Spark image

Precision, accuracy, fair tests and controls

When you do an experiment you should ask yourself:
What am I trying to do?
How am I going to do it?

but also

1. How good is my apparatus?
2. How good an experimenter am I? and
3. How accurate an answer do I need?

Try and answer the third one before you start there is no point is spending lots of money or time if all that you need is a rough idea of the size of a quantity or an approximate proof of a law.

To improve the first you should buy, or make, better (more accurate) apparatus.
To improve the second you will need practice - take more readings as a check and work out an average result from these.

Ask yourself if your experiment is good enough to convince other people that the results that you have obtained are correct and reliable.

Can other people get the same results by doing the same, or a different, experiment? (Remember the cold fusion debate!).

Control experiment
Always have a control experiment to check that the surroundings have not affected your results and if they have how you can make allowances for this. (Example melting ice due to the air temperature alone without a heater in a latent heat experiment).

See: Errors (16-19 text)

Fair tests

Make sure that your experiment is a 'fair test' of the things that you are trying to measure. In other words be sure which quantity is going to stay fixed while you change another and see what effect it has on a third. For example if you wanted to investigate how heat travelled down a selection metal rods you would keep the size and length of the rods the same as well as the temperature of the hot end while changing the type of metal.

Quoting an answer and accuracy

Don't quote an answer to an unreasonable amount of accuracy. Decimal places are not so important as what are called significant figures.

For example let's imagine that you have measured the mass of a small stone and found it to be 0.0232 kg. This looks very accurate to four places of decimals. However if you quote the answer in grams it is 23.2 g just the same value but now to only one decimal place.

Both answers are accurate to three significant figures one part in 232. It is the size of the last figure compared with the other figures in the number that matters.

If you are sat on by an elephant it is not important whether or not the elephant has a fly on its back - you are still squashed flat. However to another fly the first fly is important!


Beware the use of your calculator say you wanted to find the density of the stone. You measured its mass (23.2 g) and then found its volume say 1.7 cm3 accurate to one part in 17. Working out its density using a calculator you would get 13.64705882 g/cm3. This would look as though you could measure it to one part in ten thousand million! In this case the answer should be quoted as 13.6 g/cm3 at the very best and probably simply as 13 g/cm3.
The more things that you have to measure to get your answer the more inaccurate your answer is likely to be.

© Keith Gibbs 2013