Design is where a good question becomes a plan somebody else could follow. The test is brutally simple: could another student pick up your method, with no conversation with you, and get the same data? Everything on this page is about passing that test — and about justifying each choice rather than just stating it.
📘 What you need to know
Identify and justify your independent, dependent and controlled variables.
Use at least five values of the independent variable, spread across a range you can defend.
Repeat each value at least three times, and five is better, because living material varies so much.
Repeats let you calculate a mean, which reduces the effect of random error and exposes anomalies.
Write the method in enough detail to be replicated exactly — real volumes, real apparatus, real times.
Structure it as materials, then safety, then procedure.
Not every investigation is hands-on: databases, simulations and surveys are valid approaches.
Run a pilot before committing to the full thing.
Naming your variables
Every design starts here, and the three types are easy to state and easy to get slightly wrong.
Variable
What it is
In a yeast respiration study
Independent (IV)
The single thing you deliberately change
Temperature of the water bath
Dependent (DV)
The thing you measure, to see how the IV affected it
Volume of carbon dioxide produced per minute
Controlled (CVs)
Everything else that could plausibly affect the result, held constant
Yeast concentration, glucose concentration, pH, volume, time
You are not just listing these. For each one you say how you will control it and why it matters — which is the whole subject of the next page.
How many values, and how many repeats?
Two numbers decide whether your data can show anything: how many points across your range, and how many times you measure each one.
The two numbers do different jobs. Adding more values shows the shape of the relationship; adding more repeats tells you how much to trust each point.
Justifying the range
Stating the range is half a mark. Explaining why that range is the other half, and it always comes from your background reading.
WORKED EXAMPLE
Justify a temperature range of 10 °C to 50 °C, in 10 °C steps, for a yeast respiration investigation.
Why the bottom of the range
Below about 10 °C respiration is so slow that too little gas is produced to measure reliably in a lesson.
Why the top of the range
Above about 50 °C the enzymes of respiration denature, so the rate collapses.
Why five values in between10, 20, 30, 40 and 50 °C gives five points, enough to show a rise, a peak and a fall — the shape you expect.
The range brackets the expected optimum on both sidesThis is the key idea: choose a range that will actually contain the interesting behaviour, not one that stops just before it.
Why repeats matter so much in biology
Living material is naturally variable. Two potato cylinders from the same potato are not identical, let alone two potatoes.
Repeats let you calculate a mean, which reduces the effect of random error.
They also let you spot anomalies — a reading that disagrees with its own repeats is visible, where a single reading is not.
They give you a spread to quote, so you can put error bars on your graph.
WORKED EXAMPLE
Each run of your experiment takes about 3 minutes. You have one 60 minute lesson. Work out whether five values with five repeats is feasible, and what to do if it is not.
Step 1: total runs5 values × 5 repeats = 25 runsStep 2: total time25 × 3 = 75 minutes
That is more than the lesson, before any setting up.
Step 3: the options
Drop to three repeats: 5 × 3 = 15 runs = 45 minutes. Or run several tubes in parallel in the same water bath, keeping five repeats.
Three repeats fits; parallel runs keep fiveDo this arithmetic during design, not on the day. It is also exactly the kind of planning that earns credit.
Writing a method that can be replicated
The single most common weakness in a design is vagueness. “Add some starch solution” cannot be repeated, so it cannot be checked, so it is not science.
The right-hand column is not longer because it is padded. Each extra word is a number somebody would otherwise have to guess.
WORKED EXAMPLE
Rewrite this instruction so it could be replicated: “Put some pondweed in water and shine a light on it.”
What is missing?
How much pondweed, what water, which lamp, how far away, for how long.
The rewrite
Cut a 5.0 cm length of pondweed and place it, cut end upwards, in a boiling tube containing 20 cm3 of 1% sodium hydrogencarbonate solution, measured with a 25 cm3 measuring cylinder.
And the lamp
Place a 12 W LED lamp at 10 cm from the tube, measured with a ruler, and leave for 2 minutes to settle before counting bubbles for 1 minute.
Quantities, apparatus, distances and times all statedThe hydrogencarbonate is there so carbon dioxide never becomes limiting — and saying that turns a step into a justified choice.
Creativity counts too
Design credit is not only for detail. It is also for solving a measurement problem neatly. Timing a colour change by eye depends on whose eye it is; using a colorimeter to follow the same change gives a number anybody would get. Finding the more objective measurement is exactly the kind of thinking that separates a good design from a routine one.
How to lay the method out
Putting materials first is not just tidiness — it forces you to decide every concentration and size before you start writing steps.
Investigations that are not hands-on
Laboratory work is the usual route, but it is not the only valid one. Three alternatives count fully:
Database investigations — pull data from an established biological database, such as DNA sequence records or conservation status listings, then process and analyse it to find a trend. Your skill shows in the question and the processing.
Simulations — use a simulation to collect data on something too slow or too complex for a school lab, like population dynamics or evolution over many generations.
Surveys — design a questionnaire to gather data on human physiology or on people’s ecological awareness. The design of the questions is where the rigour lives.
These are not the easy option. A database study still needs a focused question, controlled comparisons, and honest discussion of how the data was originally collected — which you did not control.
Run a pilot first
A pilot study is a small-scale trial run: one or two conditions, one repeat, just to see whether the method behaves. It takes a lesson and saves a fortnight.
🧩 What a pilot tells you
Is the reaction the right speed? If it finishes in five seconds you cannot time it; if nothing happens in twenty minutes you cannot finish. Adjust concentrations accordingly.
Is the range sensible? If every value gives the same reading, your range is too narrow or in the wrong place.
Is the measurement workable? Bubbles that come as a froth cannot be counted, and you would rather find that out now.
How long does one run really take? Feed that number back into your feasibility sum.
Say in your report that you ran a pilot and what you changed as a result. It shows the design was thought about rather than copied, and it explains why your final method looks the way it does.
💡 Exam tip
Justify, do not just state. Every range, every apparatus choice, every organism needs a “because”.
Five values of the IV, and at least three repeats — five if you can manage it.
Choose a range that brackets the behaviour you expect, so the interesting part is inside it.
Give concentrations, volumes, times and apparatus sizes in the method.
Include a dedicated safety, ethics and environment section.
Mention your pilot and what it changed.
Draw your blank results table during design. It forces you to notice what you forgot to measure.
⚠ Common mix-up
Stating variables without saying how they will be controlled. The “how” is the mark.
Three values of the IV, which cannot show the shape of a relationship.
Repeating the whole experiment once and calling it repeats. Each value of the IV needs its own repeats.
A method written as a memory aid rather than as instructions for someone else.
Safety tacked on as one sentence at the end.
Choosing a range that stops before the optimum, so the graph never turns.
Skipping the pilot, then discovering the problem when there is no time left.
Up next: Controlling Variables — calibration, constant conditions, representative sampling, and the control run that proves it was your independent variable doing the work.
Want this explained one-to-one?
Book a free session with an experienced IB Biology tutor and get your trickiest topics made simple.
