In April 2018, the Connecticut Office of the Child Advocate (OCA) released a follow-up report to its investigation of the 2017 death of Matthew Tirado, a teenage boy who had been severely abused and neglected by his mother. Because homeschooling was tangentially related to the situation (Matthew himself was not being homeschooled at the time of his death, but his sister was), the OCA decided to collect data on homeschooling families’ involvement in child abuse and neglect investigations.
In the three years since it was released, some have cited this report in attempts to tighten restrictions on homeschooling families. However, there were numerous issues with the report: its methodology, its data set, and its underlying assumptions about how the child welfare system works.
The Tirado case
Matthew Tirado died in February 2017. He was 17 years old. Investigators concluded that Matthew was beaten and starved by his mother.
The OCA released two reports about Matthew’s death. The first was primarily focused on significant, multiyear failures on DCF’s part to properly investigate and remediate Matthew’s situation. But there was a section toward the end that discussed Matthew’s educational background.
Matthew had autism and intellectual disabilities. Throughout his life, his school attendance had been sporadic, and the family had been investigated several times by the Department of Children and Families (DCF) while Matthew was attending public school. The report concluded that the Hartford Public Schools (HPS) needed to do a better job identifying and reporting abuse and neglect among its students.
The initial report contained just a few scattered references to homeschooling. Investigators found that in December 2016, Matthew’s mother withdrew his younger sister from school to homeschool her. Matthew was not withdrawn to be homeschooled; he was enrolled in public school at the time of his death. But the initial report concluded that if the DCF had been notified of Matthew’s sister’s withdrawal from school to be homeschooled, the case might have turned out differently:
HPS officials did not notify DCF of the child’s withdrawal from school despite HPS having made five child protection reports to DCF within a recent 18 [months] regarding this child and her brother, Matthew. A call by HPS officials to DCF at this time, while the Juvenile Court case was still pending, could have potentially altered the trajectory of the family’s case, perhaps leading to DCF’s ongoing involvement and efforts to see the children.
A few months after its initial report on Matthew Tirado, the OCA issued its follow-up report that focused almost exclusively on homeschooling. To prepare that report, the OCA asked six Connecticut school districts to provide them with the names of all children who had been withdrawn to be homeschooled between 2013 and 2016 (according to the Connecticut Department of Education, there are 205 districts in 2021).
The OCA concluded that “across the six districts, 139 out of the 380 students withdrawn to be homeschooled during the previous three academic years lived in families that were the subject of at least one prior accepted report to DCF for suspected abuse or neglect.”
Unlike a traditional study, which lays out its methodology and findings for peer review, the OCA report didn’t delve into its data set, nor did it explain how it arrived at the “139 out of 380” number. Several Freedom of Information Act (FOIA) requests were made for this underlying data, but so far as we know, none were ever answered. Whatever the data was, it remains sealed in the proverbial “black box,” hidden from view.
Even so, there are good reasons to dispute the two main conclusions drawn from the follow-up report:
- The OCA saw these data as illustrative of the need for a “safety net” to protect homeschooled children from abuse and neglect.
- Other commentators have taken their conclusions further, seeing the numbers as suggestive of exceptionally high rates of abuse and neglect in homeschooling families.
A safety net?
For proponents of homeschooling regulation, the fact that Matthew’s mother was able to homeschool his sister—despite a history of DCF involvement—signals a gap in child protection. From this point of view, children attending school are automatically safer than those being homeschooled because they are away from their (potentially abusive) home environments almost daily and because school personnel are mandatory reporters with training in identifying abuse and neglect—the mandatory reports act as a safety net.
But this approach raises questions. First, whatever one thinks about the argument in the abstract, the fact is that this safety-net approach wouldn’t have saved Matthew, who was not only enrolled in public school at the time of his death but was also the subject of a DCF investigation when his mother withdrew his sister from school.
More importantly, the safety-net argument makes some key assumptions about the child welfare system that are simply not true. In a perfect world, perhaps communication like this between schools and DCF could identify victims of abuse in time to save them. But again, Matthew’s death isn’t attributable to a lack of communication between school officials and DCF: the Tirados had been the subjects of multiple DCF investigations based on the reports of school officials for years.
The reality is that child welfare departments are notoriously overworked and underfunded. They are tasked with investigating a broad range of maltreatment types, from benign to highly urgent, using vague protocols that sometimes treat parents as welfare recipients and other times as criminals under investigation. Relying on one bureaucracy (child welfare) to work with another (public education) to supervise and keep all children safe is an unrealistic and unworkable solution.
How many at-risk kids?
The report claimed that in these six school districts (five of which were unspecified), 139 out of the 380 students withdrawn to be homeschooled during the previous three years lived in families that were the subject of at least one prior “accepted report” to DCF for suspected abuse or neglect. The implication—which was not lost on the news outlets who covered the report in 2018—was that a significant number of homeschoolers lived in abusive or neglectful homes.
