Table of Contents
- What the numbers say about family preservation
- Household structure matters
- How family formation has shifted
- Program outcomes and service retention
- State-level child welfare signals
- What to watch when reading the data
What the numbers say about family preservation
Family preservation statistics are most useful when they are read as a set of related signals instead of isolated headline figures. The source data here shows how household structure, parenting patterns, and child welfare outcomes all point to the same core issue: keeping families stable is shaped by both demographic change and service design.
A useful way to read this topic is to separate household-level trends from program-level results. The first group tells you how family life is changing in the United States overall. The second group shows what happens when prevention, voluntary services, and family preservation programs are available to families who are already under stress.
Household structure matters
The broad national picture starts with how Americans are living. In 2024, about 64% of U.S. households were family households (Census Bureau), while 29% of all U.S. households were one-person households in 2024, equal to 38.5 million one-person households (Census Bureau). That combination matters because family preservation efforts are always operating inside a larger housing and household environment.
The long view is even more revealing. In 1974, 79% of U.S. households were family households (Census Bureau), compared with 64% in 2024 (Census Bureau). That is a substantial shift in how household life is organized. If family preservation programs are trying to keep households intact, they are doing that work in a context where fewer households now fit the traditional family-household model than they did a generation ago.
A compact comparison helps show the direction of change:
| Indicator | Earlier figure | Later figure | Source |
|---|---|---|---|
| U.S. households that were family households | 79% in 1974 | 64% in 2024 | Census Bureau |
| U.S. families living with their own children under age 18 | 54% in 1974 | 39% in 2024 | Census Bureau |
| Adults ages 18 to 24 living in their parental home | 57% in 2024 | 57% in 2024 | Census Bureau |
| Adults ages 25 to 34 living in their parental home | 16% in 2024 | 16% in 2024 | Census Bureau |
| One-person households | not stated for 1974 in this dataset | 38.5 million in 2024 | Census Bureau |
The table shows more than a single trend. It shows that household composition is changing at several levels at once. In 2024, 81% of nonfamily households were homes with someone living alone (Census Bureau), which reinforces how common solitary living has become within the broader household mix.
Family preservation statistics also connect to who is living with whom across the life course. In 2024, 57% of adults ages 18 to 24 lived in their parental home (Census Bureau). Among adults ages 25 to 34, 16% lived in their parental home in 2024 (Census Bureau). Those two figures suggest that the transition into independent household formation is not uniform across young adulthood. For anyone tracking family preservation, that matters because extended dependence, delayed household formation, and multi-stage family transitions can affect both risk and support patterns.
How family formation has shifted
Family preservation is not only about child welfare intervention. It is also about the shape of family formation itself. The dataset gives two important markers: who is counted in family households and when people marry.
In 2024, 74% of family households were married-couple households (Census Bureau). Among family households without a married couple, 68% had a female householder in 2024, while 32% had a male householder (Census Bureau). Those shares matter because they show that a meaningful portion of family life is now organized around household structures other than married couples. When a family preservation program is designed around a narrow model of family structure, it risks missing how many households actually look in practice.
Another useful marker is age at first marriage. The estimated median age to marry for the first time was 30.2 years for men in 2024 (Census Bureau). For women, the estimated median age to marry for the first time was 28.6 years in 2024 (Census Bureau). That is relevant to family preservation because later marriage often means a longer period in which adults form households, parent, or coordinate care before marriage enters the picture at all.
The share of all U.S. families living with their own children under age 18 also shifted sharply. In 1974, that share was 54% (Census Bureau). In 2024, it was 39% (Census Bureau). That is a major demographic change, and it gives context to any discussion of preserving family stability. A preservation strategy cannot assume the same baseline household pattern that existed decades ago.
What emerges from the census figures is not a single story of decline or progress. It is a story of diversification. Households are smaller, more varied, and more likely to include people living alone or adults living in parental homes than many older family models would suggest. For family preservation advocates, the practical lesson is straightforward: the definition of “family at risk” must match the current structure of family life.
Program outcomes and service retention
The most direct family preservation statistics in this dataset come from service outcomes. These figures are useful because they show what happens after families are offered support. They also show that service completion can be a major dividing line.
New Jersey provides one of the clearest examples. About 85% of all cases served by the department are voluntary (Casey Family Programs). That matters because voluntary models often depend on engagement rather than coercion. If a family chooses to participate, the service model may have a better chance of keeping children safely at home.
In New Jersey, 86% of children offered Family Preservation Services remained at home 12 months after receiving services (Casey Family Programs). That figure is one of the strongest direct preservation indicators in the dataset. It suggests that a substantial majority of children served in that setting did not enter out-of-home placement during the follow-up period. For a family preservation statistic, that is exactly the kind of outcome the field is trying to achieve.
New York City offers a different kind of evidence: service volume and completion. In FY2017, 19,494 families received prevention services (Casey Family Programs), representing 44,445 total children (Casey Family Programs). Those are large numbers, and they show the scale at which family support systems can operate when prevention is built into child welfare practice.
The gap between completion and non-completion is especially important. Families that completed preventive services had a 2.6% repeat maltreatment rate (Casey Family Programs). Families that did not complete preventive services had a 14.3% repeat maltreatment rate (Casey Family Programs). That is a wide spread. It does not prove every causal detail on its own, but it strongly indicates that completion is associated with a much better outcome on this measure. If a family preservation program is measured only by enrollment, it misses one of the most consequential parts of the story. Completion appears to be a key outcome driver.
Oklahoma adds another dimension. In state FY2018, 2,086 families were served by the agency’s voluntary services (Casey Family Programs). In the same broader service environment, 80% of children served by intensive-level services remained at home during the first three years of the waiver demonstration project (Casey Family Programs). Those children also had no re-entry or re-referral within 12 months (Casey Family Programs). That combination suggests that intensive support can preserve placement stability when the service model is strong enough to hold families through a longer period.
