Test Accommodations: What the Assessment Manual Permits
You have the kit open, the student in front of you, and a plan in the file that says extended time. Nothing in that plan tells you what extended time does to the score you are about to report. For that, the only authority that matters is the test accommodations section of the assessment manual in your hand — and the manuals do not agree with each other.
Some publishers print a table of approved changes. Some print a paragraph and hand the decision to you. One publisher's three reading products have three different approved lists. And several of the instruments educational therapists use most do not publish their rules anywhere a practitioner can read without buying the manual.
This post reports what the published documents say. It is not about entitlement. Nothing here is legal advice. Whether a particular student is owed a particular change, and what a school plan must contain, is a question for that student's school team and, where it matters, for an attorney — not for a test manual and not for this article.
The short version
The measurement field's reference text draws a hard line between two words. An accommodation keeps score comparability. A modification changes the construct, and therefore changes what the score means.
The line is not drawn by the change itself. It is drawn by whether the change touches the thing the test is measuring.
Extended time is the clearest case. On an untimed measure it is nothing. On a fluency measure it is a modification, because speed is the construct.
Publishers mark the record differently. One wants a letter on the booklet cover. One wants a note on the test record. One asks for a sentence in your report.
The same accommodation — directions in the student's first language — is approved by one widely used screener and unapproved by its closest relative.
Where the manuals are silent, nobody has filled the gap. The measurement literature's own summary says validity questions are frequently not addressed by the people making accommodation decisions at all.
Two words that are not interchangeable
The Standards for Educational and Psychological Testing, published jointly by AERA, APA and NCME and still the field's reference text, settles the vocabulary in one sentence.
"As used in the Standards the term accommodation is used to denote changes with which the comparability of scores is retained, and the term modification is used to denote changes that affect the construct measured by the test."
And then it says what follows from that:
"With a modification, the changes affect the construct being measured and consequently lead to scores that differ in meaning from those from the original test."
So the two words are not two degrees of the same thing. They are a claim about whether your score still means what the manual says it means. An accommodation is a change you can make and still read the norm table. A modification is a change after which the norm table is answering a different question.
The chapter carries a footnote warning that federal disability law uses both words differently from the way the Standards use them. That footnote is the source of a great deal of confusion in team meetings, and it is the reason a change that is correct to provide can still be a change that costs you your normative comparison. Those are two separate questions with two separate answers.
The Standards also set out the accommodation types: changes in "administration setting, presentation, interface/engagement, and response requirements," and "the addition of individuals to the administration process (e.g., readers, scribes)."
The rule underneath all of it
Standard 6.1 of the same document states the principle that every publisher's accommodation policy is an application of:
"In general, the same procedures should be followed as were used when obtaining the data for scaling and norming the test scores."
That is the whole of it. A standard score is a statement about how this student performed relative to a sample of students who were tested a particular way. Change the way, and you have weakened the only thing that makes the comparison legitimate. Standard 6.1 says so directly: "departures from standard procedures may jeopardize the validity or complicate the comparability of the test score interpretations."
Standard 6.4 extends the same logic to the room: "In general, the testing conditions should be equivalent to those that prevailed when norms and other interpretative data were obtained."
This is also why the question is not answerable in the abstract. It depends on what the particular instrument was built to measure, which is a fact about that instrument and lives in that instrument's manual. The same change is an accommodation on one subtest and a modification on the next one in the same kit.
Extended time is the case that proves the rule
The Standards name extended time specifically, and they do not treat it as benign:
"Construct underrepresentation is a primary threat to the validity of test accommodations. For example, extra time is a common accommodation, but if speed is part of the intended construct, it is inappropriate to allow for extra time in the test administration."
The sentence after that one is the part worth copying into your own notes:
"Scores obtained on the test with extended administration time may underrepresent the construct measured by the strictly timed test because speed will not be part of the construct measured by the extended-time test."
Read that against the instruments on your shelf. Every fluency measure you own is a strictly timed test. A timed word-reading task with the clock removed is not a slower version of the same measure; it is an accuracy measure with no norms of its own.
