Homeschooling’s 

Invisible Children

Methods

The Homeschooling’s Invisible Children database is a project of the Coalition for Responsible Home Education (CRHE). The database contains instances of abuse and neglect in homeschooling environments. We collect data, code it along various qualitative and quantitative measures, and analyze the results. We then use those analyses to inform our advocacy work. 

We also make our data and codebook available so that other researchers can do their own analyses. For the data, see our cases page. To contact the HIC research team, email research@responsiblehomeschooling.org.

Data sources

We identify cases via monitoring of online news sources through Google News, and supplement data with documentation from court records, obituaries, and other publicly available sources. We also crowdsource cases from researchers, journalists, and other members of our network. We include incidents that took place in the time period ranging from the mass legalization of homeschool in the 1970s and 1980s until the present, although the bulk of cases occurred during or after the year 2000. 

There are two major consequences of our identifying data this way: 

  1. By definition, cases found this way are those that were severe or unusual enough for a news outlet to consider them newsworthy. Consequently, this database likely overrepresents incidences of severe abuse and neglect in homeschool settings while underrepresenting less severe cases. 
  2. The cases in this dataset are nonetheless almost certain to undercount cases of this nature. Cases can take years or decades to come to light; some never do. When they do, reporting does not always mention that the victims were homeschooled. This dataset does not include cases that have not come to light or cases where the victims were homeschooled but were not reported as such, with exceptions mentioned below.

While these data can speak to the nature of some manifestations of abuse and neglect in homeschool settings, they cannot be taken to be representative of abuse/neglect in homeschooling as a whole, nor can they be taken as a complete accounting of cases of this nature. We encourage researchers to interpret the results appropriately. 

Inclusion criteria

To be included in the database, a case must meet all the following criteria, as outlined in our reports of findings and codebook:

  1. Location: The incident must have occurred in the United States or its territories (American Samoa, Guam, the Northern Mariana Islands, Puerto Rico, and the U.S. Virgin Islands).
  2. Homeschooling status: At least one victim age 5-18 must have been homeschooled at the time of the incident:
    • Both legal homeschooling and claimed homeschooling count, regardless of whether they are following their state’s homeschool requirements. The reason for including claimed homeschooling is that claiming homeschooling provides parents with social cover: they use claimed homeschooling to discourage peers and family from expressing concern about the children.
    • If the victim is homeschooled but that fact is not reported in media coverage, written confirmation from an authority involved with the case counts as confirmation they were homeschooled.
    • For cases added after the beginning of 2026, being unenrolled in a no-notification state counts, even if they are not stated to be homeschooling. No-notification states are currently: Alaska, Connecticut, Idaho, Iowa, Michigan, Missouri, Illinois, Indiana, New Jersey, Oklahoma, Texas, Wyoming.
      • Being enrolled but truant/chronically absent does not count, unless it leads to unenrollment and homeschooling (or, in a no-notification state, just unenrollment).
    • Exception: we will include (but flag) virtual schooling / cybercharter cases (excluding remote schooling during the pandemic), due to their risk factors resembling those of homeschooling.
  3. Sources: The incident must be publicly documented in at least two third-party sources, at least one of which states that the child was homeschooled (or was claimed to be homeschooled, or was unenrolled in a no-notification state). However, a non-fatality case may be included with only one source if it is comprehensive and credible (e.g., national news outlet or state/local outlet affiliated with a wider network).
    • The sources must include the abuse/neglect as a central focus (including if the reporting is on coming to light of the abuse/neglect).
    • Self-report/memoir may be included as sources, but do not count toward the two needed sources.

Data coding

Volunteers and employees who code cases follow specific coding procedures that have been discussed and documented by the research team to ensure consistency.

We record basic information about each case, such as location, dates, how the case came to light and under what circumstances, history of contact with social services, and information about fatalities and withdrawal from school if applicable. We also record information about forms of abuse and neglect (e.g., food deprivation, imprisonment, medical neglect, and so forth), and look at whether some children were singled out for worse or different abuse than other children. For cases involving adopted or fostered children, we record some additional related information. 

Review process

Cases coded prior to July 2024 were systematically and comprehensively reviewed for a publication; cases from then through October 2025 were re-reviewed for a database update and variable check. Cases since October 2025 are spot-checked by an employee or volunteer who did not code the original case.