A Diagnosis Isn’t An Accommodation

A challenging conversation I have with clients isn’t with an applicant who has been denied an accommodation. Instead, it’s with someone I’ve just evaluated, whose diagnosis is real, whose suffering is real, whose history is real - but whose accommodation request, if they submit one, is unlikely to be granted. They came in believing the evaluation would confirm that they need extra time on the Bar, or a separate testing room for the USMLE, or frequent breaks on their Board exam. The data tells a different story…and so does the law.

This is one of the most uncomfortable truths in high-stakes accommodations work, and it’s one that we, psychologists, have done a poor job of articulating publicly: a real diagnosis and a defensible accommodation request are not the same object. They aren’t even the same kind of object. Diagnosis lives in clinical taxonomy. Accommodation lives in disability law. They overlap, often substantially, but they are not interchangeable, and high-stakes review is one of the few systems in clinical practice that forces us to confront that distinction.

Put plainly: getting a diagnosis answers the question “Do you have this condition?” Getting an accommodation answers a separate question: “Does this condition limit you enough, by the standard the law uses, to require a change in how the test is given?” Those are different questions and often have different answers.

I’ve spent much of my early career on both sides of this. I conduct evaluations for professionals seeking accommodations on high-stakes exams, and I review documentation submitted by applicants to credentialing bodies. From the reviewer’s seat, I see what happens when clinicians collapse the two: when “you have ADHD” is treated as equivalent to “you qualify for extended time,” and the report is built on that assumption. From the evaluator’s seat, I see what happens when applicants are told the same thing by well-meaning providers and arrive at my office for an evaluation expecting their diagnosis to translate, on its own, into an accommodation. It often doesn’t. Understanding why requires getting clear on what each of these things actually is.

Two different objects

A DSM-5-TR diagnosis is a clinical judgment that a person meets a set of criteria for a recognized condition. The criteria require clinically significant distress or impairment, but the threshold for “clinically significant” is, in practice, set against the demands of typical adult life, specifically relationships, job, daily functioning, etc. A person with ADHD whose symptoms cause real impairment at work, in relationships, or in self-regulation meets the clinical criteria. That’s the clinical question, and it is the one most clinicians are trained to answer.

The DSM-5-TR is the reference book clinicians use to decide whether someone meets the criteria for a condition. It asks: “Do your symptoms cause meaningful problems in your life?” It does not ask whether your symptoms place you outside the general adult population in a way that requires legal protection.

A disability under the ADA Amendments Act (ADAAA) is a different determination. It requires that a person have a physical or mental impairment that substantially limits one or more major life activities. The operative phrase is substantially limits. The 2008 amendments to the ADA - the ADAAA - explicitly broadened that standard, and both the ADAAA and the Equal Employment Opportunity Commission define disability as a substantial limitation in major life activities compared to the average person in the general population. Not most people in your professional cohort. Not most people with your educational background. Not most people who took the exam you’re preparing for. Most people.

The legal comparison group is “most people” (i.e., the broad adult population). Not other medical students, not other lawyers in training, not other applicants for the same exam. That choice of comparison group is doing more work than almost any other piece of the analysis.

Those two standards, specifically DSM clinical impairment and ADAAA legal impairment, do overlap, but they are calibrated to different reference groups. A person can meet the first without meeting the second. That is not a failure of either system; instead, it’s the system working as designed.

What ‘substantiality limits’ actually means

Here's the part that trips people up. The clinical standard and the legal standard aren't the same thing, and the distance between them is bigger than most people realize - including many psychologists doing these evaluations.

Harrison, Lovett, and Gordon (2013) surveyed 119 clinicians who were actively diagnosing students applying for postsecondary accommodations. On a 30-item true/false questionnaire about diagnostic and legal standards, the average score was 69 percent, which is only modestly better than guessing. More than half of the surveyed clinicians indicated that a person would qualify as disabled under the ADA on the basis of average reading scores combined with a high IQ; more than a third believed the ADA was designed to guarantee that an applicant would “perform at his or her best.” Neither of those is correct. Clinicians who diagnose for accommodations frequently do not understand the legal standard their reports are being measured against, and the documentation reflects that gap.

The general-population standard exists for a reason. Accommodations on high-stakes exams are not designed to remediate every form of human variability in cognition or attention. They are designed to level the playing field for people whose functional capacity is meaningfully constrained relative to the broader population, in ways that would otherwise prevent a fair demonstration of ability. The standard is comparative, and the comparison is intentionally broad.

