Assessment Modification Without Changing the Target
How one OpenSciEd transfer assessment became a family of versions - open response, scaffolded organizers, word banks, digital forms, checklists, rubrics, and alternate pathways - without abandoning the science I actually wanted students to demonstrate.
THE BIG IDEA
In this chapter, you will see...
FOCUS | WHAT TO EXPECT |
|---|---|
THE ORIGINAL TRANSFER TASK | OpenSciEd asks students to apply a healing model to growth: child and adult growth plates, interacting body systems, plant cells, and plant growth. |
A SCAFFOLD LADDER | Several paper versions move from open response to task decomposition, sentence stems, word banks, and partially completed responses. |
DIGITAL ALTERNATIVES | Google Forms versions vary question type, writing demand, amount of prompting, and the degree to which students must generate versus recognize an answer. |
CHECKLISTS & RUBRICS | The criteria for mastery are made visible before and after the assessment so students know what a complete explanation actually requires. |
WHAT STAYS CONSTANT | Cell division, evidence, structure/function, system interaction, growth over time, and transfer to a plant remain the conceptual center. |
THE REAL DESIGN PROBLEM | How far can I modify access before I have changed what the assessment is measuring? |
What stays protected
Growth as a transfer phenomenon • cells and cell division • bone growth plates • blood and nutrient delivery • skin/muscle/nerves growing with bone • plant cells • evidence from images • structure/function • interacting systems • explanation from unfamiliar evidence
What is still being tested
Which supports are access supports versus content supports; when multiple choice becomes recognition rather than transfer; whether one common rubric can span several versions; how much scaffolding should disappear over time; and whether a single “test” is even the right way to think about this assessment.
Return to top ↑Lesson 13: Assessment Modification Without Changing the Target
The lesson where differentiation stops being a worksheet problem and becomes an assessment-design problem.
Lesson 13 is the transfer task. That matters because transfer is exactly where assessment modifications become difficult.
If I simplify the science too aggressively, I may no longer be assessing transfer. If I leave the original task untouched for every student, I may end up measuring reading load, writing stamina, executive function, language production, or the ability to organize several ideas at once more than I am measuring the biology.
Over the years, I have made a lot of versions of this assessment. Some are only slightly different from the original. Others deconstruct the prompt, add sentence frames, add word banks, convert sections to multiple choice, provide graphic organization, or shift the task into Google Forms. I also use checklists, rubrics, answer keys, and self-reflection tools. Looking at them together, I do not see a pile of accommodations. I see a record of trying to answer one recurring question:
What is the minimum barrier I can remove while still requiring the student to show the science?
The assessment is supposed to be hard - but for the right reason
A transfer assessment should create productive difficulty. Students should have to recognize familiar science inside a new context. They should have to decide which ideas matter, connect evidence to a claim, and explain how cells and systems interact. I am not trying to remove that difficulty.
What I am willing to remove is unnecessary friction: a giant undifferentiated prompt, too many simultaneous directions, a writing demand that overwhelms the science, a vocabulary-retrieval burden that is not itself the target, or a response format that prevents a student from showing what they know.
What OpenSciEd is trying to assess
OpenSciEd is explicit that Lesson 13 is a transfer task. Students apply what they figured out about healing to a related phenomenon: growth. They compare x-rays of children and adults, notice gaps at growth plates, and use their healing model to explain how bones grow. The lesson then extends the transfer again to plant growth. The teacher edition also draws a clear boundary: the purpose is not to teach the full biology of development, genetics, or cartilage. The purpose is to see whether students can apply the cellular and systems model they already built.
That boundary is useful because it clarifies what should survive every modified version. The student does not need a more detailed theory of growth. The student needs to use familiar unit ideas in a new setting.
The original structure is already broader than one question
The lesson begins with a short shared investigation of child and adult bones, then shifts to individual assessment. Students explain what happens at a growth plate, how other systems must grow along with bone, what plant microscope images reveal, how a plant changes over time, what cellular process must support those changes, and what additional evidence would be useful.
Modification #1: Separate the construct from the format
The first step in modifying this assessment is deciding what I am actually assessing. For me, the non-negotiable construct is not “write several long paragraphs.” It is the science underneath those paragraphs.
