Multi-Site Expansion of an Early Childhood Intervention and School Readiness
OBJECTIVES: To evaluate the impacts of the expansion of an evidence-based full- and part-day early childhood development program on multiple indicators of school readiness, attendance, and parental involvement for a large cohort of low-income children.
METHODS: This study involved the end-of-preschool follow-up of a nonrandomized, matched-group cohort of 2630 predominantly low-income, ethnic minority children who enrolled in the Midwest Child–Parent Centers (CPC) or alternative preschools in the fall of 2012 in 31 schools in Chicago, Illinois. The program provides comprehensive education, family support, and health services. In the preschool component assessed in this study, 1724 children aged 3 to 4 years in all 16 Chicago centers enrolled in the program. The comparison group included 906 children of the same age who participated in the usual preschool services in 14 matched schools.
RESULTS: Relative to the comparison group who enrolled in the usual preschool services and adjusted for covariates, CPC participants had higher mean scores on all performance-based assessments of literacy (59.4 vs 52.4; P = .001), socioemotional development (57.0 vs 51.8; P = .001), and physical health (34.5 vs 32.1; P = .001). They also had higher ratings of parental involvement in school (5.3 vs 4.0; P = .04). Group differences also translated into higher rates of meeting national assessment norms. Program estimates were similar for children attending new and established CPCs and according to age, race/ethnicity, and family income status.
CONCLUSIONS: The findings show that expansion of the program to new schools and more diverse populations is feasible and effective in promoting school readiness skills and parental involvement.
- CPC —
- Child–Parent Centers
- TS —
- Teaching Strategies GOLD Assessment System
What’s Known on This Subject:
Although the benefits of early intervention are well documented, relatively few children are enrolled in high-quality publicly funded programs, especially the most vulnerable. Most previous evidence relies on efficacy trials rather than on routine programs at the population level.
What This Study Adds:
This study provides the first findings of the contemporary scale-up of one of the most evidence-based early childhood programs. Findings illustrate the feasibility and benefits of scaling effective preventive interventions to reduce achievement gaps and later health disparities.
The behavioral sources of achievement and health disparities begin early in life. Economic disparities of at least 10 percentage points are evident in prenatal care quality, which increase in magnitude during childhood as measured according to school readiness, reading proficiency, household food insecurity, delinquency, and high school graduation.1–4 These factors predict economic well-being, mental health, and health behavior.5,6
In recognition of the major role that social determinants play in well-being, Healthy People 2020 includes early and middle childhood as a key context for policy and programming.7 This approach reflects the inherent connection between the school achievement gap and health disparities. Because of their demonstrated impact on well-being, early childhood interventions are at the forefront of prevention for improving educational success and health. Life-course studies indicate that participation in high-quality center-based programs at age 3 and 4 years links to higher levels of school readiness and achievement, as well as higher rates of educational attainment and socioeconomic status as adults, and lower rates of crime, substance use, and mental health problems.8–11
Although publicly funded preschool programs such as Head Start and state pre-kindergarten serve an estimated 42% of US 4-year-olds, most provide only part-day services and only 15% of 3-year-olds are enrolled.12 These rates plus differences in quality, intensity, and comprehensiveness may account for the finding that only about one-half of children entering kindergarten have mastered the cognitive skills needed for school success.13,14 Large differences are also found among economic groups.
Child–Parent Center Program and Scale Up
The Child–Parent Center (CPC) Education Program is a school-based public program with strong evidence of benefits for children and families.15 Routinely implemented in the Chicago Public School District since 1967, the program provides comprehensive educational and family support services to children ages 3 to 4 years in high-poverty neighborhoods with continuing services up to third grade. In a series of reports in the CLS (Chicago Longitudinal Study), an economically disadvantaged cohort of 1500 program and control group children born in 1980 has been followed up to age 30 years. Participation beginning in preschool was found to eliminate the achievement gap in school readiness and early performance, reduce rates of child maltreatment and school remedial education, reduce rates of felony arrest and substance abuse, and increase rates of high school graduation.15–19 Economic benefits were found to exceed costs by a ratio of 7-to-1 or higher.20–22 Although these findings provide a foundation for expansion, the contemporary context of early childhood programs are different than in the 1980s.
