📖6 min read · 1,131 words
This is the plain-language version. Same content, written for a quick read.
You sit closer to students than anyone else in the building. You hear the confused whisper. You see the moment a student checks out. That means you hold real evidence about whether students are learning, if you know how to gather it and get it to the teacher.
This brief covers two tools: opportunities to respond (OTR) and checks for understanding (CFU). These are terms teachers and BCBAs use. Knowing them, and using them, makes you a more effective partner.
---
An opportunity to respond is any moment when a student produces an active response: says an answer, writes on a whiteboard, holds up a card, points to a choice, gives a thumbs up or down.
More responses = more practice + more evidence + less off-task behavior.
| Content type | Target per student per minute |
|---|---|
| New content | 4–6 responses (aim for about 80% correct) |
| Review / practice | 8–12 responses (aim for about 90% correct) |
These are targets to aim at, not a bar you're failing. Even one response every two or three minutes beats you doing all the talking. But if a student goes five minutes without producing a response, that's a flag.
Counts: student says, writes, points, gestures, or manipulates something in response to a question.
Does not count: student watches you demonstrate, listens to your explanation, or copies a model.
Choral response. Ask the whole group, then give a signal ("Three, two, one,") for everyone to answer at once. Good for one-right-answer questions.
Individual whiteboards. Every student writes, then holds up on cue. You see every answer at once.
Wait time. After asking a question, count silently to five before calling on anyone. This gives students, especially slower processors, time to form a real answer. It's harder than it sounds. Practice it.
Cold calling with a system. Don't always call on students who raise their hands. That gives you evidence about one or two students. Use popsicle sticks or a name card to call randomly so everyone knows they might be next.
Turn and share. Pair students; have them tell each other the answer. Circulate and listen. You get evidence from multiple students at once.
Frequent short turns. Instead of one long response, ask for many short ones. "What's step one? Good. Step two?" gives you more data than "Walk me through the whole thing."
---
A check for understanding is a deliberate pause to find out if students actually got what you just taught, before you move on.
The bad version: "Does everyone understand?" Students who don't understand are the least likely to say so. This tells you nothing.
Hinge questions. A question where the wrong answer tells you the specific misunderstanding. Example: "The character felt nervous because (a) she had done this before, (b) she had never done this before, (c) she forgot her homework." A student who picks (a) has a different problem than one who picks (c). The wrong answer is the data.
Exit-ticket style. Before you move to the next section, ask one or two questions in writing. Sort responses quickly: got it / almost / not yet. Use that to decide whether to reteach.
Targeted cold call. Call on the student you're least sure about, not the most confident one. Their answer is the most useful data point.
Ask for a different form. "Can you draw that?" "Can you put that in your own words?" "What would happen if …?" Surface-level correct answers can hide shallow understanding. Asking for a different form surfaces it.
Treat errors as data. When a student makes an error, notice the type: careless, consistent pattern, vocabulary confusion, or conceptual gap? The pattern across errors is what you report to the teacher.
- Most students got it → move on.
- One or two didn't → flag for the teacher; reteach in the next available moment.
- Most didn't → stop. Reteach the key step. Check again. Tell the teacher before the next session.
---
This is where paras lose the most value: the observation stays in your head and never reaches the teacher in a form they can use.
Specific, not general.
- ✗ "She struggled."
- ✓ "She got steps 1–3 independently but broke down at regrouping every time. When I showed that step she got the rest."
Behavioral, not interpretive.
- ✗ "He doesn't care."
- ✓ "He stopped responding after about eight minutes. When I gave him a shorter task he re-engaged."
Pattern, not single event. One error is noise. The same error across three sessions is a signal worth reporting.
What you tried and whether it worked. "I tried the sentence frame and she could complete it. Without the frame she got stuck." This tells the teacher what scaffolds to plan for, or start fading.
Right after the session if something important happened. In a short written note if the teacher isn't available. One or two specific, dated sentences is enough.
---
Accepting "yes" when you ask "do you understand?" Follow up with a question that requires demonstration: "Can you show me the next step?"
Answering for the student. When you answer for them, you've taken away the OTR and collected false evidence. Wait. Prompt with the least intrusive cue first. The student's inability to respond is still useful data.
Only checking at the end. If students were lost in minute five, you spent thirty minutes building on a gap. CFUs should happen during the lesson, not just at the end.
Reporting only problems. Teachers also need to know what scaffolds are working, which students hit mastery, and what can be faded. Positive evidence matters too.