ASSESSMENT FEATURE | IS THIS THE TARGET? | MY DECISION |
|---|---|---|
Recognize that bones are made of cells | Yes | Protect it. |
Use cell division to explain increasing tissue | Yes | Protect it. |
Explain how blood supports growing tissues | Yes | Protect the interaction. |
Connect skin, muscle, nerves, and bone through structure/function | Yes | Protect the systems reasoning. |
Use microscope/plant evidence to infer cells and division | Yes | Protect the evidence-to-claim move. |
Generate every term from memory | Sometimes | Reduce if vocabulary retrieval blocks the larger explanation. |
Write one long unstructured response | No | Change freely. |
Hold six subparts in working memory at once | No | Chunk or list the expectations. |
Spell or phrase every response conventionally | No | Allow models, labels, short responses, or oral clarification when appropriate. |
This distinction sounds obvious, but it changes the entire modification process. Instead of asking, “How do I make this easier?” I can ask, “Which barriers are irrelevant to the claim I want to make about this student’s understanding?”
A modification is strongest when it changes access more than it changes evidence.
One target, several ways to reach it
That is why I am comfortable allowing words, pictures, or a combination. A labeled model can reveal structure, function, interaction, sequence, and mechanism just as clearly as a paragraph - sometimes more clearly. I still require labels and captions because the model has to communicate scientific reasoning, not just look correct.
Modification #2: Build a scaffold ladder instead of one “modified test”
The paper versions I have made form a useful continuum. They are not perfectly linear, but they show several distinct ways to reduce cognitive and language load while keeping the same phenomenon and many of the same ideas.
LEVEL | WHAT CHANGES | WHAT STUDENTS STILL HAVE TO DO |
|---|---|---|
Open response | Large prompts, few cues, broad response spaces. | Select relevant ideas, organize them, explain the mechanism, and transfer independently. |
Task decomposition | The prompt is broken into explicit criteria: bone structure, filling the gap, blood, timeframe; then skin, muscle, nerves, and system interactions. | Generate the science, but with the organization problem reduced. |
Guided organizer | Subquestions appear directly under each task: What changes? Why? What evidence? | Answer smaller reasoning moves and assemble them into the larger explanation. |
Sentence stems | Partial syntax and conceptual slots are supplied. | Complete the relationship: structure → function → change → reason. |
Word bank / fill-in | Key vocabulary is available and response generation is narrower. | Choose appropriate concepts and place them into a causal explanation. |
Recognition-heavy alternate | More multiple choice and short responses; fewer extended explanations. | Demonstrate core concepts with reduced writing, though the amount of true transfer evidence decreases. |
The critical point is that these versions do not all provide the same quality of evidence. A student selecting “cell division” from four choices has demonstrated something different from a student independently using cell division to explain an unfamiliar growth phenomenon. That does not make the first response useless. It means I need to be honest about what each version can support me in claiming.
The scaffold ladder changes the source of difficulty
The more scaffolded versions are not simply shorter copies. Each one moves a different amount of work from the student to the page. That is the design variable.
STUDENT MUST GENERATE | PAGE PROVIDES | TRADE-OFF |
|---|---|---|
Open response: vocabulary, organization, causal sequence, system connections, evidence, and explanation. | Almost nothing beyond the phenomenon and the question. | Strongest evidence of independent transfer; highest language and executive-function demand. |
Task decomposition gives students the categories but not the answers. Sentence frames provide the grammatical and conceptual skeleton. Word banks reduce retrieval. Fill-in-the-blank formats narrow the possible response even further. Multiple-choice versions shift some of the task from production to recognition.
This is why “same standard” does not mean “same test”
Two students can work toward the same standard while producing different kinds of evidence. My job is to know what evidence I am getting. If a student requires word banks to access the content, I can still learn a great deal about whether they understand that cells divide, blood delivers needed materials, and body systems must grow together. I just should not pretend that the response demonstrates the same level of independent retrieval and organization as the open version.
Modification #3: Make success criteria visible before the assessment
One of the simplest modifications may also be one of the most defensible: tell students what a complete response needs to contain.
The checklist I use does not provide the answers. It names the dimensions of the answer. For the growth-plate item, students are reminded to address bone structure, how the gap fills, blood interaction, and timeframe. For the surrounding systems, they have to address skin, muscle, nerves, structure/function, and blood. The plant section asks for a claim, image evidence, cell division, visible structural change, and the conditions needed for growth.
This is a useful distinction. A checklist can reduce the executive-function demand of remembering the whole rubric while preserving the intellectual demand of answering it. In fact, for many students it makes the assessment more valid because an omitted component is less likely to reflect “I forgot that part of the prompt” and more likely to reflect what the student can actually explain.
The rubric also changes feedback
The companion rubrics let me score the same conceptual components after the assessment and leave targeted comments such as “good job with cell division,” “too general about blood,” “you missed the root cells dividing,” or “your timeframe was too short.” That moves feedback away from a mysterious overall grade and toward the anatomy of the explanation.
