Early Childhood Integrated Data System (ECIDS)

Introduction 

What is ECIDS?
The Early Childhood Integrated Data System—the first-ever initiative in Massachusetts to link participation across Massachusetts state programs serving children birth-to-five—is a powerful tool that allows us to explore key policy questions about the birth-to-five (B-5) population and the services provided to them. Among its many uses, ECIDS gives us the power to:
  • Delve deeper into questions about the demographics (race, gender, language spoken, family income, location) of the children enrolled in each of these programs.
  • Examine the overlap of participation [1] across more than one program in ECIDS.
  • Identify potential gaps in services to inform cross-sector collaboration and policy changes to better support children's well-being and academic performance.
Of the many programs that Massachusetts offers to the birth-to-five population, six are currently included in ECIDS (Figure 1). There are also programs for this age group that do not currently provide data to ECIDS [2], such as Head Start, Early Head Start [3], MassHealth services for young kids, and other programs that are funded by other child-serving departments (e.g.: DMH, DCF, DDS, etc.).
The Origins of ECIDS 
Funded by the Federal Preschool Development Grant B-5, ECIDS was enabled through the development of the PDG Data Use Licensing Agreement (DULA), a data-sharing agreement between participating agencies that provides the legal terms and conditions for sharing and integrating child-level data to inform and improve the administration and impact of programs serving young children and their families. This pivotal agreement, signed in July 2020 by participating agencies, emerged as a result of collaborative efforts between agencies, including The Children’s Trust (CT), Department of Early Education and Care (EEC), Department of Elementary and Secondary Education (DESE), Department of Public Health (DPH), Executive Office of Education (EOE), Executive Office of Health & Human Services (HHS), and Executive Office of Technology Services and Security (EOTSS). The signing of the DULA marked a significant milestone toward building ECIDS, giving us the first-ever view of the total unduplicated children who are enrolled in these services, and the ability to glean insights into essential policy questions about birth-to-five programs and the children who participate in these programs.
Data Use Licensing Agreement Analytical Questions  
As part of their collective undertaking while developing the DULA, participating agencies identified seven analytical questions that ECIDS may answer. These seven questions were crafted to address questions ranging from the demographics of children participating in these programs; the number of program participants screened by the Ages and Stages Questionnaire [4] and the outcome of the screening; the number of program participants receiving special education services in public preschool; and other data related to third grade MCAS scores, and chronic absenteeism in first, second, and third grade. See the Appendix for a list of the analytical questions in the DULA.
Programs that Report Data to ECIDS 
The programs included in ECIDS support children’s healthy development and well-being through high-quality childcare, home-based services, and special education in public preschools. Programs vary in the age ranges they serve and their eligibility criteria, as some programs are designed to reach low-income families or underserved communities, while other programs are available to families whose children meet a specific set of developmental needs. ECIDS contains program participation data starting July 2015, and these data points are updated quarterly. This report includes program participation data from July 2015 to September 2022. Figure 1 below describes these programs in greater detail.
The following is a list of agencies that manage these programs and provide data to ECIDS:
Interpreting ECIDS Data
While ECIDS is a powerful tool that helps answer key questions, it is essential to know its limitations to avoid making incorrect or inappropriate interpretations. ECIDS does not allow us to:
  • Determine cause and effect (e.g.: program effectiveness) [5].
  • Draw inferences about the driving forces behind trends without understanding the broader societal context, the program context, or the type of service provided.
Furthermore, knowing the context of the data is critical to understanding the data.
  • Program context: Figure 1 describes the programs that report data to ECIDS, including whether the program is means-tested. While these programs—particularly Home Visiting and the Commonwealth Preschool Partnership Initiative—do not all fall neatly into means-tested versus not means-tested, this report separates programs in this manner to avoid comparing programs that differ in crucial ways. Of note, Home Visiting and CPPI are not means-tested programs, but they tend to follow the same trend as means-tested programs because the eligibility criteria are designed to reach underserved and low-income communities.
  • Broader societal context: As ECIDS data shed light on questions of equity, it is important to keep in mind the role that past and current structural inequalities played and continue to play in service utilization among people of color [6].
Understanding Differential Privacy
The ECIDS dashboard displays data treated to ensure differential privacy (DP), which is a statistical approach that adds “noise” to aggregated data to prevent the re-identification of child-level data when cell sizes meet a particular threshold. The ECIDS Board of Directors went through a year-long process to identify an approach to the data that enables multiple filters to be applied in order to focus on a particular policy question, including the ability to drill down into race and other demographics to understand the data with an equity lens. However, a tradeoff was made, so with DP-treated data, the dashboard is only able to display the true count within a margin of error. The smaller the number of kids we are examining, the larger the margin of error, so programs with fewer children, such as CPPI and Home Visiting, have higher margin of errors. Refer to the Appendix for the margin of error tables of each figure.

