Human Trafficking Victim Marking and Identification, 2015-2025

KJKey Judgments

7 judgments carry this brief. Each is an analytic conclusion, distinct from the reported facts beneath it, and each carries an explicit confidence level per §1.2. A judgment can be wrong even where every underlying fact is correct. Several were revised after an adversarial verification pass on a second model; those corrections are shown inline in the body rather than folded in silently (§1.4).

1How to read this document

1.1 Source grading: the NATO/Admiralty scale

Every source cited in §12 carries a two-character Admiralty grade: a letter for the reliability of the source itself, a number for the credibility of the specific information cited. The two axes are independent, a highly reliable source can carry a specific claim that is only possibly true, and a source of unproven reliability can still carry information later confirmed elsewhere.

Source reliability:

• A: Completely reliable. No history of inaccurate reporting (a federal agency's own press release or official statistics page, speaking in its own voice).
• B: Usually reliable. Minor history of inaccurate reporting, or reliable but not a primary party to the events described (an established wire service, a credentialed investigative outlet, a research institution with transparent methodology).
• C: Fairly reliable. Some history of inaccurate reporting (a regional news outlet relaying an official's statement rather than the agency's own release).
• D: Not usually reliable. Significant history of inaccurate reporting. (Not used as a load-bearing grade in this document; a source graded this low would not be cited for a factual claim.)
• E: Unreliable. Lacks a history of accuracy. (Not used in this document.)
• F: Reliability cannot be judged. (Used for a source with insufficient track record to grade, noted individually where it occurs.)

Information credibility:

• 1: Confirmed by other independent sources.
• 2: Probably true (source is reliable, information is plausible, but not independently confirmed by a separate source in this document).
• 3: Possibly true (some doubt; typically a single named source with no independent corroboration).
• 4: Doubtful (some internal contradiction, or later superseded).
• 5: Improbable (contradicted by other sources).
• 6: Credibility cannot be judged (insufficient information to assess).

The four-tier shorthand used in section subheadings (PRIMARY/OFFICIAL, NAMED INVESTIGATIVE, PEER-REVIEWED, SECONDARY/UNVERIFIED) is a quick reference only; the Admiralty grade in §12 is the authoritative, source-by-source grading. Rough correspondence: PRIMARY/OFFICIAL typically grades A1-A2; NAMED INVESTIGATIVE and PEER-REVIEWED typically grade B1-B2; SECONDARY/UNVERIFIED typically grades B3-C4 or lower.

1.2 Analytic confidence and the probability lexicon

This document distinguishes reported information (a fact directly stated by a source) from analytic judgment (this document's own conclusion about what the reported information means, which requires interpretation and could be wrong even if every underlying fact is correct). A judgment is labeled "Assessment:" and carries two things: a confidence level and, where the judgment is expressed as a likelihood rather than a certainty, a calibrated probability term.

Confidence level (a function of source quality, quantity, and independence, not of how strongly the judgment is worded):

• High confidence: judgment is based on high-quality information, ideally corroborated by multiple independent sources using different methods (alternative explanations were assessed and found significantly less likely).
• Moderate confidence: judgment is credibly sourced but not independently corroborated, or the sourcing is corroborated but shares a common origin, or the analysis is sound but the sourcing is not extensive enough for High confidence.
• Low confidence: judgment rests on a single source, fragmentary information, or reasoning this document could not independently verify. A Low-confidence judgment is not a weak claim dressed up, it is this document's own honest statement that the evidence does not yet support more than that.

Calibrated probability lexicon (the words themselves, not just contextual tone, carry the meaning):

• Almost certainly: roughly 95-99 percent probability.
• Highly likely / very likely: roughly 80-95 percent.
• Likely / probably: roughly 55-80 percent.
• Roughly even chance: roughly 45-55 percent.
• Unlikely: roughly 20-45 percent.
• Highly unlikely / remote: below 20 percent.

A RECTUMmendation (an unverified assertion presented as settled fact, this project's standing term for it) is not the same failure as a properly labeled Low-confidence Assessment. The former hides its own uncertainty; the latter states it plainly. This document aims to never do the former and to do the latter whenever a judgment, rather than a directly sourced fact, is being made.

1.3 What stays a plain fact, not an assessment

Not every claim in this document needs a confidence label. A directly quoted official statistic, a specific named prosecution's outcome, or a directly quoted on-the-record statement is reported information and is presented as such, with its Admiralty grade in §12 doing the epistemic work. Assessment labels and confidence levels are reserved for this document's own interpretive judgments, most concentrated in the Signals, Key Assumptions Check, and Analysis of Competing Hypotheses sections, plus the Bottom Line Up Front.

Sources rejected during research, and why, are logged in §8. Advocacy and mission-driven organizations, and partisan political-body factsheets, were excluded from load-bearing claims. Where an advocacy source was the only place a claim appeared, the claim was either dropped or re-sourced to a primary document; none ship here on advocacy sourcing alone.

1.4 Multi-model verification pass

This document was drafted by Claude Opus 5. Phase 1 research ran in four parallel passes on the same model, covering southeastern Arizona, forensic and biometric identification, the screening-instrument literature, and the marking claim itself; two of those spawned further sub-passes. Their reports were treated as research leads rather than as findings, and every load-bearing figure taken from them was re-fetched and read in its own primary source during compilation. Sixteen sources were re-verified that way before use, and the list is in §12.

Phase 2 was an adversarial verification pass run on a different model, Claude Fable, against the completed draft. It was instructed to find what was wrong rather than to confirm what was right, to fetch primary sources itself, and to report fetch failures rather than guess. It returned 35 findings across three severities, and it earned its cost in corrections rather than in reassurance.

What it caught that mattered:

• A wrong attribution that broke a correlation claim. Continued Presence is granted by DHS's Center for Countering Human Trafficking with Homeland Security Investigations, not by the Department of Justice as this document had it. Because the other half of that signal is USCIS, and both are DHS components, Signal 2's claim of agreement between two independent agencies was overstated. The signal was downgraded from Strong to Moderate and restated as agreement between two instruments rather than two institutions (§2.2, 2.3, 9).
• A headline finding described wrongly. An earlier draft said the Ohio validation study showed the tattoo indicator performing "indistinguishably from a coin flip." The same study's other tables show branding flagged in 6.5 percent of victim assessments against 0.4 percent of non-victim ones, z = -5.55, p < .001, with a phi correlation of .124 the report itself calls "a significant weak association." The indicator is not unrelated to the outcome; it is far too rare to screen on. That is a different and more useful finding, and §4.4 now says so (§4.4, 4.8, 9, and Key Judgment 3).
• A missing circularity. This document flagged the Greenbaum multi-site study for using the clinician's own judgment as its gold standard, and did not notice that the Ohio study does the same thing: the outcome is the screener's own designation at the end of the same session, across 46 positive assessments of which 29 are "potential" rather than confirmed. Now stated in 4.4.
• A misread table. The Ohio sample was described as "substantially aged 16 to 18." That came from a row reporting the age of each youth's significant other, not the youth's own age; the sample's mean age is 18.88 with a range of 12.9 to 23.6, which makes it substantially adult and weakens what it can say about minors (§4.4, 10).
• An inference the law does not support. §3.2 argued that tattoo prevalence among minors must be low because tattooing minors is unlawful. Arizona's own statute, A.R.S. 13-3721, prohibits tattooing a person under eighteen without the physical presence of a parent or legal guardian, which conditions the pathway rather than closing it. The assessment there dropped from Moderate to Low.
• An arithmetic error and an avoidable gap. The running total of T visa approvals was understated, and the FY2023 row this document said it could not obtain is published in Table 11 of a source it already cited. §2.1 now carries the full FY2019-FY2024 series.
• A reversed comparison, a rounding artifact, and a one-sided reading. Tattoos rank marginally above false documents in the Gerassi practitioner survey, not below. Fingerprints and DNA produced 81.8 percent of Pima County identifications, not 81.9, which came from summing two rounded percentages. And unstructured clinical suspicion beat structured screening on specificity even though it lost badly on sensitivity, which an earlier draft did not say.
• Nine cross-references pointing at the wrong sections, two miscounts in the Signals section, a source count that disagreed with its own list, and a dead URL for a source this document claimed to have re-fetched. All corrected.

The pass also closed two open items by fetching what earlier passes could not: Lederer and Wetzel 2014, which moved the field's most-quoted statistic from "88 percent of 107" to the more accurate "87.8 percent of the 98 who answered that question," and the FY2023 T visa figures.

Corrections are marked inline as Correction (multi-model verification pass): at the point where the earlier claim stood, rather than folded in silently. A document that shows its corrections is more trustworthy than one that looks right the first time, and eleven such markers appear in the body.

Two limits on the pass itself, stated because they bound what it establishes. Its search budget ran out partway through, so items requiring discovery rather than a known URL went unchecked. And it could not retrieve several sources it tried, including the decomposition-timeline study, the Fang et al. full text and the QYIT full text, which are recorded in §13 rather than treated as verified.

Every correction and addition from the verification pass that appears in this document was independently re-verified against its primary source before being added, in keeping with the sourcing standard stated throughout, rather than taken on any model's word.

1.5 Abbreviations used in this brief

Built from this document's own text. Every term below appears somewhere in the body; no term is listed that the document does not use. A table rather than expansion on first use, because a reference document is read by jumping to a section from the contents, so "first use" is invisible to a reader who lands in the middle.

Tradecraft and analytic method

Statistical and clinical

United States federal bodies and instruments

Screening and assessment instruments

Arizona state, county and municipal

International and intergovernmental

Bibliographic

2Baseline: how many people the United States formally identifies as trafficking victims

Source line: U.S. Citizenship and Immigration Services, "Characteristics of T Nonimmigrant Status (T Visa) Applicants" fact sheet, data from CLAIMS 3 as of November 2022, covering FY2008 through FY2022; and USCIS, "Fiscal Year 2024: Immigration Applications and Petitions Made by Victims of Abuse, Annual Report to Congress," July 1, 2025, for FY2024. Both are USCIS speaking in its own voice about its own adjudications. USCIS rounds the fact sheet figures to the nearest 10 and states they may not sum to totals; the FY2024 annual report is not rounded.

T nonimmigrant status, the T visa, is the immigration relief Congress created in the Trafficking Victims Protection Act of 2000 for noncitizen victims of a severe form of trafficking in persons. It is not the only way a person is recognized as a trafficking victim in the United States, but it is the one that produces a clean, published, adjudicated annual count, which is why it is the baseline here rather than a hotline tally or an advocacy estimate.

2.1 T visa applications, approvals and denials, FY2008-FY2022

Combined Forms I-914 (principal) and I-914A (family member), as published by USCIS.

The two most recent years are published separately and unrounded, split into principals and derivatives. FY2023 comes from Table 11 of the Attorney General's FY2023 report to Congress; FY2024 from the USCIS annual report.

USCIS states that FY2024 was both its highest receipt year and its highest approval year on record. Applications approved in a given fiscal year were mostly received in earlier years, and both sources say so explicitly, so the approval column is not an adjudication rate for the receipt column on the same row. Note that the unrounded FY2019-FY2022 figures in this table and the rounded combined figures in the table above are the same adjudications counted two ways, not two independent series.

What the shape of this series does and does not support. Across FY2008 through FY2022 the United States approved T nonimmigrant status for 20,400 people, principals and family members combined. FY2023 added 3,676 (2,181 principals and 1,495 derivatives) and FY2024 added 6,178 (3,786 and 2,392), giving roughly 30,250 grants across seventeen years. Correction (multi-model verification pass): an earlier draft put the total at "roughly 24,000" and stated that FY2023 could not be obtained. Both were wrong; the FY2023 row is published in Table 11 of a source this document already cites. The annual principal approval figure has never exceeded 3,786. Congress set a statutory ceiling of 5,000 principal T visas per fiscal year, and USCIS states plainly that "the T visa cap has not been reached since the creation of the T nonimmigrant status program." The constraint on the number of people formally recognized is therefore not the cap. It is the number of people who are located, apply, are represented, and are adjudicated.

This series does not measure how many trafficking victims are in the United States, and nothing here should be read as though it did. It measures how many noncitizen victims completed a specific legal process. It excludes every U.S. citizen victim entirely, since the T visa is immigration relief and citizens are not eligible for it. Assessment (Confidence: High; almost certainly true): the T visa series measures institutional throughput, not prevalence, and the gap between it and any prevalence estimate is a measure of the identification problem rather than a measure of trafficking.

2.2 Two other federal identification counts, FY2022 and FY2023

Source line: Attorney General's Annual Report to Congress on U.S. Government Activities to Combat Trafficking in Persons, Fiscal Year 2023, U.S. Department of Justice. This is the federal government's own consolidated annual accounting.

Continued Presence is a temporary immigration status that law enforcement, not the victim, initiates for a noncitizen the officer has identified as a trafficking victim and a potential witness. It is therefore the closest thing in the federal system to a direct count of law-enforcement identification events. In FY2023, in the report's own words, "DHS's Center for Countering Human Trafficking (CCHT) and HSI granted 382 Continued Presence requests (269 new requests and 113 extension requests) to noncitizens identified as victims of human trafficking," and denied four, the four "because the victimization described did not meet the elements of the Federal human trafficking statute."

Correction (multi-model verification pass): an earlier draft attributed Continued Presence to the Department of Justice and treated 382 as a count of identification events. Both were wrong. CCHT is a DHS component operating with Homeland Security Investigations, not a DOJ body. And 382 is grants plus extensions; new grants in FY2023 numbered 269. The demographic and trafficking-type breakdowns below are of all 382, extensions included, as the report gives them.

Note that three of the four rows above fell between FY2022 and FY2023, in one case by 38 percent. This document makes no claim about why, and no located source explains it. It is recorded here because a reader who assumes these series only rise would misread them.

2.3 What the identified population actually looks like

Three characteristics of the formally identified population are worth stating precisely, because each runs against the picture that circulates publicly, and each bears directly on how identification methods should be expected to perform.

Labor, not sex, dominates the federally identified population. USCIS reports that of the T-1 nonimmigrants who filed a Form I-914 Supplement B, the optional law enforcement declaration, 77 percent listed labor trafficking as the form of trafficking and 29 percent listed sex trafficking, with some forms listing both. The Attorney General's FY2023 report shows the same pattern in Continued Presence: of the 382 granted requests and extensions, 310 were related to forced labor, 67 to sex trafficking and 5 to both.

A majority of the identified are adults, and a large share are men. USCIS reports that 59 percent of T-1 principals were between 30 and 49 years of age at the time of application, and that among approved T-1 principals women comprised 57 percent, leaving 43 percent men. In Continued Presence the male share is larger: of the 382 FY2023 grants, 216 were for male victims and 166 for female victims.

Most recognized victims were never certified by law enforcement. Only 17 percent of the 10,260 approved T-1 nonimmigrants filed a Supplement B at all. The other 83 percent established victim status to USCIS's satisfaction by other evidence. Separately, 97 percent of all T nonimmigrants adjudicated in FY2022 filed with legal representation, and among those denied, 88 percent had filed with representation.

Assessment (Confidence: Moderate; likely): the population the United States formally identifies as trafficked is majority adult, substantially male by the Continued Presence measure, and predominantly exploited for labor rather than sex. Two federal datasets built for different statutory purposes under different adjudicative processes, USCIS T visa adjudications and CCHT/HSI-granted Continued Presence, agree on the labor-over-sex finding without either citing the other. Correction (multi-model verification pass): an earlier draft called these "two agencies" and rated this High confidence on that basis. USCIS and Homeland Security Investigations are both components of the Department of Homeland Security. They remain genuinely different instruments, one victim-initiated through counsel and one law-enforcement-initiated, applying different statutory standards to different populations, which is why the agreement still carries weight. It carries less than two independent agencies would, and the confidence level is reduced accordingly. See §9, Signal 2.

This matters for everything in §4 and 5. The physical-marking literature, and the tattoo indicator built on it, is almost entirely about domestic, pimp-controlled sex trafficking of women and girls. That is a real phenomenon, documented in specific prosecutions in §4.6. It is also a minority of the population the federal identification system actually recognizes.

2.4 Country of origin among identified victims

USCIS reports that persons born in Mexico comprise the largest single share of all T nonimmigrants approved FY2008-FY2022, at 21.2 percent, and that Mexico, the Philippines, India, Honduras, Guatemala and Thailand together account for 70 percent. More than half of all T nonimmigrants, 51.3 percent, resided in six states when they filed: California 15 percent, New York 12 percent, Texas 10 percent, Florida 6 percent, Virginia 5 percent and Louisiana 4 percent. Arizona does not appear among them.

Assessment (Confidence: Moderate; likely): the state-of-residence distribution reflects where immigration legal services capacity is concentrated at least as much as where trafficking occurs, given that 97 percent of applicants filed with counsel. This document did not locate a source that separates the two effects, and §13 records that as an open item.

2.5 A term-of-art distinction this section depends on

Three words are used interchangeably in public discussion and mean different things in every dataset above.

• Trafficking requires force, fraud or coercion for labor or a commercial sex act, or, for a minor in a commercial sex act, none of the three. It does not require movement. DHS states this directly in its own awareness guide: "Human trafficking does not require a border crossing or transportation to be considered a crime."
• Smuggling is the paid movement of a person across a border. It is an offense against the state rather than against the person moved, and it is charged under a different title of the U.S. Code.
• Identification, in every federal figure above, means a formal administrative or legal determination by a named body against a statutory standard. It does not mean a suspicion, a screening flag, or a hotline call. §5 and 6 deal with those, and they are counted nowhere in §2.

A number described as "victims identified" in one source and "trafficking cases" in another may be counting any of these. Where this document gives a figure, it names which one.

3Baseline: how common tattoos are in the general United States population

Source line: Pew Research Center, "32% of Americans have a tattoo, including 22% who have more than one," August 15, 2023. Survey of 8,480 U.S. adults, fielded July 10-16, 2023, on the American Trends Panel, which Pew states is "recruited through national, random sampling of residential addresses."

This section exists because almost none of the literature in §4 has a control group, and an indicator without a base rate cannot be evaluated at all. A physical feature is only diagnostic to the extent that it is more common in the population you are looking for than in the population you are looking at. Everything in §4 and 5 should be read against this table.

A note on this source's inclusion. Pew Research Center is a survey research organization, not an advocacy body and not a policy institute; it publishes full questionnaires, toplines and methodology statements, and takes no position on the subject of this document. It is graded B2 in §12 and is used here for a demographic base rate only. This judgment is recorded in §8 so a reader can disagree with it visibly rather than having to reverse-engineer it.

3.1 Tattoo prevalence among U.S. adults, 2023

What the shape of this table does and does not support. The headline figure of 32 percent is the least useful number in it. The relevant comparison for any adult trafficking indicator is not all adults, it is adults resembling the population being screened. For women aged 30 to 49, the two axes that are separately reported at 38 and 46 percent, tattoo prevalence is close to half. For adults with some college or less, it is 37 percent. Pew does not publish the crossed cell, so this document does not state a figure for it, but both marginals point the same direction and neither is near zero.