---
| Term | What it means |
|---|---|
| OTR | How many times per minute each student actively responds |
| CFU | A deliberate check before moving on |
| Hinge question | A CFU where the wrong answer reveals a specific misunderstanding |
| Formative assessment | Ongoing evidence-gathering used to adjust instruction, not a test |
| Wait time | Deliberate pause after a question before accepting answers (aim: 3–5 seconds) |
| Cold call | Calling on a student who didn't raise their hand, using a fair random system |
| Choral response | All students answer at the same time on a signal |
---
- Brief 04.02 (Prompting Hierarchies) and 04.03 (Prompt Fading): the scaffolds OTR lives inside
- Brief 04.06 (Error Correction): what to do when a CFU shows an error
- Brief 06.03 (Prompt Level Data) and 06.05 (IEP Progress Monitoring): data recording formats
- Brief 04.16 (Running a Small Group): where most of these moves get used
Formative Assessment Support and Opportunities to Respond
How to elicit real evidence of understanding and get it to the teacher
For paraprofessionals working directly with students in any instructional setting
The Education Endowment Foundation's Teaching and Learning Toolkit rates feedback as one of the highest-impact interventions in education: an average of six months' additional progress. But feedback only works when it's based on real evidence of where a student actually is. And in a one-teacher classroom, the teacher often cannot gather that evidence for every student while also delivering the lesson.
That's the gap the para fills. This brief names two specific tools for filling it: opportunities to respond (OTR) and checks for understanding (CFU).
---
An opportunity to respond is any moment when a student is asked to produce an active response: saying an answer aloud, writing on a whiteboard, holding up a card, pointing to a choice, giving a thumbs up or down. The research base on OTR comes largely from applied behavior analysis and direct instruction, where it is one of the most consistent predictors of student engagement and learning.
Why it matters: the more frequently students respond, the more data you have, the more practice they get, and the less time is available for off-task behavior. OTR is not about drilling. It's about making learning visible and active.
Research guidelines (from Alberto & Troutman; Carnine; the PBIS technical assistance center) suggest:
| Content type | Target OTR rate |
|---|---|
| New material (acquisition) | 4–6 responses per minute per student, aiming for about 80% correct |
| Practice / review material | 8–12 responses per minute per student, aiming for about 90% correct |
These are targets, not minimums. Even one student response every two to three minutes is dramatically better than a para doing all the talking. But if a student goes five minutes without producing a response, that's a flag.
Counts:
- Student says an answer aloud (even just a word or short phrase)
- Student writes on a whiteboard, card, or paper
- Student points to a picture, letter, or number
- Student gestures (thumbs up / down / sideways; head nod; finger count)
- Student acts out or manipulates a material in response to a question
Does not count:
- Student watches the para demonstrate
- Student listens to an explanation
- Student copies a model (unless a response to a cue follows immediately)
- Para answers the question for the student
The distinction matters: passive tasks can look like learning but produce almost no evidence and almost no practice.
Choral response. Ask the group a question and have all students respond simultaneously on a signal ("Everyone, what's the answer? Three, two, one, "). Works well for facts with one correct answer; not useful for open-ended thinking.
Individual whiteboards or response cards. Each student writes, holds up on cue. Every student produces a response at once; you see every answer. Cheap, reusable, no prep.
Wait time. After asking a question, pause three to five seconds before calling on anyone. This dramatically increases the quality and frequency of responses, especially for students who process slowly. Counting to five silently is harder than it sounds: practice.
Cold calling with a system. Calling on the same students every time (or only those who raise their hands) produces zero evidence about everyone else. Use a random selection method (popsicle sticks, a name spinner, a card deck) so every student knows they may be called.
Turn and share. Pair students, pose a question, have them tell each other the answer. You circulate and listen. You get evidence from multiple students at once; students get a low-stakes rehearsal before answering publicly.
Frequent short turns. Instead of one long student response, structure for multiple short ones. "What's step one? Good. What's step two?" generates more responses, more data, and more practice than "Can you walk me through the whole process?"
---
A check for understanding is a deliberate pause to probe whether students have actually grasped what you just taught or practiced, before moving on. It's distinct from OTR in purpose: OTR is ongoing response during instruction; a CFU is a targeted assessment at a decision point.
The classic bad version: "Does everyone understand?" (Students who don't understand are the least likely to say so. This generates social information, not learning information.)
Hinge questions. A question with answer options designed so that the wrong choice reveals a specific misconception. Example: "The character feels nervous because (a) she has done this before, (b) she has never done this before, (c) she forgot her homework." If a student picks (a), that's a specific misconception. She's reading the meaning of the words wrong, not just missing the inference. A good hinge question tells you not just who doesn't know but why they don't know.
Exit-ticket format CFU. At the end of a section (not just end of session), ask students to respond to one or two questions in writing before moving on. Review quickly, sort into "got it / almost / not yet," and use that to decide whether to reteach or advance.
Targeted cold call. After a worked example or explanation, call on a student you're uncertain about, not the most confident one. Their response is the most useful data point.
Representational check. Ask the student to show understanding in a different form: "Can you draw that?" "Can you put it in your own words?" "Can you give me an example?" "Can you tell me what would happen if …?" Surface-level correct answers can mask shallow understanding; representation in another form surfaces it.