Purpose of this Report

This report is designed to provide an overview of ECIDS and demonstrate its power in producing significant insights, especially related to equity. It also seeks to generate discussions about policy topics by offering a set of questions in each section that the reader may consider exploring further.
Notably, while this report touches heavily on the first analytical question in the DULA, it is not designed to answer any of the analytical questions outlined in the DULA nor draw any conclusive findings or inferences from the data. See the Appendix for a list of the analytical questions in the DULA.

General Demographics and Trends of Birth-to-Five Children Participating in Programs that Report Data to ECIDS (July 2015 - September 2022)

Programs are reaching communities of color. Programs reporting data to ECIDS serve a large proportion of people of color relative to the Massachusetts birth-to-five (B-5) population, with means-tested programs typically serving more people of color than programs that are not means-tested. Take, for example, the demographics of CCFA participants:
  • 39% of Childcare Financial Assistance (CCFA) participants are Hispanic/Latinx versus 11% of the Massachusetts B-5 population and 27% of the Massachusetts B-5 population under the federal poverty line [7]. Only 21% of CCFA participants are white versus 74% of the population and 50% of the population under the federal poverty line (Figure 2).
  • White children make up 52-55% of Public PreK participants and 51% of Early Intervention participants compared to 74% of the Massachusetts B-5 population [8] (Figure 2).
  • 13% of CCFA participants speak Spanish versus only 9% of the overall Massachusetts population [9] (Figure 3).
Children of color are accessing more than one program that reports data to ECIDS. The percentage of children of color is larger when looking at participation in more than one program in ECIDS versus only one program in ECIDS. Take, for example, the demographics of CPPI participants:
  • Asians, Black/African Americans, and Hispanic/Latinx make up 71% of kids who participate only in CPPI and no other program, compared to 85% of Asians, Black/African Americans, and Hispanic/Latinx who participate in CPPI and at least one other program that reports data to ECIDS [10] (Figure 4).
Relative to other programs that report data to ECIDS, Early Intervention and special education in public preschool are accessed more commonly by boys.
  • Boys comprise 62% of Early Intervention participants and 70% of children with IEPs in public preschool. Other programs show a fairly even split between boys and girls, though more girls (54%) participate in the Home Visiting program (Figure 5).

The Use of Integrated Data

Integrating data allows us to gain insights into potential service gaps and barriers to accessing services. The case below demonstrates potential language and/or cultural barriers to accessing programs that report data to ECIDS.
WIC and Home Visiting are services that begin pre-birth and support families with referrals, but they have different trends.
  • WIC serves many children, and 46% of children in WIC participate in at least one other program that reports data to ECIDS (Figure 6).
  • Home Visiting serves a small number of children, and 73% of children in Home Visiting participate in at least one other program that reports data to ECIDS (Figure 6).
While further research is needed to understand these trends, we explore below three possible explanations for the difference.
1. WIC serves far more children. The percentage of children who participate in WIC and at least one other program in ECIDS may be lower than other programs because WIC targets the entire population of families with young children who meet the income-eligible threshold, and only a portion of those children may have needs that require additional services from the programs that report data to ECIDS.
  • WIC’s total population is 337,755 compared to 7,034 children for Home Visiting (Figure 6).
2. Families who receive WIC services may participate in other means-tested programs such as SNAP, fuel assistance, and MassHealth, but these programs do not report data to ECIDS. Within ECIDS, the largest overlap with WIC is CCFA and Early Intervention. Of the children who participate in WIC and at least one other program that reports data to ECIDS:
  • 30% participate uniquely in WIC and CCFA
  • 20% participate uniquely in WIC and Early Intervention
  • 11% participate uniquely in WIC, Early Intervention, and CCFA
3. There may be language and/or cultural barriers to accessing services.
  • Non-English speakers and children for whom the language is flagged as “unknown” make up 45% of all kids who participate only in WIC and no other program that reports data to ECIDS. In comparison, 27% of non-English speakers and children for whom the language is flagged as “unknown” participate in WIC and at least one other program that reports data to ECIDS (Figure 7).
  • This observed trend of lower percentages of non-English speakers participating in more than one program is consistent across means-tested programs, Home Visiting, and CPPI [11](Figure 7).