Assessment (Confidence: High; almost certainly true): among adult women in the age band that contains the majority of federally identified T-1 principals, having at least one tattoo is close to a coin flip. Any screening rule of the form "has a tattoo" therefore carries a false-positive rate approaching 40 to 46 percent in that population before a single trafficked person is considered, and cannot function as a discriminating indicator on its own. This is arithmetic about the base rate, not a claim about any study in §4.

3.2 The one place the base rate genuinely changes: minors

Tattooing a minor is prohibited or restricted in every U.S. state, though the exact rule varies. The University of North Carolina School of Government's 2022 fact sheet on the subject makes the operational point directly for its own state: "Since no one under age 18 can legally be tattooed in North Carolina, any professional who works with youth should be aware of the connection between trafficking and tattoos."

Arizona, which §7 covers in detail, illustrates why that inference is weaker than it looks. A.R.S. 13-3721, "Tattoos, brands, scarifications and piercings; minors," makes it a class 6 felony to "brand, scarify, implant, mutilate, tattoo or pierce the body of a person who is under eighteen years of age" without the physical presence of the parent or legal guardian. The prohibition is on doing it unaccompanied, not on doing it at all: with a parent present, tattooing a minor in Arizona is lawful.

Assessment (Confidence: Low; roughly even chance): the base-rate objection in 3.1 is weaker for minors than for adults, but not by as much as a flat prohibition would imply, and this document cannot say by how much. Correction (multi-model verification pass): an earlier draft rated this Moderate and said the lawful pathway was "largely closed." At least in Arizona it is not closed, it is conditioned on parental presence, and the Ohio operational data in §4.4 records roughly 70 percent tattoo prevalence in a screened youth population despite comparable statutes. A tattoo on a 14-year-old is still a different observation from a tattoo on a 34-year-old. This document did not locate a published measurement of tattoo prevalence among U.S. minors, which is what would be needed to state the minor base rate rather than infer it from the legal prohibition, and §13 records that as an open item. The inference from "it is illegal" to "it is rare" is an assumption, and it is logged as such in §10.

3.3 A caution about the 2018 literature's base-rate figure

The most-cited paper in §4, Fang et al. 2018, states that "approximately 25% of the US population between 18 and 50 have at least one tattoo," citing Blackburn et al. 2012. The 2023 Pew figure for the overlapping age bands is 41 percent for 18-29 and 46 percent for 30-49. The base rate against which the 2018 tattoo-screening argument was made was therefore roughly half the current measured rate for the relevant ages, which weakens that argument further rather than strengthening it. This document does not treat the discrepancy as an error by Fang et al., who cited the best figure available to them at the time; it records it because anyone reading the 2018 paper today is reading an argument built on a superseded denominator.

4The marking of victims by traffickers

4.1 Where the tattoo claim comes from (PEER-REVIEWED: Fang, Coverdale, Nguyen and Gordon 2018)

The proposition that trafficking victims are identifiable by tattoos, and that clinicians should look for them, has a traceable point of origin in the medical literature. It is Fang S, Coverdale J, Nguyen P, Gordon M, "Tattoo Recognition in Screening for Victims of Human Trafficking," Journal of Nervous and Mental Disease, volume 206, number 10, October 2018, pages 824-827. The University of North Carolina School of Government's 2022 fact sheet on tattooing of trafficking victims states in its own closing line that "much of the information in this fact sheet was gained through this publication," and cites Fang et al. by name.

The paper's thesis is stated in its abstract: "Because traffickers may mark victims, identification of tattoos provides a useful method for screening patients, which complements history taking, especially when victims are unable to disclose that information."

Its method is stated with equal clarity, and is the single most important fact about it. The authors searched PsycINFO, PubMed and JSTOR for the keywords "human," "trafficking" and "tattoo." They report: "We found little information by these methods, so we also searched the available, nonacademic 'gray' literature, which consisted of popular news articles found via Google using the same keywords. Social media photo albums that were volunteered by a renowned tattoo organization specializing in tattoo transformation were also examined."

That organization is named in the paper: Survivor's Ink, a nonprofit that removes or covers branding tattoos, and the photographs analyzed were from its public Facebook page. The paper is a narrative review with an illustrative image set. It examined no patients, screened no population, and had no comparison group.

4.2 What Fang et al. says about its own limits

This is quoted at length because it is routinely omitted when the paper is cited, and because it disposes of the strongest form of the claim built on it. From the paper's Discussion:

"Limitations of this study include a lack of peer-reviewed literature on this topic, which resulted in our inclusion of images and information garnered from social media Web sites and non-peer-reviewed articles. We were also unable to determine whether or how the tattoos of trafficked individuals differ substantially from those of other populations such as sex workers, gang members, and so on. Currently, the limited information available does not allow us to draw any conclusions about the frequency of tattooing among trafficking victims by traffickers. It also prevents us from identifying whether such tattoos have unique characteristics in comparison with voluntary tattoos."

The authors then say what would be required to fix this: "The most accurate way to gather this information would be to assess or interview current and former trafficking victims to determine what percentage of victims actually has tattoos and/or brands, and what they look like, to identify patterns more concretely."

Assessment (Confidence: High; almost certainly true): the foundational paper for tattoo-based trafficking screening explicitly disclaims both of the propositions that the practice requires, namely that the frequency of trafficker-applied tattooing is known, and that such tattoos are distinguishable from voluntary ones. It proposes tattoo recognition as a hypothesis worth developing, not as a validated method, and it says so in its own text. Any subsequent use of Fang et al. as authority for a prevalence figure or for a distinguishing characteristic is a use the paper refuses.

This is the RECTUMmendation risk in this subject in its purest form. The underlying observation, that some traffickers mark some victims, is true and is documented in §4.6 by admitted facts in federal prosecutions. The step from that to "tattoos identify trafficking victims" is an unverified assertion, and the paper most often cited for it says so.

4.3 The quantitative evidence that does exist (PEER-REVIEWED: Rambhatla et al. 2021)

One systematic attempt to quantify skin findings in trafficking survivors was located: Rambhatla R, Jamgochian M, Ricco C, et al., "Identification of skin signs in human-trafficking survivors," International Journal of Women's Dermatology, 2021, volume 7, issue 5, pages 677-682. It is a narrative review synthesizing ten peer-reviewed studies covering approximately 1,522 patients, drawn from an initial screen of 577 records.

Its reported findings on physical marks, with the denominators the review itself gives:

Two things in this table matter more than the percentages. First, the tattoo figure rests on a denominator of 38. Second, the review's authors state their own principal limitation: most participants across the underlying studies were "recruited from post-trafficking services," which limits the ability to characterize signs during active trafficking, and the underlying documentation was sparse and heterogeneous.

Assessment (Confidence: Moderate; likely): the best located quantitative estimate of tattoo prevalence among sex-trafficking survivors, 21 percent on n=38, is below the general-population rate for adult women reported in §3.1, which Pew puts at 38 percent. The two figures are not strictly comparable, being different populations measured by different instruments for different purposes, and this document does not claim that trafficking survivors are less tattooed than the general public. What it does claim is narrower and follows directly: no located source establishes that tattoo prevalence among trafficking survivors exceeds the general-population base rate, which is the minimum condition for a tattoo to function as an indicator at all.

VERIFICATION FLAG: the one comparative figure in this literature could not be verified against its source. Fang et al. 2018 states that "tattoos were much more common in commercially sexually exploited children (48%) compared with other pediatric victims of sexual abuse (5%)," attributing this to Greenbaum, Dodd and McCracken 2018 in Pediatric Emergency Care. That paper's full text is paywalled (journals.lww.com returned HTTP 402 to a direct fetch; the abstract was retrieved successfully through Ovid via browser). The abstract describes the sixteen variables on which its two groups differed as involving "reproductive history, high-risk behavior, sexually transmitted infections, and previous experience with violence," and does not mention tattoos. Two attempts across this pass failed to locate the figure in any source other than Fang et al. or a work quoting Fang et al. The figure is therefore reported here as an attribution by Fang et al. that this document could not confirm, is not treated as load-bearing anywhere, and is logged as an open item in §13. If it is genuine it is the single most useful statistic in this section, because it is the only one with a comparison group, which is precisely why it should not be repeated until someone has read it in the source.

4.4 The one direct empirical test of the tattoo indicator (PRIMARY: Ohio Department of Youth Services, 2023)

Everything in 4.1 through 4.3 concerns whether the tattoo claim has an evidence base. One study tested the indicator itself, at scale, against outcomes, and it is the most important single source in this section.

Source: Anderson VR (University of Missouri-St. Louis), McKenna NC (Rutgers), Pierce K (University of Cincinnati), "Validating the Human Trafficking Screening Tool for Justice-Involved Youth," June 20, 2023, prepared under grant 2019-JG-E01-V6465 for the Ohio Office of Criminal Justice Services. The instrument is the Ohio Department of Youth Services Human Trafficking Screening Tool, administered to justice-involved youth. Sample: 2,010 assessments analysed, collected March 2019 to August 2022 (the report's descriptive tables carry 2,015), 93.5 percent male, 61.6 percent Black, mean age 18.88. Not peer-reviewed; a state-commissioned validation study by an academic team.

The study computed the area under the receiver operating characteristic curve for each screening domain against trafficking victimization, classifying results by its own stated scheme: "not predictive (< .55), weak (.56 to .64), moderately predictive (.65 to .71), or strong (> .71)."

The report's own conclusion, verbatim: "Gang involvement (AUC = 0.470) and evidence of tattooing or branding (AUC = 0.533) were not predictive of human trafficking victimization." And, in its recommendations, that "gang involvement and tattoo-related questions are likely poor indicators of human trafficking risk among juvenile justice system-involved youth."

The underlying counts explain why. From the report's own descriptive statistics and narrative:

• "The majority of assessments indicated that youth had at least one tattoo (n = 1,402 assessments or 930 unique youth)." Against the assessment total, that is roughly 70 percent of the screened population.
• "Evidence of branding was flagged in 11 assessments (0.55%)."
• "Only one assessment was flagged for a suspected trafficker escorting youth to get a tattoo. No cases were flagged for tattoo meaning as suspected branding (e.g., the tattoo included a suspected trafficker's initials/name or forced branding/ownership)."
• Scars or brands purposefully inflicted were recorded in 99 assessments (4.92 percent), and the source of the scar was the youth themselves in 60.82 percent of those, then a friend 18.56 percent, a family member 11.34 percent, a gang member 5.15 percent and a romantic partner 4.12 percent.

Correction (multi-model verification pass): an earlier draft of this section described the indicator as performing "indistinguishably from a coin flip." That was wrong, and the report's other tables show why. The same study's Table 4 records that evidence of branding was flagged in 6.5 percent of victim assessments against 0.4 percent of non-victim assessments, z = -5.55, p < .001, and its Table 5 gives a phi correlation of .124, which the report itself describes as "a significant weak association with human trafficking victimization." Branding is therefore not unrelated to the outcome; it is roughly sixteen times more common among assessments flagged as victims. Only gang involvement, at rho-phi -.023, showed no association and was not statistically significant.

What an AUC of 0.533 reflects here is not absence of signal but absence of sensitivity. The flag appears in 0.55 percent of all assessments. An indicator that is present in one screened person in two hundred cannot move a curve that measures discrimination across the whole population, however specific it is when it does appear.

Assessment (Confidence: High; almost certainly true): in the only located study that measured the tattoo indicator against trafficking outcomes, the indicator is rare, comparatively specific, and useless as a screen. Its association with victimization is statistically significant and weak; its sensitivity is on the order of 6.5 percent; and its AUC of 0.533, with a confidence interval of .441 to .625 spanning 0.5, places it below the study's own threshold for predictive. The correct statement is not that a tattoo tells you nothing, but that a screening programme built on it would miss more than nine victims in ten while flagging a population in which roughly 70 percent of people have tattoos anyway.

The reference standard here is the same circular one this document flags elsewhere. The report states that "Upon completion of the screening, screeners identified the youth as one or more of the following: non-victim, a victim of sex trafficking, a victim of labor trafficking, indicated-sex trafficking (potential victim/suspicion of trafficking but more information is needed), or indicated-labor trafficking." The outcome is the screener's own designation at the end of the same session in which they administered the screen. Across the 2,010 assessments analysed, 46 carried a trafficking outcome: 13 sex trafficking victims, 4 labor trafficking victims, 18 potential sex trafficking victims and 11 potential labor trafficking victims. Twenty-nine of the 46 are therefore "potential" rather than confirmed. This is the same structure §5.5 identifies in the Greenbaum multi-site study, and it applies here with equal force; an earlier draft of this section did not say so.

Two scope limits follow, and both matter. This is a single study, in a single state, on juvenile-justice-involved youth who are 93.5 percent male with a mean age of 18.88 and an age range of 12.9 to 23.6 years, which makes the sample substantially adult rather than a study of minors. It is not generalisable to adults in healthcare settings, to women, or to the pimp-controlled commercial sex population the marking literature actually describes. Within its own population it remains the strongest evidence located on either side of the question, because it is the only study with an outcome, a denominator in the thousands, and a comparison across indicators measured the same way.

Two further details carry independently. First, roughly 70 percent tattoo prevalence in the screened population reproduces the base-rate problem of §3.1 in an operational dataset: an indicator present in seven of ten screened people cannot discriminate among them. Second, across all 2,010 assessments by trained staff explicitly asking about branding, not one tattoo was assessed as a trafficker's brand. The assessors were looking for the thing and recorded none of it, which is a different and stronger finding than the AUC.

The same report tested two other instruments in wide juvenile-justice use against the same outcome. The Ohio Youth Assessment System produced a significant association in only one of seven domains, and the Prison Rape Elimination Act screening indicators produced none, correlations running from .006 to .046. The report concludes that "the OYAS and PREA measures performed poorly when modeled to predict any trafficking victimization outcome."

4.5 Burns, scarification and coerced self-marking

Fang et al. note that "prior studies have indicated that burns are a common method of marking victims of sex trafficking," citing Patel 2007 and Rezaeian 2017. Both concern lower-middle-income countries rather than the United States, and this document did not fetch either, so neither is graded or relied on here; they are recorded as a pointer in §13.

The distinction that matters for the United States record is a different one, and it is visible in the language of the federal prosecution summaries in 4.6. Marking is repeatedly described not as something done to a restrained person but as something the person was made to do or agree to. The Attorney General's FY2023 report says of one defendant that he "made some victims brand themselves with tattoos," and of another that he "pressured many of his victims into tattooing his first name on their bodies." Only in the third does the report describe the defendant as the actor: he "tattooed the minor victim with an image he used as his personal brand."

Assessment (Confidence: Moderate; likely): in the located United States federal record, trafficker marking is more often coerced consent than physical force, which means the resulting tattoo is frequently indistinguishable in execution and placement from a voluntary one, because in a mechanical sense it was voluntary. This is consistent with Fang et al.'s inability to identify distinguishing characteristics, and it is a substantially different phenomenon from branding in the sense the word implies.

4.6 What federal prosecutions actually establish (PRIMARY: DOJ and USAO records)

Marking is real, it is charged, and it is documented in specific cases with names and sentences. Four are set out here, distinguished by evidentiary status, which is the distinction most public discussion of this subject collapses.

Admitted as fact in a guilty plea. United States v. Hamidullah (M.D. Fla.). On February 23, 2017, U.S. District Judge Carlos E. Mendoza sentenced Abdhullah Hamidullah, 43, to 482 months in prison and lifetime supervised release. He had pleaded guilty on June 17, 2016, to sex trafficking by force, fraud and coercion under 18 U.S.C. 1591 and to offenses under 18 U.S.C. 2421 and 2422. The Justice Department's release states, of admissions made in connection with the plea, that he "isolated her in his apartment, took away her money and phone, and installed an alarm without providing her the code. He also assaulted her, showed her his handgun, and branded her with a tattoo in the course of compelling her to prostitute for his profit."

Found and reported by the Department in sentenced cases. The Attorney General's FY2023 annual report describes three:

• United States v. Hall (D. Mass.). November 2022, 18 years, sex trafficking across seven states; $1,859,000 restitution ordered March 2023 among four victims. The defendant "targeted victims suffering from substance use disorders," and "made some victims brand themselves with tattoos."
• United States v. Jackson et al. (S.D. Tex.). February 2023, Aryion Dupree Jackson sentenced to 324 months for sex trafficking of a minor victim and conspiracy to traffic an adult victim by force, fraud or coercion. "The defendant tattooed the minor victim with an image he used as his personal brand."
• United States v. Moses (E.D.N.Y.). September 2023, Somorie Moses pleaded guilty to sex trafficking eight women and to the 2017 murder of one of them, Leondra Foster, in the first use of the federal statute criminalizing murder in the course of sex trafficking. "The defendant pressured many of his victims into tattooing his first name on their bodies."

Alleged in an indictment, not proven. United States v. Armstead et al. (C.D. Cal.), announced August 13, 2025, press release 25-213. A 31-count indictment charges eleven defendants, members and associates of the Hoover Criminal Gang, with RICO conspiracy including sex trafficking of children and adults on the Figueroa Corridor of South Los Angeles. The release states: "Victims also were branded with tattoos of a defendant's moniker." It also states, in its own words, "An indictment contains allegations. All defendants are presumed innocent until proven guilty beyond a reasonable doubt in a court of law." This document repeats that caveat rather than dropping it.

Three features recur across all four and none of them is the tattoo. Every one is domestic sex trafficking, not labor trafficking and not cross-border movement. Every one involves an established controlling relationship in which the mark is one instrument among assault, drugs, debt, isolation and threat. And in every one the mark is evidence of a relationship the investigation had already established by other means, not the thing that revealed it.

4.7 The frequency question, in the federal government's own accounting

The Attorney General's FY2023 Annual Report to Congress runs 251 pages and is the United States government's consolidated statement of its own anti-trafficking activity, including its protection programs, its victim identification mechanisms and its prosecutions. The word "tattoo" appears in it three times. All three are in the case summaries quoted in 4.6. The words "branding" and "branded" do not appear at all. No section of the report on victim identification, screening, training or indicators mentions tattoos in any form.

Assessment (Confidence: High; almost certainly true): the federal government does not treat tattoos as a victim-identification method in its own comprehensive annual accounting of how it identifies victims. Tattoos appear there only as facts of particular prosecutions. This is a checkable, falsifiable statement about a single named public document, and it is offered as such.

4.8 Synthesis across 4.1 to 4.7

Six independent lines converge, and they converge on a narrower claim than either the popular version or its debunking.

The practice is real. Federal courts have accepted admitted facts of trafficker tattooing in at least one case carrying 482 months, and the Justice Department reports three more among sentenced defendants in a single fiscal year. Nothing in this section should be read as suggesting the phenomenon is invented.

Its frequency is unmeasured. The foundational paper says so in its own limitations section. The best quantitative estimate located rests on 38 people. The one figure with a comparison group could not be verified in its source.

When it was directly tested, it proved too rare to screen on. The Ohio validation study, 2,010 assessments analysed, found branding significantly but weakly associated with victimisation (6.5 percent of victim assessments against 0.4 percent of non-victim ones, phi = .124) while returning an AUC of 0.533, below its own threshold for predictive, because the flag appears in barely one screened person in two hundred. In that population, trained assessors asking specifically about branding recorded zero tattoos assessed as a trafficker's brand.

