TL;DR: A concept vote should determine the next teaching move. Give learners an individual attempt, ask them to compare reasons rather than answer letters, and inspect what happens when they tackle a new question independently. A better second vote can reflect understanding, persuasion, or both. The teacher still needs to resolve misconceptions and check whether students can use the distinction in a changed case. Design the question sequence and the response branches before class, rather than improvising from a bar chart alone.
Scene one: thirty students choose three different explanations
The teacher shows two sealed, rigid boxes with the same external volume. One contains more mass than the other. Both are held completely underwater in the same tank, away from the bottom and sides. Which box experiences the larger upward buoyant force from the water? The choices are the heavier box, the lighter box, or equal forces.
Students first decide alone and write a short reason. The correct answer under these stated conditions is equal forces: the displaced volumes and fluid density are the same. The restraint matters. The question is about the force exerted by the water, not whether either box will accelerate when released. This is an original classroom design; all votes, dialogue, and classroom outcomes below are illustrative rather than observations from a study.
Now imagine that the votes divide among all three options. The instructor has several choices. They can announce the answer, repeat the relevant equation, ask students to persuade a neighbor, or investigate which quantity the class thinks determines buoyancy. Only the last two possibilities use the disagreement as information, and even those require care. A lively discussion can make an incorrect explanation more popular.
The aim of peer instruction is not to outsource the answer key to a majority. It is to make competing models available for examination. The teacher's responsibility shifts from delivering the next paragraph of content to deciding what kind of reasoning the class needs next. That decision depends on the explanations behind the votes, not just the height of the bars.
What the research supports, and what this lesson adds
Catherine Crouch and Eric Mazur's 2001 account of ten years of Peer Instruction reports improved conceptual and quantitative performance in introductory physics courses using the approach. The courses and their implementation also changed over time. The paper supports taking the method seriously; it does not establish a guaranteed gain for every question, teacher, age group, or subject.
A useful methodological distinction appears in Smith and colleagues' 2009 study of undergraduate genetics. They followed discussion and a repeat vote with a different question addressing the same concept, answered individually. Their findings supported a learning explanation rather than simple copying alone in that setting. The transferable design idea is to ask for evidence beyond agreement on the original item, not to assume that every classroom reproduces the study's results.
The physics examples here use the standard relationship between buoyant force and displaced fluid weight, described in OpenStax's treatment of Archimedes' principle. The classroom questions, response branches, and sample dialogue are original. Their purpose is to show how an instructor can distinguish several plausible student interpretations without copying a published assessment or claiming that this exact lesson has been evaluated.
That distinction between evidence and design should remain visible in course planning. A research supported method does not validate every activity carrying its name. The teacher must still choose an appropriate conceptual target, write an unambiguous prompt, support participation, and inspect the result. A discussion task can fail because the question was weak even when discussion itself is a reasonable instructional choice.
Write the disagreement you want students to resolve
The learning target is narrower than “understand buoyancy.” Students should distinguish the water's upward force from an object's weight, and distinguish equal external volume from equal submerged volume. The first box question addresses the first distinction while holding the second constant. The options are useful because they can arise from different models rather than random arithmetic errors.
A student choosing the heavier box may think the water supplies whatever force is needed to support an object. That model can work for freely floating objects at rest and fail when the same reasoning is applied to a held submerged object. A student choosing the lighter box may be comparing upward acceleration rather than upward force. A student choosing equal forces may know the displacement relationship or may simply assume that identical looking boxes must behave identically.
The teacher should anticipate those possibilities without assigning them automatically to everyone choosing a particular letter. An answer option is not a diagnosis. A learner might have misread “force” as “net force,” missed the statement about volume, or guessed. That is why the short written reason matters. It supplies information that the vote alone cannot provide.
Before class, ask a colleague to solve the question literally. Do the boxes change shape? Can water enter? Are they floating or held below the surface? Is the relevant quantity force, acceleration, or displacement? Removing accidental ambiguity is not making the question easy. It ensures that disagreement concerns the intended concept rather than an unstated physical condition.
