If you started the Diploma Programme recently, your Chemistry IA isn't called the "Internal Assessment" in the subject guide anymore. It's now a Scientific Investigation, capped at 3,000 words, and marked against four criteria instead of the older five. That's not just a name change. It changes what actually earns marks, and a lot of the advice floating around online still describes the old structure. Here's what the current version actually rewards, and where I've seen people — myself included, on an early draft — lose marks without realizing why.
The four criteria, and what each one is really asking
Research Design (6 marks). This is where most weak IAs fail before the experiment even starts. Examiners want a research question that explicitly names your independent and dependent variables, states the chemical system you're investigating, and connects to relevant background theory, not js a vague framing like "how does temperature affect reaction rate," which doesn't specify a reaction, a measurable quantity, or a range. A strong research question reads more like: "How does the concentration of hydrochloric acid (0.5–2.5 mol dm⁻³) affect the initial rate of reaction with magnesium ribbon, measured by volume of hydrogen gas produced in the first 30 seconds?.
Data Analysis (6 marks). This criterion is about whether your processing is correct and appropriately uncertain, not just whether your final number looks reasonable. That means: propagating uncertainty from your equipment through every calculated value, not just stating a percentage at the end; using appropriate significant figures throughout, driven by your least precise measurement; and presenting processed data (not just raw data) in a way that makes trends visible. A properly labeled graph with error bars beats a table of numbers every time.
Conclusion (6 marks). A common mistake here is restating your data instead of answering your research question. The conclusion criterion specifically wants you to interpret your result against accepted chemical theory.
Does your measured activation energy roughly match a literature value, and if not, why not? Chemistry examiners want to see that you understand why your result makes physical or chemical sense, not just that you can report it.
Evaluation (6 marks). This is where personal engagement now effectively lives in the new structure, and it's also where safety and ethical considerations get assessed. A strong evaluation identifies specific limitations tied to your actual method — not a generic "human error" — and proposes a specific, technically grounded improvement. "Replace manual titration endpoint judgment with a pH probe and datalogger to remove observer bias in identifying the color change" is a real improvement. "Be more careful next time" is not.
Picking a topic that actually works within 3,000 words
The word cap changes topic strategy more than people expect. A topic that requires explaining a lot of unfamiliar background theory eats into your word count before you've even started analyzing data. The investigations that tend to score well are ones where:
The chemistry is squarely within (or a clear, well-justified extension of) the syllabus, so you don't need paragraphs of extra theory to justify your method
You can generate enough quantitative data points — through a range of a single independent variable — to support real statistical treatment, like a line of best fit with a defensible gradient uncertainty
The apparatus is realistic for a school lab, so your method section can be specific rather than aspirational
Titration-based investigations, rate-of-reaction studies using gas collection or colorimetry, and electrochemical cell investigations tend to work well because they naturally produce numerical, repeatable data without requiring exotic equipment.
The mistake that costs the most marks for the least reason
Choosing a topic that's too ambitious for a school lab. I've seen IAs attempt investigations that would need equipment — a bomb calorimeter, precise spectrophotometric kinetics, controlled inert atmospheres — that the school simply doesn't have, which forces the student into a compromised method that undermines Research Design and Data Analysis simultaneously. A simpler investigation, executed with real precision and genuinely propagated uncertainty, consistently scores higher than an ambitious one executed sloppily. Chemistry examiners are marking whether you understood, controlled, and honestly evaluated what you actually did.
Before you submit
Read your research question and your conclusion side by side. If the conclusion doesn't directly and explicitly answer the research question as written — with a number, a comparison to expected theory, and a stated confidence in that answer — that's the fastest fix available to you, and it's one most students can make in an afternoon of editing rather than a re-run of the experiment.