The Fingerprint That Never Was, Part II: One in Nine
Investigation by Project Milk Carton
The federal government proved the emergency name check doesn't work — twenty-seven years ago. Then it wrote states a permission slip to keep using it.
Project Milk Carton — child welfare transparency. Part II of an ongoing investigation. Part I, "The Fingerprint That Never Was," documented how states license foster homes without the fingerprint-based FBI checks that 42 U.S.C. § 671(a)(20) has required since 2008. Part I ended on assertions we could not yet quantify: that name-based "emergency" checks materially miss criminal records, that the provisional-placement workaround is written into state law rather than improvised, and that the price of getting caught is effectively zero. For Part II we retrieved and read the primary documents themselves — the Departmental Appeals Board decisions, the federal eligibility-review reports, the state statutes, ACF's own policy manual, and the Justice Department study that settled the name-check question in 1999. Every number below comes from a document in our possession.
The Permission Slip
Start with the sentence that reorganizes everything in Part I.
The Administration for Children and Families maintains a Child Welfare Policy Manual — the official, binding Q&A that tells states what federal child welfare law requires of them. Section 8.4F, Question 7, asks the exact question this investigation is about: "Must a State complete the fingerprint-based check of national crime information databases required by section 471(a)(20)(A) of the Social Security Act before placing a child in the home of a prospective foster or adoptive parent?"
ACF's answer, on the books since January 29, 2007, begins with one word:
"No. The State is not required by Federal law to complete the fingerprint-based checks before placing a child in the home of a prospective foster or adoptive parent."
The manual continues that the check must be completed before the parent is licensed or approved, and concedes — in the federal government's own understated prose — that placing a child before completing "thorough safety checks has serious practice implications." The only consequence it names is financial: federal foster care maintenance payments may not flow "until the criminal records check has been completed" and the home is licensed. Question 8 gives the identical answer for child abuse registry checks.
It gets starker. Question 24 addresses the permanent statutory bars — the felony convictions for child abuse, crimes against children, rape, sexual assault, homicide that § 671(a)(20) says shall forever disqualify a home. Are there exceptions? "No, there are no exceptions." And then, in the same answer:
"The State... has the discretion to place the child in a home where prospective parents have been convicted of such crimes. However, the State or Tribe may not claim title IV-E foster care maintenance or adoption assistance payments in such cases."
Read it twice, because it is the entire architecture in two sentences. Federal law does not stop a state from placing a foster child with a convicted child abuser. It stops the state from billing Washington for it. The child is protected by an invoice.
This is the answer to the question Part I kept circling: how can a mandatory, no-exceptions fingerprint requirement coexist with tens of thousands of children living in unchecked homes? Because the requirement was never a placement rule. It is a payment rule. ACF says so in writing and has since 2007.
One in Nine
Part I asserted that "a name-based check is not a fingerprint check" and left it there. It turns out the federal government did not leave it there. It ran the experiment — once, definitively — and the result should have ended the practice.
In 1999, a national task force convened at the direction of the U.S. Attorney General published Interstate Identification Index Name Check Efficacy (NCJ-179358), a study built with the Bureau of Justice Statistics and the FBI. Its design has never been repeated: 93,274 Florida employment and license applicants were run through both an III name-based check and an FBI fingerprint search — every subject, both checks. The task force itself noted that before this study, no one had ever done that. As far as our review can establish, no one has since. It remains the only head-to-head federal test of the name check that states now use to clear emergency foster placements.
The results, from the report's own text:
10,673 applicants — 11.4% — had fingerprint-verified FBI criminal history records.
Of those, 1,252 were cleared by the name check. The name check missed 11.7% of the people who actually had criminal records — about one out of every nine.
Even when the applicant's name and date of birth matched the FBI's master record exactly, there was still a 10% chance of an incorrect miss. Without a perfect match, the miss rate exceeded 50%.
A central reason the check fails: people lie to it. The FBI's own analysis found 11.7% of civil applicants with fingerprint-verified records "had provided false names and/or other false identification information in their applications." A fingerprint defeats a false name. A name check is made of the false name.
Projected nationally against the 6.9 million civil background checks the FBI processed in 1997, name-check-only screening would have produced roughly 70,200 missed criminal records per year.
