How to diagnose Science misconceptions — a practical method for teachers

You cannot correct a misconception you have not identified, and "the class struggled with light" is not an identification. It is a topic. The thing you actually correct is far more specific: the students think we see an object because our eyes send something towards it. Until you can write the wrong idea down in that form — as something the student believes — you are aiming at a fog.

Diagnosis is the step that turns a vague sense that "they did not get it" into a precise, correctable target. It is also the step most often skipped, because marking pressure pushes us to total the scores and move on. This guide is about doing it efficiently: reading the marking you already have, asking one good follow-up, and telling a real misconception apart from a slip — without adding hours to your week.

Start with the marking you already have

You do not need a special test to begin. The first diagnostic pass is reading the marking with a different question in mind. Instead of "how many got this wrong", ask "which wrong answer did they give, and how many gave the same one?". A cluster of students all choosing the same distractor is a diagnosis arriving for free — they are very likely sharing one idea. A scatter of different wrong answers usually means something else: a confusing question, a careless spread, or genuinely mixed understanding. Reading marking this way is the heart of item analysis, and it is where most misconception diagnosis actually happens.

For multiple-choice and structured items, this means the distractors carry information. In a well-built question, each wrong option corresponds to a particular wrong idea. The option students reach for tells you which idea is live. If your questions were not written that way, that is the first thing to improve — which brings us to diagnostic questions.

Diagnostic questions: make every wrong answer mean something

A diagnostic question is written so that each wrong answer points at a specific misconception (Treagust, 1988). The aim is that a student cannot get it right by accident and cannot get it wrong without telling you why.

Compare two questions about shadows:

• A weak question: "What causes a shadow?" with one obviously right answer and three throwaway wrong ones. A student gets it right or wrong; you learn nothing about their thinking.
• A diagnostic question: a scenario where the distractors match real wrong ideas — the shadow is "made of" the object's colour, the shadow comes from the object glowing, light bends around the object, light is blocked. Now the wrong answer a student picks tells you which model of light they are using.

You do not have to rewrite every question. Even turning your two or three most important questions per topic into genuine diagnostic items changes how much your marking tells you.

The second tier: ask for the reason

The single most useful upgrade to a diagnostic question is a two-tier format: tier one asks for the answer, tier two asks for the reason (Treagust, 1988). It catches the two cases a plain question misses — the student who gets the right answer for a wrong reason, and the one who gets a wrong answer for a sensible reason that just needs nudging. In primary classrooms a full two-tier paper is overkill, but the principle scales down beautifully: after one key question, simply ask "why?" — in a sentence, a quick show of hands on competing reasons, or a one-line written follow-up.

A low-tech version of the same idea is Predict–Observe–Explain (White & Gunstone, 1992): before a demonstration, ask students to predict and to commit to a reason. Their predictions are a diagnosis on the spot, and — usefully for the correction stage — students who have publicly predicted are more invested in finding out why they were wrong.

Telling a misconception from a slip

This is the judgement that decides whether the student needs a reteach or just needs the question handed back. Two signals do most of the work:

1. Repetition. A slip is usually isolated; a misconception repeats — across many students at once, or across CA1, SA1, and CA2 for the same student. The catalogue of common primary misconceptions is, in effect, a list of the wrong ideas repetitive enough that researchers wrote them down.
2. Reasoning. Ask the student to explain. A slip dissolves — "oh, I misread it." A misconception holds up: the student argues for the wrong answer confidently and consistently. If they can defend it, it is a belief, not an accident.

Common diagnosis mistakes

A few traps quietly waste the effort:

• Diagnosing the topic, not the idea. "They struggled with forces" is not actionable. "They think a moving object must have a force pushing it in the direction it moves" is.
• Stopping at the mark. A right answer can hide a wrong reason; a wrong answer can hide an almost-right one. Without the second tier, you cannot tell.
• Over-diagnosing. Not every wrong answer is a deep misconception. Some questions are just badly worded, and some days students are tired. Look for the pattern before you commit a reteach to it.
• Diagnosing once. Children's ideas shift. A misconception you cleared in Term 1 can re-form by Term 3, which is why diagnosis and re-checking are part of the same loop, not separate events.

What to do with the diagnosis

A clean diagnosis points straight at the next two steps. Once you can name the wrong idea as a belief, you can plan a correction that confronts it directly rather than re-explaining around it — see correcting Science misconceptions through contrastive teaching. And because a corrected idea can quietly return, the diagnosis also tells you what your re-check should test, in a fresh context, a few days later.

An honest boundary

Diagnosis narrows the target; it does not read minds. Two students can give the same wrong answer for different reasons, and a tired child can look like a confused one. Treat a diagnosis as a strong hypothesis to act on, not a verdict — and keep the conversation with the student open, because their explanation is more accurate than any inference from a script. The point is not to label children; it is to aim your teaching at the idea that is actually in the way.

This is the part of the week MyScienceHOD is built to make lighter — surfacing which wrong answer is repeating across a class so the diagnosis is in front of you instead of buried in the pile, with you making every teaching call. If that is useful, the free Beta is open to Singapore Science teachers and departments.