TL;DR: Alternative assessment formats can preserve rigor when they require comparable evidence of the intended learning. Define the knowledge or skill being assessed, identify extra demands introduced by the format, and decide which demands belong to the learning goal. Then use common criteria, appropriate access support, and a clear review process. An oral explanation, written response, and visual model need not look alike, but they must make the relevant reasoning visible. Formal accommodations still follow the institution's procedures.
The disputed assignment
A fictional introductory environmental science course asks students to explain why a lake's oxygen level can fall after a nutrient increase. The original assessment is a handwritten essay completed in forty minutes without notes. A student asks to submit an audio explanation with an annotated diagram. The instructor worries that allowing the alternative would lower the standard or make grading incomparable.
That concern contains several questions. Is handwriting part of the learning objective? Is rapid recall essential? Does the essay assess scientific reasoning, written communication, or both? Could an audio response reveal the same causal understanding? Could a polished diagram hide a missing explanation? Answering these questions requires a definition of the assessment's purpose, not an immediate verdict about whether one format is easier.
The design hearing below is an original fictional discussion among an instructor, an assessment colleague, and an accessibility specialist. It is not a report of an actual accommodation case. The examples illustrate educational design; they do not establish legal requirements or authorize an instructor to override an institution's accommodation process.
CAST's UDL and Assessment resource distinguishes the intended construct from features that introduce unrelated barriers. Here, a construct means the knowledge or skill the assessment is meant to reveal. The challenge is to remove irrelevant demands without accidentally removing part of the capability the course actually promises to teach.
The instructor's opening statement
Instructor: “Students must explain a causal process, use evidence appropriately, and identify a condition that would change the outcome. I also want the work to be their own.”
Assessment colleague: “Those are several aims. Which are learning outcomes, which are conditions for obtaining trustworthy evidence, and which are conventions inherited from last year's assignment?”
The distinction matters. Scientific explanation is a learning target. Independent authorship concerns how confidently the instructor can attribute the response to the student. Handwriting may simply be the historical method used to collect it. Treating all three as equally essential would make the current format impossible to question, even if another format supplied better evidence of the learning target.
The instructor rewrites the outcome in observable terms: students must connect the nutrient change to biological activity and oxygen demand, distinguish an inference from an observation, and explain a relevant limiting condition. The hearing does not need to settle every scientific detail of the fictional task. It needs to identify what a satisfactory response would have to demonstrate.
This outcome is more specific than “understand the topic” and less format-bound than “write an essay.” It creates a basis for comparing possible responses. The next step is to ask whether writing itself is part of the course's intended capability. If it is, the assessment may need to preserve a written component or assess scientific explanation and writing separately.
Evidence requirement one: the causal links must be explicit
A written essay can contain the required terms without explaining their relationships. An audio response can sound fluent while making the same mistake. A diagram can display arrows whose meaning is unclear. No format guarantees that the intended reasoning is present merely because the student produced a plausible artifact.
The shared criterion should therefore identify the causal links that must be explained. A response needs to say how one process affects another and under what assumptions. The instructor can ask students using a diagram to label the arrows and provide a short explanation of the most important relationship. Students using prose or audio must make that relationship equally explicit.
The criterion should not require every student to reproduce the instructor's preferred wording. Equivalent reasoning can be expressed through different language or organization. The marker's task is to identify whether the explanation is scientifically defensible and sufficiently complete, not whether it resembles a model answer sentence by sentence.
A practical design check is to create two intentionally weak sample responses in different formats. One might use all the vocabulary but reverse a causal direction. Another might provide a correct sequence while omitting the mechanism. If the rubric catches the problem in one format but overlooks it in the other, the criterion or the assessment instructions need revision.
These samples should be original instructional examples, not disguised copies of a student's work. Their purpose is to test the assessment design before grades depend on it. The design team can then revise the evidence requirement rather than discovering the mismatch during a contested marking decision.