The spelling instruments and oral reading fluency measures already covered here all sit on one side or the other of this line, and the line runs through kits, not between them.
Standard 3.11 adds a requirement that almost nothing in this area meets. If a developer recommends a different time limit, that limit should be established by piloting, "rather than simply allowing test takers a multiple of the standard time without examining the utility of the arbitrary implementation of multiples of the standard time."
Time and a half is a convention. It is not a published, validated administration condition for the measures discussed here.
DIBELS publishes a table, and it is shorter than you would guess
The DIBELS 8th Edition Administration and Scoring Guide, 2026 Edition, from the University of Oregon, is the clearest published example of a publisher doing this properly. It states the default first:
"The DIBELS 8th Edition measures are designed to be used unmodified with all students. They have been validated with thousands of students using the DIBELS 8th Edition standardized procedures. Interpretation of student scores is only informative when students have been assessed in this standardized way."
Then it gives two criteria for using anything on its approved list. An accommodation should be used only when "an accurate score is unlikely to be obtained without the accommodation," and/or when it is specified in a documented plan for that student.
The guide's own test of whether a change is warranted is worth keeping: use it only "if there is evidence that without them, the assessment would be measuring something other than the intended reading-related skill."
Table 2.3, in the chapter on administration, lists what is approved: a quiet setting, breaks between measures, assistive technology such as hearing aids and assistive listening devices, administration directions in the student's home language, enlarged student materials, coloured overlays or filters or lighting adjustments, a marker or ruler for tracking, a whisper phone and a slant board.
Two details in that table are easy to miss and both matter.
First, approval is per measure, not per student. The table is a grid, and the cells are not all filled. Phonemic Segmentation Fluency, which has no student materials at all, carries no approval for enlarged materials, overlays or a tracking marker — because there is nothing to enlarge, tint or track.
Second, the guide draws a line around the font that is sharper than most practitioners would guess. The published forms are locked, and the guide says plainly that "changing the font type invalidates the measures." Font size may be changed as an accommodation for low vision. Font face may not. The guide explains why: the typeface was chosen for the particular forms of certain letters.
And then the part that most accommodation conversations never reach — the record:
"When approved accommodations are used, the examiner should mark an 'A' on the front cover of the testing booklet."
Anything else is a modification by definition: "Anything an assessor does that is not listed in the standardized scoring and administration and is not an approved accommodation falls under the category of a modification." The guide names five examples — extending the time on a probe, repeating practice items, giving extra models of the task, adding to or changing the directions, and offering unapproved prompts or feedback.
"When unapproved accommodations or modifications are used, the examiner should mark an 'M' on the front cover of the testing booklet. Scores are not valid in these cases and should not be entered in a data system or interpreted in relation to DIBELS 8th Edition benchmark goals and norms."
That is an unusually direct sentence from a publisher, and it is the standard against which to read every other policy below. Not "interpret with caution." Do not enter it.
Acadience and DIBELS disagree about the same accommodation
Acadience Reading began life as DIBELS Next, and the two instruments remain close relatives. Their accommodation policies are not close at all.
Acadience publishes its lists in its help centre, separately for K–6, for grades 7–8 and for Pre-K. The approved list for K–6 is four items: enlarged or larger-print materials for visual impairment, coloured overlays or filters or lighting adjustments for visual impairment, assistive technology such as hearing aids for hearing impairment, and a marker or ruler where a student cannot otherwise demonstrate the skill.
That is a shorter list than DIBELS 8th Edition's, and the difference is not cosmetic. Directions given in a student's primary language, which DIBELS approves outright, is on Acadience's unapproved list. So is untimed administration, with Acadience's own reason attached: it measures accuracy only, not fluency.
The consequence Acadience attaches to its unapproved column is as blunt as Oregon's. Scores from an approved accommodation "can be reported and interpreted as official Acadience Reading scores." Scores from an unapproved one "should not be treated or reported as official Acadience Reading scores," cannot be compared to other Acadience scores or to benchmark goals, and are useful only for measuring that individual student's growth.