A scale-up of the CPC program was begun in 2012 under an Investing in Innovation Grant from the US Department of Education. Based on a matched-group design similar to earlier studies, we assessed effects in more diverse communities and school contexts. The program model was revised to optimally address the learning needs of children and families, including the opening of full-day preschool classrooms at Chicago sites and implementing a comprehensive set of educational and family supports.23
Three major questions were addressed: (1) Is participation in the CPC expansion preschool intervention associated with higher school readiness skills, attendance, and parental involvement at the end of the year? (2) Do the effects of CPC vary between established sites and new sites beginning to implement the program? and (3) Do the effects of CPC participation vary according to age, race/ethnicity, and low-income status?
The Midwest Expansion of the Child–Parent Center Education Program, Preschool to Third Grade (Midwest CPC) is an intervention project of the Human Capital Research Collaborative at the University of Minnesota and 8 school districts and educational partners. Midwest CPC was implemented for a 2012 preschool cohort of 2364 to be followed up to third grade and beyond. The goals of the project are to implement the program with high levels of fidelity; assess impacts on child and family well-being; and facilitate expansion and sustainability. Institutional review board approval and informed consent were obtained through the University of Minnesota and participating school districts. The focus of the present report is on 1724 children in 16 Chicago public school sites, which is the largest district and the location of the largest preschool expansion (Supplemental Information).
Sample and Design
The original sample included 1006 children aged 3 to 4 years in all 10 existing CPCs and 718 children in 6 new CPCs in underrepresented areas of the city. Although all schools serve high proportions of low-income families, the new centers were more diverse in economic status and ethnicity, including a larger concentration of Latino families. The control group included 906 children of the same age who enrolled in preschool programs in 14 non-CPC schools that matched the CPC schools as closely as possible. Due to poor matches and insufficient records, two additional control schools were excluded (Supplemental Information). Similar to reports in the CLS, all children in the control group participated in the usual early childhood services available in the Chicago school district for part of the day (Supplemental Information).24
We assessed the impact of 2 measures of CPC participation. For any preschool, children enrolled in the program at age 3 or 4 years in 2012–2013 were compared with children who did not participate but who enrolled in the usual preschool program in the control sites (either Head Start or state-funded pre-kindergarten). Thus, the added value of CPC preschool above and beyond the typically implemented program was assessed rather than the impact of preschool compared with no preschool. This approach may provide a conservative bias. Given our focus on the effects of participation, children in both groups attended preschool for at least 3 months and began no later than January 2013.
As a dosage measure, full-day preschool compared children enrolled in the CPC program for the entire school day of 7 hours and children in a part-day CPC program for 3 hours per day or in the usual preschool program in control sites. Table 1 shows the pattern of participation and data collection in the study.
Midwest CPC Intervention
The CPC intervention in the expansion project is designed to enhance early childhood development in multiple domains of health and well-being. Located within or near elementary schools, the program provides educational and family support services between the ages of 3 and 9 years (preschool to third grade). Within a structure of comprehensive services (education, family, health, and social services), 6 major components are included23,25: (1) collaborative leadership team led by the head teacher; (2) effective learning experiences (eg, small classes, a literacy-rich instruction); (3) parental involvement and engagement; (4) aligned curriculum across grades; (5) continuity and stability; and (6) a professional development system of teacher coaching and site support.15,20 The preschool program is 3 (part-day) or 7 hours per day, 5 days a week during the school year (Supplemental Information).