Evidence requirement two: separate the explanation from the presentation
Accessibility specialist: “Which aspects of presentation affect the claim you want to make about learning?”
Instructor: “I need to understand the answer. I do not need an accomplished broadcaster or graphic designer.”
That answer suggests a useful boundary. An audio response must be intelligible enough to evaluate, but vocal polish should not become an unannounced criterion. A diagram must communicate relationships, but decorative quality should not earn scientific credit. A written response must make its meaning accessible, but handwriting neatness should not dominate unless it is genuinely relevant to the learning outcome.
The boundary is not always simple. In a professional communication course, adapting an explanation to an audience may be central. In a language course, written expression may be the construct itself. The same support that removes an irrelevant barrier in one course could change the target in another. The instructor must reason from the stated learning goal rather than adopt a universal rule about preferred formats.
A common rubric can therefore contain separate dimensions. Scientific reasoning may be shared across all formats. Communication may assess whether the explanation is coherent and appropriate to the audience, with mode-specific instructions. Technical production quality may be excluded unless the course explicitly teaches and assesses it. Separating these dimensions prevents a polished artifact from masking weak understanding.
The false shortcut: equal minutes must mean equal difficulty
The instructor proposes that all students receive forty minutes, regardless of format. Equal time is administratively simple, but it does not establish equivalent evidence. Producing a diagram, recording an explanation, and handwriting an essay can involve different planning, motor, editing, and technical demands.
Conversely, allowing unlimited time for one format does not automatically preserve comparability. If the assessment is intended to examine timely reasoning under a defined constraint, changing the constraint may alter the claim. The correct response is to identify whether speed belongs to the construct and why, not to assume that time must always be identical or always flexible.
For this fictional course, the stated outcome concerns causal explanation rather than rapid performance. The instructor decides that a narrow timed handwritten task was not necessary for this particular claim. The assessment can use a broader completion window with a short follow-up question to clarify reasoning. That is a design decision for this example, not an instruction to abolish timed assessments generally.
The follow-up also needs accessible design. A spontaneous oral question might introduce a new barrier if the original concern involved speech, processing time, or interaction conditions. The instructor should decide how clarification can occur without simply moving the irrelevant demand to another stage. Written clarification, a scheduled conversation, or an approved support arrangement may be appropriate depending on the goal and institutional procedures.
A common rubric can describe different artifacts
The hearing drafts four shared criteria: causal accuracy, use of evidence, treatment of a limiting condition, and clarity of the explanatory structure. Each criterion describes observable features of the reasoning. None awards points merely for selecting a particular medium.
For causal accuracy, the highest level might require correct relationships with a justified explanation of the mechanism. A middle level might identify the main relationships but leave an important link unexplained. A lower level might confuse correlation with mechanism or reverse a relationship. The exact descriptors must match the course's taught content and expected level.
The Carnegie Mellon Eberly Center's rubric guidance describes rubrics as explicit criteria and performance descriptions. In this example, the important design choice is to define the evidence at each level before examining whether a particular format makes it easier or harder to observe. A rubric cannot fix an unclear construct simply by adding more rows.
Mode-specific instructions then tell students how to expose the common evidence. Audio submissions might identify the evidence being discussed at each point. Diagrams might include a key and explanatory annotations. Written responses might use headings or another coherent organization. These instructions are routes to the same evidentiary requirement, not separate hidden standards.
Before release, the instructor marks one original sample in each mode. A colleague marks the same samples independently. Their disagreement can reveal that a descriptor rewards visual density or verbal fluency rather than the intended reasoning. Revising the rubric at this stage protects students from discovering the actual standard only after submission.
The objection about effort
Instructor: “What if the audio option takes less effort?”
Assessment colleague: “Less effort doing which part?”
An assessment can impose effort that contributes little to the intended learning. Removing that effort may improve the quality of evidence. A student who spends less time overcoming a handwriting barrier may have more capacity to demonstrate scientific reasoning. The assessment should not preserve an irrelevant obstacle merely because someone else has already learned to manage it.
