Texas diagnosticians occupy a specific lane. We evaluate. We interpret. We make eligibility recommendations that follow students for years. And we do it inside a legal and procedural framework that most of our colleagues — including the ones who schedule our ARDs — do not fully understand. When a full multidisciplinary team is involved, speech, OT, PT, and others each contribute their own data streams to that same framework. What ties it all together — whether the eval involves one evaluator or six — is convergent evidence.

In 2012, two attorneys — Torin Togut and Jennifer Nix — published a law review article called The Helter Skelter World of IDEA Eligibility for Specific Learning Disability. I came to this paper through Mahnaz "Nazzie" Pater-Rov — a TEDA presenter and moderator of the Educational Diagnostician Clubhouse — who brought Togut himself in for a live discussion with practicing diagnosticians. That kind of professional exchange is rare and worth seeking out.

The title is accurate. They looked at the post-2004 IDEA landscape and described what they found: a structural collision between RTI and Child Find that produced inconsistent eligibility standards across states, delayed evaluations, and students who fell through the gap depending entirely on their zip code. Fourteen years later, the paper reads like a diagnosis of a problem that hasn't fully resolved — except in Texas, where the convergent evidence framework — multiple independent data sources pointing toward the same conclusion — gives us something more defensible than what most states landed on.

The 2004 Amendments and What Broke

Before 2004, the dominant method for identifying SLD was the severe discrepancy model. A student was eligible if there was a significant gap between their cognitive ability — typically IQ — and their academic achievement. The model was tidy on paper. It was widely understood to fail in practice.

Children had to demonstrate a substantial gap before they qualified, which meant that for early elementary students whose cognitive testing was unreliable and whose academic gaps were still emerging, the model effectively required them to wait until they had failed long enough to show the discrepancy. Daniel Reschly, writing in 2003, summarized two decades of research on the model with a damning four-part critique: it was unreliable, it was invalid, it was easily undermined in practice, and most importantly, it was harmful, because it delayed treatment to the years when reading remediation becomes substantially more difficult.

The 2004 IDEA reauthorization tried to fix this. Congress amended the law to remove the requirement that states use severe discrepancy and to permit alternative approaches — most prominently, Response to Intervention. The intent was to break the wait-to-fail cycle. The execution, as Togut and Nix document at length, was a compromise. Congress allowed states to use severe discrepancy, or RTI, or both, or neither. State educational agencies were given discretion. The discretion was used inconsistently. By the time of their paper, the country had an incoherent mix of SLD eligibility standards that varied by state — and sometimes by district. A student in Boise might not be eligible under criteria that would qualify the same student in Atlanta. The standard for one of the most commonly identified categories in special education varied not just by state but, in some places, by district.

The Helter Skelter, by the Numbers

By 2010, approximately thirteen states had adopted RTI as mandatory for SLD identification in specific grades or subjects. Other states retained severe discrepancy. Others permitted both. None of the major federal agencies — OSEP, OCR, the Department of Education — had issued binding guidance that would force convergence. The variability was, structurally, the policy. The consequences are measurable: a 2025 NCLD white paper documented that SLD identification rates in 2022–23 ranged from 1.58% of students in Idaho to 6.08% in Puerto Rico — and Idaho's eligibility criteria were found by OSEP that same year to be so restrictive they were out of compliance with IDEA. The same zip code problem Togut and Nix described in 2012 is still producing winners and losers.

Some states never moved away from discrepancy. California's regulations permit all three IDEA methods — discrepancy, RTI, and PSW — but offer no state-level guidance on which to use, leaving the choice to individual districts. In practice, discrepancy remains common, with districts varying widely on how the threshold is calculated. New Jersey permits all three methods as well, leaving the choice to the evaluation team, with discrepancy remaining a live option.

