Identifying Student Misconceptions: How to Shift from Grading to Diagnosing

You mark a stack of quizzes on a Sunday evening. Three students land on the same score: 6 out of 10. You write the number, move to the next paper, and plan a revision lesson for Monday. It feels responsible. It feels thorough.

But here is the uncomfortable question. What did those three students actually get wrong? And would one revision lesson fix all three?

Most teachers already sense the answer. A mark tells you how much a student scored. It rarely tells you why. And when you cannot see the why, you end up reteaching everything to everyone, hoping something sticks. That is not a failure of effort. It is a failure of design.

This guide is about fixing that design. It draws on a Suraasa session led by Manish Pandey, Teacher Career Mentor and Associate Director at Suraasa, who walked teachers through a practical method for identifying student misconceptions through the questions we write. His method is called FIND, and by the end of this article you will know how to use it in your own classroom, whatever you teach.

Pandey was clear from the start that this is not an argument against marks. Schools need them. Parents expect them. Boards require them. As he put it, the problem is what marks cannot do, and that is something teachers can design their way out of.

If you would like to hear the full argument in his own words, including the worked fractions example, watch the session below. Manish Pandey walks through each step of FIND with real classroom data you can borrow.

The Flaw of Numerical Grading: Why Equal Scores Hide Unequal Needs

Pandey built his session around a grade 7 mathematics teacher named Mira. She gave her class a fractions quiz. Three of her students, Kabir, Sana and Riya, each scored 6 out of 10.

On a mark sheet, they look identical. In reality, they are three different learners with three different problems.

• Kabir understood the concept but made calculation slips. His thinking was sound. His execution was careless.
• Sana misapplied a rule. She carried whole-number thinking into fractions and treated the denominator as if a bigger number always meant a bigger value.
• Riya could pick the right answers, but she could not explain her reasoning. She lacked the conceptual vocabulary to say why an answer was correct.

Now picture the standard response. Mira notices the class average is middling, so she plans a full reteach of the fractions chapter. Kabir sits through forty minutes of content he already understands. Sana hears the rule again, but nobody names the specific belief that is tripping her up. Riya practises more procedures, when what she needs is language and explanation.

One lesson. Three students. Almost no impact on any of them.

This is the core problem with relying on scores alone. A number compresses very different kinds of thinking into one figure. It records performance. It does not explain it. And if you cannot see the thinking, you cannot plan the next move with any precision.

If you have ever felt that your revision lessons do not move the needle, this is likely why. You were not missing effort. You were missing evidence.

The Thermometer Analogy: Measuring Symptoms vs. Diagnosing Causes

To make the distinction stick, Pandey used a medical comparison. A thermometer tells you a patient has a temperature of 102 degrees. That is useful information. It tells you something is wrong. But it cannot tell you whether the cause is viral or bacterial, and those two causes need very different treatments.

Grading works like the thermometer. Diagnosis works like the doctor.

"Reading a temperature is not a diagnosis of a problem."
Manish Pandey

A grade tells you how much a student scored. Diagnosis tells you what kind of thinking produced the answer. One measures the symptom. The other finds the cause.

This framing also clears up a common confusion between two kinds of assessment. Summative assessment records where a student has landed. Formative assessment informs what you do next. If you want a fuller picture of how these two purposes differ, our guide on assessment for learning vs assessment of learning sets them side by side. Diagnostic assessment sits firmly on the formative side. Its job is to change what happens tomorrow.

Research supports the idea that wrong answers carry information. A study in the Journal on Mathematics Education found that analysis of student responses reveals systematic misconceptions and fragile conceptual understanding rather than random error. That phrase, rather than random error, matters. When a student gets something wrong, there is usually a reason, and the reason tends to repeat across students. Errors have patterns. Patterns can be read.

The FIND Framework: An Overview of Diagnostic Assessment

FIND gives that reading a structure. It has four steps:

Notice what is happening here. Most of the diagnostic work happens before a single student sees the question. Focus and Identify are design steps. Notice and Decide are interpretation and action steps. The quality of the second half depends entirely on the first.

