1. Introduction

In many countries, early childhood education programs are the core of educational disadvantage policies. They aim at young children who have normal intelligence and capacities, but because of unfavorable circumstances at home are not able to realize their full potential. Their home situation is usually characterized by low-educated and jobless parents, low family income, immigrant or minority background, and deprived neighborhoods. As a consequence, such children start their school career with considerable arrears, which in most cases they are not able to catch up with. This is detrimental to their educational and life chances; in the end the gap between the lower and higher socioeconomic classes is only widening^[1]. The idea has taken root that educational arrears can best be prevented when the children are still young, preferably before they enter elementary school^[2]. This has led to the development and implementation of a variety of compensational stimulation programs.

2. Early Childhood Education Programs

Early Childhood Education (ECE) programs usually focus on children from around 2 to 6 years of age. Most programs are carried out in preschools and the kindergarten groups of elementary schools, where the emphasis typically is on stimulating cognitive development, especially language development. Regularly, such programs also have a parent component which includes teaching the parents the skills needed to stimulate the cognitive, linguistic, and motor development of their children^[3].

ECE programs have been in existence since the 1960s, and meanwhile, numerous studies into their effectiveness have been conducted. It can be concluded that effects in general are small at most, and usually disappear (“fade-out”) once the program has stopped; in a few cases, minor effects reappear after some time (“sleeper effect”). In addition, recent studies show smaller effects than earlier studies, and so do large studies as compared to small-scale studies. However, a critical reading reveals many methodological flaws, and that cherry-picking and neglecting the many null and even negative effects are quite common. Moreover, ECE is a broad and complex field with lots of variation. Target groups may differ, as do the children’s ages, contents of the program, duration and intensity of the intervention, quality of the program and the teachers, scientific support from developers, the period the program is offered, available alternatives, financial costs, and parental backgrounds^[4]. What appears to be crucial is a high quality of the program, and its continuation in elementary school^[5]. A big problem is that in most synthesizing studies, such differences are ignored and lumped together, resulting in an amalgam of findings of low validity, and much ambiguity regarding their generalization^{[6][7][8][9][10][11][12][13][14]}.

3. The Big Three

Probably the most quoted, even today, are three early so-called model programs (“The Big Three”), viz. Perry Preschool, Abecedarian, and Chicago Child-Parent Center^[15]. The main reasons for their fame are that two of them constituted true experiments (Randomized Controlled Trials (RCT), the gold standard of effect studies), and that the participating children have been followed for several decades. Advocates of ECE programs are of the opinion that these model programs have laid the foundations for later programs^[16]. They claim that they are highly effective and have yielded many positive effects on both cognitive and non-cognitive outcomes. According to them, these programs may serve as an example and their effects can be generalized to later similar programs^[17][18].

In some recent publications, the Perry and Abecedarian studies were critically examined, particularly regarding their validity and generalization of their outcomes^[19][20][21]. The studies’ findings are often summarized as follows: significant effects occurred on educational achievement, employment, health, and criminal behavior, and these effects sustained well into adulthood. However, there were many limitations to the studies: the programs date from the 1960s and 1970s, that is, in an entirely other era; the experiments were carried out at just one center; the samples were very small and because of the studies’ longitudinal character, became smaller per measurement round; many of the evaluations were carried out by the project leader; the target group was exceptional: low-IQ children (“mentally retarded,” according to Perry’s project leader), from African-American families (many single mothers) with no or little education and often no job, living in high-poverty neighborhoods; parental participation constituted a critical component; the teachers were highly overqualified and received continuous coaching; the teacher-child ration was very low; the programs’ duration and intensity was decidedly unusual, and so were their costs. All things considered, these limitations raise serious doubts about the reliability and validity of the studies’ findings and the possibility of generalizing them to other programs, in other contexts and times.

4. The Chicago Child-Parent Center

In this entry, the focus is on the last of The Big Three, viz. the Chicago Child-Parent Center (CPC) program. The main question is whether CPC leads to positive effects, and if it does, whether these can be generalized to similar programs today. To answer this question, a literature review was conducted.