Conclusion

This report only scratches the surface of important insights that ECIDS can provide about the demographics of children who participate in programs that report data to ECIDS. Many other analyses are possible with ECIDS data, such as examining participation trends over time within programs and across programs by race, language, gender, age, and gateway cities to understand if there have been any changes to the demographic of kids that programs are reaching. These analyses allow us to learn about opportunities to improve access to services and coordination of services to address children’s well-being and academic needs.
By Ana Vasconcelos, Oct. 18, 2023

Footnotes

[1] Participation and enrollment are defined as a child enrolled in at least one program for at least one day.

[2] The data presented in this report can only show a child’s participation within programs that report data to ECIDS. They do not show if children are enrolled or not enrolled in other programs outside of programs in ECIDS.

[3] Some children in Head Start and Early Head Start are represented in CPPI and CCFA.

[4] The Ages & Stages Questionnaires (ASQ) is a series of developmental screening tools used to assess the developmental progress of children from birth to 6 years old. These questionnaires are designed to help parents, caregivers, educators, and healthcare professionals identify potential developmental delays or areas of concern in children's development. These questions are on hold as statewide ASQ data are not yet available.

[5] Program data demonstrate correlation, not causation. For instance, we cannot determine a program’s effectiveness by merely observing that more children who participate in more programs do not receive special education services in school later. Besides program participation, many other factors may influence a child’s participation in special education services in school, so it is impossible to determine causation from this observed trend.  

[6] For a few examples of how structural inequalities can impact people of color’s access to services and participation trends, see recent reports by diversitydatakids.org, Child Trends, or the Early Childhood Technical Assistance Center.

[7] Data for the Massachusetts birth-to-five population and birth-to-five population under the federal poverty level are from the Census tables B17001(A-I) “Poverty Status in the Past 12 months by Sex by Age” ACS 5-year estimates subject tables 2015-2020. Due to the way in which the Census Bureau calculates race/ethnicity, census totals for people of color are slightly inflated relative to the agency counts. In addition, research suggests the way the Census Bureau measures poverty undercounts the number of people living in poverty.

[8] This trend [of lower proportion of people of color and girls] in Early Intervention and special education services is observed across multiple states. For examples of existing research on this trend, refer to this report by the Early Childhood Technical Assistance Center.

[9] Refer to the ACS survey for more information.

[10] The data do not show if these children are participating in programs that do not report data to ECIDS.

[11] Home Visiting and CPPI are not means-tested programs. However, they tend to follow the same trend as means-tested programs because the eligibility criteria are designed to reach underserved and low-income communities. Refer to Figure 1 for more information on the eligibility criteria.

Appendix

Analytical Questions in the DULA
1. Of the number of children in Massachusetts who are age five (5) and under, how many are participating in one or more programs for [period of time] including all participation and termination dates by demographics?
2. Of children who participated in programs, how many scored meeting expectations or above on the third grade English Language Arts Massachusetts Comprehensive Assessment System (ELA MCAS) by demographics?
3. Of children who participated in these programs, how many of those children were screened using Ages and Stages Questionnaire (ASQ) by demographics, how many of those screenings resulted in [potential concern, strong concern, or no development concerns], and did those children score proficient or above on the third grade ELA MCAS by demographics?
4. Of children who participated in at least one program, how many received an Individualized Education Plan (IEP) or Individualized Family Support Plan (IFSP) in [pre-school, kindergarten, first grade, second grade, third grade, ever?] by demographics, and how many accepted/did not accept the IEP or IFSP by demographics?
5. Of children who participated in at least one program and were referred to another program, how many [did not accept the referral, enrolled in the program] by demographics?
6. Of children who participated in programs, how many participated in programs that have met or exceeded quality benchmarks recognized by the state and how many of those children had ASQ screenings that resulted in [potential concern, strong concern, or no development concerns]?
7. Of children who participated in at least one program other than public education, how many were chronically absent in [first grade, second grade, third grade]?
Standard Error Tables
Interpreting Standard Error Tables
To interpret standard errors, refer to the examples below:
  • Example with count: Table 1 shows that 5,319 of children participate in CPPI with a standard error of 222. This can be interpreted as: “Roughly 70% of the time, the true number of children who participate in CPPI will fall in the range of 5,319 children ± 222 children.".
  • Example with percentage: Table 4 shows that 51% of Hispanic/Latinx kids participate in CPPI with a standard error of 5%. This can be interpreted as: “Roughly 70% of the time, the true percentage of Hispanic/Latinx children who participate in CPPI will fall in the range of 51% ± 2.55% (e.g., 0.05 X 0.51).".
Note that CPPI and Home Visiting have larger standard errors compared to other programs because fewer children participate in these programs, and the smaller the sample size of the population, the larger the standard error to protect against the re-identification of the child.
Browse public data  |  Subscribe to our newsletter  |  Contact us