That is a shift in how many of us were trained to think about assessment. We often write the question first and think about errors afterwards, while marking. FIND flips that order. You think about errors first, then write questions that will catch them.

Let us take each step in turn.

Step 1 (Focus): Why Precise Targets Yield Actionable Classroom Evidence

Pandey's first point here was blunt.

"If the target is vague, the evidence will be vague."
Manish Pandey

Consider the targets teachers commonly write. "Assess fractions." Or slightly better: "Check whether learners can compare fractions." Both sound reasonable. Neither tells you what evidence you are looking for.

What does it mean to compare fractions? Does a student need to compare fractions with the same denominator? The same numerator? Different both ways? Do they need to state the answer, or explain it? Without that precision, any question you write could be about anything, and any answer you get back is hard to interpret.

A diagnostic target names three things:

1. The concept (for example, the relationship between denominator and part size)
2. The skill (comparing two fractions with the same numerator)
3. The expected evidence (the student explains how a larger denominator creates smaller parts)

Here is the difference in practice:

The right-hand column is longer. It takes a few extra minutes to write. But every question you build from it now has a purpose, and every answer you get back can be read against a clear standard.

If this kind of target-writing feels familiar, it is closely related to backward design, where you decide what evidence of learning looks like before you plan instruction. Our guide to backward design lesson planning goes deeper on that process. The verbs you choose also matter. "Explain" asks for something very different from "identify", and Bloom's taxonomy can help you pick the verb that matches the thinking you actually want to see.

Step 2 (Identify): Mapping the Six Common Misconception Categories

Once your target is sharp, you predict how students might miss it. This is the heart of identifying student misconceptions before they show up on paper.

Pandey offered six categories to guide that prediction:

Go back to Mira's three students and you can place each of them. Kabir's calculation slips are a process error. Sana's whole-number thinking is a misapplied rule. Riya's inability to explain her reasoning points to partial understanding and vocabulary gaps. Same score, three different categories.

Pandey's practical advice is modest and achievable: before drafting answer choices, brainstorm two or three specific misunderstandings your students are likely to have. You do not need all six every time. You need the ones most likely for this target and these students.

Your own experience is the best source here. Think back to last year's errors. What did students consistently get wrong in this topic? Which mistakes surprised you? Those memories are diagnostic gold. They just need to be written down before you write the question.

Researchers have built entire instruments around this idea. One project described in ERIC is working to create a formative concept inventory that uses a psychometric model to predict whether a student exhibits one or more misconceptions across many items. Another team developed a fractions misconception diagnostic test covering representations, ordering and comparing, equivalence, the number line, and all four operations. You do not need a research team to apply the same principle. You need a clear target and a short list of predicted errors.

Step 3 (Notice): Engineering Distractors that Reveal Student Thinking

This is where the question itself changes. In a multiple-choice item, the wrong options are called distractors. In many tests, distractors are almost random: a number one digit off, a misspelled word, an option nobody would choose. They exist to fill space.

Pandey argued that distractors should do something far more useful. Each one should act as a window into a specific line of reasoning.

"A normal question checks correctness. A diagnostic question reveals thinking."
Manish Pandey

Here is how Mira redesigned one question. She asked students to compare 3/8 and 3/5. Then she built each wrong option around a misconception she had predicted:

• Option A captured denominator-only thinking. A student choosing it reasons that 8 is bigger than 5, so 3/8 must be bigger. This is whole-number bias, exactly Sana's error.
• Option C captured numerator-only thinking. A student choosing it looks at the matching numerators and concludes the fractions must be equal or cannot be compared.
• Option D captured a confused comparison, where the student mixes up which part of the fraction drives size.

The correct answer, that 3/5 is greater because fifths are larger parts than eighths, sits alongside these. Now when a student picks a wrong answer, you know which wrong idea they hold. The wrong answer has become data.