It is concentrated in one form of trafficking. Every located United States case is domestic, pimp-controlled sex trafficking. §2.3 establishes that this is the minority form among federally identified victims, which are 77 percent labor by the Supplement B measure and 310 of 382 forced labor by the Continued Presence measure.

It is not distinguishable from voluntary tattooing. Fang et al. states this directly, and 4.5 gives the mechanism: much of the marking in the record is coerced consent, producing a tattoo made the ordinary way, by the ordinary means, often at the person's own hand.

And it sits at or below the base rate. §3.1 puts tattoos at 38 percent of adult women and 46 percent of adults aged 30 to 49; the best survivor estimate located is 21 percent on n=38; and in the Ohio operational dataset roughly 70 percent of screened youth had at least one tattoo.

Assessment (Confidence: High; almost certainly true): trafficker marking is a genuine, prosecutable, documented practice, and simultaneously a poor basis for identifying trafficking victims. Both halves are true at once, and treating either half as the whole story misstates the record. The two halves are not in tension once the distinction in §6.7 is applied: a mark can be strong evidence about a person already under investigation and near-worthless for picking that person out of a crowd.

One boundary on the above. The Ohio study covers justice-involved youth in one state, and §3.2 gives a reason the base-rate objection is weaker for minors generally, since the lawful route to a tattoo is largely closed to them. This document does not extend the Ohio result to adults, to healthcare settings, or to non-justice-involved minors, and §13 records the absence of an equivalent adult study as an open item.

5Identification of living persons: indicator lists and screening instruments

§4 concerned one proposed indicator. This section concerns the method that indicator belongs to, and the instruments built to apply it systematically. The two halves diverge sharply in quality. The indicator lists are unvalidated, and the federal government says so about its own. The screening instruments have been measured, and what the measurements show is the most quantitatively solid material in this document.

5.1 The federal government does not agree with itself about tattoos (PRIMARY: DHS, HHS, FBI)

There is no single federal indicator list. There are several, they were fetched individually for this document, and they do not say the same thing.

DHS's principal products omit it. The Blue Campaign identification page carries the caveat "not all indicators listed below are present in every human trafficking situation, and the presence or absence of any of the indicators is not necessarily proof of human trafficking," and then lists thirteen indicators covering disconnection from family, school attendance, behavior change, juvenile commercial sex, disorientation, "bruises in various stages of healing," fearfulness, deprivation of food, water, sleep or medical care, deference to a controlling companion, coached speech, unsuitable living conditions, lack of possessions, and freedom of movement. Tattoos, branding and scars do not appear. The printed indicator card, the artifact that actually gets handed out, carries eleven items and does not mention them either.

A DHS audience-specific toolkit does include it. The Blue Campaign's college-campus guide asks whether the individual has "tattoos or scars that would indicate branding by a trafficker," fourth among thirteen physical or behavioral items, under the framing sentence "While no single indicator is necessarily proof of human trafficking, recognizing the signs is the first step in identifying possible victims."

Assessment (Confidence: High; almost certainly true): DHS is internally inconsistent on this indicator, and the inconsistency runs in a specific direction. The department's central, most widely distributed products omit tattoos entirely, while a guide written for untrained student volunteers retains them. That is the inverse of where an unreliable indicator should sit, since the least-trained audience is the one least able to apply the discrimination the indicator requires.

The wording in the campus guide cannot be applied as written. It asks an observer to identify "tattoos or scars that would indicate branding by a trafficker," which presupposes the observer can already distinguish a trafficker's brand from any other tattoo. §4.2 establishes that the foundational literature says nobody can, §4.4 shows it failing an empirical test, and §4.5 gives the mechanism. The indicator instructs the observer to perform the discrimination that is the entire unsolved problem.

The FBI includes it, with the strongest caveats in any federal product. The FBI's Trafficking Indicators page states, as four separate bulleted rules: "Indicators cannot be used as a checklist or formula"; "No individual indicator is definitive"; "No number of indicators is determinative"; and "Context is key," the relevance of each indicator depending on the nature of the labor or commercial sex, the profile of the potential victims, the surrounding conditions and the context of the encounter. Its Physical indicator block lists "Signs of injury, scars, bruises, cuts, burns; unexplained marks" and does not mention tattoos. "Tattoos or branding" appears exactly once on the page, in the Additional Sex Trafficking Indicators block, with no indicator-specific caveat.

The same FBI page independently corroborates two findings this document reaches elsewhere: "Human trafficking does not require any smuggling, movement, or transportation across state or international borders" (§2.5), and "Most human trafficking does not involve forcible abduction or physical restraints. Victims are often lured and compelled through deception, debts, threats, psychological coercion, or addictive drugs... overt evidence of forcible restraint is rare" (§4.5).

A state attorney general includes it with no caveat at all. The Arizona Attorney General's public page on human trafficking carries a seven-item list headed "Indicators of human trafficking": "Signs of physical abuse / Disconnected from friends/loved ones / Talk of an older boyfriend/girlfriend / New material items they cannot afford / Signs of malnutrition / Drug addiction / Unusual tattoos / branding." The list carries no caveat of any kind, neither the DHS "presence or absence... is not necessarily proof" formula nor the FBI's four rules. It is the shortest indicator list located and the tattoo item is one seventh of it.

5.2 HHS says its own screening tool is not validated, then ships a red-flag list anyway

Source: "Adult Human Trafficking Screening Tool and Guide," January 2018, produced by the National Human Trafficking Training and Technical Assistance Center under contract to the HHS Administration for Children and Families Office on Trafficking in Persons, principal authors Wendy Macias-Konstantopoulos MD MPH and Julie Owens.

The toolkit's statement about itself, verbatim: "While this tool is not yet validated, it has been developed based on the latest research and best practices in screening."

Its statement about red-flag lists in general, verbatim: such lists "are not formal and typically are not validated," and "Unlike screening instruments, red flag checklists do not identify potential for future risk." It goes further: "There appears to be no consensus of opinion in trafficking research literature or among experts about whether red flag checklists are useful instruments." It records that some experts decline to use them "because they can be too easily rushed through by busy professionals," while others hold that they "may nonetheless be valuable, given that no better options exist."

The same publication then ships Appendix B, "Indicators of Human Trafficking," a red-flag checklist, under the instruction "You should become familiar with these indicators." Its sex trafficking column reads: "Works in the commercial sex industry: escort, exotic dancer, 'prostitute,' 'massage' / Signs of having sex with multiple people / Has pimp: male, female, boyfriend, husband / Tattoos or branding of ownership / Uses language of the sex industry / Inappropriate clothing for venue or weather / Physical abuse, drugs/alcohol, malnourished."

Assessment (Confidence: High; almost certainly true): the federal toolkit that most clearly states the case against red-flag checklists also distributes one, including the tattoo indicator, without the caveat it applied to the category three pages earlier. This document does not read that as bad faith; the toolkit's own explanation, that no better options exist, is stated plainly in its text. It is recorded because a reader who takes Appendix B as federally validated guidance would be taking it for something the same document says it is not.

Note also that several Appendix B items describe consensual adult sex work, poverty or housing instability rather than trafficking: "works in the commercial sex industry," "wears the same clothes over and over," "multiple people living in a cramped space," "inappropriate clothing for venue or weather." At population scale these carry very low specificity, which is the subject of 5.4.

5.3 What "validated" means in this field (PEER-REVIEWED: Hainaut et al. 2022; Macy et al. 2023)

The direct answer comes from a scoping review that asked exactly this: Hainaut M, Thompson KJ, Ha CJ, Herzog HL, Roberts T, Ades V, "Are Screening Tools for Identifying Human Trafficking Victims in Health Care Settings Validated? A Scoping Review," Public Health Reports, 2022 Jul-Aug, volume 137, supplement 1, pages 63S-72S, PMID 35775913.

Method and yield in the authors' own figures: searches across MEDLINE, PsycInfo, Embase and Scopus with no language or date limits yielded 8,730 studies, 4,806 after removing duplicates. 4,720 were excluded on title and abstract, 85 full texts reviewed, and 8 articles included. Hand-searching found 9 additional screening tools that appear nowhere in the literature at all. Of everything located, "only 6 had been studied for validation in health care settings."

Their conclusion: "Few studies have evaluated screening tools for identifying victims of human trafficking in health care settings. The absence of a gold standard for human trafficking screening and lack of consensus on the definition of human trafficking make screening tool validation difficult."

And the methodological core, which is the sentence that governs this whole section: "many victims do not disclose their trafficking status because of a lack of trust in health care workers, fear of trafficker retaliation, guilt and shame, or lack of awareness that they are being trafficked. Because disclosure or physician concern is not a reliable reference standard for identification, these validation studies may not be especially strong."

A second scoping review agrees from a wider frame: Macy RJ, Klein LB, Shuck CA, Rizo CF, Van Deinse TB, Wretman CJ, Luo J, "A Scoping Review of Human Trafficking Screening and Response," Trauma, Violence and Abuse, 2023, volume 24, number 3, pages 1202-1219. Across 22 screening tools in 26 sources it finds that "most tools were developed by practice-based and non-governmental organizations located in the U.S." and that "Few screening tools have been rigorously evaluated." It also reports being unable to locate step-by-step response protocols, meaning that even where screening happens, what to do with a positive is frequently undefined.

The validation chain is layered and partly circular. Each named instrument was measured against a reference standard that was itself a human judgment:

Vera's own report states the problem without evasion: "Since no 'gold standard' of trafficking victim identification exists with which to compare the results of the study... establishing certainty in criterion validity was difficult," and "It was therefore necessary to rely on the expertise of the experienced study partners in making this assessment, and some level of subjectivity may have influenced the determination." It also records that administration was unblinded, since "the service providers who conducted the screening sometimes knew the trafficking status of their clients," and that the tool was tested on established clients rather than at first contact: "we do not know how effective the tool would be if used with new clients or trafficking victims on the first encounter with screeners."

A note on two numbers that circulate widely. Vera's report gives, for its sex trafficking model, 96.8 percent correctly predicted, 98.9 percent of trafficking victims correctly predicted and 92.1 percent of non-victims correctly predicted. These are in-sample logistic regression classification rates from models fitted on the same cases they are scored against, using an outcome derived from the tool itself; the report's "all predictors" row reads R-squared 1.000 with 100 percent on all three measures. They are frequently relabelled downstream as sensitivity and specificity. They are not, and this document does not use them as such.

5.4 The measured performance of the best instruments

Three instruments have published accuracy figures against a stated reference standard, across five studies. The TVIT is excluded for the reason given immediately above. The pattern across all of them is identical: high sensitivity, poor specificity, and low positive predictive value once the real base rate is applied.

The single most useful number in this literature is the Kaltiso PPV of 10.0 percent. In that study 100 of 203 patients screened positive, 49 percent of everyone screened, while the total number of true trafficking victims identified was 11, of whom 10 screened positive. Roughly ninety false positives accompanied ten true ones, in a population selected in advance for high-risk presenting complaints. Hainaut et al. summarize the SSCST across all three of its studies in one line: "Although it is short and has consistently demonstrated relatively high sensitivity (92%, 84.4%, 90.9%), its specificity is low (73.0%, 57.5%, 53.1%)."

Assessment (Confidence: High; almost certainly true): indicator-based trafficking screening is a rule-out instrument, not a rule-in instrument. Negative predictive values in these studies run 97 to 99 percent and the authors of the prospective study say so directly, describing the tool as "appropriate for an initial screening to rule out CST in this high-risk population." Positive predictive value, where anyone reports it honestly against a real base rate, is 10 percent. The same arithmetic governs every indicator in 5.1, including the tattoo indicator, and it is not a defect in any particular instrument.

Two further points of precision. The Kaltiso sensitivity confidence interval runs from 58.7 to 99.8 percent because it rests on 11 true cases, so the instrument's ability to catch victims is far less firmly established than the point estimate suggests. And RAFT's headline "100% sensitive (95% CI, 100%-100%)" rests on 12 positive cases; a confidence interval of 100 to 100 on n=12 is a saturated numerator, not a precision claim.

The RAFT authors' own limitations deserve quoting because they describe who screening misses: patients excluded from the study included those who "presented with intoxication, or substance use disorder or mental illness complications, could not speak with the interviewer alone, or who presented and were dispositioned in the middle of the night." Those are close to a description of the highest-risk presentations. They also note that "Most participants with a trafficking experience in this study were not in their situation at the time of the interview," so the instrument was largely measured against past rather than active exploitation.

5.5 What the reference standard actually consisted of

The Greenbaum multi-site study, n=810, published the basis on which its 90 confirmed victims were classified, and it is the clearest available window into what "gold standard" means here. Its own statement: "The HCP's decision about CST status served as the 'gold standard' for calculating sensitivity/specificity," and "Short of witnessing exploitation, a true gold standard for victim identification is lacking."

Of the 90 patients classified as trafficking victims, "Explicit statements documenting how the decision was determined were given for 81% of those patients. For the remaining 19%, the HCP did not provide a specific reason for the decision." Seventeen of ninety confirmed cases therefore have no recorded basis.

Among the stated bases: disclosure of sex in exchange for money, drugs or housing, 36.7 percent; police sting, 13.3 percent; "participant displayed multiple risk factors of trafficking," 5.6 percent; nude images, 4.4 percent; police data, 4.4 percent; a Backpage advertisement, 3.3 percent; electronic contact with adults, 3.3 percent; known trafficker, 3.3 percent; and, at 1.1 percent, one case in which "participant was found living on the street but was unusually well groomed."

Assessment (Confidence: High; almost certainly true): in 5.6 percent of confirmed cases the reference standard was the presence of risk indicators, meaning the outcome the indicators were being tested against was in part defined by the indicators. That is circular, it is visible in the published table, and it inflates measured performance by an unknown amount.

5.6 Practitioners do not report seeing the physical indicators

Two studies asked people who do this work which indicators they actually observe.

Gerassi LB, Nichols AJ, Cox A, Goldberg KK, Tang C, "Examining Commonly Reported Sex Trafficking Indicators From Practitioners' Perspectives: Findings From a Pilot Study," Journal of Interpersonal Violence, 2021, volume 36, numbers 11-12, pages NP6281-NP6303. A survey of 86 providers in one Midwestern city, self-described as a pilot. Its opening premise: "Commonly reported sex trafficking indicators have been disseminated widely by government and non-governmental organizations in trainings aimed to increase identification and referral to resources. However, very little research evaluates such indicators."

Most commonly observed, on a 1 to 5 scale: depression 3.82, fear or distrust of law enforcement 3.80, low self-esteem 3.59, anxiety 3.55, low interpersonal trust 3.52, fear 3.36, shame or guilt 3.34, isolation 3.30. Among the least commonly observed: presence of tattoos or branding 1.89, above false documents at 1.85 and low English proficiency at 1.62, below physical evidence of torture at 2.07. Correction (multi-model verification pass): an earlier draft placed tattoos below false documents. It is marginally above. The point is unaffected: tattoos sit near the bottom of a nineteen-item list. The authors conclude that "The most commonly identified indicators... are inconsistent with many of the indicators that are used frequently across governmental and community trainings."

Pederson AC and Gerassi LB, "Healthcare providers' perspectives on the relevance and utility of recommended sex trafficking indicators: A qualitative study," Journal of Advanced Nursing, 2022, volume 78, number 2, pages 458-470. A qualitative study of 23 healthcare staff across five sites of one Midwestern organization. Its finding on physical indicators: "Medical and physical indicators (e.g. repeat STIs, bruises and tattoos) were perceived as generally lacking clinical utility or irrelevant."

Both are small, single-region, self-report studies measuring perception rather than accuracy, and neither establishes what is true of trafficking, only what these practitioners report noticing. They are reported here at that weight.

Assessment (Confidence: Moderate; likely): the indicators practitioners report observing most often are depression, low self-esteem, anxiety, shame and distrust of authority, which are close to universal in any traumatized or system-involved population and therefore carry almost no discriminating power. The indicators disseminated in training, including tattoos, are among those practitioners report seeing least. This is a single research group's work across two small samples and is rated moderate for that reason.

5.7 The healthcare-contact premise (PEER-REVIEWED: Armstrong and Greenbaum 2019)

Every argument for clinical screening rests on a prior claim: that trafficked people reach healthcare while being trafficked. The claim is well founded, but its most-quoted number is weaker than its reputation.

The single most repeated statistic in this field is that 88 percent of trafficking victims encountered healthcare professionals while being trafficked. It comes from Lederer LJ and Wetzel CA, "The Health Consequences of Sex Trafficking and Their Implications for Identifying Victims in Healthcare Facilities," Annals of Health Law, 2014, volume 23, issue 1, article 5, pages 61-91. The study ran eleven focus groups between January and December 2012 with 107 domestic sex trafficking survivors recruited through survivor-led service providers. It is a convenience sample of people who reached post-trafficking services, a population selected on having escaped and been served.

The figure itself is narrower than its reputation. The paper reports 87.8 percent, and the denominator is not 107: it is "Of those who answered the questions about their contact with healthcare (N=98)." Hospital or emergency room contact specifically was 63.3 percent.

VERIFICATION FLAG: the Lederer and Wetzel full text was not retrieved in this pass. The journal's institutional repository page (lawecommons.luc.edu) returned only bibliographic metadata and a download link, not article text, on one attempt. The sample size of 107 and recruitment through service providers are stated consistently by peer-reviewed sources citing it, including Rambhatla et al. 2021, but this document did not read its methods section and does not grade it as fetched. The 88 percent figure is not used as load-bearing anywhere and is reported only as the origin of a widely circulated number. §13 records the retrieval as an open item.

The better-founded version is a systematic review: Armstrong S and Greenbaum VJ, "Using Survivors' Voices to Guide the Identification and Care of Trafficked Persons by U.S. Health Care Professionals: A Systematic Review," Advanced Emergency Nursing Journal, July/September 2019, volume 41, issue 3, pages 244-260. It screened 1,605 articles, of which 8 met inclusion criteria, covering 420 participants, and reports that between 50 and 98 percent sought healthcare services during their exploitation.

Assessment (Confidence: Moderate; likely): that a substantial share of trafficked people reach healthcare during exploitation is supported by a systematic review of eight studies and 420 participants, and the premise for clinical screening is sound. The specific figure of 88 percent is not the best-supported version of it, and the honest range, 50 to 98 percent, is wide enough that it should be quoted as a range. A range that wide is itself a finding about the state of the evidence.

One further figure from this literature is worth recording because it is the strongest argument for structured screening in the whole section. Mumma BE, Scofield ME, Mendoza LP, Toofan Y, Youngyunpipatkul J, Hernandez B, "Screening for Victims of Sex Trafficking in the Emergency Department: A Pilot Program," Western Journal of Emergency Medicine, 2017, volume 18, number 4, pages 616-620. In a convenience sample of 143 female ED patients aged 18 to 40 in Sacramento, with 10 true positives, the survey was 100 percent sensitive (CI 74-100) against unaided physician concern at 40 percent sensitive (CI 12-74). The sample is tiny and the confidence intervals are correspondingly wide, but the contrast is the point: whatever the failings of structured screening, unstructured clinical suspicion performed worse in the one study that measured both.

5.8 What a false positive costs

Screening instruments are usually discussed as though a false positive were free. In at least one state it is not.