Then the FBI's appendix did the arithmetic that belongs in this investigation. Among the missed record-holders were 259 child-care applicants — projected at roughly 777 people per year, in Florida alone, who would have walked into child-care work carrying criminal histories a name check failed to surface. Their 560 charges included 58 for assault, 67 for dangerous drugs, 13 for commercial sex offenses, 7 for sexual assault, 5 for other sex offenses, one homicide, and one kidnapping. The FBI's own conclusion: name-based-only screening "would allow some of the most vulnerable portions of our population to have continual exposure to persons with criminal backgrounds."
The task force's recommendation was categorical — fingerprint searches are "highly preferable" — and the National Crime Prevention and Privacy Compact that took effect the same year requires fingerprints for noncriminal-justice III searches.
Now hold that next to ACF's own policy manual. Section 8.4F, Question 30, authorizes a name-based check of the national databases in exactly two situations: a documented physical disability that prevents fingerprinting, and prints so poor the FBI rejects them — and even then the manual warns that alternatives are "for limited and case-specific situations" and that it is "not acceptable" to substitute a name check because of "the State's lack of technological capacity or use of improper techniques."
Nowhere — not in the statute, not in the regulation, not in the policy manual — does the federal government authorize a name-based check because the placement is urgent. The emergency name check that states run every night has no home in written federal policy. It exists in the gap between "finally approved" and "placed," the gap ACF's Question 7 opened with the word "No."
Place First, Check Later: The Statutes Say So
Part I treated the emergency workaround as agency practice. Part II can now report it as codified law, from the statute and regulation texts themselves. We retrieved and read the operative provisions for nine states. Eight of them write place-first-check-later directly into law. One — a genuine counter-example — refuses to.
Missouri (RSMo § 210.482): for emergency placements, the court or children's division "may request" — permissive, not mandatory — a name-based check through the state system and the FBI's Interstate Identification Index. Fingerprints from every adult in the home are due within 15 calendar days of the name check; a child "shall be removed immediately" if an adult refuses.
Arizona (A.R.S. § 8-514.02): emergency placement on "a preliminary state and federal name-based background check," fingerprints within 15 calendar days, immediate removal on refusal.
Florida (Fla. Stat. § 39.0138(5)): emergency placement on "a name-based check of criminal history records"; fingerprints due to state law enforcement within 7 calendar days; if they aren't provided within 15 days, the department must seek a court order to remove the child — removal is not automatic.
California (Welf. & Inst. Code § 361.4, as amended 2024): emergency placement on a state-only name checks through CLETS — not a national check at all. The fingerprint clearance must merely be initiated within 10 calendar days of the CLETS check or 5 business days of placement; the section sets no deadline for completing it. And subdivision (b)(6) lets a court authorize the placement "regardless of the status of any criminal record exemption" if it finds no risk — a judicial override of the record itself.
Michigan (MDHHS foster care policy FOM 722-03B, eff. March 2024): a name-based state check (ICHAT), central registry, and sex-offender registry screen before emergency placement; fingerprints "within 30 days of placement." The authorizing statute, MCL 722.954a, is silent on criminal checks entirely.
Ohio (OAC 5180:2-42-18): in an immediate-removal case, the agency may place the child with a relative "if there are no known safety concerns" — no criminal check of any kind is required before the child is physically in the home; assessments begin "no later than the next business day." A companion rule (OAC 4501:2-10-09) authorizes name-based NCIC access for certified exigent circumstances, fingerprint cards due in 15 days.
New York (18 NYCRR § 443.7) — the furthest end of the spectrum: emergency approval of a relative home requires an expedited home study, the relative's own signed statement about their history, and a check of the agency's own records — no criminal-history check at all at the moment of placement. Fingerprint cards go out the day of placement and are due back within two weeks. The emergency approval runs 90 days — and subdivision (h) expressly allows the placement to continue past the 90th day while the criminal-history check is still incomplete. The exception is written into the rule.
South Carolina (S.C. Code Regs. § 114-550): emergency kinship placement for up to 90 days on a state-only criminal check, with out-of-state abuse-registry history covered by the caregiver's own attestation — and the agency may extend another 90 days if removal would be "detrimental to the best interests of the child." Hold that against the study above: a state-only check is precisely the tool the 1999 task force found inadequate, and Florida's data showed why — only 45.5% of that state's disqualified applicants could be identified from state records alone. The rest were visible only to the national fingerprint check that the emergency placement postpones.