Seven indicators were assessed at the end of the preschool year by using the Teaching Strategies GOLD Assessment System (TS).26 TS is a performance-based assessment designed for children from birth through kindergarten and comprises 66 items measuring mastery on 38 objectives in 9 domains of child development. We reported outcomes for 6 of the domains assessed with 49 items: literacy (12 items), oral language (6), math (7), cognitive development (10), socioemotional (9), and physical health (5). Each item (indicator) is rated by teachers from 0 (not yet meeting objective) to 9 (full mastery of objective) denoting the level of mastery observed for the objective (Supplemental Table 6). The mean of the scale is set at ∼36 months of age. The raw scores summed across items for the 6 subscales plus the total score for all domains were analyzed. Dichotomous scores measuring performance at or above the national average were also assessed. As a widely used assessment in early childhood settings, TS has shown strong reliability and validity in measuring school readiness that is predictive of school achievement and performance.26–28
Teachers were blinded to intervention conditions, and the assessment, being routinely administered by schools, avoided the reactivity of measurement that is common with new assessments used just for research purposes. A significant advantage is that teachers observe children’s behavior over a period of 4 to 6 weeks before recording their assessments. This method enhances the ecological validity of the scores.24,29
The study used 3 indicators of attendance in the preschool program from official school administrative records. Average daily attendance was the percentage of total available days of enrollment that a child was in attendance. Chronic absence was a dichotomous indicator of whether a child missed 10% or 20% of the total possible school days or more. Absences are not only an indicator of whether the child (and family) are meeting basic school requirements but reflect health problems, illness, and adverse experiences in the family that prevent regular attendance. Economic factors related to transportation and employment instability also correlate with chronic absence. There is extensive literature that attendance and chronic absence predict not only academic achievement but social and emotional adjustment as well as health behavior.30,31 Attendance in preschool is also highly correlated with elementary school attendance.30
The study used 3 indicators of parental involvement in children’s education. For parental involvement in school or the center, classroom teachers rated on a 10-point scale the “percent of parents who participated in school events and activities from January to the end of the year.” A rating of 1 designated that <10% of families in the classroom participated and a rating of 10 designated that ≥90% of families in the classroom participated in school events and activities (range, 1–10; mean ± SD, 5 ± 2.2). The rating for each class was assigned to each individual child, which reduces response bias and “halo” effects found in ratings of individual children. A dichotomous indicator at or above the mean of 6 was also assessed. The third indicator was parent ratings of their own involvement in the school. A mid-year survey item asked “So far this year, about how often have you participated in school or center activities?” Categorical responses were coded as follows: 0 = never, 1 = less than once a month, 2 = once a month, 3 = two or three times per month, 4 = once a week, and 5 = more than once a week (range, 0–5; mean, 2.19 ± 1.56). Previous studies show that parental involvement ratings by teachers are valid indicators of parenting practices and are a mechanism of long-term effects of early intervention.16,18
Data were analyzed by using generalized estimating equations in SPSS (IBM SPSS Statistics, IBM Corporation, Armonk, NY).32 This model accounts for the multilevel, nested design of children clustered within schools.33 It also provides robust estimates of marginal means under deviations from multivariate normality and misspecification of the correlated data structure.34 Findings are reported as adjusted means or percentages and group differences controlling for the influence of the following factors: child’s gender, race/ethnicity, subsidized lunch status, age in months, special education, school size, and fall scores (school readiness or attendance). These covariates were measured at preschool entry from school administrative records and parent surveys. Continuous variables were analyzed with the generalized linear model via maximum likelihood (average attendance, TS subscales, and parental involvement). Dichotomous variables were analyzed with probit regression (chronic absences, TS at/above national norms). All estimates account for clustering or nesting of observations according to site. The generalized estimating equation model provides consistent estimates of marginal means under a wide variety of modeling assumptions (distribution of error terms, correlated structure, and clustering).34,35 SEs and significance tests were adjusted for variation among sites by using the Huber-White/sandwich correction.32,33
As observed in previous studies,16–21 coefficients were robust to alternative covariates and model specifications. Given the relatively high levels of fidelity of program implementation, site-to-site variation in outcomes was modest. School-level achievement, site differences, and new versus established CPC status, however, were accounted for in the models. Multiple imputation of missing data TS subscales was based on the expectation-maximization algorithm after determining that scores were missing at random.36 Findings were consistent across a range of analyses and for different sample sizes. Adjusted group differences at the 0.05 probability level were emphasized. To test subgroup effects, program interaction terms included age, race/ethnicity, and family income status. Within the program group, the effect of established and newly opened sites was also assessed. The statistical significance of subgroup effects was set at 0.05, emphasizing those with overall effects. Standardized mean differences were reported to show the relative magnitude of changes across outcomes.
In the fall of 2012, total enrollment in Chicago sites was 2630, with 1724 in the CPC program and 906 children in the matched comparison schools. Nearly 60% of each group was 4 years old; the remainder were 3 years of age. One-quarter of CPC participants attended full-day classes. The sample is more heterogeneous in both ethnic composition and geography than in previous years. Compared with the existing CPCs, the new CPCs (expansion sites) have a large percentage of Hispanic families.
