Child care cost and quality. Future of Children,6(2), 62−82.
Henriques, D. B. (2006, October 8). Religion trumps regulation as legal exemptions grow. : The New York Times.
Hofferth, S. L. (1996). Child care in the United States today. Future of Children,6(2), 41−61.
Huston, A. C., Chang, Y. E., & Gennetian, L. (2002). Family and individual predictors of child care use by low-income families in different policy
contexts. Early Childhood Research Quarterly,17, 441−469.
Jacobson, L. (2005, February 9). More governors want to rate early-childhood programs. Education Week,24, 17.
Kagan, S. L. (1991). Examining profit and nonprofit child care: An odyssey of quality and auspices. Journal of Social Issues,47,87−104.
McCartney, K., & Rosenthal, R. (2000). Effect size, practical importance, and social policy for children. Child Development,71(1), 173−180.
Mocan, H. N. (2001). Can consumers detect lemons? Information asymmetry in the market for child care. Cambridge, MA: National Bureau of
Economic Research.
409L.S. Sosinsky et al. / Journal of Applied Developmental Psychology 28 (2007) 390–410

Morris, J. R., & Helburn, S. W. (2000). Child care center quality differences: The role of profit status, client preferences, and trust. Nonprofit and
Voluntary Sector Quarterly,29, 377−399.
National Child Care Information Center. (2002). Tiered strategies: Quality rating, reimbursement, licensing. Vienna, VA: Child Care Bureau,
Administration for Children and Families, U.S. Department of Health and Human Services.
National Institute of Child Health and Human Development Early Child Care Research Network. (1996). Characteristics of infant child care: Factors
contributing to positive caregiving. Early Childhood Research Quarterly,11, 269−306.
National Institute of Child Health and Human Development Early Child Care Research Network. (1997). In G. J. Duncan & J. Brooks-Gunn (Eds.),
Consequences of growing up poor (pp. 100−131). New York: Russell Sage Foundation.
National Institute of Child Health and Human Development Early Child Care Research Network. (2000). Characteristics and quality of child care for
toddlers and preschoolers. Applied Developmental Science,4,116−135.
National Institute of Child Health and Human Development Early Child Care Research Network. (2001). Nonmaternal care and family factors in
early development: An overview of the NICHD Study of Early Child Care. Journal of Applied Developmental Psychology,22, 457−492.
National Institute of Child Health and Human Development Early Child Care Research Network. (2001). Before head start: Income and ethnicity,
family characteristics, child care experiences, and child development. Early Education & Development,12, 545−576.
National Institute of Child Health and Human Development Early Child Care Research Network. (2002). Child-care structure →process →outcome:
Direct and indirect effects of child-care quality on young children's development. Psychological Science,13, 199−206.
National Institute of Child Health and Human Development Early Child Care Research Network. (2003). Does quality of child care affect child
outcomes at age 4 1/2? Developmental Psychology,39, 451−469.
National Institute of Child Health and Human Development Early Child Care Research Network. (2006). Child-care effect sizes for the NICHD Study
of Early Child Care and Youth Development. American Psychologist,61,99−116.
Neugebauer, R. (2000). Religious organizations taking proactive role in child care: Trends in church-housed child care. Child Care Information
Exchange,133,18−20.
Neugebauer, R. (2006). For profit child care: Four decades of growth. Child Care Information Exchange.
Phillips, D. A., Howes, C., & Whitebook, M. (1991). Child care as an adult work environment. Journal of Social Issues,47(2), 49−70.
Phillips, D. A., Howes, C., & Whitebook, M. (1992). The social policy context of child care: Effects on quality. American Journal of Community
Psychology,20,25−51.
Phillips, D. A., Mekos, D., Scarr, S., McCartney, K., & Abbott-Shim, M. (2000). Within and beyond the classroom door: Assessing quality in child
care centers. Early Childhood Research Quarterly,15, 475−496.
Pianta, R., Howes, C., Burchinal, M., Bryant, D., Clifford, R., Early, D., et al. (2005). Features of pre-kindergarten programs, classrooms, and
teachers: Do they predict observed classroom quality and child–teacher interactions? Applied Developmental Science,9, 144−159.
Shonkoff, J. P., & Phillips, D. A. (2000). From neurons to neighborhoods: The science of early childhood development. Washington, DC: National
Academy Press.
Smolensky, E., & Gootman, E. (2003). Working families and growing kids: Caring for children and adolescents. Washington, DC: National
Academy Press.
Vandell, D.L., & Wolfe, B., (2000). Child care quality: Does it matter and does it need to be improved? Washington, DC: Office of the Assistant
Secretary for Planning and Evaluation, U.S. Department of Health and Human Services.
Whitebook, M., Howes, C., & Phillips, D. (1990). Who cares? Child care teachers and the quality of care in America. Oakland, CA: Child Care
Employee Project.
Whitebook, M., Howes, C., & Phillips, D. (1998). Worthy work, unlivable wages. The National Child Care Staffing Study, 1988–1997 Washington,
DC: Center for the Child Care Workforce.
Whitebook, M., & Sakai, L. (2003). Turnover begets turnover: An examination of job and occupational instability among child care center staff.
Early Childhood Research Quarterly,18, 273−293.
Whitebook, M., Sakai, L., & Howes, C. (1997). NAEYC Accreditation as a strategy for improving child care quality: An assessment by the National
Center for the Early Childhood Work Force. Washington, DC: National Center for the Early Childhood Work Force.
Zigler, E. F., Gilliam, W. S., & Jones, S. M. (2006). A vision for universal preschool education. New York: Cambridge University Press.
410 L.S. Sosinsky et al. / Journal of Applied Developmental Psychology 28 (2007) 390–410