However, effort cannot be dismissed entirely. If one option permits the student to avoid a central intellectual demand, it may no longer assess the same capability. A diagram that merely labels components would be insufficient if the goal is to explain causal relationships. The design should require those relationships explicitly rather than assuming that any alternative product counts.
The useful comparison is therefore between cognitive demands relevant to the outcome. What must the student infer, select, justify, and explain? Those demands should remain appropriately challenging across the available routes. Surface differences in production time do not settle that question, but they can prompt a closer examination of whether a route accidentally omits a required step.
Access support does not need to be a public negotiation
An instructor may offer flexible formats to the entire class while individual students still require formal accommodations. Universal design and individualized support are related but distinct. A menu of options does not automatically meet every student's needs, and an approved accommodation does not need to become a classroom debate about fairness.
The instructor should work through the institution's established accessibility process and share only the information needed to implement the arrangement. Students should not have to disclose personal details to classmates to justify a different method. The course can explain its general assessment principles without identifying who uses which support or why.
The CAST assessment tips emphasize aligning assessment with goals, reducing unnecessary barriers, and clarifying expectations. The fictional hearing translates those principles into an evidence question: what can the instructor reasonably infer from the response after the support is provided? That question is more useful than assuming that support either invalidates all comparison or solves every access issue.
If a requested change appears to alter an essential learning outcome, the instructor should discuss the specific conflict with the appropriate institutional specialists. A vague statement that “this is how we have always assessed it” does not explain the conflict. Equally, the instructor should not independently promise an arrangement that contradicts formal requirements without the relevant review.
The technology introduces its own conditions
An audio option may require a microphone, a quiet place, an accessible recording interface, and a way to upload a file. A visual option may require software or physical materials. A written digital response may involve a keyboard, screen reader compatibility, or document-format constraints. Offering a choice can create new barriers if these practical conditions are ignored.
The design team therefore specifies low-complexity submission routes and explains how students can obtain support. It does not assume that every student owns a current device or has private space at home. A campus recording room may help some students while remaining inaccessible to those who cannot attend during its opening hours.
File limits and technical requirements should serve a clear purpose. A very small upload limit can force unnecessary compression work. A proprietary format can make access dependent on paid software. The instructor should choose formats supported by the institution and test the actual submission path with the tools students are expected to use.
Technical failure also needs a recovery rule. If an upload fails before the deadline, what evidence of the attempt is useful, whom should the student contact, and how will the work be accepted? The rule should be communicated in advance. An assessment of scientific reasoning should not silently become an assessment of navigating an unreliable upload system.
A short design matrix makes the tradeoffs visible
The hearing compares the three routes without declaring one universally best:
| Route | Evidence it can expose | Demand that needs attention |
|---|---|---|
| Written explanation | Explicit argument, evidence references, revision | Writing demands beyond the stated science outcome |
| Audio explanation with notes | Verbal causal reasoning, emphasis, clarification | Recording access, speech demands, locating evidence |
| Annotated visual model | Relationships, boundaries, conditional pathways | Unexplained arrows, visual design demands, accessible alternatives |
The matrix is a starting point for design rather than a guarantee of equivalence. An instructor must still examine the actual prompt, rubric, and student work. “Audio” describes a medium, not a complete assessment method. The conditions under which it is produced and the evidence it must contain matter just as much.
The team also identifies what the matrix does not answer. It does not determine a student's formal accommodation. It does not establish that the three routes have equal statistical difficulty. It does not prove that every marker will apply the criteria consistently. Those questions require additional processes or evidence appropriate to their importance.
This restraint prevents a useful design tool from becoming an unsupported claim. The immediate goal is a defensible assessment with transparent expectations, followed by review of how it works. Equivalence is a reasoned design objective that should be examined, not a label automatically conferred by putting several options on a page.