Here’s an example. An adult presents for evaluation. They’re bright, accomplished, anxious, but exhausted. Their processing speed on a major cognitive battery falls at the 30th percentile. Their reading fluency is at the 40th percentile. Their working memory is at the 45th percentile. They describe years of feeling slower than their peers, specifically taking longer to finish assignments and suspecting something has always been off. They have ADHD. The diagnosis is supportable.

Are they substantially limited in a major life activity compared to most people in the general population? The 30th percentile is statistically and quantitatively Average. By definition, roughly thirty percent of the general population scores at or below that level. Many of those people navigate adult life, including complicated and challenging high-stakes exams, without accommodation. Some of them are sitting in the same testing center on the same day as your client, performing comparably, and they have never been evaluated or granted an accommodation.

A percentile tells you what fraction of the population scores at or below your score. A 30th percentile score means about 30 of every 100 people score at or below that level. That is statistically average and, thus, where a meaningful slice of the general population sits. It is not, on its own, a substantial limitation.

This is not an abstract concern. Harrison and colleagues (2021) examined 63 medical students who had been diagnosed with a reading disability and granted accommodations. When their reading scores were re-interpreted against the general population, as the ADAAA requires, only ~8% showed an actual normative impairment. The remaining ninety-two percent had been classified as disabled because their scores were compared against same-grade or same-program peers (i.e., other medical students), not against most people. That comparison-group choice, made quietly in the back end of an evaluation, was doing almost all of the diagnostic work.

This is the construct-validity problem in plain language. A score that looks low when compared to a person’s own estimated ability, or to an unusually high-functioning peer group, can simultaneously be average when compared to the general population, and the law asks the second question, not the first. Both observations are true; they are not in conflict. They’re just answers to different questions.

Construct validity is the question of whether a test actually measures what it claims to measure, and whether the comparison it makes is the right one for the decision being made. Comparing a medical student’s reading speed to other medical students might tell you something interesting about that group. It does not tell you whether the person is substantially limited compared to most people - which is the question the law is asking.

The intra-individual trap

Much of clinical neuropsychology is built around intra-individual analysis. In other words, comparing a person’s scores to their own estimated potential and identifying meaningful weaknesses. That framework has legitimate diagnostic utility, particularly when a marked drop from expected performance signals something clinically important (e.g., learning disability, acquired injury, cognitive change, etc.).

“Intra-individual” simply means “within one person.” It is the comparison of someone’s scores against their own other scores (e.g.,”Your working memory is much lower than your reasoning ability”). That can be useful clinically. It is not the same comparison the accommodations standard requires.

It is, however, frequently misapplied in accommodations contexts, and the empirical literature on this is now substantial. Maddocks (2018) examined the normative sample for a widely used cognitive assessment and found that ~61% of non-disabled individuals scored below the 25th percentile on at least one academic cluster. Most healthy adults, in other words, will produce at least one low average (or lower) score in a sufficiently broad assessment. Brooks and colleagues (2009, 2011) document the same pattern across other major batteries. Treating that statistical fact as evidence of impairment is, simply, a misreading of the data.

Sparks and Lovett (2009) made the same point a different way. They re-applied five different sets of objective diagnostic criteria for learning disability (LD) to the assessment files of 378 U.S. college students who had already been diagnosed with LD and granted accommodations. More than half failed to meet any of the five sets of criteria. Fewer than ten percent met sets of criteria that required academic achievement at least one standard deviation below the population mean. In a separate study of 50 medical students diagnosed with ADHD and seeking COMLEX accommodations, Joy and colleagues (2010) found that only 14 percent of the files contained evidence that the applicant met all five DSM criteria for the disorder. Most files documented only one or two. The diagnoses had been made and the accommodations had been granted, but the supporting record didn’t meet a basic standard of completeness.

The discrepancy framework for LD fares no better. Maddocks (2018) also found that between 89 and 92 percent of gifted but otherwise non-disabled individuals would meet at least one common discrepancy criterion for a learning disability. The discrepancy approach, applied to a high-functioning population, identifies the highly talented/gifted as disabled at rates that are diagnostically implausible. The DSM-5 and DSM-5-TR have correspondingly moved away from discrepancy as a definitional feature of specific learning disorder, which is a shift many clinicians have not fully integrated into their practice (Harrison & Sparks, 2022).