That conclusion is severely misleading. An “accepted report” is a term of art in Connecticut: when a report is first made, it is vetted by DCF to determine if the allegation, if true, would meet the statutory definition of abuse or neglect. If so, it is “accepted” for further investigation.
For example, if someone alleged that a newborn had been left alone at home, that report would be accepted—even if it wasn’t true—because if the allegation were true, the child would be neglected. Conversely, if someone alleged that a parent had given their child blue ice cream instead of green ice cream, that would not be accepted—even if it were true—because giving your child blue ice cream instead of green is not considered abuse or neglect in Connecticut law.
The key point is this: if one looks solely at how many “accepted” reports are made against homeschooling families—as the OCA’s report did—all it will reveal is how many times people made allegations about those families. It tells us absolutely nothing about whether the allegations were true. In fact, it can’t tell us that, because reports are “accepted” before they are ever investigated.
To answer the question of how many of the allegations in the OCA’s report were true, the OCA would have had to use another term of art: of those “accepted” abuse and neglect reports, how many were “substantiated”? A report is “substantiated” if, at the end of the investigation, DCF concludes that there is at least some evidence that a child was abused or neglected. But the OCA’s report never stated how many of the “accepted” reports were “substantiated,” and the underlying data it relied on was never released.
Without this clarification, the OCA’s data does nothing to quantify the actual danger posed to homeschooled children. On the contrary, national data collected by the US Department of Health has consistently found that the vast majority of “accepted” reports are ultimately determined to be “unfounded.” Of the 21,617 Connecticut children who were investigated in 2019 (the most recent year for which data is available), nearly 60% (12,958) were the subjects of “unsubstantiated” allegations.
Not all risk is abuse or neglect
There is one more gap in the data. While the OCA report provides number of accepted reports, it doesn’t shed any light on the type of abuse or neglect alleged by the reporter. There’s no doubt that the kind of abuse that gave rise to the report in the first place—the tragic, horrific, death of a child—is both heinous and criminal. But an allegation of abuse or neglect covers acts that range from a one-time error in judgment (such as briefly leaving children alone in a car), to poverty-related failures to provide adequate shelter or supervision, to physical and sexual abuse.
Again, drawing from national statistics, most reports that are substantiated are for neglect. Connecticut reported 8,042 child victims in 2019. Of those victims, the overwhelming majority—6,917—were victims of neglect. The next-highest category (psychological maltreatment) accounted for 2,570 children. The numbers for the categories of physical abuse (434), sexual abuse (376), and medical neglect (245) were significantly lower.
It needs to be said that children can be victims of more than one type of abuse (in Connecticut, for instance, there were 8,042 child victims, who were assigned 10,542 maltreatment types, which is more than one per child). Matthew Tirado would have been one of those children. But thankfully, substantiated reports of physical abuse, sexual abuse, and medical neglect—the types of reports that tend to be more severe and life-threatening—are much rarer than reports of neglect.
The OCA’s report doesn’t delve into this distinction explicitly, which is unfortunate—especially given the seven case examples that follow. These, like Matthew’s story, are incredibly troubling. But outside of the seven case examples cited in the report, there are no details as to how many of these accepted reports were for life-threatening situations, and how many were not. Or how many reports were based on some evidence, and how many were based on misunderstandings or were simply false reports made by people with axes to grind.
That last possibility may seem farfetched, but unfortunately, it happens more than we’d like to believe. Just last month, a federal court in Connecticut allowed two parents to proceed with a lawsuit in which they accused a school social worker of reporting them to DCF for educational neglect. What was that neglect? The family had said they were thinking about withdrawing their child to homeschool.
There’s a lot left to be done in the case—the parties are about to start collecting evidence for trial, so the parents still have to prove the worker made a deliberately false report about them—but the judge warned that if this was true, the social worker may have violated the parents' constitutional rights.
Unfortunately, what the school social worker did in that case is not all that far removed from the course of action advocated by the OCA: the OCA’s initial Report in 2017 concluded that the district “should have immediately contacted the DCF Careline with a new report” when Matthew’s sister was withdrawn from school.
The subsequent 2018 report didn’t repeat that recommendation per se, but it did call for a “regulatory framework in Connecticut that will minimally ensure a child [who] is withdraw[n] from school for the purpose of homeschooling is receiving an education and is making progress in instructed areas,” and that “DCF should consider . . . what processes it may best employ” when investigating families who have withdrawn to homeschool.
This underscores the chief misconception about the report: while the report borrows the lot of the verbiage of a dispassionate research study, at its core, is really a work of advocacy. It is designed to advocate for changes, which are summarized in its “strongly recommend[ed]” suggestions at the end. But as hard as it tries, it is not a research study. If we try to give the report the same weight as peer-reviewed studies, or the voluminous data collected by the US Department of Health and Human Services—as other commentators have done—it falls well short of the mark.