The authorship question belongs in the design
The instructor remains concerned that an unsupervised recording or polished diagram might include substantial outside assistance. That concern also applies to many written assignments. Choosing handwriting may change the available assistance, but it does not by itself define what independent understanding means or how the assessment will demonstrate it.
The course should state which resources and forms of collaboration are allowed. If students may use editing tools, examples, or AI assistance, the instructor needs to identify what work they must still perform and explain. If particular assistance is prohibited, the rule should be clear and connected to the learning purpose rather than introduced after a suspicious-looking submission appears.
A short process note can sometimes help: which sources were consulted, what changed during revision, and which part remained uncertain? But process documentation can also create workload and access demands. Require it when it supplies useful evidence, not as an indiscriminate surveillance layer attached to every assignment.
A carefully chosen follow-up question may reveal whether the student can apply the explanation to a changed condition. That question should assess the intended reasoning and respect approved support. The goal is to obtain relevant evidence of understanding, not to stage an adversarial performance in which stress becomes an unannounced criterion.
Review the first set of submissions before standardizing the method
After the assessment, the instructor reviews the distribution of responses and the kinds of marking difficulty encountered. If one mode repeatedly conceals a required link, the next prompt may need clearer instructions. If a mode requires excessive technical troubleshooting, the submission route may need simplification. These are design findings, not immediate proof that students choosing the mode know less.
Comparing average grades across modes can raise questions, but it cannot by itself establish which mode is easier. Students selected their routes, and those groups may differ in preparation, preferences, access needs, or other characteristics. A grade difference in this setting is descriptive evidence with several possible explanations.
The instructor can inspect matched features of the work: whether causal links are explicit, whether evidence is accurately used, and whether the limiting condition is addressed. Marker notes can identify recurring ambiguity in the rubric. Student feedback can reveal practical barriers. None of these sources is complete, but together they can guide a more focused revision.
A change should be proportional to the problem. If audio submissions are hard to navigate, a brief outline may solve the issue without imposing a written essay on top of the recording. If diagrams omit mechanisms, require explanatory annotations rather than banning diagrams altogether. The remedy should address the missing evidence while preserving the access benefit where possible.
Students need enough information to choose a route
A choice is not useful when students discover its demands only after committing to it. The instructor provides a small example of the expected evidence in each mode and explains the practical requirements. Students can then distinguish a route that suits their way of expressing understanding from one that merely looks easier in the assignment menu.
The course also states whether and when students can change routes. A student who discovers that recording is inaccessible should have a clear way to discuss an alternative, while the instructor needs a manageable submission process. The rule should distinguish a practical barrier from a request to restart after receiving substantive evaluative feedback. Those situations raise different fairness and workload questions.
Choice itself can impose planning demands. Some students may prefer a recommended default rather than evaluating several unfamiliar production methods. Offering a clear standard route alongside meaningful alternatives can reduce that burden. The default should be a helpful starting point, not a signal that students using another route will be judged as less serious.
Finally, the instructor checks that examples do not accidentally privilege one format. A polished model essay beside a rough audio sketch may imply unequal expectations even when the rubric is shared. Comparable demonstrations should make the reasoning visible at a similar level of completeness, while preserving the natural differences between the media.
The hearing's decision
The fictional course adopts three routes for this assessment, with common criteria for the scientific explanation and clear mode-specific instructions. It retains the institution's formal accommodation process, provides accessible submission support, and plans a review after the first use. It does not claim that every assessment in the course should use the same options.
The instructor also discovers that the original timed essay had combined several aims without naming them. In a later assignment specifically designed to assess written scientific communication, writing may appropriately become a central requirement. The decision about access will then begin again from that different construct rather than mechanically copying this assessment's menu.
That is the central discipline: define what the evidence must establish, distinguish necessary demands from inherited conventions, and inspect whether the chosen routes actually reveal the learning. Rigor resides in the quality of the intellectual work and the defensibility of the inference. A format earns its place by helping the instructor see that work clearly.