Practical Skills » Analysing and Evaluating Experimental Data

From results to a conclusion

The spec says you may be tested on your ability to analyse and interpret data and draw conclusions that are consistent with the evidence. It also says you may be tested on your ability to communicate your findings using technical language, relevant calculations and graphs. You will often be given a small table of results, and you must say what it shows.

A good conclusion has three parts:

1. The pattern. Say how the dependent variable changes when the independent variable changes. Use numbers from the table.
2. The type of link. Say whether it is a steady increase, a decrease, or no link at all. If one quantity doubles when the other doubles, say so.
3. The physics. Explain the pattern using what you know, with the right technical words.

Only say what the evidence supports. If you tested parachute canopies from 10 cm to 40 cm wide, do not claim anything about one 2 m wide.

Key terms:

• Conclusion: a statement of what the results show, based only on the evidence.

How reliable are the results?

The spec asks you to assess the reliability of an experimental activity. Here, a set of results is reliable when repeat readings are close together, so you can trust the mean.

• Repeat the reading at least three times for each value of the independent variable, then work out the mean. Repeats show you how much the readings vary.
• Spot anomalous results. You met these in Investigating Current and Voltage. One reading far from the others is anomalous. Repeat it or leave it out of the mean, and say that you did.
• Check the spread. Times of 0.60, 0.61 and 0.59 s are close, so they are reliable. Times of 0.60, 0.85 and 0.45 s are not, so the mean cannot be trusted.

Errors, accuracy and validity

The spec says to evaluate data and methods, taking into account factors that affect accuracy and validity. First, the key words.

Key terms:

• Reliable: repeat readings agree closely.
• Accuracy: how close a measurement is to the true value.
• Valid: an experiment is valid if it measures what it set out to measure and controls the other variables.
• Random error: an error that changes from reading to reading, so repeat readings scatter above and below the true value.
• Systematic error: an error that shifts every reading by the same amount in the same direction.

You have met the zero error before. If an ammeter reads 0.02 A with nothing flowing, every reading is 0.02 A too high. Subtract 0.02 A from each reading.

The resolution of an instrument also limits how accurate a result can be. A timer that reads to 0.1 s cannot show a difference of 0.01 s. The range and the number of readings matter too. Two readings cannot show a pattern, and five values spread over a wide range show it much better than five values bunched together.

For validity, check the control variables. If a student tests how the mass of a trolley affects its speed but also changes the ramp height, the test is not valid. Two things changed at once, so the conclusion is not safe.

Evaluating a method and suggesting improvements

To evaluate, you look for weaknesses, say how each one affects the results and suggest a fix. A fix must be specific and must match the problem. Use this pattern: problem, effect, improvement.

Problem	Effect on results	Improvement
Timing a fast fall by hand	Reaction time gives a random error in short times	Use light gates and a data logger, or time a longer drop
Only three heights tested	Pattern is hard to see	Test at least five heights over a wider range
Only one reading at each height	No way to spot an anomaly	Repeat each three times and take a mean
Ruler read at an angle	Parallax (the scale seems to move when your eye is not level with it) gives a random error	Read with your eye level with the mark
Meter reads 0.02 A at zero	Systematic error in every reading	Correct each reading, or replace the meter
Two variables changed together	Result is not valid	Change one variable and control the rest

Vague answers such as "be more careful" or "do it again" score no marks. Name the change and say what it improves.

Worked example: a second table

A student drops a metal ball into a tank of oil and times how long it takes to fall 0.50 m. The oil is at different temperatures. Each time is the mean of three readings.

Oil temperature (°C)	Mean fall time (s)
20	2.4
30	1.9
40	1.5
50	1.3
60	1.2

Conclusion: as the temperature rose from 20°C to 60°C, the mean fall time decreased from 2.4 s to 1.2 s. The ball fell faster in hotter oil, so hot oil resists the ball less than cool oil. The time fell by 0.5 s between 20°C and 30°C but by only 0.1 s between 50°C and 60°C, so the effect gets smaller as the oil warms up.

Evaluation: with times as short as 1.2 s, a reaction time of about 0.2 s is a large part of the reading. Light gates would be more accurate. The ball must be released from the same position each time, and the oil must be kept at its temperature while the ball falls, otherwise the test is not valid.

Exam-style question

A student investigates how the extension of a spring depends on the force. She hangs masses from the spring and measures the extension with a ruler. She takes one reading for each force.

Force (N)	Extension (cm)
1.0	2.1
2.0	4.0
3.0	6.2
4.0	9.5
5.0	10.2

(a) Identify the result that looks anomalous. [1 mark]

(b) Describe the relationship between force and extension shown by the other results. [2 marks]

(c) The masses hang from a holder of weight 0.2 N that she forgot to include, so every force she recorded is 0.2 N too small. State the type of error this causes and explain why repeating the readings would not remove it. [2 marks]

(d) Suggest one improvement to her method and give a reason. [2 marks]