A rubric should not merely justify the score after the fact. It should reveal what scientific completeness looks like.
Modification #4: Use digital forms for a different kind of evidence
The Google Forms versions are not just electronic copies. They let me rebalance question types. I can combine selected response, short answer, sentence frames, and a few higher-value open responses. That makes it possible to sample more concepts without requiring every concept to be expressed in a paragraph.
One version, for example, asks students to identify which blood component transfers nutrients and oxygen, explain why the growth-plate gap disappears, predict what happens when blood vessels do not work properly, connect muscle and nerve function to growth, and then answer open-ended questions about blood supporting all systems. The second half uses microscope images and a growing plant to ask for claims, evidence, cell division, and similarities/differences between plants and humans.
Recognition is useful - but it is not the same as explanation
Multiple choice can be an excellent accessibility tool and a useful way to check component knowledge. It can also reveal misconceptions efficiently. But I do not want an assessment of transfer to become only recognition. If every hard idea appears inside the answer choices, students may succeed by identifying familiar language rather than constructing a model of the phenomenon.
That is why I prefer a hybrid. Selected-response items can establish whether the pieces are present; a smaller number of open responses can show whether the student can connect the pieces.
Modification #5: Sometimes the content really does need to change
Not every modification can stay at the access level. Some students need a substantially reduced assessment. The lower modified Google Form does exactly that: it uses direct multiple-choice questions about growth plates, cell division, nutrients, plant cells, roots, and plant growth. It also extends into simple comparisons such as what roots do and what plants and bones both need to grow.
That version is not equivalent evidence to the advanced or open-response version. I do not think it helps anyone to pretend otherwise. What it can do is give a student a fairer opportunity to demonstrate foundational unit understanding when the full transfer task would produce mostly missing work, random responses, or evidence about reading/writing limitations rather than biology.
Modification has to be tied to the purpose of the decision
If I am deciding whether a student can independently transfer a systems model, I need independent transfer evidence. If I am deciding whether the student understands that living things are made of cells, cells divide, blood transports materials, and growing structures need resources, a more supported format may be appropriate. The same student may need both kinds of information at different times.
QUESTION TO ASK | WHY IT MATTERS |
|---|---|
What claim am I trying to make from this score? | The assessment must actually provide evidence for that claim. |
Which barriers are part of the science? | Those should usually stay. |
Which barriers come from language, writing, memory, or organization? | Those are candidates for support. |
Does the support cue the process or supply the answer? | Cues are usually safer than answers when transfer is the target. |
Can I triangulate with another source of evidence? | Notebook work, discussion, models, previous tasks, and oral explanation can prevent one format from becoming the whole story. |
A modified assessment does not have to prove everything. It has to be clear about what it does prove.
What I will be experimenting with this year
I have more versions of Lesson 13 than I need. That is useful historically, but not necessarily useful operationally. My next step is to turn the pile into a cleaner assessment system.
EXPERIMENT | QUESTION I AM TESTING | SUCCESS WOULD LOOK LIKE |
|---|---|---|
Three core versions | Can I replace six paper versions with a clear advanced/core/scaffolded progression? | Teachers and students can immediately see what changes and what remains constant. |
Common mastery map | Can every version point back to the same small set of science targets? | Different formats still produce interpretable evidence against common criteria. |
Checklist before test | Does giving the checklist in advance improve completeness without inflating content knowledge? | Fewer omitted components; explanations are still genuinely student-generated. |
Hybrid digital assessment | Can selected response handle component knowledge while open response handles transfer? | Less writing fatigue with no loss of systems reasoning. |
Model option | Can labeled/captioned models substitute for some written explanations? | Students communicate mechanism and interaction accurately with less language burden. |
Post-assessment reflection | Does asking students to estimate performance and explain why improve metacognition? | Students can name what they understood, missed, and should study next. |
What I am protecting, compressing, and still questioning
PROTECT | COMPRESS | STILL TESTING |
|---|---|---|
Transfer to growth; cell division; systems interaction; structure/function; evidence; child/adult x-rays; plant microscopy and plant growth; at least some student-generated explanation. | Redundant writing; giant prompts; repeated directions; vocabulary retrieval when it is not the target; long-form responses for every concept; grading language that hides the specific misconception. | How much recognition can coexist with transfer; when word banks cross the line into answer-providing; whether all versions should share one score scale; how quickly scaffolds should be faded. |
The bigger point
Differentiation in assessment is not a separate project from good assessment design. It forces the teacher to become precise about the construct. What exactly am I measuring? What evidence would convince me? Which difficulties are essential, and which are accidental? Once those questions are answered, accommodations and modifications become less arbitrary.