Keep the first vote private enough to mean something
Students need a chance to form their own account before encountering the room's apparent consensus. A private response can be collected electronically, on a card, or through a written note. The device is secondary. What matters is whether the instructor can see a distribution and whether students have committed to a reason they can examine during discussion.
Do not display a nearly unanimous histogram while students are still deciding. Late responders may treat it as evidence about the expected answer. Similarly, asking the most confident student to speak before everyone has responded can turn the first vote into a measure of that student's influence. The initial attempt is useful because it captures a starting model, not because private thought must always precede all collaborative learning.
The instructor can ask for a reason in ordinary language: “What feature of the situation determines your answer?” This is less demanding than a polished explanation but more revealing than “How confident are you?” Confidence can be recorded separately if it will inform a teaching decision. It should not replace the reason or become another number gathered without a purpose.
Participation arrangements also need an alternative for students who cannot use the main response system. A technical failure should not make someone's reasoning invisible. A paper response or a quiet written submission can preserve the instructional function without stopping the room. Decide in advance how those responses enter the teacher's view rather than treating them as exceptions after the vote closes.
Choose a teaching branch from the reasons in the room
Branch A: the votes are divided and the reasons are usable
The teacher sees several distinct accounts in the written responses. This is a good opportunity for comparison. The prompt should ask students to examine a relationship, not merely change a partner's vote. “Explain which quantity the water responds to in this situation, then identify a condition under which your answer would change” gives the pair something to test.
A sample exchange might begin this way:
Lena: I chose the heavier box because it needs more support.
Omar: But it is being held underwater. Does needing support determine the force from the water, or the extra force from whatever holds it?
Lena: I was imagining the boxes floating. If both displace the same amount of water here, perhaps the water's contribution is equal.
The exchange is productive because it finds the hidden condition in Lena's model. Omar does not merely announce a formula. He separates two physical arrangements that Lena had merged. The teacher can listen for this kind of distinction while circulating, rather than counting how many groups appear animated.
Pairs should not be required to reach consensus. The second response remains individual. A learner can disagree with a partner and still benefit from articulating the disputed assumption. Requiring a common answer may hide unresolved reasoning and make the more assertive participant's view look like collective understanding. The teacher wants to know what each student can now defend.
Branch B: almost everyone chooses the same wrong answer
A strongly concentrated incorrect vote can mean the class shares a misconception, but it can also indicate an unclear question or a missing prerequisite. Sending students into discussion without changing the information available may simply rehearse the same mistaken explanation. First inspect a few reasons and ask a neutral clarification question.
For the boxes, the instructor might ask students to identify the water volume displaced by each fully submerged box. If they cannot connect external volume to displacement, a brief diagram or physical demonstration may be needed before discussion. If they correctly identify equal displacement but still argue that the heavier box has more buoyancy, the conversation can focus on why they are treating weight as the controlling variable.
This is a limited intervention, not a full lecture inserted before the class has had any chance to reason. The teacher supplies the missing representation or isolates the conflict, then asks students to reconsider. The intervention should leave a meaningful inference for them to make. Otherwise the next vote merely confirms that they heard the teacher's answer.
Do not adopt a universal percentage rule for when discussion is worthwhile. The same distribution can arise from very different problems. A class may have enough conceptual resources to resolve a disagreement even when few initially select the correct option. Another class may need explicit instruction despite a superficially balanced vote. The explanations and the prerequisite knowledge should guide the branch.
Branch C: almost everyone is correct, but the reasons are thin
A high initial success rate can conceal a weak question. If students answer “equal” because both boxes are drawn at the same size, they may fail as soon as the drawing changes. Ask for a counterfactual before moving on: which change would make the forces unequal while the boxes remain fully submerged?
One student might propose changing a box's mass. Another might propose changing its external volume or placing it in a fluid of different density. These alternatives reveal whether the correct original answer came from the intended relationship. The teacher can ask students to explain why one change matters and another does not, making the distinction visible without repeating the original item.