And the counter-example, reported with the same care: Pennsylvania (23 Pa.C.S. § 6344) contains no provisional path for foster parents at all. The FBI fingerprint-based check is part of the approval package, and the statute's own 45-day provisional-hire mechanism expressly excludes "a prospective adoptive parent or a prospective foster parent" from its reach. Texas's fetched materials likewise require the FBI check before a foster home is verified, with no emergency day-window anywhere in the documents we hold — though our Texas set covers licensed homes only, so we make no claim about Texas kinship practice. It is possible to write the statute correctly. Two states in our sample did.
The pattern across the other eight is uniform in structure and heterogeneous in laxity: place the child tonight on a name check, a state-only check, or no check; collect fingerprints on a clock that ranges from 15 days to never-quite-enforced; and vest the remedy — removal — in the same overloaded agency that needed the bed. Part I quoted a Hawaii audit finding provisional placements stretching 314 days. Part II adds: in New York, the regulation itself says the placement may outlive the check.
️
The Price of Getting Caught
Part I argued the federal enforcement machinery cannot detect background-check failure and barely punishes it when it does. We have now read the enforcement record itself — the eligibility-review reports and the Departmental Appeals Board decisions where states and ACF actually fight over money. The record is worse than Part I claimed, in five documented ways.
First: the regulation announces its own blindness, with numbers. 45 C.F.R. § 1356.71 — the eligibility-review rule — states its statistical design in plain text: when a state's 80-case sample shows eight or fewer ineligible cases, "ACF can conclude with a probability of 88 percent that in a population of 1000 or more cases the population ineligibility case error rate is less than 15 percent and the title IV-E agency will be considered in substantial compliance." That is the certification standard for the program that pays for foster care in America: 88% confidence that no more than roughly one case in seven is bad. Pass it, and no one looks again for three years.
Second: when reviewers do catch licensing and background-check failures, the bill is a rounding error. We read six federal eligibility-review reports end to end:
New Jersey, 2003 secondary review — the one catastrophic failure on record: 56 of 150 cases in error (37.33%), including 26 cases in homes that were unlicensed, expired, or licensed by no state at all, and 3 with no criminal background check completed. Because both error rates crossed 10%, the extrapolated penalty fired — the only time in our document set it ever has: $6,220,853 clawed back.
Idaho, 2013: 3 error cases, including one home approved without the state's own safety measures — total disallowance $20,661.
Mississippi, 2017: 11 of 80 cases in error — non-substantial compliance — six on licensing/safety grounds, two with no documented background safety checks at all. The Children's Bureau wrote, in its own report, that the failure "posed a risk to the children by placing them in potentially dangerous settings." The disallowance: $56,007.
Massachusetts, 2019 secondary review: 4 error cases, three involving payments for children in unlicensed homes — $61,551.
New Jersey, 2008 secondary review: 9 of 10 error cases were unlicensed-home cases; 4 of 10 had no documented criminal history check on the foster parents — this after two failed reviews and two federal improvement plans. The disallowance: $94,449.
Idaho, 2018: $139,899, driven mostly by court-paperwork errors.
Set aside the 2003 New Jersey outlier and the modern price of a failed review — including reviews that found children in unchecked, unlicensed homes — runs between twenty and one hundred forty thousand dollars, against a program the Congressional Research Service prices at $9.7 billion for FY2024. Mississippi's penalty for placements the government itself called potentially dangerous was 0.0006% of one year of the program.
Third: the appeals record shows the machinery litigates paperwork, not safety. In DAB No. 1797 (2001), New Jersey was caught with 33 of roughly 80 sampled cases in unlicensed or unapproved homes; the disallowance was $191,348, and the Board went out of its way to note that ineligibility "does not necessarily mean... that the proper social work was not done" — the inquiry is documentary by design. Four of the unlicensed-home cases were rescued by a sworn statement attributing the lapses to "administrative or clerical delays." In DAB No. 2360 (2011), Michigan flipped from noncompliance to substantial compliance — and erased $67,264 of a $74,318 disallowance — by winning two cases on a fair-hearing technicality. Two of eighty cases was the entire distance between "failing" and "nothing happens for three years." Meanwhile the decisions where real money moved — DAB No. 2529's $704,847 against West Virginia — turn on cost-allocation-plan methodology for training expenses. The machinery is fully capable of ferocity. It reserves it for accounting.