Why "Wait to Fail" Persisted

The severe discrepancy model survived 2004 because, in some sense, it was easy. The math was clear. A district could administer a cognitive battery and an achievement battery, compute the gap, and produce a number. The number was defensible because it was a number. But the numerical clarity papered over the underlying flaw, which is that ability-achievement discrepancy is a poor proxy for the cognitive processing weaknesses that actually define a learning disability. A bright student with dyslexia might not show a severe discrepancy because their compensatory strategies and verbal abilities masked the underlying decoding deficit. A struggling student with low average intellectual ability might not qualify because there was no "discrepancy" to find — they were "achieving commensurate with ability," which is another way of saying they were failing in line with expectations.

Michael P. v. Hawaii Department of Education (9th Cir. 2011), one of the cases Togut and Nix discuss, involved a fourth-grader with dyslexia who was found ineligible under Hawaii's severe discrepancy-only model. The Ninth Circuit eventually held that Hawaii's regulations procedurally violated IDEA by prohibiting any approach other than severe discrepancy. By that point, the student's fourth and fifth grade years had passed without specialized instruction. For the student, it was a loss measured in years.

Why Pure RTI Isn't the Answer Either

RTI was meant to solve the wait-to-fail problem by providing tiered, evidence-based intervention earlier in a student's school career. In its best form, it does. Togut and Nix do not dismiss RTI. They acknowledge its real strengths, including the way it integrates intervention with assessment and reduces unnecessary labeling of students whose difficulties are instructional rather than constitutional. But they also catalogue its limitations honestly, and the catalogue is worth sitting with.

First, RTI relies heavily on general education teachers to identify students with learning disabilities. Whether classroom teachers have the training and time to do this reliably is, in many districts, an open question. In practice, Tier 2 and even Tier 3 supports frequently fall to general education staff, particularly beyond elementary school, where structured intervention tiers are often the first budget casualty. Second, RTI measures student performance against peer-group expectations, not individual ability, which means students who are "achieving with their age group" may not be flagged even when their profiles include significant cognitive weaknesses. The student who reads at grade level through enormous effort, or who comprehends well despite labored decoding, can slip through unidentified — particularly when benchmark performance is the primary trigger for concern. Third, RTI conflates the quality of instruction with the presence of disability. If Tier 1 instruction is poor, students who would otherwise respond may not, and the data will look like disability when it is actually a teaching problem. The scale of that problem is not theoretical: in the only large-scale national evaluation of RTI, conducted by the National Center for Education Evaluation and Regional Assistance in 2015, 100 of 146 schools across 13 states failed to implement RTI with fidelity — and in those schools, RTI showed no positive effect on achievement.

The research base for RTI is also narrower than its widespread adoption suggests. Most of the empirical support is for early elementary reading. The evidence for middle school and high school RTI, and for math and writing applications, is much thinner. A significant amount of national educational policy has been built on a research base whose primary findings concern second graders learning to decode. Texas diagnosticians who logged significant hours in the Texas Reading Academies (HB 3) and dyslexia training over the past several years will recognize the irony.

What Texas Chose

Discrepancy required failure. RTI, poorly implemented, permitted delay. The question was whether any framework could do both jobs — identify genuine disability early and hold the line against indefinite deferral. Texas's answer was convergent evidence.

Texas's regulatory response to the 2004 amendments was less dramatic than some states but, in retrospect, more durable. TAC §89.1040 does not require districts to use any single framework. It permits the use of RTI, the use of a Pattern of Strengths and Weaknesses approach, or the use of other research-based procedures. The TEA Guidance for the Comprehensive Evaluation of SLD, updated in 2025, articulates the standard that ties these approaches together: convergent evidence. Eligibility is not established by a single score, a single framework, or a single data source. It is established by the alignment of multiple independent sources of data — formal cognitive and academic measures, curriculum-based measurement, intervention response, classroom observation, teacher and parent input — pointing toward a consistent pattern.

What matters is not which evaluation framework a diagnostician uses — it's whether the data converge.