Braille gets its own treatment, which is the honest answer rather than the convenient one: growth can be measured, scores can be compared with other braille-tested students, and they should not be reported as directly comparable to the print version. For anything the help centre does not settle, it refers examiners back to the assessment manual itself.
The Pre-K list is different again, and this is the detail that should end any assumption that a publisher has one policy. For Acadience Reading Pre-K, only two accommodations are approved: assistive listening devices, and assessment spread across two or more sessions for children with challenging behaviours.
Enlarged materials and overlays — approved at K–6 — are not on the Pre-K list. Pointing to the correct letter or picture instead of answering aloud is named as unapproved.
One publisher. Three products in the same family. Three different approved lists, and a change that is official in one and unofficial in another.
aimswebPlus draws its line at the page count
Pearson's aimswebPlus guidance on accommodations and modifications, written for one state's reading-screening programme, is the most carefully reasoned document in this group. It defines both words in the publisher's own terms, and its definition of the line is the sentence to remember:
"The distinction between an accommodation and a modification is not the change itself, but whether the change alters the test's intended measurement and design."
Its worked example is a good one. Reading the text of a maths measure aloud can be an accommodation, "because it provides access to test content without altering the math skills being assessed." Reading the text of a passage on a reading measure aloud provides the same access and "would significantly alter the expectation of the reading skills being measured."
Then come the tables, and they are where the real specificity is. Three findings are worth lifting out.
Large print has a threshold, and it is physical. The tables permit reproducing stimulus pages at increased font size, then add: "Large changes affecting page count may invalidate norms." The limit is not the point size. It is whether the material still fits the page the way it did when the norms were collected.
Braille is available and still breaks the percentile. Pearson can supply a braille version of several fluency measures, and using it "is only valid for measuring individual performance but would invalidate national percentile scores." Both halves of that sentence are true at once, and reporting only the first half is how a score ends up overstated.
Sign language splits by direction. Signing may supplement oral directions on most measures. Signed responses are a different matter: on a phonological awareness task, the guidance notes the test "requires students to orally produce letter sounds," so a signed response is not an appropriate accommodation for that measure. The change is permitted on the input side and not on the output side, because the output is the construct.
The guide also distinguishes flexibility that is not an accommodation at all. Most administration and scoring rules specify waiting three to five seconds for a student to respond, and using that range according to your knowledge of the student "ensures a consistent yet individualized approach across sessions without being considered an accommodation."
Untimed measures let students work at their own pace by design. Knowing which of your adjustments were never accommodations in the first place is half of this problem.
And on documentation, one line that applies to every instrument in this post: "Any deviation from the standard administration procedures must be documented and carefully considered when interpreting test results."
Translation is a modification in every document that addresses it
The aimswebPlus guidance permits repeating directions or answering a student's questions in their native language, with or without an interpreter, as an approved accommodation. Translating test content is a different act:
"Due to differences between languages (e.g., letter-sound correspondence and difficulty level of words is frequently not equivalent across languages), translation or bilingual administration of test items or reading passages is a modification that invalidates direct interpretation of national percentiles."
Standard 3.12 of the Standards makes the same point from the developer's side: a translated and adapted test requires its own evidence of validity for each linguistic and cultural group it is used with, and Standard 3.13 says a test "should be administered in the language that is most relevant and appropriate to the test purpose."
This is the same ground as the question of whose norms a multilingual student is being compared to, and it resolves the same way. An on-the-spot translation is not a shortcut to a comparable score. It is the construction of an untested instrument, in the room, without norms.
The Woodcock-Johnson hands the decision to you, in writing
Riverside Insights takes a different approach from Oregon's table. Its support documentation points examiners to Chapter 3 of the WJ IV Examiner's Manual, which "includes a section on general accommodations based on the population an examiner may work with" — including, per Riverside's own help article, individuals with attentional and behavioural difficulties, and young children.