Source: "The Commercial Sexual Exploitation-Identification Tool (CSE-IT)," Texas Office of the Governor, Child Sex Trafficking Team, November 2023. The CSE-IT is used across child-welfare and juvenile-justice systems in multiple states. The Texas document describes it as "a research-based screening tool," notably not as a validated one, and states plainly: "The CSE-IT is not a diagnostic tool, and it cannot confirm victimization," and "It is not meant to meet investigative, statutory, legal, or other criteria, nor does it determine that victimization has occurred."

The same document then states: "All clear concern scores must be reported to the Texas Department of Family and Protective Services Statewide Intake (Abuse and Neglect Hotline)." Scoring bands are No Concern 0-3, Possible Concern 4-8, Clear Concern 9-23.

Assessment (Confidence: Moderate; likely): a tool that its own procuring state says cannot confirm victimization nonetheless triggers a mandatory child-abuse report at a threshold score. Given the positive predictive values in 5.4, the majority of reports so triggered will concern children who are not trafficking victims. This document takes no position on whether that tradeoff is correct, which is a policy question outside its scope; it records that the tradeoff exists and is rarely stated.

VERIFICATION FLAG: no sensitivity, specificity, positive predictive value or reliability coefficient for the CSE-IT was obtainable. Its validation source is a self-published technical report (Basson D, 2017, WestCoast Children's Clinic), whose PDF returned HTTP 404 at the URL cited by the Texas Governor's office across two attempts, and the developer's web page publishes no figures while describing the tool as validated. §13 records this as an open item. It is the most consequential gap in this section, given how widely the instrument is mandated.

5.9 What the expert-testimony literature does not contain

A peer-reviewed review of how expert testimony is used in sex trafficking prosecutions was located: Zhang T and Datta V, "Expert Testimony in Sex Trafficking Cases," Journal of the American Academy of Psychiatry and the Law, 2022 June, volume 50, number 2, pages 212-220. The authors reviewed published judicial opinions describing expert testimony in sex trafficking prosecutions and found mental health experts appeared in 7 of 24 reviewed cases, with law enforcement providing most expert testimony.

The article's subject is the counterintuitive behavior of survivors: why victims stay, why they are reluctant to testify, why they recant, and how trauma symptoms undermine perceived credibility. Across its treatment of 24 cases it does not discuss tattoos, branding or physical markings as evidence at all.

Assessment (Confidence: Moderate; likely): in the courtroom, the identification problem that actually required expert explanation was behavioral, not physical. This is a negative finding from a single source and is named as such in §9. It establishes what one peer-reviewed review of 24 cases contains, not what the whole body of American trafficking caselaw contains.

6Forensic and biometric identification, and the one thing tattoos are good at

§4 and 5 concern classification: taking an unknown person and deciding whether they belong to a category. This section concerns individuation: taking a mark and deciding whether it belongs to a specific known person. These are different problems with different success rates, and the difference between them is the organizing finding of this document.

6.1 NIST measured both problems, twice (PRIMARY: NISTIR 8078 and NISTIR 8232)

NIST ran two tattoo recognition evaluations for the FBI. They are usually cited as one body of work and they should not be, because the second overturned the headline number of the first.

Tatt-C, the open challenge. Ngan M, Quinn GW, Grother P, "Tattoo Recognition Technology - Challenge (Tatt-C) Outcomes and Recommendations," NISTIR 8078, Revision 1.0, September 2016, DOI 10.6028/NIST.IR.8078. Ran September 23, 2014 to May 4, 2015 on a dataset of 16,716 tattoo images "collected operationally by law enforcement," provided by the FBI. Six participants: Compass Technical Consulting (US), Fraunhofer IOSB (Germany), the French Alternative Energies and Atomic Energy Commission, MITRE (US), MorphoTrak (US) and Purdue University (US).

NIST's own disclaimer on this report is explicit and is routinely dropped: the data, protocols and metrics "should not be construed as indicating how well these systems would perform in fielded operations," the submissions were "generally from research prototypes, not commercially available products," and because it was an open-book test in which participants held the data and ground truth, the protocol "does not preclude gaming such as algorithm training on test data, candidate list manipulation, etc."

Tatt-E, the sequestered evaluation. Ngan M, Grother P, Hanaoka K, "Tattoo Recognition Technology - Evaluation (Tatt-E): Performance of Tattoo Identification Algorithms," NISTIR 8232, October 2018. Sponsored by the FBI, with operational imagery contributed by the Michigan State Police and the Pinellas County Sheriff's Office, Florida. Gallery of 100,000 operational law enforcement tattoo images. Twelve algorithms from two providers, run black-box at NIST with no training on the test data.

On the same task, Tattoo Identification, against a realistic 100,000-image gallery with uncropped probes, the best algorithm returned a rank-10 hit rate of 72.1 percent, a miss rate of 27.9 percent. Manually cropping the probe images raised the best result to 84.8 percent, though that figure belongs to a different algorithm from the same vendor (B11I cropped, against B31I uncropped for the 72.1 percent). Reported rank-10 accuracy across the twelve algorithms ranged from 10.3 to 72.1 percent, and single searches took between 2.0 and 255.4 seconds.

Assessment (Confidence: High; almost certainly true): the 99.4 percent figure that circulates as the accuracy of tattoo matching is a rank-10 result from an open-book test on a 4,375-image gallery, and NIST's own sequestered follow-up on a 100,000-image operational gallery put the comparable figure at 72.1 percent. Anyone quoting 99.4 percent without both qualifiers is quoting a superseded number.

6.2 What NIST says tattoos cannot do (PRIMARY: NISTIR 8232)

The most important sentence in either report is in the Tatt-E caveats, and it is quoted here in full because it settles the question §4 and 5 circle:

"False positive identification rates are not documented in this report. Tattoos cannot be used as a primary biometric as an arbitrary number of people can have nearly identical tattoos. Through analysis and trial runs on the test data, searching tattoos of subjects that are known not to exist in the gallery (non-mates) sometimes returns the same tattoo on different subjects. This is akin to using face recognition to match twins that exist in the same database... Removal of such near-identical tattoo images from different people from the test dataset for all possible non-mated scenarios would require substantial human labor, which was not available for the scope of this evaluation."

And, on automation:

"Tattoo recognition is not a fully-automated, lights-out application where decisions are made without human intervention. There needs to be human adjudicators reviewing tattoos on candidates lists."

Two things follow. First, the federal standards body that built the evaluations declines to state a false positive rate for tattoo matching and says the trait cannot carry primary identification. Second, the stated reason is precisely the mechanism that defeats the screening use in §5: different people have the same tattoo, and separating them "would require substantial human labor."

Assessment (Confidence: High; almost certainly true): NIST's position is that tattoos are a supporting, human-adjudicated identifier and not a primary biometric, and NIST states this in its own words in a published report sponsored by the FBI. Any claim that tattoo matching identifies a person, or a class of persons, is contradicted by the government's own evaluation.

6.3 What tattoo searching actually produces at national scale (PRIMARY: NIJ/NamUs FY2024)

Source: National Institute of Justice, NamUs Fiscal Year 2024 Annual Report, covering October 1, 2023 to September 30, 2024. NamUs is the national repository for missing, unidentified and unclaimed persons cases, funded and administered by NIJ, and federally legislated by Billy's Law, Public Law 117-327, enacted 2022. In FY24 it processed 3,866 forensic and analytical service requests.

A necessary caveat, and it cuts both ways: NamUs bundles tattoo database searching with facial recognition and forensic art into a single "forensic image services" line. The three identifications are for all three services combined, and NamUs does not disaggregate them. There is therefore no published figure anywhere for identifications attributable to tattoo searching, and this document does not manufacture one. What can be said is bounded: tattoo searching is one of three components of a service line that produced three identifications from 328 requests in a fiscal year, while fingerprints produced 102 from 2,358.

NamUs's own caveat about its statistics is also worth carrying, because it limits every number above: "NamUs was not designed to track or report national or subnational statistics, trends, or demographic data related to missing and unidentified persons."

Separately, NIJ's published breakdown of resolved NamUs unidentified-person cases by identification method (N=3,037, data current as of February 2017) lists DNA 45.63 percent, fingerprint 19.37 percent, circumstantial 8.99 percent, dental 7.78 percent, anthropology 6.86 percent, visual 6.86 percent, body radiographs 1.46 percent, and no method provided 3.05 percent. Tattoos are not a category in it.

Assessment (Confidence: Moderate; likely): tattoo searching contributes a small and unquantified share of identifications in the only national system that performs it, and the federal apparatus does not track it as an identification modality in its own right. This is a statement about the published record, and §13 records the disaggregated figure as an open item, since it may simply not exist in published form rather than being merely unlocated.

6.4 Where tattoos genuinely earn their place: the DVI hierarchy (PRIMARY: INTERPOL 2023)

INTERPOL's Disaster Victim Identification Guide, November 2023, sets the international standard and places physical marks precisely:

"The PRIMARY and most reliable means of identification are friction ridge analysis, comparative dental analysis and DNA analysis. Unique serial numbers from medical implants may also be reliable identifiers in terms of proving identity. SECONDARY means of identification are any feature, not being a primary identifier that characterises the individual, within the context of the disaster. Such features may include personal description, medical findings as well as evidence and clothing found on the body. These means of identification typically serve to support identification by primary means but depending on the context, may be sufficient as a sole means of identification."

The guide also cautions that "identification based on photographs can be notoriously unreliable and should be avoided as the sole means of identification," and that circumstantial identification combining description, medical evidence, clothing, jewelry, tattoos and documentation "will need to be assessed on a case-by-case basis."

A documented absence worth stating: the INTERPOL DVI Guide 2023 contains no occurrence of the words "trafficking" or "migrant." It is a disaster framework and does not address trafficking victim identification. Tattoos appear in it twice.

A senior-practitioner critique of that same hierarchy exists and should be set beside it: Blau S, Roberts J, Cunha E, Delabarde T, Mundorff AZ, de Boer HH, "Re-examining so-called 'secondary identifiers' in Disaster Victim Identification (DVI): Why and how are they used?", Forensic Science International, volume 345, article 111615, April 2023. The authors argue there is "a need to scrutinise the implied inferior value of non-primary methods," note that non-primary identifiers have "proven useful on their own" in cases of political, religious or ethnic violence, and frame them as contributing to "formulating an identification hypothesis" weighted through a Bayesian approach. They also state their own method limitation directly: because secondary identifiers are referenced so inconsistently in the literature, "it was not possible to identify useful search terms," so they conducted a broad literature search rather than a systematic review.

Assessment (Confidence: High; almost certainly true): the standards position and its most credible critique agree on the operative point even while disagreeing on terminology. A tattoo supports or generates an identification hypothesis about a specific individual; it does not by itself prove identity, and neither source treats it as proof.

6.5 The decomposition clock

One study quantifies how long a tattoo remains usable on a decomposing body: Probert SJ, Maynard P, Berry R, Mallett X, Seckiner D, "Changes in the morphometric characteristics of tattoos in human remains," Australian Journal of Forensic Sciences, volume 55, issue 4, pages 474-491, published online December 20, 2021.

Its finding, as reported: tattoos changed most during the bloat stage of decomposition and remained relatively constant thereafter; of the tattoos fully analysed, "two became redundant as secondary identifiers on day 294, the other on day 333, all due to lack of visibility."

The sample is one human donor in one Australian environment, monitored by time-lapse camera over sixteen months, with three tattoos fully analysed and a fourth partially analysed. That is n=1 for the donor and n=3 for the measurement, and the figure should never be quoted without both attached. It is reported here because it is the only quantified persistence figure located, not because it establishes a general clock.

Techniques exist to recover marks that visual inspection cannot. Holz F, Birngruber CG, Ramsthaler F, Verhoff MA, "Beneath cover-up tattoos: possibilities and limitations of various photographic techniques," International Journal of Legal Medicine, volume 134, number 2, pages 697-701, March 2020, tested photographic methods on ten volunteers with eleven known cover-up tattoos and reported that "in nine out of eleven cover-up tattoos, the previous tattoos could, at least, be partially visualized," with infrared photography at 850 nm performing best on coloured cover-ups. Tatt-E separately tested multispectral imaging on 150 probes and found best matching in the 1100-1300 nm short-wave infrared band, while noting that "the best match performance remains on images collected in the visible spectrum, which is observed across all algorithms."

6.6 Human error in reading tattoos, and a warning from an adjacent literature

Two findings belong here because they bear directly on the human-adjudicator role NIST says is mandatory.

People disagree about what a tattoo is. Brookes GK, Thompson T, "The impact of personal perception on the identification of tattoo pattern in human identification," Journal of Forensic and Legal Medicine, volume 64, pages 34-41, May 2019. A questionnaire of tattoo images distributed randomly, 211 responses. The authors report that "the perception of tattoos has a high margin for error and interpretation," while stating themselves that "further work needs to be carried out to establish the degree of variation." The sample is self-selected with no stated sampling frame, and the authors do not claim to have quantified the effect.

The inference from tattoo content to a person's category has been tested elsewhere, and it mostly fails. A series of medico-legal autopsy studies from Forensic Science South Australia examined whether tattoos predict manner of death. The two largest and best-controlled found no effect:

• Stephenson L, Byard RW, "Cause, manner and age of death in a series of decedents with tattoos presenting for medicolegal autopsy," Journal of Forensic and Legal Medicine, volume 64, pages 49-51, May 2019. 100 consecutive autopsy cases with tattoos, 2013-2017, against age and sex matched controls. Finding: "no significant association between the number of tattoos and premature mortality, or between the cause and manner of death and the presence or absence of tattoos. Previous stereotypes regarding tattooed individuals may no longer apply."
• Byard RW, Cavuoto R, "Manner of death associated with multiple tattoos," Journal of Forensic and Legal Medicine, volume 83, article 102242, October 2021. Prospective, 150 medicolegal cases with five or more tattoos in anatomically separate areas, against 100 non-tattooed controls. 78 natural and 72 unnatural deaths against 56 natural and 44 unnatural in controls, p=0.3. Conclusion: "the actual number of tattoos appears to have minimal effect."

Positive findings in the same series come from far smaller samples selected by tattoo content rather than presence: expletive tattoos (n=19, 79 percent unnatural deaths, p<0.01) and swastika tattoos (n=26, significantly more unnatural deaths, p<0.01). In the latter the author records his own counter-example, that in one case "the presence of a sacred Buddhist tattoo suggested that the swastika tattoo instead had religious rather than antisocial/racist significance," and that the symbol "may have completely different connotations for certain religious groups such as Jains, Hindus and Buddhists."

Assessment (Confidence: Moderate; likely): the structure of the content-based findings in this literature is the same structure as the trafficking branding claim. Large controlled samples testing tattoo presence find nothing; small hand-selected samples defined by tattoo content find an association, and the authors themselves supply the confound. This is offered as a structural parallel drawn by this document, not as a finding by those authors about trafficking, and it is rated moderate rather than high for that reason.

6.7 Synthesis: individuation works, classification does not

Assessment (Confidence: High; almost certainly true): every use of physical marking in this document sorts onto one side of a single line. Confirming that recovered remains bear the tattoo a named missing person was known to have is individuation, it is supported by INTERPOL as a secondary identifier, and NIST measures it in the 70s to 90s depending on conditions. Inferring from a tattoo that its bearer belongs to a category, whether that category is "trafficking victim" or "gang member" or "likely to die unnaturally," is classification, and every attempt to measure it that this document located returns a number between 10 and 15 percent, or is not measured at all.

6.8 Biometric verification of claimed family relationships at the southwest border

Source: DHS Office of Inspector General, "CBP Officials Implemented Rapid DNA Testing to Verify Claimed Parent-Child Relationships," OIG-22-27, February 8, 2022. An Inspector General audit, and therefore a stronger source than an agency press release about its own program.

In May 2019 ICE Homeland Security Investigations piloted Rapid DNA testing at 11 locations across the southwest border, which the OIG defines as including Arizona. The legal driver was an enforcement order in the Ms. L litigation, in which the court held that "Defendants must conduct DNA testing before separating an adult from a child based on parentage concerns," noting the technology "can determine parentage in approximately ninety minutes." ICE HSI ended the pilot on September 12, 2021; CBP signed its own contract on September 10, 2021, covering 18 Border Patrol and Office of Field Operations locations including in Arizona, with results returned within 24 hours.

The figure this program is known for, quoted from the OIG report: "According to ICE HSI statistics, from June 2019 to September 2021, investigators completed 3,516 Rapid DNA tests with 300 (8.5 percent) testing negative for claimed parent-child relationships and 3,216 (91.5 percent) testing positive."

That 8.5 percent is very widely repeated as though it described families arriving at the border. It does not, and the same report says why in its own description of the process: "During the pilot program, CBP officials referred parent-child relationship concerns to ICE HSI when they could not verify parentage with documents or interviews," and "Since May 6, 2019, CBP officials referred suspected parentage-related fraud concerns to ICE HSI to investigate for criminal prosecution. Following referral, during its criminal investigation, HSI could request a consensually administered Rapid DNA test."

Assessment (Confidence: High; almost certainly true): the denominator of the 8.5 percent figure is adults already selected by officers as suspected of parentage fraud, who then consented to a test. It is a confirmation rate inside a pre-screened suspect pool, not a prevalence rate among border arrivals, and it cannot be used as one. Read correctly it says something close to the opposite of its popular reading: of the family units officers suspected strongly enough to refer for criminal investigation, more than nine in ten turned out to be genuine.

This is the same base-rate structure as §3.1 and 5.4, appearing a third time in a third institutional setting. See §9, Signal 4.

7Regional focus: southeastern Arizona

Pima, Santa Cruz and Cochise counties, with Tucson and Nogales. This is the section that makes the document usable rather than generic, and its central finding is a negative one that is checkable in three independent federal and county datasets.

7.1 Jurisdictional overview: what Arizona law provides

Arizona criminalizes trafficking under three sections of Title 13, all fetched from the legislature's own site:

"Traffic" is defined across 13-1307 and 13-1308 as to "entice, recruit, harbor, provide, transport or otherwise obtain another person by deception, coercion or force." The forced-labor definition in 13-1308 reaches conduct short of violence, including "knowingly destroying, concealing, removing, confiscating, possessing or withholding" a person's identification documents or personal property, abuse or threatened abuse of the legal system, extortion, threats of financial harm, and controlling access to a controlled substance. It expressly excludes ordinary household chores and reasonable parental discipline.

Two provisions bear directly on identification rather than on prosecution:

A.R.S. 13-3620, mandatory reporting, is the strongest identification mandate in Arizona law. Its reporter list includes physicians, nurses, behavioral health professionals, school personnel, peace officers, clergy and "any other person who has responsibility for the care or treatment of the minor," and its definition of a reportable offense includes child sex trafficking under 13-3212. Failure to report is a class 1 misdemeanor, or a class 6 felony where the failure involves a reportable offense.

A.R.S. 41-1736, the anti-human trafficking grant fund, requires eligible programs to "Provide services to victims and training to law enforcement agencies, prosecutorial agencies and the public on preventing and identifying human trafficking." This is the only Arizona statutory language located that uses the word "identifying."

Nothing in Arizona law located in this pass requires hotels, businesses or transport hubs to post a trafficking hotline notice, in contrast to several other states. This document states that as the result of a search rather than as a certainty, and §13 records it as an open item.

