Methodology

The NHVRC team relied on data from multiple sources to develop the national summary of home visiting participants and state profiles. The team gathered quantitative data from publicly available datasets; Maternal, Infant, and Early Childhood Home Visiting Program (MIECHV) administrative data; evidence-based model administrative data; and NHVRC surveys.

The 2019 Home Visiting Yearbook combines 2018 data from various sources to describe—

  • Home visiting in each state through model data
  • The federal contribution to home visiting through MIECHV administrative data
  • Who could potentially benefit from home visiting through data from the American Community Survey (ACS)

Model and MIECHV Data

Data Collection Updates

Since the release of our inaugural yearbook in July 2017, more models and states have been willing to engage with our request for data. For example, all 15 evidence-based models operating in the United States in 2018 shared counts of the number of home visits they provided and of children and families served. The data collection process for the 2019 Home Visiting Yearbook was also more streamlined, partially as a result of increased enthusiasm for the NHVRC’s products and experience gleaned from previous data requests.

For the 2019 Yearbook, we also engaged with other models that have demonstrated a contribution to home visiting but have not received a designation of evidence based from the Home Visiting Evidence of Effectiveness (HomVEE) project. These models are referred to throughout as emerging models. Nine out of nine emerging models responded to our request for data. Recognizing that this list is not comprehensive, we did not combine data from emerging models with data from the evidence-based models in the 2019 Yearbook. Rather, we compiled data from nine emerging models and presented them separately. See the “Expanded Data Collection” section of this appendix for more information on how we selected these models.

Sample and Recruitment

The team collected data from various stakeholders to capture comprehensive information about home visiting at the local, state, and national levels. As with past yearbooks, we reached out to all evidence-based models operating in the United States in 2018 and state MIECHV agencies, and worked with the Administration for Children and Families to gather data on Tribal MIECHV programs. The team received data from—

  • State and territory MIECHV agencies (54 of 56)
  • Evidence-based models (15 of 15) (Source: Models include 15 models operating in the United States in 2018 that met HomVEE criteria for evidence of effectiveness at that time: Attachment and Biobehavioral Catch-Up (ABC), Child First, Early Head Start Home-Based Option (EHS), Family Check-Up (FCU), Family Connects, Family Spirit, Health Access Nurturing Development Services (HANDS), Healthy Families America (HFA), Home Instruction for Parents of Preschool Youngsters (HIPPY), Maternal Early Childhood Sustained Home-Visiting (MECSH), Minding the Baby, Nurse-Family Partnership (NFP), Parents as Teachers (PAT), Play and Learning Strategies (PALS), and SafeCare/SafeCare Augmented.)Go to footnote #>1
  • Emerging models (9 of 9) (Source: Models represented in the emerging models numbers include Baby TALK, Following Baby Back Home, HealthConnect One’s Community-Based Doula Program, Maternal Infant Health Outreach Worker Program (MIHOW), Nurses for Newborns, Parent-Child Assistance Program (PCAP), ParentChild+, Team for Infants Exposed to Substance abuse (TIES) Program, and Welcome Baby.)Go to footnote #>2
  • National Tribal MIECHV program (1 of 1)

Model Administrative Data

We contacted each of the 15 home visiting models operating in the United States in 2018 that met HomVEE criteria for evidence of effectiveness at that time: ABC, Child First, EHS, FCU, Family Connects, Family Spirit, HANDS, HFA, HIPPY, MECSH, Minding the Baby, NFP, PAT, PALS, and SafeCare/SafeCare Augmented. The 2019 Yearbook contains model profiles for two other evidence-based models operating internationally (Early Start in New Zealand and Healthy Beginnings in Australia) but does not include their service numbers in the data presented.

The team sent emails inviting each model to share data on the characteristics of participants served in 2018 and a list of the local agencies that served them. To the extent possible, we requested that participant demographic data mirror MIECHV administrative data required for federal reporting, so we could align model data with data shared by state and tribal MIECHV agencies.

The full data request included the following variables:

Local agency characteristics
  • Agency names and addresses
  • Geographic service areas
  • Total number of full-time equivalent (FTE) home visitors implementing the model at the end of 2018
  • Total number of FTE supervisors implementing the model at the end of 2018
Participant characteristics
  • Total number of children served in 2018
  • Total number of families/households served in 2018
  • Total number of home visits completed in 2018
  • Caregiver ethnicity
  • Caregiver race
  • Caregiver educational attainment
  • Child age
  • Caregiver age
  • Child insurance status
  • Primary language exposure of child
  • Low-income status

Not all models were able to provide data for each variable, but we accepted the data that these models had available. The following number of models shared administrative data:

