# How Science Works - What is Precision?

Discussion in 'Science' started by SENIOJ00, Sep 9, 2011.

1. ### SENIOJ00

What is precision?
Some exam boards adopt the view that the essence of precision is related to equipmet used for making measurements. The lack of clarity and consitency often leads to teachers misrepresenting presion. A typical definition is "

How small a measurement is. Millimetres are more precise that centimetres, beacause the are smaller". Although this is useful when considering how to improve the precision of a set of measurement it is not in itself a definition of precision.
Just as accuracy is the closeness of a measurement (or result) to the true value, precision is the closeness of a set of repeat measurements (or results). Where a result is a mathematical value derived using one of more experimental measurements.
So what?
The definitions of accuracy and precision are introduced in order to evaluate experimental measurements and results.
The evaluation of accuracy is facilitated by a consistent approach across exam boards. The main hurdle is often the concept if true value, which is sometimes unknown, exisitng as a concept.
The evaluation of precision does not require any reference to measurement equipment but only reference to other measurements in the same set.
I am not sure why the defiinition precision is sometimes based on the nature of measurement equipment rather than the closeness of the measurements.but differences in the scientific and everyday use of the the terms accuracy and precision may be a contributing factor.
Thus when evaluating accuracy, the comparision is between individual measurements or results (or the average of a set of measurements or results and the true value. Accuracy can be improved by using more accurate equipment. For the mannufacture of a 25cm3 pipette involves a more accurate comparision to the true value 25ml, than does the manufacture of a 25cm3 measuring cylinder. This the accuracy of an experiment involving the measurement of 25cm3 would be improved by using a pipette instead of a measuring cylinder.
When evaluating precision, the comparison is between and individual result or measurement and the measurements or results in the same set. This is often expressed as the range of % range about the mean value. Precision is improved by using equipment which gives measurements and results closer to each other.
Thus many suggestions for modification of experimental methods or equipment have an impact on both accuracy and precision.

It is of course important that students understand thal although the ideal is a accurate and precise measurements and results, that both accurate and inprecise, inaccurate and inprecise and inaccurate but precise measurement/results do exist.
A more accurate and precise use of the definition of precision in the glossary's of science text books and teacher generated resources would improve the ability of students to understand and evaluate experimental metods and equipment.

2. ### doctord

The ASE book "The language of measurement: Terminology used in School Science investigations" ISBN 9780863574245 is useful when looking at terminology for science students. I think that all the exam boards are using it in their GCSEs now (certainly OCR Science B and AQA are).
There is an excerpt from the book on the Getting Practical website http://www.gettingpractical.org.uk/Books.php

3. ### Kevinberry03

D looks a lot my attempts to get treble 20 on a night out!!

4. ### rick_h

The people to ask are NPL, then it's Official!
They used to have a very good sheet summarising the argument, but I can't immediately find it. Try searching their website.
The dartboard poster is here:
Sadly, whatever NPL says, you'll have to use what your exam board wants to hear. The examiners reports are the best source for their "right" answer.

5. ### AnonymousNew commenter

I used to work as a scientist in the NHS doing lots of tests. We might get a sodium result of 135, 136, 134 mmol / l. Fantastically precise results - but not accurate if the actual level is 128 mmol/l.
Or of course - we get 136 and 120 mmol / l. Mean level - 128 mmol / l. Spot on for accuracy but would you trust a machine that gave those results?
Or 127, 128, 129 mmol / l - mean level 128 mmol / l and very precise results as well.
It was fun getting the monthly QA results and seeing how your results compared to other labs.

6. ### blazerStar commenter

So why do ISA exams nearly always ask how you could imrove the precision of your measurements and the answer is always to use a ruler/meter/measuring cylinder with smaller unit divisions?

7. ### OrionNew commenter

Precision - the scale that you use!

8. ### blazerStar commenter

Some ISAapers now also include questions on plus/minus values. So a thermometer reas to plus/minus 2 degrees what does that mean? That is really accuracy I suppose.

Best way to demontrate precision is using two metre rules, one scaledd in cm and te other in mm.

I also have a metre rule scaled in mm where the first 8cm have been sawn off. Get the kids to measure something (a page in their book) using the rule. They get a very precise answer but unless they spot the end of the rule is missing the answer is very inaccurate. Then you can get into the zero error lesson as well!

9. ### richardtatesonNew commenter

I think if we are using the new 'language of measurement' definitions from ASE
(selected bits at: http://store.aqa.org.uk/sciencelab/AQA-GCSE-SCIENCE-GLOSSARY.PDF)
then sawing the end off the ruler is still a good way of demoing how a precise answer might not be accurate
but......the cm versus mm scale idea is now a demo of resolution, not precision

If there is anyone out there who wants to have a go explaining to me what the difference (in new definitions) is between precision and repeatability I'd love to hear it

10. ### richardtatesonNew commenter

Nice! I've tried in the past posing the question to the class of which 2 of these 4 archers (or whatever) they would send to the olympics (if you had to choose 2 of 4).
Nobody wants D and everyone wants A, but do we take B and get them to work on better precision or C and work on 'calibration' or 'zero error' or whatever you want to call it?
By the way, D looks a little too accurate to me, if you imagine where the mean of those points will be. I'd move the right-most spot on to the bullseye, or even to the left of the bull to ensure that the mean is clearly far from the centre.