Processing speed is the other place this goes wrong, and it goes wrong often. Lovett (2019) administered measures of cognitive processing speed and timed academic skills to 447 students with prior LD diagnoses, and found that the two correlated only modestly, with roughly six percent shared variance. Cognitive processing speed and real-world academic fluency are, empirically, not the same construct. A low score on the WAIS processing speed index does not predict impairment in reading rate, math fluency, or written output. Yet reports routinely invoke a low processing speed score as the rationale for extended time on high-stakes exams, and the rationale doesn’t exactly hold up under scrutiny, nor is it necessarily aligned with the empirical evidence.

Six percent shared variance is a way of saying these two things - cognitive processing speed and timed academic skill - mostly travel separately. In other words, knowing one tells you very little about the other. A low cognitive processing speed score is not a substitute for actually measuring how fast someone reads, writes, or completes math under time pressure.

A processing speed score that is two standard deviations below an examinee’s estimated overall ability sounds substantial in clinical terms. But if that score still falls in the average range relative to the general population, and if direct measures of timed academic performance fall in the average range as well, it does not, on its own, establish a substantial limitation under the ADAAA. The discrepancy may indicate inefficiency relative to the person’s potential. But it doesn’t, by itself, indicate disability relative to the comparison group the law cares about.

Quality reviewers know this. Many evaluating psychologists do not foreground it clearly, and the result is reports that read as clinically thorough but legally thin. The recommendation for extended time hangs on a discrepancy that, when examined against the population standard, does not support the request. The applicant is then surprised - and the evaluator is often equally surprised (and upset!) - when the request is denied.

The late-identified, high functioning professional

Now on to one of the most common and least openly discussed patterns in adult accommodations work. A bright, accomplished adult, perhaps a fourth-year medical student, a Bar exam applicant, a motivated graduate student preparing for the GRE, presents for evaluation. They describe a lifetime of feeling like they were working harder than their peers for the same outcomes. They’ve managed and, at times, even excelled. They have advanced degrees and professional licensure on the horizon. And they’re exhausted by the effort it has taken to get this far.

The diagnostic question is often legitimately answerable, but it deserves more rigor than it usually gets, particularly in adult-onset presentations. Adult ADHD has an estimated base rate of roughly five percent in the general population (Harrison & Edwards, 2023). Approximately half of children diagnosed with ADHD no longer meet criteria as adults; importantly, longitudinal data suggest that previously non-symptomatic adults who first endorse ADHD symptoms in adolescence or young adulthood are most often experiencing those symptoms for reasons other than ADHD (e.g., normal cognitive variability, a co-occuring mood or anxiety disorder, sleep disruption, substance use; Harrison & Edwards, 2023). The DSM requires that a sufficient number of impairing symptoms have been present before age twelve. A first-time adult presentation, in a person with no documentation of childhood symptoms and a sustained record of high functioning, is not what the diagnostic criteria were built to capture.

A “base rate” is how common a condition is in a population. About 5 out of every 100 adults have ADHD. When something is uncommon, even a good test will produce many false positives — people who look positive on the test but don’t actually have the condition. That mathematical reality matters more than most clinicians and applicants realize.

The screening literature compounds this. Harrison and Edwards (2023) conducted a systematic review of the diagnostic accuracy of self-report ADHD measures and found that, while their negative predictive values are excellent (typically above 96 percent), their positive predictive values in clinical samples mostly fall below 20 percent. At a five percent base rate, even a well-built screener can produce a false-positive rate of roughly 86 percent. In plain language: when these scales say “yes,” they are wrong far more often than they are right at the base rates that actually exist in the adult population. A screening tool was never meant to be a diagnosis, and treating a positive screen as confirmation is one of the most common, most consequential errors in this work.

“Positive predictive value” answers a question patients actually care about: if the test says I have this, how likely is it that I actually do? When that number is below 20 percent in clinical samples, it means the screener is wrong more than four out of five times when it says “yes.” “Negative predictive value” answers the opposite question: if the test says I don’t have it, how likely is that to be right? Those numbers are excellent. The takeaway is that these scales are good at ruling ADHD out and poor at ruling it in.

It also turns out that these scales are remarkably easy to manipulate, even unintentionally. Harrison, Edwards, and Parker (2007) randomly assigned college undergraduates to one of two groups: an “honest normal” condition and a “faking” condition. The faking group was given the DSM criteria for ADHD and instructed to feign the disorder in order to access academic accommodations. Their performance was compared with an archival sample of 72 students with carefully established ADHD diagnoses. The fakers produced symptom rating scale scores that looked more pathological than the genuine ADHD students’, and they performed poorly enough on timed reading and processing speed tasks to qualify for extra time. Standard symptom checklists could not distinguish them from a real clinical sample. This wasn’t a study of sophisticated malingerers; it was undergraduates handed the DSM and a goal. The implication is straightforward: a brief clinical interview and a self-report rating scale, without performance and symptom validity testing, collateral informant data, or historical records, cannot reliably tell a real ADHD presentation from a motivated imitation of one.