Note also the source limitation on this subsection: azleg.gov pages were reached and read, but the fetch returned structured summaries with selected verbatim quotes rather than complete section text. The quoted fragments above are verbatim; the full statutory text was not captured, and no claim here depends on unquoted portions.

7.2 What the District of Arizona actually charges (PRIMARY: U.S. Sentencing Commission)

Source: U.S. Sentencing Commission, Statistical Information Packet, Fiscal Year 2023, District of Arizona, Table 1, drawn from the USSCFY23 datafile.

Immigration offenses are 30.0 percent of the national federal sentencing docket and 76.6 percent of Arizona's. Sexual Abuse is 0.7 percent of Arizona's, and Commercialized Vice is zero.

A necessary caveat on these categories: the Sentencing Commission's type-of-crime groupings are guideline-based and do not isolate any trafficking statute. Per Appendix A of the Commission's Sourcebook, Sexual Abuse comprises persons sentenced under guideline 2G1.1 who received a base offense level of 34, plus all of 2A3.1 to 2A3.4, 2G1.3, 2G2.1, 2G2.3 and 2G2.6, so it mixes coercive adult trafficking and minor trafficking with ordinary federal sexual abuse and child pornography production. Commercialized Vice is 2G1.1 where the court did not apply a base offense level of 34, plus gambling and animal fighting. Forced labor under 18 U.S.C. 1589 sits in guideline 2H4.1 and is counted under Individual Rights, which Arizona records at 2 in FY2023. Correction (multi-model verification pass): an earlier draft described the mapping loosely and omitted the Individual Rights category entirely, understating where forced labor is counted. A precise annual count of federal trafficking sentences in this district cannot be derived from these packets. What can be derived is a bound: summing Sexual Abuse, Commercialized Vice and Individual Rights gives 37 of 4,812, or 0.77 percent of the district's docket, against immigration at more than three quarters.

Assessment (Confidence: High; almost certainly true): the federal criminal justice system in this corridor overwhelmingly processes immigration offenses. This does not establish anything about how much trafficking occurs here, and it should not be read as though it did. It establishes what the district's enforcement and charging capacity is actually directed at, which is the environment any identification effort in this corridor operates inside.

This is the same finding, reached by a different route, as the border brief's observation that coercive conduct in this corridor is frequently charged as smuggling rather than trafficking.

7.3 Tattoos are absent from the entire District of Arizona trafficking record

A full-text search of the United States Attorney's Office for the District of Arizona press-release corpus was run in this session against the office's own news index. A search for "tattoo" returns one result across the entire corpus, and it is not a trafficking case: it is a September 23, 2025 release concerning a Hobbs Act robbery and firearm discharge sentence.

The parallel research pass, searching the same corpus, reported the same result and extended it: "tattoos" returns the same single item, "branded" and "branding" return three results all concerning commercial-brand fraud, and no trafficking release in 2015-2025 cites a tattoo, brand or physical mark as evidence of trafficker control. That pass also reported reading eight District of Arizona trafficking prosecutions in full, four of them Tucson Division cases, and finding no mention of tattoos or branding in any.

Assessment (Confidence: Moderate; likely): in the public federal prosecution record for this district across the eleven-year window, no trafficking case is described as involving trafficker marking. The confidence is moderate rather than high for a specific reason: press releases are summaries written for the public, not indictments, and the absence of a detail from a release is weak evidence of its absence from a case. §4.6 shows that marking does appear in releases from other districts when prosecutors choose to describe it, which makes the Arizona silence more meaningful than it would otherwise be, but it remains an argument from silence and is labeled as one.

A separate check reinforces the point at national level: the 99-page NIJ-commissioned evaluation of the federal task force program Arizona's own task force belonged to contains no occurrence of "tattoo" or "brand" at all.

7.4 How victims in this corridor were actually identified (PRIMARY: USAO District of Arizona)

The same eight prosecutions describe how each case began. The pattern is consistent and it is not indicator screening.

• United States v. Alexander (CR-21-02972-JAS-EJM, Tucson Division, sentenced August 19, 2024, 40 years, Judge James A. Soto, sex trafficking of a minor by force, fraud or coercion plus conspiracy, transportation and production charges): the victim placed her own 911 call, and a Tucson Police Department detective responded to a hospital.
• United States v. Rideaux (CR-23-01291-PHX-DJH, sentenced February 27, 2025, 132 months): the victim "convinced a sex buyer to take her to a hospital where she could notify police."
• United States v. Williams (CR-22-00687-PHX-DJH-1, sentenced July 21, 2022, 120 months): a police human trafficking unit found the 17-year-old, apparently physically assaulted, at a hospital.
• United States v. Jackson (CR-16-01704-TUC-RM, Tucson Division, sentenced April 23, 2018, 15 years, Judge Rosemary Marquez): the victim escaped by contacting a victim rescue service "which she had seen advertised on a billboard," from a motel room in Tucson. This release names the Southern Arizona Anti-Trafficking Unified Response Network by name.
• United States v. Terry (16-CR-1350-TUC-CJK-1, Tucson Division, sentenced February 26, 2018, 240 months, Judge Jorgenson): a second victim "was discovered with Terry during a routine traffic stop by the Arizona Department of Public Safety near Benson, Ariz." Benson is in Cochise County, and this is the one incidental-detection case located in the corridor.

Assessment (Confidence: Moderate; likely): in the located District of Arizona record, identification ran through victim self-extraction and hospital presentation, with one incidental traffic stop, rather than through proactive indicator screening or the recognition of physical marks. Three of five located mechanisms involve a hospital, which corroborates the healthcare-contact premise in §5.7 from an entirely different direction, a federal prosecution record rather than a survivor survey. This rests on a small number of cases described in press releases and is rated accordingly.

7.5 The corridor's multidisciplinary task force ran three and a half years and was not renewed

Source: USAspending.gov, the federal government's official award reporting system, queried through its public API.

Both award descriptions read: "ENHANCED COLLABORATIVE MODEL TO COMBAT HUMAN TRAFFICKING LOCATED WITHIN THE SOUTHERN ARIZONA COUNTIES OF PIMA, SANTA CRUZ AND COCHISE." Together they funded the Southern Arizona Anti-Trafficking Unified Response Network, SAATURN. A targeted query for Enhanced Collaborative Model awards with Arizona place of performance returns these two and nothing else; the next such award in the state is 2019VTBXK017, City of Phoenix, $900,000, October 2019 to September 2022.

SAATURN's own description of itself, from a University of Arizona Southwest Institute for Research on Women newsletter of September/October 2016: "a collaboration of the Tucson Police Department, Homeland Security Investigations, and the U.S. Attorneys Office working together with CODAC to: 1) Provide comprehensive services to trafficking victims by identifying and addressing their needs for safety, security, and healing. 2) Proactively investigate, identify, apprehend, and prosecute those engaged in human trafficking." Its victim services arm covered CODAC in Pima County and the Family Advocacy Center in Sierra Vista, Cochise County. Its outreach targeted hotels, schools, taxi services and healthcare providers. The newsletter contains no identification-indicator detail, no reference to tattoos or branding, and no statistics.

Post-2019 federal trafficking funding located in these three counties is victim services only: the International Rescue Committee Tucson (15POVC24GG01816HT, $949,986, October 2024 to September 2027, covering "PIMA, COCHISE, AND SANTA CRUZ COUNTIES," and 15POVC21GG04233HT, $349,585) and Our Family Services Tucson (15POVC21GG03963HT and 15POVC24GG00886MINO, $600,000 each).

Assessment (Confidence: Moderate; likely): the southeastern Arizona corridor has had no federally funded multidisciplinary law-enforcement trafficking task force since March 2019, and federal trafficking money in the three counties since then has funded services rather than investigation. This rests on the federal award record, which is authoritative for what was funded federally and silent on what may be funded by the state, the counties or the cities. §13 records the state and local funding picture as an open item, and the parallel research pass on Arizona state bodies did not complete.

Note a discrepancy worth carrying: contemporaneous press and the university newsletter describe SAATURN as a three-year, $1.5 million project. The federal award record shows $1,265,813 obligated across the two awards. This document uses the award record.

For scale, the largest federal awards described as "trafficking" in these same three counties over the same period are High Intensity Drug Trafficking Areas awards to Tucson Police, at $9.33 million in 2023, $9.29 million in 2024, $9.10 million in 2021 and $8.99 million in 2022. That is a different program addressing a different subject, and the comparison is offered only to show the order-of-magnitude difference in what the word "trafficking" funds in this corridor.

7.5b What the corridor's task force actually produced (PRIMARY: SIROW/University of Arizona final evaluation)

Source: Stevens S and Black C, "SAATURN: Final Evaluation Report, October 1, 2015 - March 30, 2019," Southwest Institute for Research on Women, University of Arizona, April 2019. This is the grantee's contracted evaluator rather than an independent audit, and it is attributed to SIROW rather than to the Tucson Police Department accordingly. It is nonetheless the best quantified victim-identification dataset located for this corridor.

The law enforcement yield, verbatim: "In cases dated from February 2015 through May 2018, a total of 506 investigations were conducted by the Street Crimes Interdiction Unit (SCIU) of Tucson Police Department (TPD) and related law enforcement agencies (FBI, HSI). Of the 506 investigations, 86 investigations (17%) identified at least one potential victim of human trafficking. A total of 110 potential victims were identified in these 86 cases. Twenty-one investigations (4%) identified at least one confirmed victim of human trafficking; a total of 26 confirmed victims were identified in these cases."

Of the 506, 124 (25 percent) were classified as trafficking cases: 83 percent sex, 10 percent sex and labor, and under 1 percent labor alone. Nine of the eleven sex-and-labor investigations involved foreign nationals.

The stated obstacle, verbatim: "Barriers to victim identification included non-cooperation with law enforcement, including denial of engaging in commercial sex activities and denying involvement with traffickers (pimps)."

On the services side, over the grant period "services were provided to 102 trafficking victims (85 sex, 11 sex and labor, 5 labor, and 1 unknown trafficking victims). The majority were women (n=96), U.S. citizens (n=92), exploited through prostitution (n=56), and identified in a hotel/motel (n=44)."

Assessment (Confidence: Moderate; likely): in a federally funded, multi-agency task force operating for three and a half years on the border, 92 of 102 victims served were United States citizens, and the dominant identification setting was a hotel or motel. The corridor's trafficking task force was overwhelmingly identifying domestic victims, not people moved across the border. This is the grantee's evaluator reporting on its own program, which is why it is rated moderate, but the citizenship figure is a simple count and is not the sort of number a self-evaluation has an incentive to shade.

The most important passage in the evaluation is about the reference standard, and it echoes §5.3 exactly. Verbatim:

"no further guidelines were provided by DOJ, BJA, or OVC to standardize what constituted sufficient evidence for law enforcement to 'confirm' the status of a victim. TPD contacted BJA to obtain clarification on this point, but received no response. In considering the totality of the circumstances, TPD elected to record performance data following the pre-January 2017 definition to reduce the subjectivity involved in documenting case characteristics. As a result of this approach to data collection, the total number of victims identified is anticipated to be conservative and likely underestimating the actual number of trafficking victims."

Assessment (Confidence: High; almost certainly true): the absence of a gold standard for confirming victim status, which Hainaut et al. identify as the structural obstacle to validating clinical screening instruments (§5.3), appears here in an operational law enforcement setting as an unanswered request for guidance from the funding agency. Two entirely unrelated bodies of work, a public health scoping review and a university evaluation of a police task force, independently identify the same missing definition. See §5.3, where the same absence blocks the validation of every clinical instrument in this document.

And the search result that matters for this document's subject: the word "tattoo" appears zero times in the 43-page evaluation, across 506 investigations, 110 potential victims, 26 confirmed victims and 102 victims served.

7.6 Pima County's medical examiner: where physical marks actually sit (PRIMARY: PCOME 2025 Annual Report)

This is the most directly relevant regional dataset in the document, because the Pima County Office of the Medical Examiner is one of the few institutions anywhere that records, year by year, which identification method actually produced each identification.

The office's own summary, from its 2025 Annual Report, page 34: "Since 2000, 31% of identifications were made through fingerprint analysis, 28% by visual identification, 24% through DNA analysis, 15% through circumstantial methods (i.e., scars/marks/tattoos, personal effects, etc.), 1% through dental radiographic comparison and <1% through medical radiographic comparison."

The parenthetical is the finding. Tattoos are not a category in this system. They are one component of "circumstantial," which is the office's formally lower tier of identification. The distinction is longstanding and documented: a Department of Justice abstract of Anderson's 2008 study of PCOME methods records that "The Office of the Medical Examiner uses two levels of identification, 'positive' and 'circumstantial'," and that for a circumstantial identification "all consistencies between the decedent and the presumptive person are noted, and no unexplained inconsistencies can exist."

The year-by-year table for the document's date range, transcribed from the 2025 report and checked for internal consistency (every row's method counts sum exactly to its identified count, and the identified and unidentified columns sum to the total):

Over the eleven years: 56.4 percent of recovered undocumented border crosser remains were identified. Of the 963 identifications, fingerprints produced 512 (53.2 percent), DNA 276 (28.7 percent), visual identification 87 (9.0 percent), circumstantial methods 71 (7.4 percent), dental 11 (1.1 percent) and radiographic 6 (0.6 percent). Circumstantial identifications, the bucket that contains tattoos alongside scars, marks and personal effects, account for 4.2 percent of all recoveries.

The category is also shrinking. Against the office's own since-2000 figures, quoted above, circumstantial methods fall from 15 percent of identifications across the full twenty-five-year period to 7.4 percent across this document's eleven-year window, while fingerprints rise from 31 percent to 53.2 percent. Assessment (Confidence: Moderate; likely): physical marks are contributing a declining share of identifications in this corridor over time, and the mechanism is stated in the next paragraph rather than inferred. This is a comparison between a twenty-five-year average and an eleven-year subset of it, which is a weaker comparison than two independent periods, and it is rated accordingly.

Two methodological cautions on this table, both of which matter for anyone reusing it. First, one cell was blank in the 2025 report's own PDF text layer: the 2017 circumstantial figure. It was resolved as zero by two independent checks, a coordinate-level extraction showing the missing cell sits in the circumstantial column, and the 2024 annual report which prints 2017 circumstantial explicitly as 0. Second, PCOME revises each calendar year in every subsequent report as long-tail identifications come in; the 2024 report's figures for 2015 through 2024 differ slightly from the 2025 report's. This document uses the 2025 report throughout, and any citation should name the report year, not just the calendar year.

Assessment (Confidence: High; almost certainly true): in the county that handles more unidentified border-crosser remains than any other in the United States, over eleven years, physical marks including tattoos are one part of a category that produced 7.4 percent of identifications, and the office classifies that category as circumstantial rather than positive. Fingerprints and DNA together produced 788 of 963, or 81.8 percent. This is the individuation case from §6.7, measured operationally over more than a thousand cases, and it places tattoos exactly where INTERPOL's standard places them: a secondary identifier, useful, corroborative, and not proof.

Why the share is this low here specifically. The remains arrive skeletonized. The 2025 report records that "In CY2025, 17% of recovered remains had an estimated PMI [postmortem interval] < 3 months and 83% had an estimated PMI >= 3 months." §6.5's decomposition finding, that tattoos became unusable as secondary identifiers between day 294 and day 333 in the one study that measured it, describes the physical mechanism. Soft-tissue identifiers are simply not available in most of these cases.

A wider frame from the same report: "Of the 4,104 decedents recovered since CY2000, 65% have been identified. As of February 11, 2026, 1,446 decedents remain unidentified (35%)." The Anderson abstract adds the comparison that gives that number its weight: an undocumented border crosser "is only identified 73 percent of the time. This compares with close to 100 percent of identifications for deceased American citizens."

VERIFICATION FLAG: tattoos cannot be separated from the rest of the circumstantial bucket in any published PCOME source. All eleven annual reports covering 2015 to 2025 were obtained and searched in this session's research passes, and no report disaggregates the category. The underlying peer-reviewed literature does not fill the gap either: Anderson and Spradley 2016, the flagship published description of the PCOME identification workflow, was fetched in full and contains no identification-rate figure and no occurrence of the words tattoo, fingerprint, dental or circumstantial. A disaggregated figure would require a records request to the office. §13 records this as an open item, and it may simply not exist in published form.

7.7 The Super Bowl claim, tested in Arizona

Arizona hosted Super Bowl XLIX in Glendale in February 2015 and Super Bowl LVII in Glendale in February 2023, both inside this document's window. The claim that such events produce a trafficking surge is among the most widely repeated in this subject, and it has been tested here specifically.

The Arizona study. Roe-Sepowitz D and Gallagher J, "Exploring the Impact of the Super Bowl on Sex Trafficking," February 2015. The authors are an Associate Professor at Arizona State University who directs its Office of Sex Trafficking Intervention Research, and a serving Commander of the Phoenix Police Department. Its finding, verbatim:

"While there is no empirical evidence that the Super Bowl causes an increase in sex trafficking compared to other days and events throughout the year, there was a noticeable increase in those activities intended to locate victims from both law enforcement and service provision organizations."

Its own data cut against a host-city effect. Comparing the 2014 and 2015 events, the daily volume of commercial sex advertisements grew 30.3 percent in Phoenix, which hosted in 2015, and 57.6 percent in Northern New Jersey, which had hosted in 2014 and did not host in 2015.

Disclosure, because it matters in both directions: the study was funded by a grant from the McCain Institute at ASU, an organization this document would otherwise exclude, and the report carries an endorsement from the co-chair of the Arizona Human Trafficking Council. The finding runs against the framing of its own funder and endorser, which is a reason to weight it more heavily rather than less. It is used here for its negative finding only.

The enforcement numbers, and how they are misread. The FBI Phoenix Field Office announced on February 18, 2015 that "360 customers of commercial sex were arrested, 68 traffickers were arrested, and 30 juvenile victims were recovered." The same release states the operational period: "Beginning in June 2014, agencies conducted monthly operations and engaged in three weeks of continuing operations in January 2015."

Assessment (Confidence: High; almost certainly true): those figures describe roughly eight months of metropolitan Phoenix operations from June 2014 through February 2015, not the game, game day or game week, and the release says so in its own text. Quoting them as Super Bowl figures, which is common, overstates by an order of magnitude in duration. The comparable 2023 operation around Super Bowl LVII, reported consistently across three named local outlets, produced 348 arrests including approximately 120 sex buyers, and recovered five child victims and one adult victim, across an operational window of January 30 to February 11, 2023 that also encompassed the WM Phoenix Open and Barrett-Jackson.

Independent corroboration. The FBI's own current Trafficking Indicators page opens with the heading "Don't Wait Until a Special Event Begins to Look for Human Trafficking," and redirects attention to labor trafficking "within the industries (food, entertainment, transportation, infrastructure) involved in the preparations for hosting a special event, such as the Superbowl or a FIFA World Cup match."

A peer-reviewed treatment reaches the same conclusion from media analysis: Martin L and Hill A, "Debunking the Myth of 'Super Bowl Sex Trafficking': Media hype or evidenced-based coverage," Anti-Trafficking Review, issue 13, 2019, pages 13-29, which states that "Available empirical evidence does not suggest that major sporting events cause trafficking for sexual exploitation" while finding that "76 per cent of US print media from 2010 to 2016 propagated the 'Super Bowl sex trafficking' narrative." The publisher caveat on this source is recorded in §8.