Fourth: the enforcer cannot count its own enforcement. GAO-06-649 (2006) reported that of six ACF regions interviewed, three had disallowed zero dollars across fiscal years 2000–2004 — and that "ACF central office officials told us they could not provide reliable data on the amount of claims that are disallowed each year." No national tally of eligibility-review disallowances exists in any document we hold, and consistent with our honesty rule, we will not manufacture one. The point is sharper without it: neither will HHS.
Fifth: when the safety-outcome track finally bites, it bites softly, slowly — and then the enforcer retreats. The companion penalty regime, the Child and Family Services Review, is the track that measures whether "children are, first and foremost, protected from abuse and neglect." In Round 3 (FYs 2015–2017), per the Children's Bureau's own aggregate report, only 3 of 38 states achieved that outcome. On the licensing systemic factor, 11 of 38 were in conformity, and on Item 34 — the criminal-background-check item — 12 of 38 states could not demonstrate their check requirements were actually functioning, with the aggregate report naming the failure in words that could headline this series: "Not having information to know whether policies requiring criminal background checks were being followed."
And the one state penalized to conclusion? Idaho failed Safety Outcome 1 in a 2016 review. The withholding letter arrived November 2022. The Departmental Appeals Board upheld it — $935,712, computed as 1% of a partial funding pool per failed outcome — in September 2023 (DAB No. 3110). Six-plus years, from a finding that children were not protected from abuse and neglect to a penalty smaller than many single settlements Part I catalogued. And in March 2023, before the decision even issued, ACF announced in writing that it would stop continuing to withhold penalties for every Round 3 state that failed its improvement plan. The enforcer looked at the one tool that had finally drawn blood and sheathed it — for everyone.
The Arithmetic, Assembled
Put the four document sets together and the system describes itself:
A child can lawfully be placed tonight with an adult whose only screening was a name check (eight states' statutes, on their face) — or no criminal check at all (New York, Ohio).
That name check, in the only both-checks federal study ever conducted, cleared one of every nine people who had an FBI criminal record — and more than half of those whose identifiers didn't perfectly match, which is precisely the population with a reason to smudge its identifiers.
Federal policy does not forbid any of this. ACF's manual says "No" — the check need not precede the child (CWPM 8.4F Q/A #7), and even a disqualifying felony blocks only the billing, not the placement (Q/A #24).
The audit instrument certifying the system passes at 88% confidence that the error rate is merely under 15% (45 C.F.R. § 1356.71), fires its real penalty only when two error rates both cross 10% — once in our entire document set — and otherwise bills the state five figures (six federal review reports).
The outcome track found 3 of 38 states protecting children from abuse and neglect, 12 of 38 unable to show their background checks function (CFSR Round 3 aggregate) — and its penalty mechanism was voluntarily stood down (DAB No. 3110, footnote 11).
As of September 30, 2022, per AFCARS Report No. 30, 368,530 children were in foster care, and 285,158 of them — 78% — were living in the relative and non-relative family foster homes that § 671(a)(20)'s checks exist to clear. Those children are the population on the other side of every "No" in this article.
What Part II Adds to the Fix List
Part I's five fixes stand. The documents behind Part II compel three more:
Close the Question 7 gap. ACF should amend the Child Welfare Policy Manual — no statute required — to state that a name-based or state-only check satisfies nothing under § 671(a)(20), that its only sanctioned uses remain the two in Q/A #30, and that federal financial participation for an emergency placement ends on day 15 if fingerprint results are not documented. Missouri and Arizona already run a 15-day clock; a federal floor merely universalizes what the strictest states codified.
Publish the enforcement ledger. GAO told HHS in 2006 it could not reliably count its own disallowances. Require ACF to publish, annually, every eligibility-review disallowance by state, amount, and error type — including the background-check and licensing subtotals. The numbers exist in the review reports; we compiled six of them by hand. The agency should not be harder to audit than the states it audits.
Re-run the 1999 study. The only head-to-head test of name checks against fingerprints is twenty-seven years old. Every state statute in this article leans on the name check's adequacy. If the miss rate is still one in nine — or worse, in an era of easier identity manipulation — Congress should know it before the next reauthorization. If it has improved, states deserve that answer too. Either way, a quarter-century-old sample of 93,274 Floridians should not be the entire evidentiary basis for the screening that stands between a quarter-million children and the adults they sleep near tonight.