Even when the diagnostic question is answered carefully, the accommodations question is a separate matter. To meet the ADAAA standard, the limitation has to be substantial relative to the general population, not relative to the applicant’s own ability or trajectory. A person whose career history includes admission to selective programs, sustained academic success, and high-functioning professional performance is, by most reasonable measures, performing within or above the general population range, even when their internal experience is one of effortful compensation.

This is the part of the conversation our field has been quiet about (but shouldn’t be). The diagnosis can be real, the suffering can be real, and the accommodation case can still be difficult to make. That is not a contradiction. It’s the predictable consequence of two different standards applied to the same person. A late-identified, high-functioning adult who does not qualify for accommodation on a high-stakes exam has not been disrespected by the system, and they have not been failed by their evaluator. They have been measured against a legal standard that is intentionally calibrated to a broad reference group.

The importance of performance validity testing

There's one more piece of the evaluation that can't be skipped, because the accommodations process gives people real reasons to try to look more impaired than they are. Performance validity testing (PVT) is not optional in high-stakes evaluations, and the empirical literature is unambiguous on why.

Performance validity tests are short, standalone and embedded measures that check whether a person is engaging with the testing process as designed - not whether they are intelligent, attentive, or honest in everyday life. They are designed to identify suboptimal effort during testing itself. Symptom validity tests do something similar with self-reported symptoms, identifying patterns that don’t match real clinical presentations.

Mascarenhas and colleagues (2023) examined more than a thousand post-secondary students seeking accommodations and found that, using the conservative current standard of two or more PVT failures, ~9% produced non-credible performance. Failure rates on individual freestanding measures ran considerably higher, specifically 25% on one widely used test. Earlier work by Sullivan and colleagues (2007) found single-PVT failure rates of approximately 22 percent for combined ADHD and LD presentations, and nearly 50 percent for ADHD alone. Marshall and colleagues (2010) found higher non-credible performance rates in adults seeking a first-time ADHD diagnosis (24%) than in those with a pre-existing diagnosis (17%), a pattern consistent with what one might expect when external incentives are highest.

Despite this, validity testing remains underused. Harrison, Lovett, and Gordon (2013) noted that in their own consulting work, fewer than ten percent of applicants for postsecondary accommodations had documentation that included any formal assessment of symptom credibility. That number has improved in some settings since, but it remains the case that the majority of reports submitted to credentialing bodies still do not include the methodological safeguards the empirical literature has made standard.

None of this is an indictment of applicants. Instead, it’s a documented reality of the incentive structure within which adult accommodations evaluations occur, and it’s one of the central reasons credentialing reviewers have grown more skeptical of psychological reports submitted without PVT data. Including embedded and freestanding performance validity measures, and reporting the results transparently, is not a sign of distrust. It is the basic methodological requirement of an evaluation whose conclusions will be relied upon in a high-stakes legal context.

What this means for clinicians

If you’re evaluating an adult for high-stakes accommodations, the task is not to answer one question. It is to answer two, separately and clearly.

The first question is diagnostic. Does this person meet DSM-5-TR criteria for a condition? That answer is based on clinical interview, longitudinal history, validated measures, collateral data, performance validity testing, and your clinical judgment. The answer can be yes even when the answer to the second question is no.

The second question is functional and legal. Does this person have a documented impairment that substantially limits a major life activity relative to the general population, in a way that would prevent a fair demonstration of ability under standard testing conditions? That answer requires you to anchor your analysis in age-based, general-population norms rather than grade-based or program-based norms. It requires you to specify which functions are affected, how they are affected, and how those effects map onto the actual demands of the exam in question.

A report that answers only the first question is a clinical document. A report that answers both, clearly and separately, is an accommodations-focused report. Reviewers can tell the difference quickly, and the gap between the two is where most denials live.