Washington gives a more cautious signal. In 2017, 31.2% of children served by Family Voluntary Services were involved in a screened-in CPS intake after case closure (Casey Family Programs). That figure does not fit the same success pattern seen in New Jersey or Oklahoma, but it is still useful. Family preservation outcomes vary by program design, family needs, and local system context. A serious reading of the data has to make room for variation rather than forcing every jurisdiction into the same narrative.
State-level child welfare signals
The state outcome figures in this dataset are not all direct family preservation measures, but they matter because they show the environment in which preservation work happens. If maltreatment, foster care entry, or reentry changes, preservation programs are operating against a moving baseline.
New Jersey shows several positive shifts between 2019 and 2023. Child maltreatment victims fell from 5,132 in 2019 to 2,951 in 2023 (New Jersey Child Welfare Outcomes). The foster care entry rate fell from 1.4 in 2019 to 0.8 in 2023 (New Jersey Child Welfare Outcomes). Children in foster care on 9/30 fell from 4,458 in 2019 to 2,804 in 2023 (New Jersey Child Welfare Outcomes). At the same time, children receiving monthly visits were 97.9% in 2019 and 98.0% in 2023 (New Jersey Child Welfare Outcomes). Children receiving visits in the home were 97.9% in 2019 and 96.6% in 2023 (New Jersey Child Welfare Outcomes).
That cluster is important because it shows that lower foster care involvement can coexist with very high visit compliance. It also suggests that family preservation is not only about keeping children out of care; it is about maintaining consistent contact and support once services are in motion.
For older children in New Jersey, reunification and adoption percentages shifted downward. Reunification for children older than age 12 at entry was 53.8% in 2019 and 47.8% in 2023 (New Jersey Child Welfare Outcomes). Adoption for children older than age 12 at entry was 6.8% in 2019 and 2.8% in 2023 (New Jersey Child Welfare Outcomes). Those changes are worth noting because they remind readers that aggregate improvements do not necessarily mean every permanency pathway moved in the same direction.
North Dakota provides another useful comparison. Children subject to an investigated report alleging maltreatment declined from 7,295 in 2018 to 4,764 in 2022 (North Dakota Child Welfare Outcomes). Total child maltreatment victims declined from 2,097 in 2018 to 1,132 in 2022 (North Dakota Child Welfare Outcomes). Child fatalities were 8 in 2018 and 6 in 2022 (North Dakota Child Welfare Outcomes). Those are large reductions in counts, and they frame the scale of child welfare involvement more broadly.
North Dakota also shows how the profile of victims can change. The share of victims under age 1 rose from 12.3% in 2018 to 19.7% in 2022 (North Dakota Child Welfare Outcomes). Alaska Native / American Indian children were 23.7% of child victims in 2022 (North Dakota Child Welfare Outcomes). Emotional abuse was identified in 65.4% of child victims in 2022 (North Dakota Child Welfare Outcomes). Physical abuse was identified in 78.4% of child victims in 2022 (North Dakota Child Welfare Outcomes). These figures highlight that family preservation and child safety work must be tailored to the composition of the affected population, not just the total count of cases.
Placement and exit outcomes also matter. In North Dakota, children maltreated while in foster care were 0.15% in 2018 and 0.00% in 2021 (North Dakota Child Welfare Outcomes). Reunification accounted for 57.1% of foster care exits in 2018 and 54.6% in 2022 (North Dakota Child Welfare Outcomes). That suggests reunification remained the dominant exit path even as the overall case mix shifted.
West Virginia adds a different pattern. Child maltreatment victims fell from 6,946 in 2018 to 5,510 in 2022 (West Virginia Child Welfare Outcomes). The foster care entry rate fell from 14.0 in 2018 to 12.7 in 2022 (West Virginia Child Welfare Outcomes). Children in foster care on 9/30 fell from 7,437 in 2018 to 6,952 in 2022 (West Virginia Child Welfare Outcomes). Children receiving monthly visits were 93.8% in 2018 and 91.3% in 2022 (West Virginia Child Welfare Outcomes). Reunification accounted for 56.3% of foster care exits in 2018 and 48.0% in 2022 (West Virginia Child Welfare Outcomes).
What to watch when reading the data
A strong family preservation reading should keep three layers separate. First, the household layer: family households, one-person households, parental-home living, and marriage timing. Second, the service layer: voluntary cases, prevention services, completion rates, and placement stability. Third, the child welfare layer: maltreatment counts, foster care entry, reunification, adoption, and reentry.
These layers are connected, but they do not mean the same thing. A rising share of one-person households does not by itself prove that family preservation is weakening. A high reunification rate does not automatically mean preventive services are sufficient. A lower foster care entry rate does not tell you whether the families receiving services were less complex, better supported, or simply counted differently. The value of the statistics is in the pattern they create together.
For policy readers, the clearest practical signal in the dataset is the difference between completion and non-completion of preventive services in New York City: 2.6% versus 14.3% repeat maltreatment (Casey Family Programs). For household readers, the strongest structural signal is the long shift from 79% family households in 1974 to 64% in 2024 (Census Bureau). For state systems, the most useful comparison is probably the combination of lower maltreatment counts and lower foster care entry with high visit compliance, as seen in New Jersey (New Jersey Child Welfare Outcomes).
Family preservation statistics are most persuasive when they stay anchored to the exact setting that produced them. National household data tells one story. Local service data tells another. State child welfare outcomes tell a third. Taken together, they show that preservation is not a single metric but a system of related measures that must be read in context.