What We Could Not Verify
Consistent with this series' practice, the gaps in our documentation, plainly:
Two intended audit retrievals (a 2018 and a 2023 federal audit file) returned the wrong documents; Part I's Washington, Hawaii, and Maryland audit findings therefore rest on Part I's original sourcing and are not re-verified here.
Alaska, Nebraska, Iowa, Mississippi, and Wisconsin emergency-placement statutes could not be retrieved (blocked pages, wrong documents); our state survey claims exactly nine states and no more. Part I's "about 35 states" figure for provisional licensure rests on its original secondary source.
Two Departmental Appeals Board decisions (Nos. 2379 and 2386) could not be retrieved and are not cited.
No national aggregate of Title IV-E eligibility-review disallowances exists in any document we hold; per GAO, HHS itself could not produce one. We state only the per-review figures we read.
Missouri's manual reportedly labels its emergency name check "Purpose Code X"; the statute we hold (RSMo § 210.482) authorizes the name-based NCIC/III mechanism but does not use that phrase, so we describe the mechanism rather than the label.
The 1999 III study reflects 1999 data and Florida applicants; it is cited here as what it is — the only such federal study ever conducted, not a measurement of current miss rates.
If you enjoyed this work and feel encouraged by independent investigations like this, please consider becoming a paid subscriber to The Constitutional Republic on Substack or donate at
Project Milk Carton Inc. Your gift keeps our site alive, spotlighting high-risk children!
Your support helps us continue:
investigative reporting
public records research
educational transparency projects
community outreach
and future investigations like this series
Most importantly, please share this article.
Independent journalism survives when ordinary people help spread information powerful institutions would prefer remain ignored, misunderstood, or buried beneath complexity.
Because silence protects systems.
Informed communities protect children.
ALL paid subscriptions are tax deductible Project Milk Carton | 501(c)(3) | EIN: 33-1323547
Sources
42 U.S.C. § 671(a)(20) — full statutory text (Office of the Law Revision Counsel)
45 C.F.R. § 1356.71 — Title IV-E foster care eligibility reviews
ACF Child Welfare Policy Manual, § 8.4F Q/A #7, #8, #12, #14, #24, #30, #34, #36; § 8.3A.8c Q/A #6, #8, #12
Interstate Identification Index Name Check Efficacy: Report of the National Task Force to the U.S. Attorney General (SEARCH/BJS, July 1999, NCJ-179358), incl. FBI Appendix D
HHS Departmental Appeals Board: DAB No. 1797 (New Jersey, 2001); DAB No. 1949 (MD/NY/NJ/SC, 2004); DAB No. 2360 (Michigan, 2011); DAB No. 2529 (West Virginia, 2013); DAB No. 3110 (Idaho, 2023)
Title IV-E Foster Care Eligibility Review final reports: New Jersey secondary (2003); New Jersey secondary (2008); Idaho primary (2013); Mississippi primary (2017); Massachusetts secondary (2019); Idaho primary (2018)
GAO-06-649, Foster Care and Adoption Assistance: Federal Oversight Needed to Safeguard Funds (June 2006)
GAO-26-107592, Child Welfare: HHS Should Clarify Guidance on State Spending for Congregate Care (March 2026)
CRS Report RL32836, Child Welfare: An Analysis of Title IV-E Foster Care Eligibility Reviews (March 2005); CRS In Focus IF10590, Child Welfare: Purposes, Federal Programs, and Funding (updated Oct. 2024)
Children's Bureau, Child and Family Services Reviews Aggregate Report, Round 3: FYs 2015–2017
AFCARS Report No. 30 (Preliminary FY 2022 Estimates)
State authorities: RSMo §§ 210.482, 210.487; A.R.S. §§ 8-514.02, 8-514.03; Cal. Welf. & Inst. Code § 361.4; Fla. Stat. § 39.0138; MCL 722.954a and MDHHS FOM 722-03B; Ohio Admin. Code 5180:2-42-18 and 4501:2-10-09; 18 NYCRR § 443.7; S.C. Code Regs. § 114-550; 23 Pa.C.S. § 6344; Texas DFPS CPS Handbook §§ 7000–7124 and FAD Background Checks Resource Guide (June 2024)
Project Milk Carton is a 501(c)(3) organization. Child welfare transparency and missing children awareness. projectmilkcarton.org