The Scientific Investigation | IB DP Physics

Home › Learning Hub › IB DP Physics › The Scientific Investigation

This page replaces the old lab report format. If you have notes or older guidance referring to Design, Data Collection and Processing, and Conclusion and Evaluation, those criteria were retired with the previous syllabus. They no longer exist. What follows is the current model.

📋The essentials

Name: the scientific investigation.
Weighting: 20% of your final grade, at both SL and HL.
Marks: 24, awarded across four criteria of 6 marks each.
Length: maximum 3000 words for the written report.
Time: about 10 hours within the practical scheme of work.
Work: individual, though data may be collected in a group provided your analysis and report are entirely your own.

📈The four criteria, and where the marks really are

Research design (6). A clearly stated and focused research question, the independent, dependent and controlled variables identified, and a methodology described in enough detail to be repeated. Say why you chose the ranges and the equipment, not just what they were.

Data analysis (6). Raw data recorded with consistent significant figures and uncertainties, processed correctly, and presented in graphs that are properly labelled with units. Uncertainties must be propagated through the processing rather than quoted once at the start and forgotten.

Conclusion (6). A conclusion that actually answers the research question, justified by your own data, and compared with accepted scientific values or an established relationship. A conclusion that restates the trend without interpreting it scores poorly.

Evaluation (6). Strengths and weaknesses of the method, the relative significance of the sources of error, and realistic, specific improvements. “Use more accurate equipment” earns nothing. “The 0.5 s human reaction time dominated the 2.1 s timings, contributing about 24% uncertainty; light gates would reduce this to under 1%” earns a great deal.

Note the balance carefully. Conclusion and Evaluation together are worth half the marks. Most students spend most of their time on the practical work and write these two sections last, in a hurry, and it shows. Plan your time backwards from that fact.

🔧Practical advice

Choosing a question. It must be measurable, it must have a genuine independent variable you can vary over a decent range, and you must be able to take enough data. A good rule of thumb: at least five values of the independent variable, with three to five repeats at each. Fewer than five points makes a graph gradient meaningless.

Uncertainties. Every measuring instrument contributes one. For an analogue scale take half the smallest division; for a digital instrument take the last digit. When quantities are multiplied or divided, add percentage uncertainties; when added or subtracted, add absolute uncertainties. Show this working — it belongs to Data analysis.

Graphs. Plot the relationship that gives a straight line if you can, because a gradient is far easier to interpret than a curve. Include error bars, and draw maximum and minimum gradient lines to obtain the uncertainty in your gradient. That single technique probably earns more marks per minute than anything else in the IA.

The word count. 3000 words is a maximum, not a target; strong reports are often well under it. Tables, graphs, labels, equations, data and the bibliography do not count towards it. Rambling method descriptions do.

⚖️Academic honesty

Cite every source, including any you used for background theory or for accepted values. If you collected data as a group, say so explicitly. The analysis, the conclusion and the writing must be yours alone, and identical wording between two students is straightforward to detect.

✅Before you submit — a checklist

Does the conclusion answer the exact research question as worded?
Is every raw measurement quoted with an uncertainty and consistent significant figures?
Do all graphs have axis labels, units, and error bars?
Have you compared your result to an accepted value or established relationship?
Are your identified errors ranked by significance rather than merely listed?
Is every improvement specific enough that someone could act on it?