Practitioners have developed several structured approaches that operationalize this standard well. The RIOT framework — Review, Interview, Observe, Test — organizes data collection across those four source categories, ensuring that no evaluation relies entirely on formal testing and that each data type is gathered intentionally. A RIOT-based evaluation, done thoroughly, produces exactly the kind of multi-source convergent picture that Texas's guidance describes. The Schultz and Stephens Core-Selective Evaluation Process (C-SEP) works from a similar convergent logic, anchoring the evaluation in TEKS-aligned academic data and then selectively adding cognitive and processing measures to test specific hypotheses about the underlying pattern. Both approaches — and others like them — are paths to the same destination. None is mandated. All are defensible when the reasoning is documented and the data genuinely converge.

The implication for practice is significant: Texas diagnosticians are not locked into any single method. A diagnostician trained in PSW can build a convergent case. So can one trained in C-SEP, or one who organizes evaluations through a RIOT lens, or one whose district has developed its own structured process. The standard is the convergence of independent evidence, not the framework used to collect it.

What "Convergent" Actually Requires

The strength of the convergent evidence standard is also its discipline. It is harder than working from a single formula. A diagnostician building an SLD case under convergent evidence has to do real interpretive work: identify the pattern across multiple data streams, explain why those streams converge, address the alternative explanations — exclusionary factors, lack of appropriate instruction, language differences — and document the chain of reasoning that connects the data to the conclusion.

This means MTSS data is evidence, not gate-keeping. A student who has responded poorly to Tier 2 reading intervention generates one piece of the convergent case. A student whose phonological processing scores are consistently low generates another. A teacher who describes labored decoding and lost meaning generates another. A parent who describes a family history of reading difficulty generates another. None of these alone is sufficient. Together, they are compelling. And the moment they begin to converge, the obligation under Child Find is to evaluate — not to wait for more MTSS data.

The stakes are not abstract. The 2025 NAEP Long-Term Trend results showed that gains in reading for 9-year-olds were concentrated at the lowest performance percentiles — the students most likely to be reached by a timely evaluation and least likely to recover ground lost to delayed identification. Convergent evidence, built and acted on early, is one of the few mechanisms that actually gets to them in time.

The Child Find Guardrail

This is the piece that the convergent evidence framework gets right where some pure RTI implementations get it wrong. MTSS data is one source of evidence. It is not a precondition for evaluation. The 2011 OSEP Memorandum to State Directors of Special Education is explicit: it is inconsistent with IDEA's evaluation provisions for a district to reject a referral or delay an initial evaluation because a student has not completed an RTI framework. Texas has built this principle into its statutory structure. TAC §89.1011 and TEC §29.0041 together govern the parental request timeline: a district must respond within fifteen school days, either by securing consent or by issuing Prior Written Notice of refusal. TEC §29.004 then requires the FIE to be completed within forty-five school days of consent. These clocks exist precisely to prevent the indefinite MTSS deferral that Togut and Nix identified as a Child Find problem — and importantly, a student's lack of progression through intervention tiers is not a permissible basis for extending or pausing that forty-five-day window.

In practice, this means MTSS and an FIE can run in parallel. A student can continue receiving Tier 2 intervention while the diagnostician completes a comprehensive evaluation. The intervention data becomes one of the convergent sources in the report. The evaluation does not wait for the intervention to finish — because, structurally, Child Find does not allow it to.

Closing

The Helter Skelter World of IDEA Eligibility ends with a call for consistency and predictability — for a system in which a student in one state and a student in another have the same shot at being identified accurately, and at being identified in time. Across the country, that hope remains partly unfulfilled. In Texas, the convergent evidence framework gets us measurably closer to it than either of the two alternatives Togut and Nix critiqued. The framework is not perfect. The work it asks of diagnosticians is substantial. But it is coherent, it is grounded, and it answers the question it was built to answer.

The helter skelter persists in some states. Texas diagnosticians have a better tool. Use it well.