And then it places the decision squarely with the examiner: "the use of accommodations is reserved for the examiner based on their clinical judgment."
That autonomy comes with two obligations stated in the same breath. The examiner must determine "whether the procedures used during their administration have been altered," and — the sentence to write on the inside of the kit lid — "derived scores may not be valid for tests that were administered significantly different from the standard procedures."
Riverside also requires the examiner to "indicate on the Test Record what modifications were made," and to include in the report "how the modifications may have impacted the validity of scores."
So three publishers, three places to put the mark: the booklet cover, the test record, the report. All three want it somewhere.
The clearest statement in the whole literature is in a sensory-impairment bulletin
The most useful single document found for this post is not an accommodation policy at all. It is Lynne Jaffe's 2025 assessment service bulletin, Use of the Woodcock-Johnson V for Populations With Sensory Impairments, published by Riverside Assessments.
It restates the distinction cleanly. Accommodations "may alter the way the examiner presents a test so that its content is accessible to the examinee." Modifications "may change the content of the test and, accordingly, the construct that is being assessed."
Then it gives the most concrete example of a modification anywhere in these documents. Signing a verbal attention task does not slightly degrade it:
"The ability being tested changes from auditory working memory to visual working memory with no relation to the norms."
That is what a modification actually is, described without euphemism. Not a score to interpret cautiously. A different ability, measured against a sample that was never asked to do it.
The bulletin is equally direct about the endpoint practitioners most want to avoid: where the deviation is significant, or substantially related to the impairment, "a score should not be reported." It also notes that interpreters almost always introduce deviation — signed instructions "will almost always deviate from standardization instructions" — while refusing the lazy conclusion, since this "will not necessarily invalidate the usefulness of the test or scores."
And it closes with the sentence that belongs in every report written under these conditions: "the most important information does not come from just the scores alone, but from qualitative analysis."
Changing the mode is a deviation too, and the publisher says so
The same logic governs the question of where the student is sitting, which is why remote administration belongs in this discussion rather than beside it. Pearson's own telepractice guidance for the WIAT-4 does not present the arrangement as equivalent:
"Telepractice is a deviation from the standardized administration."
The reason is the same reason as everything above: "the normative data were collected via an in-person assessment," and "this test was not standardized in a telepractice mode, and this should be taken into consideration." The guidance asks examiners to follow in-person procedures "as much as possible," to have more confidence where there is direct evidence of task equivalence and "greater caution" where only indirect evidence exists.
It also makes the practical constraints explicit in a way accommodation documents rarely do. One subtest "can only be administered on Q-interactive." Without a response booklet, several composites simply become unavailable — "if using the response booklet is not feasible, the composite score selection is impacted." A change in conditions does not only cost you confidence in a score. Sometimes it costs you the score.
On the record, Pearson asks for the same thing Riverside does: state in the report that the test was administered this way, record "any and all atypical events," and treat "documentation of all considerations, procedures, and conclusions" as a professional responsibility.
What the published record does not tell you
Four gaps, named plainly, because the honest answer to several reasonable questions here is that nobody has published one.
The PRO-ED batteries do not publish their rules. For widely used instruments such as the CTOPP-2, the administration and accommodation guidance exists only inside the printed Examiner's Manual, which is a separately purchased component of the kit. The product pages say nothing about accommodations.
There is no help-centre article and no public guidance document to read. If you are working from one of these kits, the rules are in your manual and only in your manual — which is an argument for reading that chapter before the session rather than after it.
Comparability evidence is largely missing, by the Standards' own admission. Standard 3.11 asks developers to document validity evidence for a changed test "when sample sizes permit," and acknowledges that for many disability groups the samples do not permit it, recommending "small-sample qualitative studies or professional judgments" instead. The field's reference text is telling you that the evidence you would want often does not exist.
The measurement literature agrees, and says so about decision-makers. Benjamin Lovett's instructional module on testing accommodations, published through the National Council on Measurement in Education, reports that testing entities "often do not conduct sufficient research on accommodation provision, usage, and effects," and that "it is often unclear or unknown whether score comparability is maintained."