4.1. The Program

The original CPC was a center-based early intervention program that provided comprehensive educational and family-support services to economically disadvantaged children living in Chicago’s highest poverty neighborhoods. The program was founded in 1967 and federally funded by Title I. Since then, it has been modified to reflect the changing socio-economic diversity. Moreover, it has also been reconceptualized as a school reform model. At present, 35 centers are in operation. The overall goal is to promote educational success, leading to higher educational attainment and, eventually, to greater health and economic well-being. Project children started with CPC in preschool when they were 3 or 4 years of age and continued with an enriched curriculum in early elementary school till third grade. Main components included small classes, activity-based learning, collaborative leadership, family engagement, and alignment of instruction across ages. In addition to activities for the children, CPC comprised a variety of family services, home visits, health and nutrition workshops, and community outreach.

4.2. The Evaluation

The actual evaluation of CPC began in 1986, with the launching of the Chicago Longitudinal Study (CLS). The researchers opted for a matched-group quasi-experimental design, which compared one cohort of CPC kindergartners with that of children of the same age and living in similar socioeconomic circumstances, who until then had not participated in CPC. There were 974 CPC children who had enrolled in 1983 and finished kindergarten in 1986, this was the treatment group; 60 percent of them attended full-day kindergarten, the other half-day. The comparison group consisted of two subgroups: (a) 389 children from randomly selected schools who also graduated from kindergarten in 1986, but who participated in other all-day kindergarten programs (“treatment as usual”); (b) 176 children who started CPC services in kindergarten, but who had not attended the preschool component. The entire sample consisted of 1,539 children; 93 percent of them were African-American and 7 percent Latino; 76 percent grew up in a single-parent family; 54 percent of the mothers had not finished high school and 63 percent were unemployed. Data have been collected from a variety of birth records, K-12 school records, surveys, interviews and administrative records from education, health and earnings. The children were followed till they were 40.

4.3. Main Findings

Numerous positive intervention effects have been reported pertaining to parent involvement; kindergarten readiness; reading and math achievement; classroom adjustment; special education placement; grade retention; school dropout; high school completion; college degree attainment; social-emotional learning; employment; earning a living wage; income; economic well-being; problem behaviors; juvenile arrests; criminal justice involvement; incarceration; conviction for felonies; well-being in midlife; physical health; mental health; child maltreatment and neglect; health behaviors; obesity; smoking; diabetes; substance use; drug and alcohol abuse; cardio-vascular disease; and health insurance coverage.

As was the case with the Perry and Abecedarian evaluations, the CPC research team has a rather biased focus on positive effects, which distracts from the many null and even negative effects. In addition, the positive effects vary depending on participation phase, background characteristics, and life stage. As a result, there are various inconsistencies which require an explanation.

5. Limitations and Criticism

Several researchers have voiced sharp criticism regarding the researchers’ conclusions and have pointed to limitations regarding the generalization of the CPC findings. A selection of them are presented below.

5.1. Target Group

The CPC target group was the same as that of Perry and Abecedarian, viz. children living in socioeconomically disadvantaged families and high-poverty neighborhoods. However, in practice, this implied that virtually all children in the three programs were of African-American descent (CPC also included 7% Latinos). Obviously, this seriously restricts the generalization of the studies’ findings, as there is no information about, for instance, Whites, Latinos, Asians, and labor migrants, asylum-seekers and refugees.

5.2. Sample Size

The CLS sample has several advantages over those of Perry and Abecedarian. It is much larger (in total 1,539, 123, and 111 participants, respectively). In principle, this means that it allows a more precise estimate of the treatment effects, and it is easier to assess the representativeness of the sample and to generalize the results.

CPC has another advantage over the Perry and Abecedarian program, as the latter two were implemented at just one site each. This evidently restricts the programs’ generalization. The CPC program was carried out in 20 centers, which gives its evaluation a sounder and more credible foundation.

Though the CLS sample was large, the children came from just one cohort. By contrast, Perry and Abecedarian were very small, but their samples included 5 and 4 cohorts, respectively. Seen from this perspective, the generalization of CPC is more limited than that of the other two programs.

5.3. Effects

When working with samples, the level of significance should be adapted to sample size. With 1,539 participants, the 1 percent level (p < 0.01) is probably most adequate. In many of the CLS publications, p < 0.05 is applied, with the consequence that the researchers may claim more “significant” effects than is warranted.