The characteristics of the program and comparison groups are shown in Table 2. Children were well matched on school-level characteristics, including the percentage of children scoring proficient on state tests and the probability of program enrollment. Groups were also equivalent in age, gender, low-income status, and receipt of special education services. The major difference between groups was in race and ethnicity. This outcome was due to a few control schools with large Hispanic populations. These differences were taken into account, and program impacts according to race/ethnicity were also estimated. Fall assessment performance was also generally similar between groups, including for the total scale.
Through monitoring, observations, and staff interviews, we evaluated the extent to which the sites implemented the program requirements and with high levels of fidelity. Overall, the 16 sites successfully implemented the program requirements, including establishing the leadership teams, maintaining small class sizes, and providing comprehensive child development and family services (Supplemental Information). The average number of hours of instruction for the year was 552 with the majority of time being devoted to literacy and math. There was nearly an equal split between teacher-directed and child-initiated activities (Supplemental Table 7).
Overall Effects of CPC Participation
Table 3 shows the group differences and P values for program and comparison groups on the outcomes after adjustment for the covariates.
Relative to the comparison group participating in the usual preschool programs, CPC participants had significantly higher mean scores at the end of the preschool year for all 6 domains and the total score. The largest of these differences were for literacy (59.4 vs 52.4; P = .001), cognitive development (59.4 vs 53.5; P = .001), math (37.6 vs 33.5; P = .001), and socioemotional development (57.0 vs 51.8; P = .001). CPC participants were also more likely to perform at or above the national average for literacy (78.0% vs 57.2%), math (73.9% vs 59.7%), cognitive development (67.7% vs 49.3%), and physical health (75.6% vs 59.3%). For the total score across all 6 domains, preschool participants had a mean score that was 27 points higher (286.3 vs 259.3; P = .001). The standardized mean differences are shown in Fig 1. Their rate of meeting the national norm was also higher (69.6% vs 52.2%; P = .033). The latter finding corresponds to a group difference of 0.47 SD. Findings based on full imputation of TS scores showed similar benefits (Supplemental Table 8).
In contrast to school readiness, CPC participants had statistically equivalent rates of daily attendance and chronic absences (missing 20% of days) but significantly higher rates of chronic absences defined as missing ≥10% of school days (60.2% vs 47.9%; P = .040). This outcome was reported with or without fall absences taken into account and may reflect the greater economic disadvantages of the CPC schools. The high rates of chronic absence for both groups reflect, in part, that preschool attendance is not mandatory whereas school-age attendance is mandatory.
CPC participants had higher rates of parental involvement in school as rated by teachers (5.3 vs 4.0; P < .001). The percentage of CPC parents with high involvement in school was also significantly greater (58.6% vs 19.5%; P < .001). This finding was corroborated for the parent ratings, which were based on a smaller sample.
Full- and Part-Day Preschool Versus the Comparison Group
School Readiness and Attendance
Table 4 shows that CPC full-day and part-day groups had higher mean scores and rates of proficiency than the comparison group across all subscales. This finding indicates that the impact of CPC is not due to just the full-day group. As illustrated in Fig 1, standardized mean differences were larger for the full-day group. For attendance, no consistent differences were found, although the CPC full-day group had higher rates of attendance and lower chronic absences than the part-day group.
Both CPC groups had higher levels of teacher-rated parental involvement than the comparison group. The CPC full-day group had a higher mean rating than the part-day group only for the parent-rated measure (2.5 vs 2.2; P = .015).
We assessed whether program effects for particular subgroups differed from other subgroups of participants (difference-in-differences).
New and Established Sites
As shown in Table 5, no pattern of differences was found in estimated effects between newly opened and established CPCs. Standardized mean differences were similar for the total school readiness scale (Fig 1). The only differences were that impacts on the percent meeting national norms on total school readiness were greater for children in newly opened sites. Standardized mean differences on parental involvement also favored the new CPC sites.