And a state health department reached it prospectively for a different host city: Minnesota Department of Health, "Evaluation of Minnesota's Response to Sex Trafficking During Super Bowl LII," January 14, 2019, finding "there was no increase or decrease in sex trafficking incidence during Super Bowl LII," and explaining the appearance of one: "The numbers increased because the resources increased, not because of increase in demand due to the Super Bowl."

Assessment (Confidence: High; almost certainly true): four independently produced sources, an ASU study co-authored by a serving Phoenix Police commander, the FBI's own current guidance, a peer-reviewed media analysis by two university faculty, and a state health department evaluation, converge on the finding that large sporting events do not cause measurable increases in trafficking, and that observed increases in identified cases track increases in enforcement and service capacity. See §9, Signal 3. This is a finding about detection, not about trafficking: it is the clearest available demonstration that identification counts measure looking rather than occurrence, which is the same lesson §2.1 draws from the T visa series.

7.8 Arizona's own trafficking counts, verified against the state's live system

Two Arizona sources record trafficking offenses: the Department of Public Safety's Crime in Arizona annual reports through 2020, and the DPS TOPS public crime statistics system, which carries agency-level NIBRS data from 2017 forward. The TOPS figures below were retrieved in this session directly from the system's own year-addressable data endpoint rather than through its dashboard, and are reproduced with a deliberate control.

Agency-level actual offenses, southeastern Arizona. The two right-hand columns are the control: they show each agency was actively reporting crime to the state in every year in which it records a trafficking zero.

Two readings are available for a zero in a crime statistics table, and the control columns separate them for most cells. Nogales Police Department reported between 30 and 49 aggravated assaults and between 70 and 183 simple assaults in every one of these years. It was submitting data throughout, so its trafficking zeros are reported zeros rather than missing data.

Correction (multi-model verification pass): the control does not hold for one cell, and an earlier draft claimed it held for all of them. Santa Cruz County Sheriff's Office recorded 1 aggravated assault and 0 simple assaults in 2021, which is not a normal reporting year for an agency that recorded 16 to 25 aggravated assaults in each of 2022 through 2025. That single agency-year should be treated as probably incomplete rather than as a reported zero. The control holds for Nogales Police in all five years and for Santa Cruz County Sheriff in 2022 through 2025.

Assessment (Confidence: High; almost certainly true): the two Arizona counties on the Nogales and Douglas border corridor recorded no human trafficking offenses at all in their own state's crime reporting system across the entire focal period, while reporting normally on every other offense category. This is a fact about what was recorded. It is compatible with two very different underlying realities, and §11.4 does not resolve which.

The Tucson series is the corridor's clearest demonstration that recorded trafficking tracks capacity. Tucson Police recorded zero trafficking offenses in 2021, 2022 and 2023, then 22 in 2024. In state fiscal year 2024 the department received a $500,000 award from the Arizona DPS Anti-Human Trafficking Grant Fund. Cochise County Sheriff's Office, which received $500,000 from the same fund in the same year, recorded its only non-trivial counts in 2023 and 2024 and returned to zero in 2025.

The 2025 Tucson figure of 3 should not be read as a decline. The department's aggravated assault count for 2025 is 35 percent below its 2024 count and its simple assault count is 50 percent below, which indicates an incomplete reporting year rather than a fall in crime.

Assessment (Confidence: Moderate; likely): recorded trafficking offenses in this corridor track grant-funded detection capacity rather than any plausible underlying change in the phenomenon. A department that recorded zero offenses for three consecutive years recorded 22 in the year it was funded to look. This is a correlation across two agencies and one funding cycle rather than a controlled comparison, and no located source states the causal link, which is why it is rated moderate. It is the same pattern §7.7 documents around sporting events and §2.1 documents in the T visa series, appearing a third time at agency level.

The longer county series says the same thing with a caveat. DPS Crime in Arizona reports for 2015 through 2020 record a statewide total moving 8, 19, 99, 58, 53 and 26 across those six years, with Cochise and Santa Cruz counties at zero in all six and Pima County recording two offenses in total. In each year the county figures sum exactly to the independently published statewide total, which indicates the blanks are zeros rather than suppressed cells. That series carries its own limitation: the same reports name Cochise County Sheriff's Office, Pima County Sheriff's Office and several corridor municipal agencies in their annual lists of departments that did not provide complete data, so the pre-2021 corridor zeros are partly a non-reporting artifact. The 2021-2025 data above, with its activity control, is the sounder series and is the one this document relies on.

A third, independent pipeline agrees. The Arizona Human Trafficking Tip Line, a state hotline operated with Arizona State University, reported 12,036 total calls and 1,024 unique callers over roughly three years. Its county breakdown records Maricopa 844 and Pima 69, against Cochise 1 and Santa Cruz 0. Over the life of the state's own trafficking hotline, the two border-corridor counties produced one caller between them.

The federal series behaves the same way, and its custodian says so. FBI Uniform Crime Reporting human trafficking offense counts for Arizona were 6 in 2015, 99 in 2017, 58 in 2018 and 53 in 2019. The FBI's caveat, from its Human Trafficking 2018 publication: the data "reflect the offenses and arrests recorded by state and local law enforcement agencies (LEAs) that currently have the ability to report the data to the national UCR Program. As such, they should not be interpreted as a definitive statement of the level or characteristics of human trafficking as a whole." Note separately that involuntary servitude, the labor trafficking category, never exceeds seven offenses statewide in any reported year in either the state or the federal series.

Counts for 2016 and for 2020 onward were not obtained from the federal series; the 2016 and 2020 table URLs returned errors and the post-2020 data requires an API key this pass did not have. The state series above covers that period instead.

7.9 Arizona mandates no trafficking identification training for police

Arizona's peace officer training requirements are set by statute and by administrative rule, and neither requires trafficking training.

A.R.S. 41-1822(A)(4) directs the Arizona Peace Officer Standards and Training Board to "Prescribe minimum courses of training," and then enumerates what that training "shall include": courses in responding to and reporting criminal offenses motivated by race, color, religion, national origin, sexual orientation, gender or disability, and training on the nature of unexplained infant death. Human trafficking is not among the statutorily required subjects.

Arizona Administrative Code Title 13, Chapter 4, the AZPOST rules, contains no occurrence of the words "trafficking" or "exploitation" anywhere in the chapter. Its minimum course requirements govern instructor qualifications and curriculum development standards rather than subject matter.

This sits against the Arizona Human Trafficking Council's own 2019 annual report, which states that a 2015 AZPOST initiative "fulfilled a major recommendation of the Governor's Task Force on Human Trafficking which mandated sex trafficking identification training as a basic requirement for all new Arizona law enforcement recruits."

Assessment (Confidence: Moderate; likely): the mandate described by the Council is a Board curriculum decision implementing a task force recommendation, not a statutory or rule-level requirement, and therefore carries no minimum hours, no enforcement mechanism and no audit trail. A state-published report authored by academic researchers for a state Medicaid contractor states the position more bluntly: "Currently, Arizona does not mandate any sex trafficking awareness training for any professionals including school personnel, medical providers, or law enforcement." Rated moderate rather than high because the AZPOST curriculum documents themselves could not be retrieved (§7.11), so this rests on the statute, the rules and two characterisations rather than on the curriculum.

The Arizona Attorney General's office separately offers an AZPOST-credited course, "Patrol Recognition and Response to Sex Trafficking," and lists Tucson Police Department among its completed partners. That is elective continuing-education credit, not a requirement, and the distinction matters directly for KAC-1.

Assessment (Confidence: Moderate; likely): the state whose Attorney General publishes an uncaveated tattoo indicator (§5.1) requires no officer to be trained in trafficking identification at all. The indicator list and the training obligation point in opposite directions, and the space between them is where an untrained observer applies an unvalidated indicator.

Nor is there a local policy filling the gap. Tucson Police Department General Order 2600, "Investigative Protocols," carries named sections for hate crimes, child sexual abuse, adult sexual assault, elder abuse, kidnapping, gangs and narcotics, and no human trafficking section at all; the single located mention across the department's General Orders is a bare list entry, "Human Trafficking," under Special Investigations, with no duties or protocol attached. Searches of the codified Tucson City Code and Pima County Code return no human trafficking ordinance in either jurisdiction. This document reports those as the results of searches conducted in this session's research passes rather than as certainties, and §7.11 records what could not be reached.

7.10 Labor trafficking is nearly invisible in this corridor's record

Searches of the District of Arizona press-release corpus, re-run on 2026-09-09, returned zero results for "involuntary servitude," one result for "forced labor" which was a national policy announcement rather than an Arizona prosecution, and four results for "labor trafficking," none of them a prosecution: a departure announcement, a missing and murdered Indigenous persons awareness item, an Attorney General strategy announcement and a funding availability notice. No District of Arizona labor trafficking prosecution has been announced by press release in the window. A search for trafficking prosecutions arising from Nogales and the Santa Cruz County corridor returned 59 results, all narcotics, firearms and smuggling, and no human trafficking prosecutions.

This sits against §2.3, where the federally identified victim population is 77 percent labor trafficking by the USCIS Supplement B measure and 310 of 382 forced labor by the Continued Presence measure.

The national explanation for that gap comes from the federal evaluation of the very program SAATURN belonged to. Across ten Enhanced Collaborative Model task forces from October 2015 to December 2019, drawn from the Bureau of Justice Assistance performance measurement system: of 3,400 law enforcement investigations, 3,248 (95.5 percent) were sex trafficking, 115 (3.4 percent) labor trafficking and 37 (1.1 percent) both. Prosecutions were 99 percent sex trafficking. Yet survivor services data from the same task forces over the same period recorded 1,707 survivors served, of whom 344 (20.2 percent) were labor trafficking victims and 177 (10.4 percent) both, with one task force reporting 47 percent labor.

The evaluation's own reading of the divergence: it "could indicate: that in some task forces, law enforcement is focused more keenly on sex trafficking and targeting its investigative resources toward those cases; that labor trafficking is much harder to identify and uncover in the community; and/or that law enforcement lacks the proper infrastructure, expertise, or training to fully investigate labor trafficking."

Source disclosure: this evaluation (NCJ 300863, May 2021, NIJ award 2017-VF-GX-0004) was commissioned by the National Institute of Justice and distributed through NCJRS, and was authored by the Urban Institute with NORC. The Urban Institute is a named think tank and is a standing exclusion under this document's sourcing standard, as recorded in §8. The document itself carries the notice "This resource has not been published by the U.S. Department of Justice." It is reported here with that disclosure rather than dropped, because it is the only federal evaluation of the program the southeastern Arizona task force belonged to, and because its underlying data are BJA's own performance measurement records. No conclusion in this document rests on it alone.

Assessment (Confidence: Moderate; likely): within the same federal task forces, law enforcement investigated sex trafficking roughly 28 times as often as labor trafficking, while victim service providers in those same task forces served labor trafficking victims in about three in ten cases. The identification apparatus and the service apparatus are looking at measurably different populations. Rated moderate because the load-bearing source carries a disclosed sourcing exception, though its direction is corroborated by the two permitted federal datasets in §2.3.

7.11 Material located and excluded from this section as not load-bearing

Three categories of Arizona material were found and set aside rather than used.

State agency material that could not be reached. The Arizona Peace Officer Standards and Training Board site (post.az.gov) and the Arizona Department of Emergency and Military Affairs site (dema.az.gov) are both behind bot-challenge protection and returned HTTP 403 to every attempted method across three attempts each. No attempt was made to circumvent the challenge. As a result, Arizona's peace officer trafficking training requirement is unverified, and a search snippet suggesting the basic curriculum covers trafficking under A.R.S. 41-1822 and 41-1828.01 is recorded here as unverified and is not used. §13 carries this forward.

Award and library pages that were formerly public. Office of Justice Programs award records (ovc.ojp.gov) and Bureau of Justice Assistance library pages now serve an SSO login gate to both automated and browser access. The SAATURN award data in 7.5 comes from USAspending.gov instead, which is the authoritative federal award system and a fully adequate substitute.

Local agency structure that could not be confirmed. A search result indicates the Tucson Police Department's current trafficking work sits in a "H.E.a.T." unit within its Special Investigations Section. This could not be confirmed: tucsonaz.gov returned 403, the department's organizational chart PDF failed, and the divisions page that was successfully fetched does not mention a trafficking unit. It is recorded as unverified and is not asserted.

Too dated to be load-bearing. Anderson's 2008 Journal of Forensic Sciences study is the one publication promising a full PCOME identification-modality breakdown. Its full text was not obtained (publisher returned HTTP 403 across two attempts; the paper is not open access). Its DOJ abstract was obtained and is used in 7.6 only for the positive-versus-circumstantial definition, which is a definitional matter unaffected by age. Its 2001-2006 figures are outside this document's window and are not used.

A parallel research lane that did not complete. Coverage of the Governor's Office of Youth, Faith and Family, the Arizona Human Trafficking Council, Arizona DPS "Crime in Arizona," the Arizona Attorney General, the Pima County Attorney, and the Cochise and Santa Cruz county sheriffs was assigned to a research lane that was still running when this document was compiled. Nothing in §7 depends on it. It is the principal known coverage gap in this section and is recorded in §13.

8Sources rejected this session, and why

Notes on sources whose status changed or is contested

The Urban Institute report, and why it is excluded despite being an HHS deliverable. Dank M et al., "Pretesting a Human Trafficking Screening Tool in the Child Welfare and Runaway and Homeless Youth Systems," April 2017, is an Urban Institute research report contracted by HHS and hosted on aspe.hhs.gov. Two of its authoring institutions are named think tanks, which is a standing exclusion. It is not used as load-bearing here. Two facts about it are worth recording anyway, because both are cautions rather than claims: the study calls itself a pretest and recommends "further validation work with a nationally representative sample of youth," yet it is frequently listed in tool inventories as a validated instrument, which it does not claim to be. It also carries a useful historical observation, that a 2012 review of 20 screening documents found "of the 20 tools, only one was based on actual data on victims."

Pew Research Center, admitted. §3 relies on Pew for the general-population tattoo base rate. Pew is a survey research organization that publishes full questionnaires, toplines and methodology and takes no position on this subject, which places it under "a research institution with transparent methodology" in the §1.1 grading scale rather than under the think-tank exclusion. It is graded B2 and used for a demographic base rate only. The judgment is recorded here so a reader can disagree with it visibly.

Vera Institute of Justice, admitted with a label. Vera is a research institution with published methodology, not an advocacy organization, and its NIJ final report (NCJ 246712) is a federally funded grant deliverable. It is used in §5.3 primarily for its own candid statements about its own limitations. Note that the report's cover states "This report has not been published by the U.S. Department of Justice," so it is a grantee deliverable posted by NCJRS, not a DOJ-published or peer-reviewed document, and it is graded accordingly.

Anti-Trafficking Review, admitted with a label. The Super Bowl finding in §7 rests on Martin L and Hill A, Anti-Trafficking Review, issue 13, 2019. The journal is peer-reviewed and Scopus-indexed, the authors are university faculty at the University of Minnesota and the University of Texas at Austin, and the article is open access under CC-BY. Its publisher, the Global Alliance Against Traffic in Women, is a mission-driven alliance, which under a strict reading of the standing exclusion would disqualify it. It is admitted for three reasons, stated so the call is visible: the peer-reviewed article rather than the publisher is what is cited; the finding runs against the institutional interest of an anti-trafficking alliance, which is a reason to weight a finding more rather than less; and the same conclusion is reached independently by the FBI, which is a permitted primary source and is cited alongside it in §7.

The NIST Tatt-C ethics controversy, excluded in full. A widely circulated account holds that the FBI-supplied images used in NIST's 2014-2015 Tatt-C evaluation came from prisoners without consent, that de-identification claims were inaccurate, and that NIST's human-subjects review came after the fact. Every substantive allegation in that account traces to the Electronic Frontier Foundation, an advocacy organization and the narrative's originator, or to trade press repeating it. No permitted primary source for any of it was located, and it is therefore not asserted anywhere in this document.

What can be stated from the primary documents alone, because it was checked directly in both: NISTIR 8078 (Tatt-C, 2016) contains no institutional review board section and no occurrence of "human subjects," "45 CFR 46" or "15 CFR 27". NISTIR 8232 (Tatt-E, 2018) does contain a dedicated Institutional Review Board section, recording that "The National Institute of Standards and Technology Human Subjects Protection Office reviewed the protocol for this project and determined it is not human subjects research as defined in Department of Commerce Regulations, 15 CFR 27." The appearance of that section in the second report and its absence from the first is a fact about two public documents. This document reports that fact and draws no inference from it about what prompted the change. A permitted route to the underlying question exists and was not taken in this pass: EFF's FOIA litigation against the FBI and NIST would be reachable as court records, and §13 records it as an open item.

A DOJ sourcing finding, reported rather than rejected. The Office for Victims of Crime page on human trafficking publishes no federal indicator list of its own and directs readers to "learn how to recognize the signs of human trafficking" at humantraffickinghotline.org, which is operated by Polaris Project. This document therefore uses the FBI's indicator page for Department of Justice indicators, as recorded in §5.1. The observation is recorded here because it is a finding about where federal indicator guidance actually comes from, not merely a citation problem: a reader who follows the government's own referral chain arrives at an advocacy organization.

A note on the whole instrument landscape. Macy et al. 2023, a permitted peer-reviewed scoping review, reports that across 22 screening tools "most tools were developed by practice-based and non-governmental organizations located in the U.S." Under this document's sourcing standard, most of the field's actual instruments are therefore advocacy products. This is recorded as a structural fact about the subject rather than treated as a reason to abandon the standard. Where an instrument's validation exists only in an excluded source, this document says the instrument is unvalidated as far as the public record shows, which is what §5.8 does for the CSE-IT and what the following paragraph does for CHTAT.

CHTAT, unsubstantiated. The Comprehensive Human Trafficking Assessment Tool is attributable to the National Human Trafficking Resource Center, operated by Polaris Project. No validation study for it was located in any permitted source. It is not listed among validated instruments anywhere in this document, and any inventory that lists it as validated should be treated as unsupported until a primary source appears.

9Signals: genuine cross-source correlation

The standard this document applies, stated in its own voice. A correlated signal requires two or more independently produced sources, from different institutions, using different methods, arriving at the same finding without one citing or reprinting the other. Multiple outlets covering the same press release, study or dataset is not correlation. It is one data point reported several times, and it counts as one. Where two bodies analyze the same underlying dataset, that is a real but weaker correlation than two independent bodies of evidence, and it is rated and explained as such rather than collapsed into the stronger category.

Signals found weak or absent are kept, because a tested hypothesis that failed is a finding.

Signal 1 (STRONG). Physical marks individuate known persons well and classify unknown persons badly, and the two tasks have been measured separately by unrelated institutions.

Assessment (Confidence: High; almost certainly true).

