Is There a Foster Care-to-Prison Pipeline? Dee Wilson Says the Honest Answer Is Harder Than Either Side Wants
Dee Wilson’s October 2026 Sounding Board takes on a phrase most of us in this work have used without examining it. The foster care-to-prison pipeline. He concludes that the research does not support it as a blanket claim, and that the reason it persists is that two very different groups of children keep getting counted as one.
I have been reading Dee for years. He spent 26 years inside Washington State’s system, from caseworker up to regional administrator, then ran a research institute at the University of Washington and spent six years at Casey Family Programs. When he says the evidence is being misused, he is not defending the system. He has spent a career describing its failures. He is saying the argument is being made badly, and that badly made arguments lose.
The book that prompted it
The commentary starts with Claudia Rowe’s 2025 book Wards of the State, which follows six foster care alumni, now aged eighteen to fifty-six. These are not interview subjects. Rowe built relationships with them over time, and the stories are devastating. All six were removed between ages five and thirteen after severe maltreatment by parents or stepparents. All moved through multiple placements. All were rejected repeatedly. Several were homeless and hungry for long stretches. Several were abused in care itself. Two were convicted of murder, one of them a sixteen-year-old girl.
Dee does not dispute any of it. His objection is to a claim the book makes about scale, namely that of the more than 400,000 American children in state care, half were headed for locked cells. He traces that figure to a misreading of the Chapin Hall Midwest Study, which followed seventeen-year-olds aging out of care in four states. Those young people are not a random sample of foster children. They are disproportionately kids whose permanency plans already failed, with severe behavioral and educational needs. The Midwest Study is a damning indictment of how four states prepare teenagers to leave care. It cannot tell you what happens to a six-year-old placed this year.
Two studies, opposite findings, and why both can be right
The pipeline claim rests heavily on Joseph Doyle’s Illinois research, which found that foster placement roughly tripled later arrest, conviction and imprisonment rates compared to similar children left at home. That study has been cited for nearly two decades.
In 2022, Jason Baron and Max Gross ran the same method on about 120,000 Michigan CPS investigations from 2008 to 2016 and found the opposite. Children placed in foster care were 25 percentage points less likely to be arrested by age nineteen, with similar reductions in conviction and incarceration.
Here is the part Dee insists on, and it is the whole argument. Both studies look only at what researchers call marginal cases, where a stricter investigator would remove and a more lenient one would not. By design, that method excludes the children in Rowe’s book. Those cases are severe enough that any investigator would have removed. So Rowe and Baron and Gross are describing two different populations, and neither one describes foster care as a whole.
That is not a technicality. It is the reason the public argument goes in circles.
What actually predicts the bad outcome
Dee points to a third body of work that gets far less attention and matters more.
An Australian longitudinal study by Barber and Delfabbro followed 235 children entering care. Within it, about fifty conduct-disordered children were moved again and again, living in conditions the authors describe as effectively homeless while in care. The authors concluded that family-based foster care does not work for disruptive children and that the field urgently needs something else.
But the same study found that most of the children liked their foster families, felt cared for, and that the experience supported their wellbeing.
Rubin and colleagues, using national NSCAW data, found roughly one in five foster children never achieved nine continuous months in a single placement. Stability predicted every permanency outcome – reunification, adoption, guardianship.
So the predictor is not foster care. It is whether a child gets stable early. Children who stabilize in the first year generally do well. Children who cannot be stabilized cycle into residential care, then detention, lose contact with family, and get more disruptive as they go. Dee is blunt about what drives that: for those young people, anger is a defense against despair. As a CASA Guardian ad Litem two of my caseload children had over 15 placements each while in my care and placement disruption was common in general.
The uncomfortable finding
Baron and Gross also argue that prevention services without removal show little evidence of meaningful impact. They cite Michigan’s Safe Care program, where only 5 to 10 percent of families took it up and 1 to 2 percent completed it. Their position is that removal and court involvement are often what motivate a parent to engage at all.
I would ask readers of this blog to sit with that one. It cuts directly against where Minnesota law has moved. Family First redirected money toward voluntary prevention services. Minnesota’s African American Family Preservation Act raises the bar for removal statewide on 01/01/2027. If the Michigan finding holds, we are expanding the voluntary path at exactly the moment the evidence questions it.
Dee would be the first to say one Michigan study does not settle it. His closing point is that foster care is an abstraction. Fifty states run fifty systems, and findings from Illinois or Michigan may say nothing about Minnesota.
What I take from it
Stop arguing about whether foster care is good or bad. It is the wrong question and it has cost us twenty years.
The right question is which children cannot be stabilized, and what we are prepared to build for them. Dee’s answer is specific: systems without a cadre of professional foster parents for the 5 to 10 percent who need them will keep failing those children, and volunteer therapeutic programs will not close the gap.
That is a budget question and a workforce question. It is answerable. We are not answering it.
Read the full commentary at deewilsoncon.com, where Dee has archived the Sounding Board since 2009. This summary is mine and the analysis is his.