  • Thirteen models shared local agency information: ABC, Child First, EHS, Family Connects, Family Spirit, HANDS, HFA, HIPPY, MECSH, Minding the Baby, NFP, PAT, and SafeCare/SafeCare Augmented.
  • Fifteen models shared service numbers: ABC, Child First, EHS, FCU, Family Connects, Family Spirit, HANDS, HFA, HIPPY, MECSH, Minding the Baby, NFP, PALS, PAT, and SafeCare/SafeCare Augmented.
    • Twelve of the models provided data on the number of home visits completed: ABC, Child First, Family Spirit, HANDS, HFA, HIPPY, MECSH, Minding the Baby, NFP, PALS, PAT, and SafeCare/SafeCare Augmented.
    • Thirteen of the models provided data on the number of families served: ABC, Child First, Family Connects, Family Spirit, HANDS, HFA, HIPPY, MECSH, Minding the Baby, NFP, PALS, PAT, and SafeCare/SafeCare Augmented.
    • Eleven of the models provided data on the number of children served: ABC, Child First, EHS, Family Spirit, HANDS, HFA, HIPPY, Minding the Baby, NFP, PALS, and PAT.
  • Nine models shared participant data: Child First, EHS, HANDS, HFA, HIPPY, Minding the Baby, NFP, PAT, and SafeCare/SafeCare Augmented.
    • Ethnicity includes data from Child First, EHS, HANDS, HFA, HIPPY, Minding the Baby, NFP, PAT, and SafeCare/SafeCare Augmented. Child First, HANDS, HFA, HIPPY, Minding the Baby, NFP, PAT, and SafeCare/SafeCare Augmented reported ethnicity for adult participants. EHS reported ethnicity for children and pregnant caregivers.
    • Race includes data from Child First, EHS, HANDS, HFA, HIPPY, Minding the Baby, NFP, PAT, and SafeCare/SafeCare Augmented. Child First, HANDS, HFA, HIPPY, Minding the Baby, NFP, PAT, and SafeCare/SafeCare Augmented reported race for adult participants. EHS reported race for children and pregnant caregivers.
    • Educational attainment includes data from Child First, EHS, HANDS, HFA, HIPPY, Minding the Baby, NFP, and PAT.
    • Child age includes data from Child First, EHS, HANDS, HFA, HIPPY, Minding the Baby, NFP, and PAT.
    • Child insurance status includes data from Child First, EHS, HANDS, HFA, HIPPY, NFP, and PAT. Public insurance includes Medicaid, Children’s Health Insurance Program (CHIP), and TRICARE.
    • Primary language includes data from Child First, EHS, HFA, HIPPY, NFP, and SafeCare/SafeCare Augmented. EHS reported primary language for children and pregnant women. SafeCare/SafeCare Augmented reported languages spoken in the home. Child First, HIPPY, and NFP reported primary language of children. HFA reported primary language of adult participants.

Although models do not uniformly report data, the NVHRC team combined as much of the data we received as possible. These data represent the most comprehensive summary of home visiting services provided by evidence-based home visiting models across the nation. We aggregated data across models and then used the summarized data to create—

  • The NHVRC National Profile featuring model data on service numbers and participant demographics
  • NHVRC State Profiles featuring model data on service numbers and participant demographics by state and ACS data on potential beneficiaries by state
  • NHVRC Model Profiles featuring model data on service numbers, participant demographics, survey information on model requirements, and geographic information on where models operate

MIECHV Administrative Data

MIECHV legislation requires awardees to report data yearly to the federal government. These data include information such as the number of home visits conducted, number of participants served, and participant demographics. The team asked MIECHV agencies in each state to share a copy of this administrative data report. Most were able to share data, but a few territories were not.

The following number of agencies supplied MIECHV administrative data:

  • State MIECHV agencies (54 of 56)
  • National Tribal MIECHV program (1 of 1)

We used the state MIECHV administrative data reports to produce the MIECHV State Data Tables.

Expanded Data Collection

Back in 2018, we broadened our description of the home visiting landscape by expanding data collection to include emerging models. The following details our process for identifying the nine models first included in the 2018 Home Visiting Yearbook and included again this year.

We started by creating a list of potential models to include. Models were included if they met one of the following criteria:

  • Reviewed by HomVEE but had not yet reached HomVEE evidence-based status
  • Being evaluated through MIECHV as a promising approach
  • Recognized as evidence based in either the Substance Abuse and Mental Health Services Administration’s National Registry of Evidence-based Programs and Practices (NREPP) or the California Evidence Based Clearinghouse for Child Welfare (CEBC)

We brought this list to members of the Advisory Committee for their expert feedback and for suggestions of additional models. After receiving their feedback, which included Welcome Baby and First Born, we refined the list following the process below.

  • We first removed models that were not operating in the United States or were no longer being implemented anywhere. This resulted in 23 models. Of these, 21 were listed by HomVEE, NREPP, or CEBC.
  • We then reviewed the models to determine if they exclusively served prenatal women and children 0–5 years. This ensured we could have accurate counts of families with children in the target age group for early childhood home visiting. This step narrowed the list to 17 models.
  • We then sent the list to the Advisory Committee for final review. This resulted in a final list of 13 models to contact.