For several decades now, most researchers have agreed that presenting p-values is not enough, and that it is imperative to also present sowme sort of effect size (ES), which gives an idea of the magnitude of an effect. Nevertheless, nearly all the Perry and Abecedarian analyses rely on significance only. In the CLS studies, presenting effect sizes is more common, but unfortunately not universal. This hampers the way effects can be interpreted: not only may the number of significant effects be over-estimated, it also remains ambiguous how strong they are.

5.4. Validity

The main threats to the external validity and generalization of findings are selection bias in the program and non-random attrition from the sample. Longitudinal studies such as the CLS, per definition suffer from ever increasing attrition. Compared with Perry and Abecedarian, CPC has a major advantage. While the former studies only had very small samples, the CPC sample is considerably bigger and more robust.

If deemed necessary, the CPC researchers (chiefly Arthur Reynolds and colleagues) used sophisticated statistical techniques to correct for selection and attrition. According to them, possible problems consequently have been solved adequately. Cook and Wong (2007: 14)^[8], however, are of the opinion that the CPC evaluation “depends on an opaque matching procedure and on data analyses (Heckman-type selection models and propensity scores) that have routinely failed to recreate similar effect sizes to an experiment on the same topic.” Therefore, Besharov et al. (2011, p. 3-2)^[15] persist that “high levels of missing data, the absence of a true randomized experiment, and various self-selection biases raise considerable uncertainty about the findings.” As a consequence, analyses by Reynolds et al. have resulted in rather large differences in estimated effects depending on the specific data sets used and their completeness, and the statistical methods used.

5.5. Changing Backgrounds

Background characteristics were collected at the start of the CLS study, for instance, pertaining to the family’s socioeconomic status and the neighborhood. However, such variables are not per se static; in the course of decades, they may change and as a result influence the intervention’s effects. Complicating is that such developments do not apply to all participants in the same way; therefore, some sort of correction is needed. Though the researchers attempted to statistically control for site variation, Besharov et al. (2011)^[15] are not confident that this captured the many differences and changes over time.

In the Perry and Abecedarian evaluations, data collection began at the moment the programs were implemented. This was not the case for the CPC evaluation, however. The program study started in 1983, but it was not until 1986, that, with the launching of the CLS cohort, the main background variables were retrospectively collected. This may influence data reliability negatively.

5.6. Business as Usual

Part of the comparison group consisted of children who had not participated in the preschool CPC program, but in another intervention program (like Head Start), or had not participated in any program. Reynolds et al. (2016)^[22] are of the opinion that by comparing these groups, the analyses test the impact of CPC over and beyond the other childhood services, and that this results in a conservative bias compared with previous studies. This may be so, yet, it would be relevant to check what the children in this comparison group exactly had done until they entered CPC, that is, in what alternative program, or what activities in case they didn’t participate in any program, and for how many years and with what intensity.

5.7. Ethics

The National Academies of Sciences, Engineering, and Medicine (2023)^[5] criticize the terminology used in the original CPC, the deficit perspective (instead of building on the children’s strengths), the definitions of quality, and the lack of attention to the participants’ racial backgrounds. They also challenge the ECE programs’ emphasis on adapting the individual child and his/hers parents to prevailing social structures, without tackling these structural barriers that are responsible for the existence and the preservation of unequal opportunities in the first place.

5.8. Comparisons

The CPC sample consists of one treatment group and two control groups, which allows for numerous comparisons^[4]. Though Reynolds et al. show that for several comparisons the background characteristics do not differ, proof for all comparisons is lacking. Furthermore, there was considerable movement among the groups, such as children who participated in CPC but who nevertheless were included in the comparison group. Yet another point is that in the numerous analyses different numbers of participants were involved, which makes it difficult to compare the results. It is also important that one of the main goals of ECE programs is to give children an even start when they enter elementary school. The comparison group then would be non-disadvantaged children. Regrettably, in many studies such comparisons are hardly ever made; Reynolds (2019)^[23] compares them to the national average. Therefore, it’s not sure whether this goal is achieved.

