What IB examiners actually look for in an Internal Assessment

The Internal Assessment is the most generous assessment in the Diploma and the one students most consistently underperform on. Generous, because the criteria are published, the weighting is known, you have months rather than two hours, and you can ask for feedback on a draft. Underperformed, because most students write the IA they find interesting rather than the IA the criteria describe.

Below is what the marks are actually awarded for, in the language examiners use, and the four places they most often go missing. The details differ between subjects — a Biology IA is not a History IA — but the failure modes are remarkably consistent across all of them.

Personal engagement is a criterion, not a mood

Every subject rewards evidence that this investigation is yours. That evidence is specific and documented: a decision you made and justified, a method you adapted because the standard one didn't fit your context, a reason this question matters where you are. "I have always been fascinated by genetics" is not engagement — it is an opening sentence. "I chose this local water source because the municipal quality report lists it as borderline" is.

Exploration: the research question is doing most of the work

A vague question makes every later criterion harder, because there is nothing precise for the method to be appropriate to or for the conclusion to answer. A strong research question names the independent variable, the dependent variable and the context, and is narrow enough that you can genuinely answer it with the equipment and time you have.

• Too broad: "How does temperature affect enzymes?"
• Better: "How does temperature (10–60 °C) affect the rate of catalase-catalysed decomposition of hydrogen peroxide, measured as oxygen volume produced in 60 seconds?"
• The second version tells the examiner what you will vary, what you will measure, how, and over what range — and it is now possible to judge whether your method is appropriate.

Exploration marks also cover background: enough theory to justify the method, not a textbook chapter. If a paragraph does not change what you did or how you interpreted it, it is costing you words and gaining you nothing.

Analysis: process the data, don't merely present it

This is the single largest source of lost marks. A table of raw numbers and a chart is presentation. Analysis is what you do to those numbers so they can answer the question: appropriate processing, an honest treatment of uncertainty, error bars that mean something, and a stated reason for the statistical choice you made.

Two specifics worth internalising. Uncertainties must propagate — if you measured two lengths each to ±0.05 cm, the uncertainty in the area is not ±0.05 cm². And a trend line needs a justification: fitting a straight line to something you have a theoretical reason to expect to be linear is analysis; fitting whatever curve has the highest R² is not.

Evaluation: weaknesses have to be your weaknesses

Generic limitations — "human error", "more time would have helped", "the equipment was not precise enough" — score close to zero, because they could be appended to any investigation ever conducted. A real evaluation identifies a specific weakness in this method, states its likely direction and size of effect on the result, and proposes a realistic improvement.

"The water bath drifted by up to 2 °C between readings, which would flatten the measured rate at the higher temperatures and partly explains why the 50 °C and 60 °C points are closer together than the theory predicts; a circulating thermostatic bath would hold this to ±0.2 °C" is worth several marks. "There may have been human error" is worth none.

Communication: structure is a mark, not a courtesy

Communication rewards a report that is focused, well structured, and uses subject vocabulary correctly and consistently. In practice that means: a clear section order, figures numbered and referred to in the text, units stated once and used everywhere, and a word count that respects the limit. Reports that go over the limit are not read past it, which is an expensive way to lose your conclusion.

The four places marks actually go missing

1. A research question that cannot be answered by the data collected — usually because it was written before the method and never revised afterwards.
2. Raw data presented rather than processed, with uncertainty either absent or not propagated.
3. Generic evaluation that names limitations without linking them to the actual results.
4. A conclusion that restates the results instead of answering the research question and comparing the outcome to the accepted theory or literature value.

None of those are about being clever. They are about reading the criteria as a specification, which is exactly what they are. Print them out, put them beside your draft, and go through your report criterion by criterion asking what specifically in this document earns each band. Anything you cannot point at, you have not written yet.