It’s also worth being honest about the field’s own contribution to the credibility problem. In a survey summarized by Harrison and Sparks (2022), 14% percent of psychologists who had authored disability documentation acknowledged that they would bend the rules or ignore published diagnostic criteria to help their clients secure accommodations. Almost half believed the primary purpose of the assessment was to help the client obtain the accommodations they had requested. Those numbers, which were self-reported by the clinicians themselves, explain a great deal about why credentialing bodies have become more skeptical of psychological documentation, and why the burden on the rest of the field has gotten heavier. The defensible response is not to lower the threshold further; it’s to do the harder version of the work.

This also means something for how we talk to applicants before the evaluation begins. Informed consent in this context is not just about the testing procedure. It’s also about the realistic range of outcomes. An applicant has a right to know, going in, that an evaluation may confirm a diagnosis without supporting an accommodation request, and that a careful, accurate report is more valuable to them than an advocacy-driven one that will be denied months later. That conversation is hard, but it’s also the most ethically honest version of this work.

What this means for applicants

If you’re an adult considering an accommodations evaluation, the most important thing to understand is that the evaluation is asking two questions at once, and the answers can come apart. A diagnosis, if you receive one, is meaningful. It is a clinically supported account of how your mind works. It can inform treatment, self-understanding, and life choices. It’s not a small thing.

It’s also not, by itself, an accommodation. The credentialing body or testing agency reviewing your request is operating under a different framework. They are not asking whether you have a condition. They are asking whether the condition substantially limits a major life activity for you, relative to the general population, in a way that requires modification of standard testing conditions to fairly assess your ability. Those are different questions, and the second is the one the law makes them answer.

If you fill out an online ADHD screener and it comes back positive, that is not a diagnosis, and it is not a basis for an accommodation request. The research is clear that those scales over-produce positives in adult populations and that even untrained people, handed the diagnostic criteria, can produce profiles that look like real ADHD on the same measures (Harrison, Edwards, & Parker, 2007). That is not a comment on you; it’s a comment on the tools. A thorough evaluation looks well past those screeners, and a defensible accommodations case rests on much more than them.

If your evaluator tells you that your diagnosis is supportable but your accommodation case is harder to make, that is not a failure of the evaluation, and it is not a sign that your suffering is illegitimate. It is, often, a clinician doing the harder, more honest version of this work. The alternative, specifically a report that conflates the two and produces a request that gets denied months later, is worse for you, not better. A denial letter from a credentialing body becomes part of the record, and a poorly built case is harder to repair on appeal than a well-built one is to win on first submission.

The honest conclusion

Accommodation, in the high-stakes context, is not a reward for having struggled. It’s a corrective for a documented mismatch between a person’s functional capacity and the demands of a standardized assessment, calibrated to a population standard the law has chosen. It is a narrow instrument by design, and a great deal of human difficulty, including diagnosable, treatable, life-altering difficulty, sits outside that instrument’s intended scope. That is not a failure of compassion but rather a feature of the law.

The work of separating diagnosis from disability is, in the end, the work of telling people the truth: that what they have is real, and that what they need may not be what the accommodations system was built to provide. It’s also the work of giving the system back its credibility, because every accommodation granted on a thin, conflated rationale makes the next legitimate request harder for someone whose limitation is unambiguous.

Diagnosis is not disability. That is not a denial of suffering. It’s a clarification of what each system is asking and what each system can answer. The clinicians who do this well, and the applicants who come into the process understanding it, are the ones who are best served by the outcome, whatever the outcome turns out to be.

__________

If you have questions about navigating an evaluation, the documentation process, or what to expect from a credentialing review, you can reach me at alex@lightsidepsych.com.

References

Harrison, A. G., & Edwards, M. J. (2023). The ability of self-report methods to accurately diagnose attention deficit hyperactivity disorder: A systematic review. Journal of Attention Disorders.

Harrison, A. G., Edwards, M. J., & Parker, K. C. H. (2007). Identifying students faking ADHD: Preliminary findings and strategies for detection. Archives of Clinical Neuropsychology, 22(5), 577–588.

Harrison, A. G., Lovett, B. J., & Gordon, M. (2013). Documenting disabilities in postsecondary settings: Diagnosticians’ understanding of legal regulations and diagnostic standards. Canadian Journal of School Psychology, 28(4), 303–322.

Harrison, A. G., & Sparks, R. (2022). Disability diagnoses: Seven sins of clinicians. Psychological Injury and Law.

Mascarenhas, M. A., Cocunato, J. L., Armstrong, I. T., Harrison, A. G., & Zakzanis, K. K. (2023). Base rates of non-credible performance in a post-secondary student sample seeking accessibility accommodations. The Clinical Neuropsychologist.