A few principles make distractors work:

• Build each one from a named misconception. If you cannot say what thinking a distractor represents, replace it.
• Make each one plausible. A student holding that misconception should find it convincing. If nobody would pick it, it tells you nothing.
• Avoid trick wording. You want to catch misunderstanding of the concept, not misreading of the sentence.
• Keep one misconception per distractor. If an option could come from two different errors, you cannot tell which one the student made.

Some researchers add a second layer. A study in the Journal of Education and Learning developed a two-tier multiple-choice diagnostic instrument using a certainty of response index to identify misunderstandings about changes in matter. Students answer the question, then rate how confident they are. A confident wrong answer suggests a firmly held misconception. An unsure wrong answer may simply be a guess. You can borrow a simple version of this tomorrow: add a line asking "How sure are you? Very sure, somewhat sure, guessing."

For more ways to gather quick evidence beyond multiple choice, our list of classroom assessment techniques includes exit tickets, hinge questions and other methods that pair well with diagnostic distractors.

Step 4 (Decide): Selecting the Next Instructional Move from Response Patterns

Once students respond, you stop counting and start reading.

"Do not just count the correct answer, read the pattern. The pattern is the diagnosis."
Manish Pandey

In Mira's class of 32 students, 14 chose Option A, the denominator-bias distractor. That is not a random scatter. That is nearly half the class holding the same specific belief.

Under the old approach, a weak result on fractions would trigger a full reteach of the chapter. Mira did something more precise. She paused to address that single misconception. She used visual circle models to show that a circle cut into eight pieces has smaller pieces than one cut into five. She used pizza analogies: would you rather have three slices of a pizza cut into eight, or three slices of the same pizza cut into five? The misconception, not the whole chapter, became the lesson.

Here is a simple way to tally and act on patterns:

1. Count responses by option, not just right versus wrong.
2. Map each wrong option back to the misconception it represents.
3. Judge the scale. Is it a class-wide trend or an isolated gap?
4. Choose a matched move.

The point of all this is action. Pandey made that explicit: diagnosis only matters if it changes your next move. A beautifully designed question that leads to the same lesson you would have taught anyway has not done its job.

Think again about Kabir, Sana and Riya. With diagnostic evidence, Kabir gets a quick note on checking his calculations. Sana joins the circle-model mini-lesson. Riya gets sentence starters and a partner to explain her reasoning to. Three students. Three moves. Far less wasted time for everyone.

Applying Diagnostic Design Across STEM, Humanities, and Primary Grades

Fractions make a clean example, but Pandey was clear that diagnostic thinking applies to every subject and every age. He shared parallel cases from across the curriculum.

English

A common pattern: students quote evidence from a text but do not explain how it supports their point. A score might say "partial marks for analysis." A diagnosis says the student knows how to find evidence but has not yet learned to connect it to a claim. Those need different teaching. One is about reading; the other is about reasoning and writing.

Diagnostic design idea: Give students a claim, a quotation, and three possible explanations. One explanation simply restates the quote. One connects the quote to an unrelated idea. One links the quote clearly to the claim. The option a student picks shows where their analytical thinking stands.

Science

Students often name heat as the cause of melting without describing what happens to particle movement. They have the label but not the mechanism. A diagnostic question separates those two levels of understanding.

Diagnostic design idea: Ask what happens when ice melts, with options that reflect common beliefs: particles themselves melt, particles get bigger, particles move faster and break out of fixed positions, or heat is a substance added to the ice. Each wrong choice reveals a distinct mental model.

Social science

Students can list historical events in order but struggle to link them by cause and effect. They know what happened. They cannot yet say why one thing led to another.

Diagnostic design idea: Present two events and ask how they relate. Options might include "they happened at the same time", "one came before the other", and a genuine causal explanation. A student picking a timing-only answer is showing sequence knowledge without causal reasoning.