If the reasons are already strong, do not force a long discussion because the lesson plan includes one. Move to a more demanding contrast or use the time elsewhere. Peer instruction is a way to respond to student thinking, not a ritual whose steps must occupy the same minutes regardless of what students know.
A good easy question can still serve a purpose when it establishes a shared reference for a harder case. The teacher should know that purpose. If it contributes no diagnosis, no useful contrast, and no preparation for the next inference, the question is probably expendable. More votes do not automatically produce more learning.
Listen for the moment discussion becomes a contest of authority
A group can sound confident while avoiding the conceptual issue. One person may invoke a previous grade, a familiar formula, or a claim that the teacher always chooses the middle option. Those moves can settle an answer socially without settling it intellectually. The instructor should redirect attention to the situation the question describes.
A useful intervention is: “Before deciding who is right, each person should explain what the other person's model predicts.” This makes listening observable. If a student cannot restate a partner's view, more argument may add volume without improving comparison. The instructor can then ask for a condition that distinguishes the two predictions.
Sometimes the incorrect view is voiced more fluently. Do not reward eloquence by assuming it reflects understanding, and do not ask a quieter learner to defend the entire correct model on behalf of the class. Offer a diagram, a brief writing interval, or a request to identify one relevant quantity. Participation should make reasoning accessible rather than turn verbal speed into a proxy for knowledge.
The teacher can collect an anonymous explanation to discuss publicly. This allows examination of a misconception without assigning its author a lasting classroom identity. Treat the explanation as a useful model to test, not evidence that its author lacks ability. A course that invites revision should make it possible to revise without embarrassment becoming the main event.
Scene two: the second vote improves, but the lesson is not finished
After discussion, students vote again. Suppose more choose equal forces. The teacher now has evidence that the response distribution changed. They do not yet know whether each learner can reconstruct the reasoning independently or whether the explanation will survive a changed condition. Announcing that the class has mastered buoyancy would go beyond the evidence.
Ask for one explanation and one unresolved objection. The explanation should identify equal displaced volumes in the same fluid. The objection may concern what happens when the boxes are released. Resolve that explicitly: the water's force is not the same quantity as net force, and the two boxes have different weights. A clear resolution prevents the discussion from leaving two incompatible accounts side by side.
The instructor's explanation can be short because it addresses the specific conflict that emerged. It need not repeat the entire textbook section. Show which assumption made the attractive wrong answer plausible, where that assumption fails, and which relationship controls this case. Students should be able to locate the correction in their own reasoning.
Then close the original vote. Keeping the same question open until everyone selects the expected letter can create an appearance of complete success while obscuring who still needs support. Move to a different question and let its responses test the model. A learner who changes their mind during discussion has made a useful move; the next task asks what that move enabled.
A matched question and a contrasting question answer different needs
For a closely matched question, show a single rigid box held fully submerged first in one liquid and then in a denser liquid. Keep the box's volume unchanged and ask in which liquid the buoyant force is larger. The intended reasoning is still about the displaced fluid, but the comparison now changes fluid density rather than object mass. Ask students to answer individually and explain the relevant change.
This is evidence about a nearby application, not proof of unlimited transfer. Students may still rely on the wording or a remembered discussion. The teacher can inspect whether they identify the density of the fluid rather than the density of the box. A correct choice paired with an irrelevant explanation remains an instructional concern.
Next introduce a contrast: the two original boxes are now freely floating at rest. Both are less dense on average than the water, and neither touches another object. Which experiences the larger buoyant force? Now the heavier box does, because each floating box's upward force balances its own weight. Equal external volume no longer establishes equal submerged volume.
The contrast is valuable because it prevents students from replacing one overgeneralization with another. “Same size means same buoyancy” worked only with a specified submersion condition. The lesson should make students attend to that condition. The teacher can ask them to write a paired statement: when the boxes are held fully submerged, compare displaced volume; when they float at rest, also account for how much of each box is submerged.