Then comes the finding most relevant to anyone who has sat in a team meeting: "validity issues are often not addressed by accommodation decision makers." The module's own position is that "there should be specific evidence of need before providing an accommodation."
Nobody rates accommodation policy, so you cannot shop for it. The word "accommodation" does not appear anywhere on NCII's academic screening tools chart.
The chart rates classification accuracy, technical standards and usability features — training, administration format, scoring, technology — and none of its columns reports whether a tool publishes an accommodation policy at all, let alone whether that policy is any good. The chart is still the right starting point for evidence ratings. It just does not answer this question.
And one genuine curiosity. The only published Pearson technical report found on accessibility in digital assessment, Q-interactive Technical Report 15, is about accommodations for examiners with disabilities, not examinees.
It opens by observing that "psychological tests were not designed to be accessible for examiners with disabilities," and notes that little research has examined how examiner accommodations affect examinee performance. It is a worthwhile document and it answers a question almost nobody asks — while the examinee-facing question stays in the manuals.
Flagging: the field has not settled it, and neither should you
Since a modification produces a score that means something different, the obvious next question is whether the report should say so in a way a reader cannot miss. The Standards describe the state of play rather than resolving it.
Where there is clear evidence that scores from regular and altered administrations are not comparable, "consideration should be given to informing score users, potentially by flagging the test results to indicate their special nature, to the extent permitted by law." Where there is credible evidence that they are comparable, "flagging generally is not appropriate."
And then the honest middle, which is where most real cases live:
"There is little agreement in the field on how to proceed when credible evidence on comparability does not exist."
For an educational therapist this is less fraught than it is for a large-scale testing programme, because you are usually writing a narrative report rather than issuing a score to a gatekeeper. But the principle transfers.
Describe the conditions, describe what they do to the interpretation, and do not let a standard score sit in a table with no note attached when the administration that produced it was not the administration the norms came from.
What this means for your own report
Three clauses of the AET Code of Ethics (Revision #9, 2025) bear on this directly, and none of them requires any legal interpretation.
II.I.B asks educational therapists to "select and use appropriate assessment instruments, recognizing their limitations with respect to reliability, validity, and bias." A score from a modified administration is a limitation with respect to validity, and recognising it means writing it down.
II.I.C asks them to "use only those assessment instruments for which they have been adequately trained." Adequate training in an instrument includes its administration rules. The chapter on accommodations is part of the instrument.
II.I.D asks them to "seek interpretation of assessment data from professionals in related fields (e.g. medical, psychological, speech/language, neuropsychological)." Where a sensory or physical access question is driving the deviation, that clause is the route.
The AET Fact Sheet is also worth rereading in this context. Its "Educational Therapists DO NOT" list includes "administer cognitive, intelligence, or psychological tests (unless otherwise qualified to do so)" — the parenthetical is part of the clause — while the same document describes educational therapists as skilled in administering formal and informal educational assessments "for which they are both qualified and trained to use, and/or interpreting assessment results."
Which instruments you may buy at all is a separate question, covered in the post on achievement test qualification levels.
A sequence, not a decision
What follows is the order of questions, not an answer. No article can tell you whether to make a particular change for a particular student, and this one is not trying to.
Ask what this measure is measuring. Not the subtest name — the construct. Speed, accuracy, orally produced sounds, comprehension of written text. The answer decides everything after it.
Find the manual's own section before the session. Chapter 3 in the Woodcock-Johnson. Table 2.3 in DIBELS 8th Edition. The help centre for Acadience. The accommodations tables for aimswebPlus. If your instrument has no published section, that is the finding, and your manual is the only copy.
Check whether the change is on an approved list for this measure, not for this instrument. The grids are per measure, and the cells are not all filled.
Ask whether the change touches the construct. Input or output. Access or ability. Pearson's formulation is the most usable one: not the change itself, but whether it alters the test's intended measurement and design.