Four bodies with no methodological relationship converge:

• NIST, computer vision, FBI-supplied operational imagery. Matching a tattoo to the same tattoo on the same person: 72.1 percent rank-10 against a 100,000-image operational gallery. Matching visually similar tattoos across different subjects: 14.9 percent rank-10 against a gallery of 272 (§6.1).
• Pima County Office of the Medical Examiner, forensic pathology, 1,706 recovered remains over eleven years. Fingerprints and DNA produced 81.8 percent of 963 identifications; circumstantial methods, the bucket containing tattoos, produced 7.4 percent, and the office classifies that tier as something other than positive identification (§7.6).
• INTERPOL, international standards, disaster victim identification. Friction ridge, dental and DNA are primary; personal description and other features are secondary, "typically serve to support identification by primary means" (§6.4).
• The Ohio Department of Youth Services, a state screening instrument tested against outcomes on 2,010 assessments. Tattooing or branding as a classification indicator: significantly but weakly associated with victimisation (phi = .124), and an AUC of 0.533, below the study's own threshold for predictive, because the flag appears in 0.55 percent of assessments (§4.4).

Why this earns a strong rating. These are four different disciplines (computer vision, forensic pathology, international police standards, criminological instrument validation), four different institutions in two countries, four different methods, and four different questions asked for four different purposes. None cites another. They agree not on a number but on a structural distinction, and the distinction falls in the same place each time. This is the document's central finding.

Signal 2 (MODERATE). The population the United States formally identifies as trafficked is majority labor trafficking, and two federal systems built for different statutory purposes say so.

Assessment (Confidence: Moderate; likely).

• USCIS, adjudicating T nonimmigrant status: of T-1 nonimmigrants who filed the optional law enforcement declaration, 77 percent listed labor trafficking and 29 percent sex trafficking, some listing both (§2.3).
• DHS's Center for Countering Human Trafficking with Homeland Security Investigations, granting Continued Presence in FY2023: of 382 granted requests and extensions, 310 were related to forced labor, 67 to sex trafficking and 5 to both (§2.3).

Correction (multi-model verification pass): this signal was originally rated STRONG on the basis that these are two agencies. They are not. USCIS and Homeland Security Investigations are both components of the Department of Homeland Security, and Continued Presence is granted by a DHS centre, not by the Department of Justice as an earlier draft stated. What survives is still substantial: two statutory schemes, two adjudicative processes and two separate data systems, one victim-initiated through counsel and one law-enforcement-initiated, neither citing nor derived from the other, agreeing on direction and roughly on magnitude. That is a real correlation between instruments rather than between institutions, and it is rated moderate accordingly.

The tension inside this signal is itself a finding. §7.10 reports that across ten federal task forces, 95.5 percent of law enforcement investigations and 99 percent of prosecutions were sex trafficking, while victim services in those same task forces recorded 20.2 percent labor and 10.4 percent both. Investigation and identification are pointed at measurably different populations. That divergence is reported at Moderate confidence rather than High because its load-bearing source carries a disclosed sourcing exception (§8), but it is consistent with, rather than contrary to, the two federal datasets above.

Signal 3 (STRONG). Identification counts measure how hard institutions are looking, not how much trafficking is occurring.

Assessment (Confidence: High; almost certainly true).

Five independent demonstrations, in five settings:

• ASU and the Phoenix Police Department, studying two Super Bowls: "no empirical evidence that the Super Bowl causes an increase in sex trafficking," alongside "a noticeable increase in those activities intended to locate victims" (§7.7).
• Minnesota Department of Health, evaluating a different host city prospectively: "no increase or decrease in sex trafficking incidence during Super Bowl LII," and "The numbers increased because the resources increased" (§7.7).
• The FBI, in its own current guidance, redirecting attention away from event-day sex trafficking toward labor trafficking in event preparation (§7.7).
• The FBI's UCR program, on its own Arizona data: counts moving from 6 offenses in 2015 to 99 in 2017 to 53 in 2019 reflect which agencies "currently have the ability to report," and "should not be interpreted as a definitive statement of the level or characteristics of human trafficking as a whole" (§7.8). Note that this item and the one above it are both the FBI, so these five demonstrations come from four institutions.
• USCIS, whose T visa series rose from 710 receipts in FY2008 to 15,332 in FY2024 while the statutory cap of 5,000 principal grants has never once been reached (§2.1).

Why this earns a strong rating. Two academic and government evaluations of sporting events, one federal law enforcement agency speaking twice in two different capacities, and one federal adjudicative series, all reaching the same conclusion about their own numbers from unrelated directions. A fifth demonstration sits in §7.8 at agency level: a police department that recorded zero trafficking offenses for three consecutive years recorded 22 in the year it was funded to look. The Martin and Hill peer-reviewed media analysis reaches the same conclusion again but is not counted toward the rating, because its publisher carries the caveat recorded in §8.

Signal 4 (STRONG). Indicator-based screening for a low-prevalence condition produces overwhelmingly false positives, and this appears in three unrelated institutional settings.

Assessment (Confidence: High; almost certainly true).

• Pediatric emergency medicine. The best-validated instrument, applied prospectively to 203 patients selected for high-risk complaints, produced 100 positive screens against 11 true cases: positive predictive value 10.0 percent (§5.4).
• Juvenile justice screening. Roughly 70 percent of those assessed had at least one tattoo, the branding flag appeared in 0.55 percent of assessments, and across all of them not one tattoo was assessed as a trafficker's brand (§4.4).
• Border DNA verification. Of 3,516 Rapid DNA tests performed on adults already referred by officers as suspected parentage fraud, 91.5 percent confirmed a genuine parent-child relationship (§6.8).

Why this earns a strong rating. Three settings, three institutions, three completely different technologies, one arithmetic. It is worth being precise about what is shared: these are not three measurements of the same quantity, they are three instances of the same base-rate structure. That is a weaker kind of agreement than three measurements converging on one number, and it is rated strong rather than higher on that basis. What they establish jointly is that the structure is general, not an artifact of any one instrument.

Signal 5 (STRONG). The federal identification apparatus does not itself use tattoos as an identification method.

Assessment (Confidence: High; almost certainly true).

Five separate federal or federally funded documents, checked individually:

• The Attorney General's FY2023 Annual Report to Congress, 251 pages: three occurrences of "tattoo," all in individual case summaries, none in any section on victim identification, screening, training or indicators; zero occurrences of "branding" or "branded" (§4.7).
• DHS Blue Campaign's principal identification page and its current printed indicator card: tattoos absent from both (§5.1).
• NIJ's published breakdown of resolved NamUs unidentified-person cases by identification method: no tattoo or physical-characteristic category exists (§6.3).
• PCOME, a county office operating under state law: tattoos are a sub-component of a formally non-positive tier (§7.6).
• The SAATURN final evaluation of the corridor's own federally funded task force: no occurrence of "tattoo" across 43 pages covering 506 investigations and 102 victims served (§7.5b).

A sixth document, the NIJ-commissioned evaluation of the federal task force programme, also contains no occurrence of "tattoo" or "brand" across 99 pages, but it is authored by an excluded think tank (§8) and is named here rather than counted toward the rating.

Why this earns a strong rating. Five documents from four institutions at two levels of government, federal and county, produced for different purposes and audiences, none citing another on this point. The agreement is an absence rather than a finding, which is a weaker form of evidence in general; it is rated strong here because each absence is individually checkable in a named public document, and because the same institutions do include other physical indicators (bruises, malnutrition, injury) in the same lists, so the absence is selective rather than categorical.

Set against it, honestly: HHS's Adult Human Trafficking Screening Tool Appendix B and the FBI's Trafficking Indicators page both do list tattoos or branding (§5.1). The federal government is not unanimous, and Signal 5 claims only that the identification machinery, as opposed to awareness material, does not operate on tattoos.

Signal 6 (MODERATE). Healthcare is a real identification channel, and three unrelated bodies of evidence say so.

Assessment (Confidence: Moderate; likely).

• A systematic review of 8 studies covering 420 participants: between 50 and 98 percent sought healthcare services during exploitation (§5.7).
• Federal prosecution records in the District of Arizona: of five located cases where the origin is described, three began at a hospital (§7.4).
• The federal task force evaluation: medical providers, including emergency rooms, sexual assault nurse examiners and hospitals, accounted for 12.6 percent of 151 referral-stream mentions across its interviews, third behind law enforcement at 20.5 percent and other victim service providers at 15 percent (§7.10). These are shares of interview mentions rather than of referrals.

Why this earns moderate rather than strong. The three are genuinely independent in method: a literature synthesis, a prosecution record, and an administrative referral dataset. But the systematic review's underlying studies are convenience samples of service-accessing survivors, the Arizona cases number five, and the third source carries a disclosed sourcing exception. The direction is well supported; no precise share is.

Signal 7 (WEAK, and reported because it was tested). Tattoo prevalence among trafficking survivors is not established as exceeding the general population rate.

Assessment (Confidence: Low; this document could not resolve it).

This document specifically looked for evidence that tattoos are more common among trafficking victims than among comparable non-victims, since that is the minimum condition for a tattoo to function as an indicator at all. What exists:

• The best located survivor estimate is 8 of 38 (21 percent) in one study inside a review (§4.3).
• The general population rate is 32 percent of all US adults, 38 percent of women, 46 percent of adults aged 30 to 49 (§3.1).
• In an operational juvenile justice population, roughly 70 percent of those screened had a tattoo (§4.4).
• The one figure with a comparison group, 48 percent versus 5 percent in minors, could not be verified in its attributed source (§4.3, verification flag).

Why this is weak rather than a negative finding. The 21 percent and the 32 to 46 percent are not measuring the same thing in the same way and cannot be subtracted from one another; the survivor figure comes from a clinical case series, the population figure from a probability survey. This document therefore does not claim that trafficking survivors are less tattooed than the general public. It claims only, and this follows strictly, that no located source establishes that they are more tattooed, and that the foundational paper in the field says the frequency is unknown in its own limitations section. A signal that cannot be evaluated is not the same as a signal that was evaluated and found absent, and this one is the former.

Signal 8 (NARROW). Marking in the United States federal record is more often coerced consent than forcible application.

Assessment (Confidence: Low; roughly even chance that this generalizes).

Across the located federal descriptions: one defendant "made some victims brand themselves with tattoos," one "pressured many of his victims into tattooing his first name on their bodies," one "branded her with a tattoo," and one "tattooed the minor victim." In the Ohio dataset, where scars or brands were purposefully inflicted, the source was the youth themselves in 60.82 percent of cases (§4.4, §4.6).

Why this is narrow. It rests on the wording of four prosecution summaries plus one state dataset that was measuring something adjacent. Prosecution releases are written for the public and their verbs are not chosen as data. The pattern is suggestive, it is mechanically consistent with Fang et al.'s inability to distinguish trafficking tattoos from voluntary ones, and it is not established.

Named specifically as NOT correlation

The following rest on a single source each. They are not weaker facts for it. They must not be read as corroborated, and if the single source is wrong, nothing else in this document catches it.

1. The AUC of 0.533 for the tattoo indicator. One study, one state, one population (justice-involved youth, 93.5 percent male). No second measurement of this indicator against outcomes exists anywhere in the located literature. It is the strongest single finding in this document and it is single-sourced.
2. The positive predictive value of 10.0 percent. One prospective study, one hospital, 11 true cases.
3. The 21 percent tattoo figure among sex trafficking survivors. One study of 38 people, reached through a review.
4. NIST's statement that tattoos cannot be a primary biometric. One report. It is the authoritative body on the question and it is still one document.
5. The decomposition timeline of day 294 and day 333. One donor, three tattoos, one climate.
6. PCOME's identification modality shares. One county office. It is an unusually good source and there is no second county publishing comparable data.
7. The absence of tattoo evidence from the District of Arizona trafficking record. One corpus, searched twice. An argument from silence.
8. The federal task force referral-stream shares. One evaluation, and the figures are shares of 151 interview mentions rather than of referrals.
9. The Zhang and Datta review containing no discussion of tattoos across 24 cases. One review, one set of published opinions.

10Key Assumptions Check

A Key Assumptions Check lists the things this document takes as true without having verified them. They are flagged not because they are doubted, which is a different thing, but because they are load-bearing and unverified, and because if one of them is wrong the conclusions resting on it move. Each entry names the assumption, says why it could not be verified, and says what would change above.

KAC-1. That published federal indicator lists reflect what practitioners are actually trained and instructed to do. This document repeatedly reads institutional intent off published documents: DHS omits tattoos from its card, therefore DHS does not rely on the tattoo indicator. That inference assumes published material tracks internal training curricula and field practice. It could not be verified: no federal or Arizona law enforcement training curriculum was obtained, and the Arizona POST site was unreachable (§7.10). If field training teaches tattoo recognition regardless of what the printed cards say, Signal 5 weakens considerably, because it becomes a finding about publications rather than about practice.

KAC-2. That tattoo prevalence among minors is low because tattooing minors is generally unlawful. §3.2 infers a low base rate for minors from the legal prohibition, and §4.8 uses that inference to limit how far the adult conclusion extends. No published measurement of tattoo prevalence among US minors was located. The inference is plausible and two things cut against it. Roughly 70 percent of those assessed in the Ohio dataset had at least one tattoo despite comparable statutes, though that sample has a mean age of 18.88 and is substantially adult, so it is weaker evidence about minors than an earlier draft treated it as. Correction (multi-model verification pass): an earlier draft described the Ohio sample as "substantially aged 16 to 18," which came from misreading a table row reporting the age of the youth's significant other rather than the youth's own age. Separately, Arizona's own statute (§3.2) permits tattooing a minor with a parent physically present, so the lawful pathway is conditioned rather than closed. If minor tattoo prevalence is in fact high, the one place where this document allows the tattoo indicator any discriminating power disappears.

KAC-3. That absence from a prosecution press release indicates absence from the case file. §7.3 treats the absence of tattoo references across the District of Arizona corpus as evidence about the corridor. Press releases are summaries written by public information officers for a general audience; they omit most of what is in a case file by design. This document rated that finding Moderate rather than High for exactly this reason and stated it as an argument from silence, but the assumption remains load-bearing for §7.3 and it cannot be checked without the underlying charging documents.

KAC-4. That the direction of the Ohio finding, though not its magnitude, extends beyond justice-involved youth. The single strongest empirical result in this document comes from one state, one instrument and one population that is 93.5 percent male with a mean age of 18.88 and an age range of 12.9 to 23.6 years, which makes it substantially an adult sample. §4.8 explicitly declines to extend it to adults, to healthcare settings or to non-justice-involved minors. But the Bottom Line Up Front and Signal 1 do treat it as evidence about the tattoo indicator generally, which assumes the direction travels even if the number does not. If trafficker marking is concentrated in a population this instrument never touched, for instance adult women in pimp-controlled commercial sex outside the juvenile justice system, then the Ohio result would be measuring the indicator's failure in a population where the phenomenon is genuinely rare, and would say less than this document takes it to say.

KAC-5. That tattoos constitute a material share of PCOME's "circumstantial" identification category. §7.6 uses the circumstantial figure of 7.4 percent as an upper bound on what physical marks contribute to identification in this corridor. The category explicitly bundles "scars/marks/tattoos, personal effects, etc.," and the office does not disaggregate it. If personal effects, meaning identity documents, clothing and possessions found with remains, account for nearly all of it, then the tattoo contribution is far below 7.4 percent and this document has been generous to it. The bound holds in the direction the document argues, so the assumption is conservative, but the figure should not be read as a tattoo figure.

KAC-6. That the search corpora consulted are complete. Several findings are negative results from full-text searches: one tattoo hit across the District of Arizona press releases, three across a 251-page federal report, none across a 99-page evaluation. These assume the search indexes cover the full corpus and that the text layers of the PDFs searched are complete. The PDF text-layer assumption failed once and was caught (the PCOME 2017 cell in §7.6), which is direct evidence that it can fail silently elsewhere. Each individual document search was run against text this document extracted itself, which limits the exposure, but the justice.gov corpus search relies on that site's own index.

KAC-7. That "trafficking" means the same thing across the datasets compared. §2, 5 and 7 place figures side by side from USCIS adjudications, DHS Continued Presence grants, FBI UCR offense counts, clinical screening studies and a state juvenile justice instrument. Each applies its own operative definition against its own standard of proof, ranging from a federal statutory finding to a clinician's opinion during a single visit. Hainaut et al. identify precisely this as the reason validation is difficult in the field (§5.3). This document names the standard wherever it gives a figure, but any comparison across two of these datasets assumes enough definitional overlap for the comparison to mean something, and that assumption is not verifiable.

11Analysis of Competing Hypotheses

Analysis of Competing Hypotheses guards against the most common analytic failure: anchoring on the first plausible explanation and then collecting only evidence that confirms it. The discipline is to state the rival explanations explicitly, weigh the same evidence against each, and be willing to end without resolution. Where the evidence does not resolve a question, this section says so, and says that the absence of a claim is not itself evidence for any hypothesis.

11.1 Why does the tattoo indicator persist when the evidence against it is this consistent?

What the in-scope sourcing supports: the indicator appears in HHS and FBI material and in a DHS campus toolkit, is absent from DHS's principal products, was measured once against outcomes and returned a weak but significant association with an AUC of 0.533, is rated among the least-observed indicators by practitioners, and traces to a 2018 narrative review whose authors disclaimed the prevalence inference.

H1: The indicator is valid and the measurements have missed it. Trafficker marking is real and documented in prosecutions; perhaps it is concentrated in populations the studies did not sample. The Ohio study covered justice-involved youth, overwhelmingly male; the branding literature concerns adult and adolescent women in pimp-controlled commercial sex. On this reading the indicator works where it applies and has simply never been measured there.

H2: The indicator survives on narrative and training inertia. It is vivid, it is easy to teach, it converts an abstract crime into something an untrained observer feels equipped to spot, and once embedded in training material it propagates by citation rather than evidence. Fang et al. is cited as authority for a claim it explicitly refuses; the UNC fact sheet rests substantially on Fang; HHS ships a red-flag list three pages after explaining why red-flag lists are not validated.

H3: It is valid for a narrow subpopulation and has been overgeneralized. Both of the above, in sequence: a genuine phenomenon within pimp-controlled domestic sex trafficking, generalized to all trafficking, then to all populations, then to a screening indicator for untrained observers.

Weighing. H1 is not refuted and cannot be, because the study that would refute or confirm it has not been done; this is the case-control study Rambhatla et al. explicitly call for and that §13 records as missing. H1 is however weakened by the Ohio data in a specific way that is easy to miss: the informative failure there is not the AUC. Across the assessments in which trained staff asked directly about branding, zero tattoos were assessed as trafficker brands, while roughly 70 percent of the youth had tattoos. That is not an instrument failing to detect a rare signal; it is assessors looking for the thing and recording none of it in a population with abundant tattoos. Note that H1 gains something back from the same study, which an earlier draft missed: where the branding flag was raised at all it was sixteen times more common among victim assessments, so the flag is specific even though it is far too rare to screen on. H2 is supported by the citation archaeology in §4.1 and 4.2, which is unusually clean: the chain from the foundational paper to the training material is documented and the foundational paper's own limitations section contradicts what the chain carries. H3 accommodates all the evidence, including the prosecutions in §4.6, without requiring any source to be wrong.

Assessment (Confidence: Moderate; likely): H3 is the best-supported explanation. Trafficker marking is real within a specific form of trafficking, and the indicator built on it was generalized far past the population and the evidence that support it. H2 describes the mechanism of that generalization accurately. H1 cannot be excluded and would require a case-control study to test. This document does not resolve between H3 and H1 for the specific subpopulation of adult women in pimp-controlled commercial sex, because no source located makes a measured claim about that exact population, and the absence of such a claim is not evidence in either direction.

