Graphs, conclusions and evaluation
In a minute
After the readings come the marks for using them: calculating, plotting a graph with a , reading a or intercept, and writing a conclusion backed by numbers. Then you judge the experiment and suggest improvements.
You'll be able to
- Process readings with a calculator, and plot a graph with labelled axes, a good scale and a .
- Find a gradient or an intercept from a graph, using and .
- Write a conclusion that quotes the data, and decide whether two results are equal within ±10%.
- Spot an , tell random from , and say why a conclusion may be uncertain.
- Evaluate a method, including its , and suggest useful improvements.
Before you start3 quick questions on what this builds on
Revise first: 1.2 Motion, 7.2 Planning investigations and 7.3 Taking and recording measurements
Processing data and plotting graphs
Readings often need a calculation first, such as a mean or .
Give calculated values to 2 or 3 significant figures, no more than the least precise reading. Give a mean to its readings' decimal places.
Try it first, then reveal one step at a time.
- Both readings have 2 significant figures, so give 2.
Try it first, then reveal one step at a time.
- Add the readings and divide by 3: .
- The readings have 2 decimal places, so the mean does too.
A good graph has five features. 1. Each axis labelled quantity / unit, such as current / A.
2. A scale where the points fill more than half the grid each way. 3. An easy scale, such as 1, 2 or 5 (or 0.5, 20…) per large square, never 3 or 7.
4. Each point plotted to within half a small square. 5. One best-fit line.
It needn't pass through every point or the origin. Rule it if the trend is straight; otherwise draw a smooth curve.
Gradients, intercepts and reading values
Reading a value between your plotted points is . Extending the line beyond them, for example back to an axis, is . Always read from the line, not from one point.
To find a , draw a triangle covering more than half the line. A big triangle keeps reading errors small.
A gradient's unit is the unit ÷ the unit. The intercept is the value of where the line meets , and it has the unit. If the scale doesn't start at 0, don't read it where the line crosses the drawn axis.
Try it first, then reveal one step at a time.
- Take two points on the line, far apart: and .
- , in (m/s) ÷ s, which is .
- It tells you the trolley gains about 0.35 m/s of speed every second.
The gradient of a speed–time graph is the acceleration, so this trolley's acceleration is (1.2).
Try it first, then reveal one step at a time.
- is outside the readings, so extrapolate: extend the line back to the speed axis.
- It meets the axis at , the intercept.
- Check: . The trolley was already moving when timing started.
Conclusions and the ±10% rule
A conclusion answers the experiment's question, then gives the evidence from the graph or table.
Be as exact as the data allow. A straight line is a linear relationship. Directly proportional needs a straight line through the origin.
Try it first, then reveal one step at a time.
- Check both conditions, straight and through the origin, and quote the graph.
- No. The graph is a straight line, but it does not pass through the origin: it meets the speed axis at . So the speed increases steadily with time but is not directly proportional to it.
Real readings rarely agree exactly. At IGCSE, two results agree within the limits of experimental if they differ by 10% or less.
Find the difference, then give it as a percentage of one value: the value predicted by theory, if there is one. Then say whether they agree, quoting both values.
Try it first, then reveal one step at a time.
- Difference: , and .
- Compare with 10%, then state the conclusion with both values.
- Yes, the results support the prediction. The periods and differ by about 3%, which is less than 10%. So they agree within the limits of experimental accuracy, even though the mass changed.
Anomalies, errors and uncertainty
Check an anomaly by repeating that reading. If you can't, leave it out of the mean or best-fit line, circle it, and say so.
Try it first, then reveal one step at a time.
- is far from the other three, so it is the anomaly.
- Leave it out: , then round.
Measurement error is the difference between a measured value and the true value. It comes in two kinds.
Causes include hand timing with a stopwatch, or a flickering reading. Repeating and taking a mean reduces their effect.
A is the classic cause. Repeats don't help, because every repeat is shifted too. Correct the readings or the instrument instead.
Try it first, then reveal one step at a time.
- Is every reading wrong the same way, or do they scatter?
- This is a systematic error: every reading is too high. Repeating and taking a mean does not remove it, because every repeat is too high by the same amount. Subtract from each reading.
A conclusion carries more with few readings, a small , widely scattered points, or values close to the 10% limit.
Evaluating and improving an experiment
To evaluate a method, ask: was it a fair test? If a changed, the results can't show the effect of the independent variable alone.
Then find the biggest source of error. A good improvement is specific and says how it helps.
Common ones: repeat and calculate a mean; more readings over a wider range; time many oscillations and divide; light gates instead of a stopwatch; smaller scale divisions.
Try it first, then reveal one step at a time.
- One oscillation: .
- Reaction time adds perhaps to a hand-timed reading. Over 20 oscillations, that is only about per oscillation.
Try it first, then reveal one step at a time.
- Look for a control variable that changed, then for the largest measurement error.
- Release the car from the same marked start line each time, so the distance travelled is a control variable. Use light gates connected to a timer instead of a hand-held stopwatch, so reaction time doesn't affect the times.
Common mistakes
- Drawing dot to dot or a thick line. Draw one thin best-fit line.
- Calling a line proportional "because it goes up". It must be straight and pass through the origin.
- Saying two results differ because the numbers aren't identical. Give the difference as a percentage of one value (the predicted one, if any) and compare with 10%.
- Writing "repeat it" without "calculate a mean", or "be more careful". Name the change and how it helps.
Exam tips
- "Plot": each point within half a small square, axes labelled quantity / unit.
- "Determine" a gradient: show the triangle and write down the readings used.
- "Comment" on agreement: quote both values, give the difference as a percentage of one value (the predicted one, if any), compare with 10%, then conclude.
- "Suggest" an improvement: name the change and what it improves.
Key takeaways
Practise this
387 past-paper questions on Experimental skills and investigations in the 0625 Extended bank, each with a mark scheme and a worked explanation.