This is the doing phase of your investigation — the part where you actually run the experiment you designed and gather your raw numbers. Get this stage right and everything downstream (your calculations, your graph, your conclusion) rests on solid ground. Get it sloppy and no amount of clever maths later can rescue it. This page shows you how to collect data that is accurate, sufficient, and honestly recorded.
📚 What you need to know
Your job is to gather high-quality raw data — direct measurements, no calculations
Design your raw data table before you start, with units and uncertainties in the headers
Record every reading to the precision of the instrument (keep the trailing zeros!)
Sufficient data means at least 5 values of your variable and 3 repeat trials each
Qualitative observations (what you see, hear, feel) are evidence, not an afterthought
When something goes wrong, record it — never quietly ignore or erase it
What counts as good raw data
Raw data is the set of direct measurements you read straight off your instruments — a length off a ruler, a voltage off a multimeter, a time off a stopwatch. The golden rule: your raw data table holds only these direct readings, with no calculations mixed in. Work out the averages, periods, and resistances later, on their own.
Think of raw data as the ingredients and processed data as the finished meal. You keep the ingredients on one table and cook on another. If you start doing sums inside your raw data table, an examiner can no longer see what you actually measured versus what you worked out — and that separation is exactly what they’re checking for.
Design the table before you touch the equipment
A well-designed raw data table is planned before the first reading. It must have a clear title, labelled columns for your independent and dependent variables, and — this is the bit students forget — the units and uncertainties written in the column headers, never scattered through the body of the table.
Plan the table first: title on top, units and uncertainties in the headers, and only raw readings in the body.
Record to the instrument’s precision
Every instrument can only be read so finely, and your recorded number must show that. This is a classic place to drop marks. If a metre ruler is marked in millimetres, a reading is written to three decimal places in metres (like 0.550 m, not 0.55 m). If a digital multimeter reads to two decimal places, you write 5.00 V, not 5 V. Those trailing zeros are not decoration — they tell the reader how precisely you measured.
💡 Top tips
Keep the trailing zeros: 0.550 m and 5.00 V show the instrument’s precision.
Aim for at least 5 values of your independent variable, spread across a good range.
Take 3 repeat trials per value so you can average and spot outliers.
Write units and uncertainties in headers, never after each number in the body.
Qualitative observations are evidence
Not all data is numbers. Qualitative data is what you notice while the experiment runs — the things you see, hear, or feel. These observations are genuinely useful later, because they often explain why a result came out the way it did.
A pendulum bob that swings in a slight ellipse instead of a flat plane.
A flicker on a digital meter that shows the reading was unstable.
A wire or resistor that felt warm to the touch — a sign it was heating up.
A laser beam spreading out (diffracting) as it passes through a narrow slit.
Here’s the trick that separates a top lab report from an average one: treat your observations as clues you’ll cash in later. That “the wire felt warm” note seems trivial now, but in your conclusion it becomes the perfect evidence for why your measured resistance drifted upward. Jot everything down — you can always ignore a note, but you can’t invent one after you’ve packed up.
When things go wrong — deal with it honestly
Experiments rarely run perfectly, and that’s fine. The scientific skill is to notice a problem and respond to it rather than pretend it didn’t happen. Whatever the issue, record it in your notes.
⚒ Handling issues as they happen
Unstable reading? Wait for it to settle, or record the central value and estimate the wobble as an uncertainty.
Anomalous result? Keep it, then take an extra trial so you have a reliable set. Don’t erase it — you’ll justify excluding it later.
Hard to measure? Note the difficulty. It becomes a limitation you discuss in your evaluation.
WE 1
A student times 20 swings of a pendulum three times at a length of 0.600 m and gets 31.1 s, 31.2 s, 31.0 s. At 0.800 m one trial reads 37.1 s while the others read 35.8 s and 35.9 s. How should they handle each set?
The 0.600 m set — all consistent
Three close readings, so record all three: 31.1, 31.2, 31.0 sThe 0.800 m set — spot the outlier
37.1 s sits far from 35.8 and 35.9 — flag it as anomalous, don’t delete it
Respond, don’t ignore
Take a 4th trial; if it lands near 35.9 s, the 37.1 s can be excluded later with justification
Keep everything, flag the outlier, add a trialRecording the odd value AND your response is exactly what earns credit — honesty is the skill being tested.
⚠ Common mistakes
Writing 0.55 m or 5 V — dropping the trailing zeros that show precision.
Putting units after every number in the table body instead of in the header.
Doing calculations inside the raw data table.
Erasing an anomalous reading instead of flagging it and adding a trial.
Treating qualitative observations as unimportant and not writing them down.
Quick recap: Collect raw readings only, recorded to the instrument’s precision. Plan the table first with units and uncertainties in the headers. Gather 5+ values with 3 repeats each. Write down qualitative observations — they’re evidence. When something goes wrong, record and respond, never erase.
You’ve now got clean, honest raw data sitting in a well-planned table. But raw numbers don’t answer a research question on their own — they need to be turned into periods, resistances, and averages, each carrying its uncertainty. That’s the next job: over on Processing Data in Physics, we’ll take these exact readings and cook them into meaningful results.
Want your lab data to earn every mark?
Book a free meeting and we’ll walk through designing raw data tables, recording to the right precision, and turning observations into evidence — the habits that quietly lift your IA grade.