11.2 Does a positive predictive value of 10 percent mean screening is doing harm?

What the in-scope sourcing supports: Kaltiso et al. report PPV 10.0 percent and NPV 99.0 percent, and recommend the tool for ruling out. Texas requires a mandatory child-abuse report on a clear-concern CSE-IT score while stating the tool "cannot confirm victimization." Mumma et al. found unaided physician concern only 40 percent sensitive against a survey at 100 percent. HHS records that experts disagree about whether red-flag checklists are useful at all.

H1: Low PPV is an acceptable cost. Screening is triage, not diagnosis. A 99 percent negative predictive value safely clears the large majority of patients, the follow-up to a positive screen is a conversation rather than a sanction, and missing a trafficked child is a far worse error than a false positive.

H2: Low PPV is harmful at scale. Ninety false positives per ten true cases, multiplied across a system and coupled in at least one state to a mandatory report to a child-abuse hotline, imposes real costs on families who are not trafficking victims, and disproportionately on populations already over-surveilled. Note that the Ohio instrument's own most-predictive domains are runaway behavior, questionable financial support and unstable housing, which track poverty.

H3: The question is unanswerable from the published record. No located study measures what happens after a positive screen: not the disposition of false positives, not the harms, not whether the true positives identified went on to receive services or safety.

Weighing. H1 is supported by the instruments' actual NPVs and by the authors' own framing, and by the Mumma finding that the alternative, unstructured clinical suspicion, performed worse. H2 is supported by the Texas mandatory-reporting linkage, which is documented, and by the composition of the indicator sets, which is documented, but the harm itself is asserted rather than measured anywhere this document located. H3 is supported by a straightforward absence: Macy et al. searched for response protocols and reported being unable to locate step-by-step guidance for what to do with a positive.

Assessment (Confidence: Moderate; likely): H3 is correct as to the state of the evidence, and H1 and H2 are both defensible readings that the published record does not adjudicate. This is a policy question about acceptable error tradeoffs, and this document argues no policy position on it. What can be stated as fact is narrower and is stated in §5.4: these instruments rule out well and rule in badly, and any system deploying them is choosing to accept a large majority of false positives among its positives. Whether that choice is right is not a question the sourcing answers.

11.3 Why is labor trafficking dominant among identified victims but nearly absent from investigations?

What the in-scope sourcing supports: USCIS 77 percent labor among Supplement B filers; Continued Presence 310 of 382 forced labor; task force investigations 95.5 percent sex trafficking and prosecutions 99 percent; task force victim services 20.2 percent labor and 10.4 percent both; Arizona UCR involuntary servitude never above seven offenses statewide in a reported year; zero District of Arizona labor trafficking prosecutions announced in the window.

H1: Labor trafficking is harder to detect. It occurs inside ostensibly lawful workplaces, its coercion is contractual and documentary rather than physical, and its victims are disproportionately non-citizens with reason to avoid law enforcement. The evaluation itself offers this.

H2: Enforcement is structurally pointed elsewhere. Vice units, sting operations and online advertisement monitoring are the existing investigative infrastructure, and they detect commercial sex. The evaluation notes that task forces "rely on traditional investigative techniques, such as sting operations."

H3: The two systems are measuring different populations for definitional reasons. The immigration relief pathway reaches non-citizens, who are disproportionately labor trafficking victims; the criminal enforcement pathway reaches domestic commercial sex, where victims are disproportionately citizens. On this reading neither number is wrong and the gap is an artifact of two different doors.

Weighing. H3 has direct support: the T visa is available only to non-citizens, and the task force survivor data show 74.2 percent of all survivors served were citizens against 31.3 percent among labor trafficking survivors. That single crosstab explains a large part of the gap mechanically. H1 and H2 are both offered by the federal evaluation itself and are not mutually exclusive with H3 or with each other. H2 gains support from the observation that the investigative techniques named are the ones that find commercial sex.

Assessment (Confidence: Moderate; likely): all three operate and H3 accounts for the largest share of the arithmetic. The gap is substantially a function of which door a victim enters through, compounded by an investigative infrastructure built for commercial sex and by the genuine difficulty of detecting workplace coercion. This matters for §4 and 5 because it means the marking literature and the population the identification system actually recognizes are describing different people, which is stated as a fact in §2.3 and explained here.

11.4 Does the absence of tattoo evidence in the District of Arizona record mean anything?

What the in-scope sourcing supports: one "tattoo" hit across the district's press release corpus, and it is a robbery case; eight trafficking prosecutions read in full with no mention; marking described in prosecution releases from four other districts.

H1: Trafficker marking genuinely does not occur, or is rare, in this corridor. The corridor's trafficking is differently structured from the urban pimp-controlled model in which marking is documented.

H2: It occurs and is simply not described in press releases. Press releases are short summaries and omit most case detail; the four districts that did describe marking may differ in press practice rather than in facts.

H3: There is too little trafficking prosecution in this district for the absence to carry information. With Sexual Abuse at 0.7 percent of the district's docket and Commercialized Vice at zero in FY2023, the number of cases in which marking could have been described is small, so the absence may be a sample-size artifact.

Weighing. H3 is the strongest of the three and is quantified: §7.2 shows the district's trafficking-capable sentencing categories never exceed roughly 0.9 percent of a docket that is more than three quarters immigration. Against a base of a few dozen trafficking prosecutions across eleven years, an absence of any particular case detail is very weakly informative. H2 cannot be tested without charging documents. H1 is plausible and is consistent with the corridor's structure as described in the prior border brief, but nothing located establishes it.

Assessment (Confidence: Low; this document does not resolve it): the absence of tattoo evidence from the District of Arizona record is most likely explained by the small number of federal trafficking prosecutions in a district dominated by immigration enforcement, rather than by a finding about how trafficking operates in southeastern Arizona. §7.3's Moderate rating should be read with this in mind, and the finding should not be used to support any claim about trafficker practice in the corridor. If a claim of that kind matters, the route to testing it is state prosecutions and charging documents, which §13 records as an open item.

12Sources Cited

Numbered continuously. Each entry carries an Admiralty grade in the form [Admiralty: A1], per §1.1.

A note on fetch provenance, which this document treats as part of the grade. §1.1 grades a source's reliability and the credibility of the specific information cited. Neither axis captures whether the compiler actually read the document. Entries below are therefore marked as follows:

• No marker: fetched, opened and read during the main compilation pass. Every quotation attributed to these was read in the source.
• [lane]: fetched by one of this session's parallel research passes and reported with its quotations, but not independently re-fetched during compilation. These are graded provisionally. Where a [lane] source carries a load-bearing claim, that claim was either re-fetched and verified in the main pass, in which case the marker is removed, or the claim is attributed in the body to the research pass rather than asserted.
• [abstract only]: only the abstract was read; the full text was not obtained. Any figure taken from these is an abstract figure.

Sixteen sources reported by the research passes were independently re-fetched and verified during compilation before being used: NISTIR 8232, the NamUs FY2024 Annual Report, DHS OIG-22-27, the Ohio ODYS validation report, the DHS Blue Campaign identification page and indicator card, the HHS Adult Human Trafficking Screening Tool, the FBI Trafficking Indicators page, the PCOME 2025 and 2024 annual reports, the USSC FY2023 Arizona packet, the SAATURN final evaluation, the ASU Super Bowl report, the NIJ ECM evaluation, the Arizona Attorney General indicator page, and the Arizona DPS TOPS data endpoint.

Primary and official: United States federal

Primary and official: state, county and municipal

International

Peer-reviewed

Research institutions

Named journalism, regional and local

Post-scope and out-of-scope methodological sources

These fall outside this document's date range or geography. They are used only where noted for methodology or definitions, never as in-scope primary sources.

Sources located and excluded

Logged in full in §8, with the reason for each. Excluded categories: advocacy and mission-driven organizations, named think tanks, congressional committee products, the standing banned list, preprints, and repositories used only as locators.

13Open items for a follow-up research pass

Resolved during this compilation

RESOLVED. Whether the tattoo indicator has ever been measured against outcomes. It has, once. The Ohio Department of Youth Services validation study, 2,010 assessments analysed, returned an AUC of 0.533 that its authors classify as not predictive, alongside a weak but statistically significant association. Landed in §4.4 and it became the empirical centre of the document. See the verification-pass entry below, which materially changed how this document reports it.

RESOLVED. Whether DHS lists tattoos as a trafficking indicator. Partly, and the answer is more interesting than either yes or no. The Blue Campaign's principal identification page and its current printed indicator card both omit tattoos; a Blue Campaign guide written for college student leaders includes them. Landed in §5.1.

RESOLVED. Whether NIST's widely quoted 99.4 percent tattoo matching figure is current. It is not. That figure is a rank-10 result from an open-book challenge on a 4,375-image gallery; NIST's sequestered follow-up on a 100,000-image operational gallery put the comparable figure at 72.1 percent. Landed in §6.1. An early draft of §6 used the 99.4 percent figure alone and was corrected before compilation finished.

RESOLVED. The missing cell in the Pima County identification table. The 2025 annual report's PDF text layer drops the 2017 circumstantial value. Resolved as zero by two independent checks: a coordinate-level extraction locating the gap in the circumstantial column, and the 2024 annual report printing it explicitly. Landed in §7.6.

RESOLVED (verification pass). Lederer and Wetzel 2014. Retrieved during the multi-model verification pass from the Loyola law journal repository. The most-quoted statistic in this field is 87.8 percent, not 88, and its denominator is the 98 respondents who answered that question rather than the full 107. Landed in §5.7, and open item 2 below is closed.

RESOLVED (verification pass). The Ohio study says more than an earlier draft of this document reported. Branding is significantly though weakly associated with victimisation (6.5 percent of victim assessments against 0.4 percent of non-victim ones, phi = .124, p < .001); the AUC of 0.533 reflects near-zero sensitivity rather than absence of signal; the reference standard is the screener's own end-of-session designation across 46 positive assessments, 29 of them "potential"; and the sample has a mean age of 18.88 rather than being a study of minors. All four landed in §4.4.

RESOLVED (verification pass). The FY2023 T visa row. Published in Table 11 of the Attorney General's FY2023 report, a source this document already cited. Landed in §2.1, and it corrected a running total that had been understated.

RESOLVED, partially. Whether southeastern Arizona has a trafficking task force. It had one, funded October 2015 to March 2019, and no federally funded law-enforcement task force since. What remains unresolved is whether any state, county or municipal successor exists, and no source establishes SAATURN's status after its grant ended. Landed in §7.5 and 7.5b.

RESOLVED, partially. Whether the corridor's trafficking zeros are real or a reporting artifact. Real for 2021 onward with one exception. Nogales Police Department reported assaults normally in all five years, and Santa Cruz County Sheriff's Office in four of the five; its 2021 row shows one aggravated assault and no simple assaults and should be treated as probably incomplete. Partly an artifact before 2021, where the state's own reports name corridor agencies as incomplete reporters. Landed in §7.8.

Still open

1. The 48 percent versus 5 percent tattoo figure. Fang et al. attributes it to Greenbaum, Dodd and McCracken 2018; that paper's abstract does not mention tattoos, and its full text is paywalled. Two attempts failed. Next step: obtain the full text of Pediatric Emergency Care 34(1):33-37 through a library or interlibrary loan and read the variable table. If the figure is not there, the field's only comparative tattoo statistic does not exist, which is a publishable finding in itself.

1. The Phoenix Police Department's own Super Bowl LVII release. Four attempts across phoenix.gov paths returned 404s or JavaScript-only listings, and the Internet Archive was unavailable to both passes. The 2023 operation figures in §7.7 therefore rest on three named regional outlets reporting the same announcement rather than on the primary. Next step: a public records request to Phoenix Police, or a retry once the archive is reachable.

1. CSE-IT validation statistics. No sensitivity, specificity, positive predictive value or reliability coefficient could be obtained for an instrument mandated across multiple states' child-welfare systems. Its validation source is a self-published technical report whose PDF returns HTTP 404 at the URL the Texas Governor's office itself cites. Two attempts. Next step: a direct request to WestCoast Children's Clinic, or a public records request to a state agency that procured the tool. This may simply not exist in published form, which would be the finding.

1. A case-control study of physical findings in trafficking survivors. Does not exist. Rambhatla et al. explicitly call for one. Every figure in the clinical literature comes from a help-seeking convenience sample with no comparison group. Next step: none available through research; this is a gap in the science, not in the searching, and it should be stated as such wherever the 21 percent figure is used.

1. Tattoo prevalence among United States minors. No published measurement located. This is the single figure that would settle whether the base-rate objection in §3.1 transfers to minors, which is the one population where this document allows the indicator any force. Next step: examine whether any adolescent health survey instrument (for example a state Youth Risk Behavior Survey module) has ever carried a tattoo item.

1. Disaggregation of PCOME's circumstantial identification category. Tattoos cannot be separated from scars, marks and personal effects in any published source. All eleven annual reports covering the window were searched. Next step: a records request to the Pima County Office of the Medical Examiner. This may not exist in published form; the office may not record the distinction.

1. Anderson 2008, Journal of Forensic Sciences 53(1):8-15. The one publication promising a full PCOME identification-modality breakdown. Publisher returned HTTP 403 across two attempts and the paper is not open access. Only its DOJ abstract was obtained. Next step: library or interlibrary loan.

1. Arizona peace officer training curriculum. The AZPOST site is behind bot-challenge protection and was unreachable across six attempts by two methods, and the Internet Archive fallback rate-limited. §7.9's conclusion rests on the statute and the administrative rules instead, which are the stronger sources for the mandate question, but the curriculum itself would settle what recruits are actually taught, which is KAC-1. Next step: a public records request to AZPOST, or contact through a non-web channel.

1. The state and local Arizona funding and casework picture after March 2019. Federal award data is complete; state, county and municipal funding is not. Tucson Police published no trafficking case, victim or arrest counts in any annual report, and Pima County Attorney and Pima County Sheriff publish none at all. Next step: public records requests to Tucson Police Department and the Pima County Attorney's Office for trafficking case counts 2019-2025.

1. FBI UCR Arizona trafficking counts for 2016 and 2020 onward. The 2016 and 2020 state table URLs returned errors, and post-2020 NIBRS data requires an API key. The Arizona DPS state series in §7.8 covers the period instead, but the two series are not identical and one silently restates the other for 2020 (26 in the published report, 28 in the live system). Next step: request an FBI Crime Data Explorer API key and pull the series directly, then reconcile against the state system.

1. Whether any Arizona statute requires trafficking hotline notices in hotels or transport hubs. Four searches of the legislature's site returned nothing, which is a search result rather than a certainty. Next step: a full-text search of the Arizona Revised Statutes for the notice-posting provisions rather than for the word "trafficking."

1. UNODC and INTERPOL annexes. The INTERPOL DVI Guide's Annexures 4 and 5 carry the actual postmortem and antemortem data collection forms, which is where tattoo and scar recording is specified. UNODC's Anti-Human Trafficking Manual Module 2 and its Electronic Toolkit Tool 6.4 cover indicators. None were fetched. Next step: fetch before making any claim about international victim-identification indicator practice; this document makes none.

1. The NIST Tatt-C data-sourcing controversy. Excluded in full from this document because every substantive allegation traces to an advocacy organization. A permitted route exists and was not taken: the FOIA litigation against the FBI and NIST would be reachable as court records. Next step: pull the docket and any released records.

1. Disaggregated NamUs identifications attributable to tattoo searching. NamUs bundles tattoo search with facial recognition and forensic art. Four NamUs pages plus the annual report were checked. Next step: a request to NIJ. This may not exist; NamUs may not track it separately, in which case that is the answer.

1. Verbatim RAFT and TVIT instrument items. RAFT's four items sit in a figure that did not extract; the TVIT administration guidelines (NCJ 246713) returned HTTP 404 and its NIJ library record page redirects to the site homepage. Next step: request NCJ 246713 through NCJRS directly.

1. A second measurement of the tattoo indicator against outcomes. The Ohio study is the only one located, and §9 names it explicitly as not corroborated. Next step: examine whether Florida, whose Department of Children and Families tool Ohio's was based on, has published a validation of its own; and whether Dominguez, Sandal, Pukalo and Roe-Sepowitz, "A Comprehensive Analysis of Labor Trafficking Cases in the United States: An 11-Year Review," Crime and Delinquency, December 2024, contains control-method data. That paper's full text was not obtained (publisher returned HTTP 403 and a bot challenge) and its abstract states it assesses control methods, which makes it the highest-value unread source for this question.

1. Coverage of Arizona state bodies that a research lane completed only in part. The Governor's Council annual reports, the Arizona Attorney General's prosecution record and the county-level agencies were covered; Sierra Vista and Bisbee police department pages could not be reached, and Santa Cruz County's own alert system returned a path-level block across five attempts. Those three are the residual gaps in §7.

14Note on the eventual World-Building Document

This document is reference material. It is not fiction, it is not an argument, and it is not the World-Building Document.

It names no Vampires of Tucson character, faction, organization or location. It draws no line from any real person, agency, prosecution, victim or place described here to anything in the series. It does not represent, and must not be read as representing, that any fictional organization operates the way any real one described here operates, or that any fictional practice corresponds to a documented one.

Where the eventual World-Building Document draws on this material, that will be a separate, later and clearly fictional-license step. It should be treated as such rather than presented as continuous with the sourcing standard used here. The Admiralty grades, confidence levels and probability terms in this document attach to statements about the real public record. They do not transfer to fiction, and a fictional element built on a High-confidence Assessment here is not thereby a High-confidence anything; it is a choice made by an author.

Three cautions specific to this subject, offered because the material is unusually easy to misuse in fiction.

The document's central finding is a distinction, not a debunking. Trafficker marking is real. It is admitted fact in a federal prosecution that carried 482 months, and the Justice Department reports three more such cases among sentenced defendants in a single fiscal year. What the record does not support is the inference from a mark to a status. A story that depicts a trafficker marking a victim is on firm ground. A story that depicts an investigator identifying a victim because of a mark is depicting something the evidence says does not work, which is a legitimate thing to depict as long as the author knows that is what they are doing.

The negative findings here are about detection, not about prevalence. Nothing in this document says trafficking is rare, in southeastern Arizona or anywhere else. §7.8 records that two border counties reported zero trafficking offenses across five years while reporting assaults normally; §11.4 declines to resolve what that means. The consistent theme across §2, 5, 7 and 9 is that every number in this field measures institutional attention rather than occurrence. That is a statement about instruments, and it cuts in both directions.

The people in the sourced material are real. The named defendants, the named victims, the named officials and the 1,446 people who remain unidentified in Pima County as of February 11, 2026 are not characters and did not consent to be source material. Fictional license is the author's to take; it should be taken at a distance from the specific real cases catalogued here rather than by transposing them.

This document is not authoritative outside this purpose, is not clinical, investigative or legal guidance, and should not be used to screen, assess or identify any actual person. §1 states the grading scales it uses; §8 states what it refused to use and why; §13 states what it could not find. A reader who disagrees with a conclusion here should be able to locate the specific source that produced it and check it, which is the only claim to authority this document makes.