IB Biology IA: planning the scientific investigation

The four criteria

Tables, graphs, equations, calculations, headings and references do not count towards the 3,000 words, according to teachers' guidance. Check your school's rules on word count.

Choosing a research question that works

Biology investigations fail most often at the question. A good question names one independent variable, one dependent variable and the organism or system: "How does light intensity, varied by lamp distance from 10 to 50 cm, affect the rate of oxygen production by Elodea, measured as bubbles per minute?" A weak question names a topic: "How does light affect photosynthesis?"

Living systems are variable, so plan for it. Use enough repeats or individuals to see a pattern through the natural spread, keep temperature, species and age constant where you can, and say which variables you could not control. The question must also be ethical and safe: the IB expects investigations to avoid harm to animals and people, so most schools steer students towards plants, microorganisms, enzymes, secondary databases or simulations.

Not every IA needs a laboratory. A study using a published database (for example gene sequences, species distribution or health data) or a simulation can score well if the question is focused and the analysis is careful.

Worked example: is a difference real?

Invented data, to show the kind of processing examiners look for. Oxygen bubbles per minute from pondweed at two lamp distances, five trials each.

1. Lamp at 10 cm: 18, 21, 19, 22, 20. Mean = 100 ÷ 5 = 20 bubbles per minute.
2. Lamp at 30 cm: 12, 14, 11, 13, 15. Mean = 65 ÷ 5 = 13 bubbles per minute.
3. Standard deviation at 10 cm: deviations from the mean are -2, 1, -1, 2, 0; squares add to 10; divide by n - 1 = 4 to get 2.5; sqrt(2.5) = 1.58. The 30 cm data give the same: squares 1, 1, 4, 0, 4 add to 10, so SD = 1.58.
4. The means differ by 7, which is more than four standard deviations, and the ranges (18 to 22 and 11 to 15) do not overlap, so the difference is unlikely to be chance.
5. A t-test confirms it: each group's variance (2.5) divided by its 5 trials gives 0.5, so t = (20 - 13) / sqrt(0.5 + 0.5) = 7 / sqrt(1) = 7. With 8 degrees of freedom the critical value at p = 0.05 is about 2.31, so the difference is significant.
6. Then link back to biology: closer light gives higher light intensity, which increases the rate of the light-dependent reactions until another factor becomes limiting.

Where Biology IA marks are lost

• A question with no range or units for the independent variable, which weakens research design from the start.
• Too few repeats or individuals for a variable living system, so the trend cannot be separated from the noise.
• Graphs of means with no indication of spread, such as error bars from standard deviation.
• Choosing a statistical test without saying why it suits the data, or using one at all when a clear graph and descriptive statistics would answer the question.
• A conclusion that repeats the results without explaining them with biology, or that claims more than the data show.
• Evaluation that says "use better equipment". Name the specific limitation, its likely effect on the data, and a realistic fix.

How one-to-one lessons help with the Biology IA

A tutor who teaches IB Biology helps the student turn an interest into a measurable question, test whether the method is practical in the school's lab or with available data, and think through controls, sample size and ethics before any data is collected. Once data exists, lessons cover how to process it: means, standard deviation, error bars, and when a test such as a t-test or correlation is appropriate, worked through on the shared whiteboard with the student's own numbers.

The IB requires the investigation and report to be the student's own work, and teachers can give only limited feedback on a draft. A tutor explains the criteria and the skills; the student designs, carries out and writes the investigation.