Early primary reading

Pandey's example here is especially useful because it challenges a common assumption. A young reader confuses "was" and "saw". It is easy to label this as carelessness. A diagnostic lens suggests something else: the child may be struggling with left-to-right directional tracking. Those are different problems. Telling a child to "be more careful" will not fix a tracking issue.

Diagnostic design idea: Track which reversals appear and how often. If a child consistently reverses word direction, that pattern points to a tracking need rather than an attention need, and calls for targeted practice with finger-pointing and directional cues.

Across all four examples the logic is the same. Ask what thinking would produce this answer. Then design questions that separate one kind of thinking from another. If you plan lessons using an inquiry model, these diagnostic questions fit naturally into the engage or evaluate phases of the 5E lesson plan model.

Leveraging AI as a Pedagogical Assistant with Guardrails

Writing good distractors takes time, and predicting misconceptions takes experience. This is where AI tools can help, as long as the teacher stays in charge.

Pandey suggested three ethical uses:

• Generating potential misconceptions. Ask an AI tool what errors grade 7 students commonly make when comparing fractions with the same numerator. Use the list as a starting point for your own Identify step.
• Drafting diagnostic distractors. Give it your sharpened target and your predicted misconceptions, and ask for answer options, each linked to one specific error, with a short rationale.
• Sorting anonymised responses. Paste in student answers with all names and identifying details removed, and ask the tool to group them by likely reasoning.

And two guardrails he stressed:

1. Anonymise student data. Strip out names, roll numbers and anything that could identify a child before you paste it anywhere.
2. Verify every output with your own judgment. AI tools can suggest plausible-sounding misconceptions that do not match your students, or distractors that test reading rather than understanding. You know your class. The tool does not.

Here is a sample prompt you could adapt:

"I teach grade 7 maths. My learning target is: students compare two fractions with the same numerator and explain how denominator size affects part size. List four common misconceptions students hold about this. For each one, write one multiple-choice distractor that a student holding that misconception would find convincing, and explain the reasoning behind it in one sentence."

Read what comes back critically. Keep what matches your experience. Cut what does not. The tool drafts possibilities. You decide what goes in front of students.

If you want to build more confidence with these tools, our guides to ChatGPT for teachers and AI in education cover prompting and responsible use in more depth. Broader research continues to examine how diagnostic tools are designed and validated, including work published in the Pegem Journal of Education and Instruction and in Heliyon, which is worth exploring if you want to go further into the evidence base.

A 14-Day Roadmap to Assessment Literacy in Your Classroom

Pandey recommended running one full FIND cycle over 14 days on a single curriculum target. Not your whole syllabus. One target. The goal is fluency, and fluency comes from doing it once properly rather than trying everything at once.

Here is one way to structure those two weeks:

A few tips as you go:

• Start small. Three well-designed questions will teach you more than twenty rushed ones.
• Keep your tally sheet. Over time, it becomes a personal record of how students in your context actually think.
• Share with a colleague. Two teachers predicting misconceptions together will catch more than one working alone.
• Expect some distractors to fail. If nobody picks an option, that is useful too. It tells you that misconception is less common than you assumed.

This kind of focused, evidence-based practice is exactly what meaningful professional development looks like. If you are planning your growth for the year, it belongs in your record. Our guide on CPD for teachers explains how to document reflective classroom inquiry like this.

From Marking to Understanding

Go back to that Sunday evening and the stack of quizzes. Three students, three scores of 6 out of 10.

With FIND, the evening looks different. You are not just writing numbers. You are reading patterns. You know that Kabir needs a nudge on checking his work, that Sana holds a specific belief about denominators, and that Riya needs words to match her intuition. Monday's lesson is no longer a guess.

Marks still matter. Schools, parents and boards will keep asking for them. But you now have a second layer of evidence underneath every mark, one that tells you what to do next.

That is the shift. From recording how much students scored to understanding how they think. From one lesson for everyone to the right move for each learner. You already care about your students' thinking. FIND gives that care a structure.

Start with one target. Run one cycle. See what your students' wrong answers have been trying to tell you all along.

For the Love of Teaching.