Build a response record that changes the next lesson
The instructor does not need an elaborate analytics platform. A compact record can include the initial distribution, the dominant explanations heard, the intervention used, and the reasoning shown on the independent question. Those fields preserve the teaching decision rather than merely documenting that an activity occurred.
For example, the record might state: “Many learners initially equated support needed with water force. Discussion separated restraint from buoyancy. On the different liquid question, several still changed the object's density instead of the fluid's density.” That observation suggests a precise next task. A generic note that engagement was high would not tell the instructor what to do next.
Avoid treating two votes on one question as a clean experiment demonstrating the effect of discussion. Students have seen the item twice, heard peers, received prompts, and perhaps inferred the teacher's preferred direction. The classroom record supports local instructional judgment. A causal research claim would require a design capable of separating those influences.
The independent follow up should also preserve uncertainty. If a student misses it, that does not prove discussion harmed them. They may have misunderstood the new wording or lacked another prerequisite. Read the reason, ask a bounded follow up, and choose support accordingly. The goal is to improve the next opportunity to learn, not to assign each student a permanent category from a small set of responses.
Compare three exit slips before choosing the next exercise
Consider three possible responses to the denser liquid question. They all select the larger force in the denser liquid, but they do not supply equivalent evidence:
- “More fluid mass occupies the volume displaced by the unchanged box.” This identifies the relationship the question was intended to test. The next task can challenge its boundary with a floating object.
- “Dense things have a greater upward force.” The choice is right, but the referent of “things” is unclear. Ask whether the student means the box or the liquid before deciding that the concept is secure.
- “My partner explained that the second liquid is the answer.” This is an honest account of answer acquisition. Ask for a prediction when the liquid order is reversed and for a reason independent of the partner's authority.
The third response is not a moral failure. Learners often borrow an explanation before they can use it independently. The teaching question is what support would turn that borrowed answer into a usable relationship. A brief comparison of what stays constant across two diagrams may be more useful than another exhortation to think harder.
The instructor could group the next activity by the unresolved distinction rather than by whether the answer letter was correct. Some students examine fluid density; others distinguish buoyant force from net force; others tackle the floating contrast. Groups should remain temporary and task specific. Once the relevant reasoning is visible, the learner moves on rather than inheriting a permanent label from an exit slip.
This reading of written work takes time. Limit the collection to information you can actually use, and sample explanations during class if a full review is impractical. A smaller number of carefully read reasons can justify a better teaching decision than a large archive of responses nobody examines. Plan the review effort as part of the lesson, rather than treating assessment as costless data collection.
Adapt the conversation when the room cannot support quick spoken pairs
A large online class can preserve the same reasoning sequence through an initial private answer, a short written exchange, and an independent follow up. The instructor should make the exchange prompt specific and keep the original question available. A chat stream full of answer letters creates a poor substitute for comparing models because it makes agreement visible while hiding the reasons.
In a multilingual classroom, allow students to sketch, annotate a diagram, or write a brief explanation before speaking. The conceptual target is the relationship among force, fluid, and volume, not fluency in spontaneous debate. The teacher can ask pairs to mark the quantity they held constant and the quantity they changed. That representation may reveal reasoning more clearly than a polished paragraph.
Students who need a quieter participation route can compare two anonymous written explanations and identify the decisive assumption. This preserves the intellectual comparison while changing the social arrangement. It is not necessary to insist that every learner persuade a stranger aloud in order to take the conceptual conflict seriously.
The essential design is a sequence of evidence and response: an initial model becomes visible, a comparison challenges it, the teacher resolves the relevant distinction, and a new case tests what the learner can now do. The best next move is the one justified by that evidence, even when it differs from the activity the instructor expected to run.
Sources
- Crouch and Mazur, 2001: Peer Instruction—Ten Years of Experience and Results.
- Smith and colleagues, 2009: Why Peer Discussion Improves Student Performance on In-Class Concept Questions — original research abstract and publication record; DOI.
- OpenStax, College Physics 2e: Archimedes' Principle — the physical relationship used in the original teaching examples.