Child and Family Characteristics
No differential effects of the program were found for ethnicity or race, low-income status, and child age (Supplemental Tables 9 and 10). As shown in Fig 1, the relatively large standardized mean differences were similar for Hispanic and black subjects, according to family income status, and for 4- and 3-year-olds.
This study reports the largest expansion of the CPCs. Although the intervention has gone through distinct phases over the course of >4 decades, the present study’s findings illustrate the positive impacts of a scaled-up program for a contemporary and more heterogeneous cohort of families. Compared with routinely implemented preschool, CPC participation was linked to greater school readiness skills and parental involvement.
The standardized mean difference for the total score of approximately one-half a SD is relatively large and is equivalent to a >0.5-year gain in proficiency skills and a 33% increase over the comparison group in meeting the national norm. Increases in parental involvement in school were even larger in magnitude. Full-day preschool provided added school readiness benefits beyond part-day services, but both CPC groups showed greater performance than the comparison group. The program in the new sites, which served higher proportions of Latino families, exhibited the same pattern of effects as those in the established sites. Program benefits were also similar according to age, race and ethnicity, and family income.
Although there is a growing literature that state pre-kindergarten and other publicly funded programs improve school readiness,8,12 the present study illustrates the feasibility and demonstrated effects of scaling up one of the most evidence-based programs. Because CPC provides more intensive and comprehensive services than most other programs, larger and more sustained effects have been found on educational, economic, and social well-being.16–20 In contrast to efficacy trials, CPC also implements a system of services for broader populations that are cost-efficient and replicable. The Midwest CPC expansion is designed to increase access to high-quality early childhood programs and provide a continuous learning experience through the early school grades.
Our findings show that the CPC program can help achieve the Healthy People 2020 early childhood goal to “increase the proportion of children who are ready for school in all five domains of healthy development: physical development, socio-emotional development, approaches to learning, language and cognitive development.”7 This goal suggests that implementation at the population level can promote positive effects. Achieving these gains requires high levels of quality. In the CPCs, not only are class sizes small and family services extensive, but there is a curriculum focus on child engagement in all domains of learning. Staff professional development activities further support the quality of implementation. Early childhood investments that prioritize these elements are more likely to promote healthy development. Enhancing school readiness skills initiates a process of cumulative advantage that leads to better health and well-being in adulthood.11,15,18
The findings of the Midwest CPC expansion provide support for increasing access to effective preschool as a strategy for closing the achievement gap and addressing health disparities. As the present study illustrates, preschool participation seems to be a particularly effective approach for strengthening school readiness. By promoting consistent improvements in readiness skills and in parental involvement, the transition to kindergarten may be smoother and more successful. These presage continuing effects into the early grades. The size and breadth of impacts go beyond previous studies.14 The positive effects of full-day preschool over part-day also suggest that increasing access to early childhood programs should consider the optimal dosage of services. In addition to educational benefits, full-day preschool benefits parents by freeing time to pursue career and educational opportunities.
As shown in previous studies,37⇓–39 improvements in early cognitive, literacy, and numeracy skills are a primary mechanism through which early childhood intervention promotes long-term effects on health and well-being; these improvements include reducing the need for remediation and treatment services and promoting educational attainment and economic self-sufficiency.19–21 For example, in the adult follow-up of the CLS, school readiness skills accounted for up to one-third of the direct effect on educational attainment, substance abuse, and crime prevention.18,38 Similarly, parental involvement in school (an indicator of the family support hypothesis) explained one-quarter of these long-term effects.
The interpretation of the positive effects should be viewed in the context of major changes in the intervention from that previously evaluated. The Midwest CPC program emphasizes 6 major elements: effective learning experiences, collaborative leadership, parental involvement and engagement, aligned curriculum, continuity and stability, and professional development. The previous model emphasized only the first 3 elements and with a lower degree of intensity. For example, enhanced elements of effective learning experiences include a curriculum balance of teacher-directed and child-initiated activities, full-day preschool, and progress monitoring of instruction.23,25
The estimated effects in the present study measure the added value above and beyond that of the usual preschool services. Because all comparison group participants enrolled in Head Start or state pre-kindergarten, the effects in this study would have been larger if a no-preschool group was included (as was typical of earlier studies).8,22 Thus, relative to conventional comparisons, the present findings are likely to be conservative. Although only 15% of the comparison group in the earlier CPC study attended other preschool programs, the entire group attended full-day kindergarten, whereas only 60% of the CPC group did.22,39 Nevertheless, we interpret the pattern of findings in both sets of studies as indicative of large effects on early childhood development outcomes.