We reached out to these 13 models, asking them to provide information on their model, service delivery information, and if available, participant demographics.

  • Nine models shared number of home visits completed, number of families served, number of children served, and participant data: Baby TALK, Following Baby Back Home, HealthConnect One’s Community-Based Doula Program, MIHOW, Nurses for Newborns, PCAP, ParentChild+, TIES Program, and Welcome Baby

We aggregated data across models and then used the summarized data to create—

  • The emerging model section of the Yearbook, featuring model data on service numbers and participant demographics
  • NHVRC Model Profiles featuring model data on service numbers, participant demographics, survey information on model requirements, and geographic information on where models operate

Surveys

Based on feedback, the NHVRC team dropped our request for state MIECHV agencies and models to complete a survey for the Home Visiting Yearbook. Some exceptions were made for—

  • Models that did not complete the survey for a prior NHVRC publication
  • Models that recently received an evidence-based designation from HomVEE
  • Models operating internationally only

The survey covered content related to program, participant, and community characteristics; service capacity and enrollment; program implementation; and funding. Models were asked to share programmatic data, not individually identifiable information. All models had the opportunity to review their program information and to include updates prior to the release of the 2019 Home Visiting Yearbook. Survey data were used to develop the model profiles.

Data Analysis

We conducted a rigorous data cleaning and analysis procedure for all data sources. For the model data, we reviewed each model dataset to determine which data elements were available among those in our initial data request. We then examined all models to determine how to combine and report data uniformly across models for state and national profiles. We then cleaned the data to ensure all reported elements were complete. Next, we combined data across models using statistical analysis software. NHVRC staff double-entered state MIECHV administrative data to ensure accuracy before the software analysis.

To maintain the confidentiality of model and state data, we conducted cell suppression of variable categories with 10 or fewer participants. These counts are replaced with a * in the data.

NHVRC data and communications teams verified the final profiles before they were presented to state and model staff for additional review. In coming years, we will continue to work with states and models to address unique data issues and questions as they arise while adhering to our systematic protocols.

American Community Survey Data and Documentation

The 2019 Home Visiting Yearbook catalogs national- and state-level information on potential beneficiaries of home visiting using information from the ACS. We first define potential beneficiaries broadly. We then examine subgroups of families who might be a higher priority for services based on several targeting criteria. ACS data were analyzed for all 50 states and the District of Columbia, but not for territories or tribal communities.

Data Source

The team relied on the 2017 ACS 5-year (2013–2017) file, accessed through the Integrated Public Use Microdata Series (IPUMS). The ACS is a nationwide, ongoing survey designed to provide data on demographic, housing, social, and economic issues. IPUMS grants access to ACS microdata, where each record represents a person.

Potential Beneficiaries of Services

We define potential beneficiaries of home visiting services as families and subfamilies with pregnant women and/or children under 6. (Subfamilies are families that live in the household of someone else.) First, we estimate the number of families and subfamilies with children younger than 6 years old who are not yet enrolled in school (that is, not in kindergarten or a higher grade). To this estimate, we add an estimate of the number of families and subfamilies that include a pregnant woman and are not otherwise counted.

Estimates of pregnant women are based on adjusted counts of families with infants because the ACS does not identify pregnancy status. Specifically, we count the number of families with infants but no other children under age 7 in first grade or higher, as a proxy estimate of pregnant women without a child under age 6 not yet enrolled in kindergarten (assuming rough stability in the number of births from one year to the next). We multiply the number of families with infants by 0.75 to account for 9-month pregnancy. (Source: We do not attempt to refine the estimate to account for (1) fetal and infant deaths, or (2) the lag in time before a woman’s pregnancy would be verified; the first adjustment would raise the estimate of pregnant women not already counted, while the second would lower it.)Go to footnote #>3

Families With High Priority For Services

To identify a subpopulation of “high-priority families,” we count the number of families with young children and pregnant women who meet one of five different economic and demographic criteria (as defined below) and the number of families that meet at least two such criteria. We conferred with the Advisory Committee to select our targeting criteria. Although other criteria could also be considered, we chose these because they align with several of the priority areas from the MIECHV legislation, they align with several of the model requirements for enrollment, and they are available in the ACS.

Targeting Criteria

We estimate the number of families with preschool children under 6 and pregnant women who meet each of the following criteria at the national and state levels:

  • Presence of an infant; that is, a child younger than 1 year old. By definition, none of the pregnant women without children under 6 meet this criterion.
  • Low income, where family income is below 100 percent of the federal poverty threshold.
  • Young mother or young pregnant woman. We define young as under 21 years old.
  • Single mother, never married.
  • Low parental education. We count the number of families in which the child’s parent(s) have not completed 12th grade. (Source: In two-parent households, we consider both parents’ educational levels; in one-parent households, we consider only that parent’s educational attainment. For pregnant women, we look at the education of the mother only.)Go to footnote #>4