Error correction as CFU. When a student makes an error, the error is data. What type of error is it? Careless? Consistent pattern? Vocabulary confusion? Conceptual gap? The error-correction procedure (see Brief 04.06) tells you how to respond; the pattern across errors tells the teacher where instruction needs to go.
If most students got it: move on.
If one or two students didn't: flag for the teacher; consider a quick reteach in the next available moment; don't hold the group back if others are ready.
If most students didn't: stop. Don't advance. That's the point of the CFU: to catch this before it compounds. Use the error correction procedure, reteach the key step, then check again. Report to the teacher.
---
This is where paras lose the most value: the observation stays in the para's head and never reaches the teacher in a usable form. The teacher makes instructional decisions based on whatever they observed during whole-class instruction, which is a small, biased sample, while you were sitting on richer data.
Specific, not general. "She struggled" → "She got steps 1–3 independently but broke down at regrouping every single time, once I showed the regrouping step she got the rest." The teacher can act on the second sentence.
Behavioral, not interpretive. "He doesn't care" → "He stopped responding after about eight minutes; when I gave him a shorter task he re-engaged." Behavioral descriptions don't put you in the role of diagnosing; they give the teacher material to work with.
Pattern, not single event. One error is noise; the same error three sessions in a row is a signal. Track simple tally data across sessions so you can say "This has happened four of the last five times."
What you tried and whether it worked. "I tried giving her the sentence frame and she was able to complete it; without the frame she got stuck." This helps the teacher know what scaffolds are working, and whether to plan for them or fade them.
Immediately after a session if something important happened, don't wait for planning time.
In a written log if the teacher doesn't have immediate availability: one or two sentences, specific, dated. Brief 06.05 (IEP Progress Monitoring) and Brief 06.03 (Prompt Level Data) cover data recording formats in more detail.
In the language the teacher uses. If the teacher talks about "mastery criteria" and "prompt levels," use those terms. If they talk about "grade-level standards" and "benchmark," use that. Speaking the same language as the teacher makes your observations more usable.
---
Asking "do you understand?" and accepting yes. Students who don't understand often say yes, either because they're embarrassed, because they don't know what they don't know, or because they want to keep the interaction moving. This is not a CFU. Follow with a question that requires demonstration: "Great, can you show me the next step?"
Doing the thinking for the student. When a student is stuck, the impulse to help is real, but if you answer for them, you've eliminated the OTR and also collected false evidence (the student looked like they succeeded). Wait. Prompt with the least intrusive cue first. If the student still can't respond, that information is valuable.
Checking only at the end. Checking understanding at the end of a session tells you what happened at the end. If students were lost in minute five, you delivered thirty minutes of instruction on a crumbling foundation. CFUs should interrupt the lesson, not just cap it.
Reporting only negatives. Teachers need to know what's working too: which scaffolds are effective, which students have hit mastery, which skills can be faded. Positive evidence shapes instructional decisions just as much as negative evidence does.
---
When your supervising teacher or BCBA uses these terms, this is what they're asking for:
| Term | What it means |
|---|---|
| OTR (opportunities to respond) | The number of times per minute each student is actively producing a response |
| CFU (check for understanding) | A deliberate probe to confirm understanding before advancing |
| Hinge question | A CFU question designed so wrong answers reveal specific misconceptions |
| Formative assessment | Ongoing evidence-gathering used to adjust instruction, not a test, not a grade |
| Wait time | The deliberate pause after a question before accepting responses (target: 3–5 sec) |
| Cold call | Calling on a student who hasn't volunteered, with a fair random system |
| Choral response | All students respond simultaneously on a signal |
| Acquisition vs. fluency vs. generalization | Stages of learning: does the student know it at all (acquisition), know it automatically (fluency), know it across contexts (generalization)? Different stages call for different OTR targets |
Knowing these terms, and being able to use them with your supervising teacher, makes the collaboration more efficient and positions your observations as usable professional data.
---
- Brief 04.02 (Prompting Hierarchies) and 04.03 (Prompt Fading): the instructional scaffold that OTR lives inside
- Brief 04.06 (Errorless Learning and Error Correction): what to do when the evidence shows an error
- Brief 06.03 (Prompt Level Data) and 06.05 (IEP Progress Monitoring): data recording formats for reporting to teachers
- Brief 04.16 (Running a Small Group): the setting where most of these moves are most used
Reading is useful, but recall is where it sticks. Three short scenarios, low-stakes, no scoring. About 3 minutes. You can stop any time.
Start the practice set →You do instructional work all day — and you need a clear line between supporting and teaching, with…
Prompting is most of what you do — and the difference between skilled and well-intentioned prompting…
You prompt students through skills they can't yet do alone — and fading those prompts is where actua…
You run programs that another adult also runs — and consistency between you is the difference betwee…