We found no effect of the overall program on attendance, which may be attributed to the challenges of an initial start-up. CPCs are located in the most disadvantaged areas of the city and enroll 3 to 5 times more children than comparison sites. The latter had higher rates of attendance at the beginning of the year. Monitoring efforts take time to be established. We found that attendance rates improved as parental involvement and outreach staff increased their time engaging families. Overall parental involvement and engagement was substantially strengthened in the program, which also expands on previous studies.15,18,19
The study has 3 limitations. First, the measures assessed program outcomes rated by teachers and parents. Although not purely objective measures of school readiness skills, TS scores are a performance-based assessment of mastery. They have the further advantage of being aligned with district and state learning standards. Direct cognitive assessments would be complementary indicators, but they were unavailable at the end of preschool. Performance-based and direct assessments correlate highly with each other, however.40–42 These factors and related indicators such as peer relations, task orientation, and executive functioning will be assessed in follow-up periods.
The second limitation was that even with the history of previous program implementation, the Midwest CPC expansion was being implemented for the first time. Delays and changes in staffing, as well as extra time for establishing the structure of operations, were unavoidable. This situation suggests that the positive findings may be conservative compared with implementation after the start-up period. Moreover, the findings address only the preschool component of the program. Continuing kindergarten and school-age services are now being evaluated. Fidelity of implementation to the program, however, has been relatively high.
Finally, the findings have limited generalizability beyond urban contexts. Despite the expansion to new underrepresented areas, most families were low-income and ethnic minority. The fact that the new sites were found to have the same impact as established sites, as well as other parallel findings, suggests a moderate degree of external validity.
We found that a contemporary expansion of the CPC program provided significant benefits in school readiness skills and parenting practices. The magnitude and consistency of the impacts are predictive of sustained effects on well-being and school success. The comprehensive and intensive approach of the program to promoting early childhood development provides a strong foundation for improved achievement and health.
- Accepted April 4, 2016.
- Address correspondence to Arthur J. Reynolds, PhD, Institute of Child Development, 51 E. River Rd, Minneapolis, MN 55455. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: All phases of this study were supported by the US Department of Education (grant no. U411B110098) and the National Institute of Child Health and Human Development (grant no. R01HD034294). Matching grants from 20 foundations also contributed to the project. Funded by the National Institutes of Health (NIH).
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
- Richie SJ,
- Bates TC
- Barton PE,
- Coley RJ.
- Katsiyannis A,
- Thompson MP,
- Barrett DE,
- Kingree JB
- US Department of Health and Human Services
- Karoly LA,
- Kilburn MR,
- Cannon JS
- Schweinhart LJ,
- Montie J,
- Xiang Z,
- Barnett WS,
- Belfield CR,
- Nores M
- Barnett WS,
- Carolan ME,
- Squires JH,
- Clarke Brown K,
- Horowitz M
- US Department of Education, National Center for Education Statistics
- Reynolds AJ
- Reynolds AJ,
- Temple JA,
- Ou S,
- Arteaga I
- Reynolds AJ,
- Temple JA,
- White BA,
- Ou SR,
- Robertson DL.
- Human Capital Research Collaborative
- Reynolds AJ,
- Hayakawa M,
- Candee AJ,
- Englund MM
- Lambert R,
- Kim D,
- Burts D
- Soderberg JS,
- Stulis S,
- Cummings K,
- Nolen E,
- McCutchen D,
- Joseph G
- Kim D,
- Lambert RG,
- Burts DC
- Snow CE,
- Van Hemel SB
- Balfanz R,
- Byrnes V.
- SPSS Inc
- Hoffman JP
- Diggle P,
- Heagerty P,
- Liang K,
- Zegler S
- Little RJ,
- Rubin DB
- Singh-Manoux A,
- Ferrie JE,
- Chandola T,
- Marmot M
- Teaching Strategies, Inc.
- Reynolds AJ,
- Englund MM,
- Hayakawa C, et al.
- Copyright © 2016 by the American Academy of Pediatrics