Decide what you will report, before you administer. If the honest answer is that no standard score can be reported, that decision is cheaper to make now than after the scoring.
Mark the record wherever the publisher asks — booklet cover, test record, report narrative — and say in the narrative what you changed and what it does to the interpretation. This is the same discipline as any other session documentation, and it is the part of the work that protects both the student and you.
Where this sits in the rest of the record
Several posts here converge on this one, and each supplies a piece of it.
The piece on achievement test qualification levels establishes that publishers, not statutes, decide who may purchase an instrument — and the same documents decide how it may be given.
The piece on remote administration found that publishers permit changes of mode with explicit limits attached, which is exactly the shape of the accommodation question.
The piece on standard error of measurement explains why a single score was never a point in the first place, which matters more once an administration has deviated.
The piece on outdated test norms made the same argument from a different direction: a norm table is a claim about a sample, and the claim has conditions.
The piece on grade equivalent scores showed what happens when a derived score is read as though it meant something it does not.
The piece on assessing multilingual students covers the language question in full, including why translation cannot be improvised.
The piece on progress monitoring tools and the one on decision rules both depend on administrations being identical across time points — which an inconsistently applied accommodation quietly destroys.
The piece on assistive technology covers the tools that are an intervention in instruction and a modification in assessment, depending only on which room you are in.
The pieces on dyslexia screening, phonological awareness, reading comprehension, vocabulary, listening comprehension and written expression each describe instruments whose accommodation rules differ from one another.
And the pieces on informed consent and psychoeducational evaluation cover the surrounding obligations, including what a family is told about what was done and why.
If you are new to the scope questions underneath all of this, the comparison of educational therapists, learning specialists and executive function coaches is the place to start.
Common questions
Does extended time invalidate a test score?
It depends entirely on whether speed is part of what the test measures. On an untimed measure, students working at their own pace is the design. On a fluency measure, the Standards are explicit that extra time is inappropriate where speed is part of the intended construct, and DIBELS 8th Edition lists extending the time on a probe as a modification whose scores should not be entered in a data system.
Can I read a test aloud to a student?
The publisher documents answer this by asking what is being measured. Pearson's aimswebPlus guidance treats reading the text of a maths measure as a possible accommodation, because it supplies access without altering the maths skill, and treats reading a passage on a reading measure as a change that "would significantly alter the expectation of the reading skills being measured."
Is giving directions in a student's first language an accommodation or a modification?
The two most widely used early reading screeners disagree. DIBELS 8th Edition lists administration directions in the student's home language, currently available in Spanish, as an approved accommodation. Acadience Reading lists directions in the student's primary language as unapproved, meaning the resulting score should not be reported as an official Acadience score. Check the manual for the instrument you are actually holding.
How do I document an accommodation?
Wherever that publisher asks you to. DIBELS 8th Edition wants an "A" on the front cover of the booklet for an approved accommodation and an "M" for a modification.
Riverside asks examiners to indicate modifications on the Test Record and discuss their effect on validity in the report. Pearson asks for a statement in the report and a record of atypical events. The Standards expect the steps to appear in the test report.
Can I enlarge a test's print?
Often, with limits that are more specific than practitioners expect. DIBELS 8th Edition permits a change of font size for low vision but states that changing the font type invalidates the measures. Pearson's aimswebPlus tables permit reproducing stimulus pages at increased font size and warn that "large changes affecting page count may invalidate norms."
If a score is not valid, should I report anything at all?
Yes — just not a standard score presented as comparable. The publisher documents consistently point the same way: Acadience says unapproved-accommodation scores can still track an individual student's growth, Riverside's sensory bulletin says the most important information comes from qualitative analysis rather than scores alone, and in some cases says a score should not be reported.
A described administration with no normative claim is more useful, and more defensible, than a number that quietly overstates what you know.
If this is the work you do
Illuminate Educational Therapy Group writes for practitioners — educational therapists, learning specialists and executive function coaches — about the parts of the work that the public-facing literature skips.
If you want posts like this one when they are published, join the educator list.




Comments