5.9. Costs and Benefits

The CPC program was relatively cheap. The total costs per child were $18,802 (in 2023 $), while those of Perry were $43,500, and those of Abecedarian $120,000^[21]. Therefore, choosing one of the three, all other things being equal, is simple.

Computing the benefits of an ECE intervention is a perilous undertaking, as the various publications of Reynolds prove (also see Rea & Burton, 2019^[24]). There are differences in outcomes between the studies. Furthermore, there is criticism of the reference points and categories chosen for the estimates. The result is a lot of ambiguity as to what the financial benefits really are^[8]. In addition, there is also confusion regarding differences found between the Big Three. Barnett (2007)^[25] points to huge differences in crime benefits across the three. Perry had enormous benefits from crime reduction, CPC had much smaller but still considerable benefits, and Abecedarian had none at all (crime cost savings (in 2023 $) were $300,744, $62,199, and $0, respectively). The question then is, what could have caused these differences?

5.10. Developments

The fact that at present, nearly sixty years after the first four centers were established, there are still 26 centers in operation, although after substantial revisions, indicates that there is a high level of trust in their success. However, it is important to realize that all the CLS analyses are based on the original CPC version of the 1983-1986 cohort. In the course of years, the original program has been modified, enhanced and reconceptualized. Therefore, any effects reported on the basis of the CLS data probably have little relation with the updated CPC version that is in use nowadays. Thus, there is no proof that the CPC/CLS findings can be generalized to the “new” CPC and other ECE intervention programs today.

The original, evaluated program was carried out some forty years ago. Since then, a lot has changed in society. The question, therefore, is how current and relevant the CLS’s findings still are today. Society is not a static entity, and neither are its citizens. The same applies to pedagogy and didactics.

6. Conclusion and Discussion

Nowadays, ECE programs constitute the central intervention of educational disadvantage policies. Though they date from several decades ago, The Big Three - Perry, Abecedarian, and CPC - today are still frequently put forward as convincing evidence that ECE programs are highly effective in preventing educational arrears of socioeconomically disadvantaged children. Moreover, proponents claim that this effectiveness can be generalized to similar programs today. On the basis of a literature review, this entry critically evaluated both claims.

The CPC researchers report numerous positive effects pertaining to education, physical and mental health, economic well-being, and crime reduction. Though they acknowledge that there are methodological issues, they are of the opinion that these have adequately been tackled. This has not convinced all critics, however.

The CPC evaluation has many advantages over that of Perry and Abecedarian, including sample, sites, and period, and the presentation of effect sizes instead of p-values, but there are also issues that have not been solved convincingly. There are still discussions regarding the non-experimental design of the study, selection bias, and non-random attrition. One insoluble problem all three programs have in common, concerns their target group. While they all claim that socioeconomically disadvantaged children are their target group, in practice participants nearly all were African-American. This implies that any research findings can solely be generalized to African-American disadvantaged children living in high-poverty neighborhoods in America. With all this in mind, it probably is justified to pass “the benefit of the doubt” as a final verdict.

Concerning this entry’s two research questions, the first answer is that yes, there is abundant evidence that CPC has resulted in many positive effects. The second answer, though, is more restricted: to the extent that effects occur, they can only be generalized to a specific part of the target group, viz. African-American disadvantaged children living in high-poverty American cities. Therefore, nothing can be concluded with regard to Whites, Latinos, Asians, labor migrants, asylum-seekers and refugees.

This study shows that while many ECE programs have been implemented, only a few were evaluated using an experimental design. Most evaluations have a retrospective quasi-experimental design, which in itself raises methodological questions. Though neither RCT designs are perfect, they are still being considered as “the gold standard”. This raises the question as to why there are no more experimental studies, which are also large enough to inspire confidence.

Often, ethical objections are being brought forward with regard to experiments with a group of children. But what these adversaries oddly enough forget is that the alternative, the reigning “business as usual” approach, is considerably more objectionable. For many years, millions of young children have been subjected and still are being subjected to all sorts of programs and interventions without any scientific evidence that they are effective.

Note

This entry is based on the peer-reviewed article: Driessen, G. (2025). The Chicago Child-Parent Center early-childhood program: A critical appraisal. Research on Preschool and Primary Education, 3(1), 39-51. ISSN 3005-2629 DOI 10.55976/rppe.32025134339-51