Return to top ↑The best modification is not the one that makes the student most comfortable. It is the one that lets the student show the most valid evidence of the target.
LESSON 13 | CONDENSED MODIFICATION GUIDE
Unit 6.6 • Injuries & Healing
PROTECT: cells • cell division • growth plates • blood/nutrient delivery • skin/muscle/nerves • structure/function • interacting systems • plant cells • image evidence • transfer
MOVE | WHY IT MATTERS | |
|---|---|---|
1 | IDENTIFY THE CONSTRUCT | Write down the science you actually need evidence for before modifying anything. |
2 | REMOVE IRRELEVANT BARRIERS | Chunk prompts, reduce writing, allow models, or provide organization when those skills are not the target. |
3 | CHOOSE A SCAFFOLD LEVEL | Open response → criteria list → organizer → sentence stems → word bank → recognition-heavy alternate. |
4 | MAKE CRITERIA VISIBLE | Use a checklist or rubric so students know what scientific completeness requires. |
5 | KEEP SOME GENERATION | Even digital/modified versions should include at least one place to explain, model, or connect ideas when transfer is the goal. |
6 | GRADE THE SCIENCE YOU ASKED FOR | Use comments tied to specific concepts and interactions, not just an overall score. |
Access support vs. construct change
USUALLY ACCESS | WATCH CAREFULLY |
|---|---|
Breaking one long prompt into smaller questions | Supplying the key causal relationship inside the question |
Providing a checklist of required components | Providing the actual scientific statements students are supposed to generate |
Allowing a labeled/captioned model | Accepting an unlabeled picture that does not communicate reasoning |
Reducing writing volume | Replacing all transfer with recognition |
Providing vocabulary when vocabulary recall is not the target | Using a word bank so narrow that the response becomes mechanical |
Return to top ↑Modify the barrier. Protect the evidence.
Lesson 13: A Practical Assessment-Modification Workflow
A way to choose among the many versions without turning differentiation into guesswork.
Start with five non-negotiable targets
TARGET | WHAT COUNTS AS EVIDENCE |
|---|---|
1. Growth requires more cells | Student uses cell division/copying to explain increasing tissue. |
2. Systems grow together | Student connects bone growth to skin, muscle, nerves, and/or blood vessels. |
3. Structure supports function | Student explains why a system must change in order to keep performing its function. |
4. Cells need materials | Student explains a delivery role for blood/nutrients or identifies analogous needs in plants. |
5. Transfer from evidence | Student uses x-rays, microscope images, or plant changes to support an explanation in a new context. |
Then choose the lightest support that is likely to work
IF THE MAIN BARRIER IS... | TRY THIS FIRST |
|---|---|
Organizing a complex response | Checklist or task decomposition. |
Knowing how to start a sentence | Sentence frames. |
Retrieving vocabulary | Word bank. |
Sustaining extended writing | Shorter responses, labeled models, or hybrid digital items. |
Reading/processing the original prompt | Chunked questions with one demand at a time. |
The full transfer construct itself | Use a genuinely modified/alternate assessment and be explicit that it provides different evidence. |
CURRENT RECOMMENDATION I would keep one common transfer phenomenon and one common mastery map, then maintain three intentional versions: (1) open/advanced, (2) core with visible criteria and limited sentence support, and (3) scaffolded/alternate with reduced writing and more recognition. The checklist and rubric can sit beside all three. |
What not to cut
Do not cut every opportunity to explain. Do not remove the unfamiliar context entirely. Do not turn cell division, system interaction, or evidence into optional enrichment. Those are the assessment.
What to cut first
Cut redundant writing, repeated directions, unnecessary vocabulary retrieval, giant response boxes, and the expectation that every target must be demonstrated through the same modality.
The question is not, “Did every student take the same test?” The question is, “What valid evidence did each version allow the student to produce?”
Sources & design notes
Primary curriculum source: OpenSciEd, Unit 6.6, Lesson 13 Teacher Edition, “How is the process of growing similar to healing?” Version dated 2/9/22. OpenSciEd identifies the lesson as a transfer task and explicitly limits the assessment boundary to applying previously developed cellular and systems ideas to growth.
Teacher-created materials reviewed for this chapter: multiple paper versions of the Growth Summative Assessment (V1-V6), answer key, key-figures page, checklist, student/teacher rubrics, and several Google Forms versions including advanced, 504, lower-alternate, and lower-modified formats. These artifacts show changes in prompt decomposition, modality, writing demand, sentence support, vocabulary support, item type, and scoring/feedback.
Return to top ↑