The data analysis round

The IOAA data analysis round hands you real (or realistic) measurements and asks you to turn them into a physical result. The data arrive as tables, light curves, spectra, images or ephemerides, and the answer is usually a number with an uncertainty: a distance, a mass, a period, a slope.

The format has been stable for years. A few examples from the official papers:

• 2007 (Chiang Mai): 3 hours, three questions.
• 2016 (Bhubaneswar): three problems of 50 marks each, 150 in total.
• 2017 (Phuket): 4 hours, two problems of 75 marks, 150 in total.
• 2018 (Beijing): 5 hours, two problems of 75 marks, 150 in total.
• 2023 (Chorzów): 3 hours, 125 marks, problems of 50 and 75.

You get graph paper, answer sheets, a summary answer sheet for final values, and a table of constants. Bring a ruler, a sharp pencil and a scientific calculator you know well. The round tests physics, but it rewards care and method at least as much as cleverness.

The official marking schemes are published with the solutions, and they are remarkably consistent. A graph typically earns marks for:

• using more than half of the graph paper,
• both axes labelled with quantity and unit,
• ticks and values on the axes (or the scale written explicitly),
• every point plotted correctly (marks drop off with each wrong or missing point),
• error bars where the data have uncertainties,
• a sensible best-fit line or smooth curve.

Numerical answers are marked against tolerance windows. In 2016 D1, for example, $P_0$ earned full marks anywhere from 7588.43 to 7588.53 μs and half marks out to 7588.38–7588.58. A value read from a carelessly drawn graph falls outside the window even when the method is right.

Errors usually carry forward without a double penalty: if part (b) uses your wrong answer from (a) correctly, (b) still earns its marks. So never leave a part blank because an earlier one went wrong. Use a reasonable value, and say that you are doing so.

• Read everything first (10 minutes). Mark which parts need a graph, which need only a formula, and which depend on earlier results.
• Do the derivations and pure formula parts early. They are quick, safe marks (for example "write expressions for $P_B$ and $r$" or "derive $\theta_c(u)$").
• Tabulate before you plot. Add computed columns ($\log P$, $1/\lambda$, $\Delta t_{\rm gal}$...) to a clean table. Plotting from a table is fast and graders can follow it.
• Draw each graph once, carefully. Choose the scale before plotting the first point. A redrawn graph costs 15 minutes.
• Keep 15 minutes at the end to copy final values into the summary sheet, with units and sensible significant figures.

When a question says "estimate", give a value and an uncertainty. When it asks "YES/NO", give the word and the one-line reason; the reasoning is often what earns the mark.